JP3409702B2 - Wheel condition estimation device - Google Patents

Description

Detailed Description of the Invention

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a wheel condition estimating device, and more particularly, to a response output of a wheel resonance system based on a vibration model of the wheel resonance system including a friction characteristic between a tire and a road surface. The present invention relates to a wheel state estimation device that estimates a wheel state from the above.

2. Description of the Related Art In recent years, research and development of preventive safety technology has been promoted due to an increase in safety consciousness for automobiles, and an antilock brake system (AB) is a typical safety device.
S) is already installed in many passenger cars.

Under these circumstances, a new principle of ABS control focusing on the tire resonance phenomenon has been proposed and studied (Japanese Patent Application No. 7-220920, etc.). The present technology applies a minute excitation having a frequency component equal to the resonance frequency when the tire is gripping to the brake pressure, and at that time, the resonance gain of the tire resonance system (small amplitude at the resonance frequency of wheel speed / brake pressure). The average brake pressure is controlled on the basis of the excitation amplitude).

It has been known that this resonance gain is a physical quantity related to the gradient of the friction coefficient μ with respect to the slip ratio S in the so-called S-μ characteristic (change curve of the friction coefficient μ with respect to the slip ratio S). It is expected to be able to estimate the frictional characteristics related to the slipperiness between the tire and the road surface during braking based on the gain.

[0003]

However, since the above-mentioned prior art is premised on that the resonance frequency of the tire resonance system is known, the resonance frequency changes immediately when the tire is replaced or when the tire air pressure changes. There is a problem that it can not cope with. For example, when the tire pressure decreases, the resonance frequency when the tire grips the road surface fluctuates to the lower side, so the value of the resonance gain becomes smaller and the peak μ just before the peak μ by comparing with the predetermined reference gain. The detection accuracy of the state of is reduced.

Further, in the above-mentioned prior art, the brake pressure is slightly excited when the resonance gain is calculated. However, depending on the structure of the brake unit, this minute excitation can be applied only during braking, and thus such a brake unit is used. In the case of a vehicle model with, there is a problem that the opportunity for estimating the road surface condition is limited during braking. In some cases, minute excitation of the brake pressure is not desirable, such as during driving or steady running.

The present invention has been made in view of the above facts, and it is possible to equally estimate the road surface state in various running states such as braking, driving, and steady running, and to reduce the resonance frequency due to tire replacement and tire air pressure fluctuations. It is an object of the present invention to provide a wheel condition estimation device that can always highly accurately estimate a friction characteristic relating to slipperiness between a tire and a road surface regardless of a change and that enables highly accurate ABS control based on the friction characteristic. To do.

[0006]

In order to achieve the above object, the invention of claim 1 is a wheel condition estimating device for estimating a wheel condition in a wheel resonance system including a friction characteristic between a tire and a road surface, detecting means for detecting a response output to vibration input to said wheel resonance system, the transfer characteristic of said wheel resonance system to the response output from the vibration input, and a physical quantity relating to ease of slippage between at least the tire and the road surface ,tire
Is represented by a vibration model including a physical quantity related to the twisting and twisting constant as an unknown element of a wheel state, and based on the vibration model, at least the unknown element that substantially satisfies the response output detected by the detecting means is estimated. And estimating means.

Further, the invention of claim 2 is characterized in that, in the invention of claim 1, it further comprises a correcting means for correcting the vibration model based on the response output detected by the detecting means. .

According to a third aspect of the present invention, in addition to the first or second aspect of the present invention, a vibration input is applied to the wheel resonance system.
It further comprises a vibrating means for vibrating .

Further, in a fourth aspect of the invention, in the invention of the third aspect , the vibration is applied to the wheel resonance system by the vibrating means.
It is characterized by further comprising a vibration input detection means for detecting a generated vibration input .

Further, the invention of claim 5 is the invention of claim 4, wherein the vibration is applied to the wheel resonance system.
The excitation inputs that can be used are random input, impulse input, and
It is characterized in that it is one of the step inputs . Well
According to the invention of claim 6, in the invention of claim 4,
Of the vibration input applied to the wheel resonance system by the vibration means.
The frequency is the resonance frequency obtained from the twisting constant of the tire.
It is characterized by being a number. Further, the invention of claim 7 is
In the invention according to any one of claims 1 to 6 , a judgment value for judging whether or not the response output is a periodic output is calculated based on the response output detected by the detecting means. The estimating means further comprises: calculating means; and selecting means for selecting an output to be input to the estimating means from response outputs detected by the detecting means based on the judgment value calculated by the calculating means. When the unknown element is estimated, the output selected by the selecting means is used. The invention according to claim 8 is the invention according to claim 7 , wherein the calculating means uses the response output detected by the detecting means and the response output as the judgment value, wherein the periodic part of the response output is A correlation coefficient with a value converted so as to change in response to a change in response output is calculated. It should be noted that the above conversion is performed by any of the even number differential, the even number difference, the even number integration, and the even number increment. Further, the calculating means calculates the judgment value each time a predetermined number of response outputs are detected by the detecting means. (Principle of the Present Invention) The principle of the present invention will be described with reference to the drawings of FIGS. Here, FIG. 1 is an equivalent dynamic model of a wheel resonance system,
2 is a friction characteristic between a tire and a road surface that defines the transfer characteristic of the wheel resonance system of FIG. 1, and FIG. 3 is a vibration characteristic from the vibration input to the response output in the transfer characteristic of the wheel resonance system of FIG. It is an illustration of an example of a model.

First, as shown in FIG. 1, a vibration phenomenon at a wheel when a vehicle is traveling at a vehicle body speed v (ω _{v in} terms of angular velocity), that is, a wheel resonance system constituted by at least the wheel and a road surface. The vibration phenomenon of is discussed with reference to a dynamic model equivalently modeled by the wheel rotation axis. The various quantities shown in FIG. 1 are as follows.

J ₁ : Inertia moment on rim side J ₂ : Inertia moment on belt side K: Twisting constant of tire T ₁ : Braking / driving torque (drive side is a positive sign) ω ₁ : Angular velocity on rim side ω ₂ : Belt side Angular velocity θ _s : rim-belt twist angle T _d : road surface disturbance T _L : tire-road surface generated force In the dynamic model of the wheel resonance system of FIG. 1, the braking / driving torque T ₁ acting on the rim is Is transmitted to the belt via the twisting constant K and acts on the road surface via the belt surface. At this time, the generated force T _L as a reaction of the braking / driving torque T ₁ acts on the wheel from the road surface, with the ground contact point between the belt and the road surface as the base point.

The generated force T _L is due to the frictional force between the tire and the road surface, and acts in the direction opposite to the braking / driving torque T ₁ . That is, the generated force T _L acts in the direction opposite to the wheel rotation direction (direction of ω ₁ ) when the driving torque T ₁ acts on the rim during driving, and when the braking torque T ₁ acts during braking. Acts in the direction of rotation of. When the road surface has irregularities, the torque of the road surface disturbance ΔT _d generated by the irregularities also acts on the tire.

Here, when the vehicle is running at a certain speed v (the value converted into the rotation system is ω _v ) and the brake is applied, a slip occurs between the tire and the road surface. The generated force T _L generated between the tire and the road surface changes like the functional relationship of FIG. 2 with respect to the slip ratio S ₁ represented by the following equation.

[0015]

[Equation 1]

Similarly, when the driver depresses the accelerator pedal to accelerate the vehicle from the time when the vehicle is traveling at a certain speed v, a slip occurs between the tire and the road surface. T _L changes with respect to the slip ratio S ₂ represented by the following equation as the functional relationship of FIG.

[0017]

[Equation 2]

Here, assuming that the rotation direction of the wheel is a positive direction, the generated force T _L between the tire and the road surface can be expressed by the following equation.

During braking: T _L = WRμ (S ₁ ) (2) During driving: T _L = −WRμ (S ₂ ) (3) where W is wheel load, R is tire dynamic load radius, and μ is It is the coefficient of friction between the tire and the road surface. Note that μ is expressed as a function of the slip ratio S ₁ or S ₂ .

As shown by the S-μ curve in FIG. 2, the generated force T _L is 0 when the slip ratio is 0, but at a certain positive slip ratio, the generated force T _L during braking has a positive peak value. Then, at a certain negative slip ratio, the generated force _TL during driving is
It can be seen that has a relationship of taking a negative peak value. In addition, at various operating points, the gradient of the generated force with respect to the slip ratio has a unique value such as a value near 0 at the peak value. Therefore, by using the gradient, the gradient between the tire and the road surface can be reduced. It can represent the ease of slipping.

Here, in the dynamic model of FIG. 1, when the braking / driving torque acting on the rim is excited with an amplitude ΔT ₁ around the average braking / driving torque T ₁ , the excitation torque component is around the wheel speed ω ₁ . Appears as the vibration component Δω ₁ .
If the road surface disturbance T _d has a vibration component ΔT _d , the vibration component Δω ₁ of the wheel speed also includes the vibration component generated by the disturbance.

Therefore, the transfer characteristic of the wheel resonance system of FIG.
The vibration model at various operating points in FIG. 2 is expressed by the following equation.

Δω ₁ = H ₁ (s) ΔT ₁ + H ₂ (s) ΔT _d (4) where

[0024]

[Equation 3]

[0025] In addition, s is a Laplace operator. Further, D ₀ is represented by the following equations D ₁₀ and D ₂₀ depending on braking and driving, respectively.

[0026]

[Equation 4]

Here, S ₁₀ and S ₂₀ are
Ω _v0 is the slip ratio at a certain operating point during driving.
Is the vehicle speed at the operating point.

From equation (7), D ₁₀ is proportional to the slope (∂μ / ∂S ₁ ) of the S-μ curve and the wheel load W at the slip ratio S ₁₀ at the operating point, and the vehicle body at the operating point It is inversely proportional to the velocity ω _v0 . The same holds for D ₂₀ when S ₂₀ is close to 0.

Up to this point, it was assumed that μ has a slip ratio dependency, but if it has a slip velocity dependency,
By redefining S ₁ = ω _v −ω ₂ and S ₂ = ω ₂ −ω _v ,

[0030]

[Equation 5]

It can be expressed as Again, D
_{10 and 20,} the slip ratio S _10,20 at the operating point S-
It is proportional to the slope of the μ curve and the wheel load W.

Since the vibration model described above represents the motion at an arbitrary operating point, a special case is also described in the case of steady running in which neither braking nor driving is performed. In the case of steady running, the operating point is the origin of the S-μ curve, and D ₀ = D ₁₀ = D ₂₀ represents the μ gradient at the origin.

Although the vibration model is expressed by a cubic system with respect to the Laplace operator s, it is considered that the quadratic is sufficient to describe the physical phenomenon called vibration. Therefore, the following equation is obtained by approximating the cubic model to the quadratic model.

[0034]

[Equation 6]

As described above, according to the above vibration model, in the wheel resonance system including the frictional characteristics between the tire and the road surface, the tire is rolled on the road surface having vibration input torque (ΔT ₁ ) to the resonance system and unevenness. Excited road surface (Δ
Wheel speed vibration (Δω ₁ ) as a response output for T _d ).
And the physical quantity D ₀ relating to the slipperiness between the tire and the road surface is included.

The state of transmission from the vibration input to the response output in the above vibration model is illustrated in FIG.

The validity of this vibration model can be shown by experimental results. FIGS. 4 (A) and 4 (B) show the transfer characteristics from the vibration input to the wheel speed vibration at that time when braking is applied by superposing the vibration input on the brake pressure P _m at a certain constant vehicle speed. , The results obtained by conducting experiments on various values of the brake pressure P _m (0.98 [MPa] to 4.90 [MPa]). 4A is a frequency characteristic (amplitude characteristic) of a ratio of wheel speed vibration to a vibration amplitude of the brake pressure P _m (resonance gain), and FIG. 4B is a brake pressure P _m.
The phase characteristics from the vibration of the vehicle to the vibration of the wheel speed are shown.

As shown in FIG. 4A, it can be seen that the peak of the resonance gain near the resonance frequency (about 40 Hz) decreases as the brake pressure increases. Further, as shown in FIG. 4 (B), the phase from the brake pressure to the wheel speed vibration changes according to the magnitude of the brake pressure P _m .
It can be seen that the phase characteristics differ greatly around the resonance frequency.

Here, by increasing the brake pressure P _m ,
Since the slip ratio and the braking force increase and the slope of the S-μ curve decreases, increasing P _m and decreasing the slope D ₀ of the S-μ curve have almost the same physical meaning. To have. Therefore, it can be expected that the same results as in FIGS. 4A and 4B can be obtained even when D ₀ is changed.

On the other hand, based on the vibration model of the present invention,
The response characteristics of the wheel speed vibration to the brake excitation input.
When calculated, the amplitude characteristic as shown in FIG.
The phase characteristic as shown in (B) is obtained. As shown in FIG.
At the resonance frequency (40 Hz),
Gradient of₀If you reduce the
It can be seen that there is a characteristic that the number of peaks decreases. Also, FIG.
As shown in (B), the slope D ₀Depending on the size of the resonance circumference
The phase characteristics around the wave number are very different,
It can be seen that it has similar characteristics to 4 (B). This
Therefore, the vibration model of the present invention is a transmission characteristic of an actual wheel resonance system.
It can be said that it shows the sex well.

6 (A) and 6 (B) show the wheels of the driving wheels (right rear wheel and left rear wheel) obtained when the vehicle actually travels on an asphalt road (dry road) and a dirt road, respectively. It is a frequency spectrum of fast vibration.

On the asphalt road (FIG. 6 (A)), the resonance peak clearly appears, but on the dirt road (FIG. 6 (B)) where the tire is running in a state close to idling due to drifting or the like. It can be seen that the resonance peak does not appear.

On the other hand, FIGS. 7 (A) and 7 (B) show the amplitude characteristic and the phase characteristic of the wheel speed vibration with respect to the tire road surface excitation input, respectively, based on the vibration model of the present invention. Also in this figure, as the slope D ₀ of the S-μ curve is reduced, the resonance peak disappears, and it can be seen that the characteristics agree well with the experimental results.

As described above, the present invention estimates the physical quantity D ₀ relating to the slipperiness between the tire and the road surface based on the vibration model that well represents the actual transmission characteristics. That is, the estimation means of the present invention, based on the vibration model including at least a physical quantity related to the slip between the tire and the road surface as an unknown element of the wheel state, at least substantially satisfy the response output detected by the detection means. The unknown element that causes the above-mentioned unknown is estimated.

Further, as in the invention of claim 1, when the vibration model further includes a physical quantity related to torsion spring constant K of the tire as an unknown component of the wheel condition, the estimating means, torsion spring constant of the tire It is also possible to estimate a physical quantity related to K or the constant. The physical quantity related to the twisting constant of the tire includes the resonance frequency of the wheel resonance system.

Next, the estimation principle of the estimation means of the present invention will be described. As shown in FIG. 3, the vibration model of the present invention is
In the case of the examples represented by the transfer function 1 and the transfer function 2, the estimation of the unknown element by the estimating means is equivalent to identifying these transfer functions.

Here, a z-transformed discretized model is often used as the transfer function to be identified. When identifying a discrete-time model, the identification accuracy of the discrete-time model depends on the sampling period. Therefore, many trials and errors are involved in obtaining an appropriate sampling period.

After the identification by the discrete time model, in order to calculate the physical quantity that constitutes the system, it is necessary to perform the inverse conversion into the continuous time model, but the calculation requires a higher function, and the calculation time and the calculation error are Increase. Moreover, the inverse transformation is not uniquely determined. Therefore, there is a problem as a model for obtaining a physical quantity corresponding to the slipperiness of a road surface.

Therefore, in the present invention, the continuous time model is identified. The identification of the continuous-time model has an advantage that there is no problem associated with the identification of the discrete-time model, and the physical quantity corresponding to the slipperiness of the road surface can be directly calculated.

For example, when identifying the transfer function of the second-order continuous-time model, it can be obtained by modifying equations (11) and (12).

[0051]

[Equation 7]

Can be the transfer function to be identified. At this time, when the excitation torque ΔT ₁ is applied to the wheel resonance system as the excitation input as in the invention of claim 3 , the road surface disturbance is minute compared to the excitation torque, and if this is ignored, the equation (4) is obtained. As a result, Δω ₁ = ΔG ₁ (s) ΔT ₁ is obtained.

For example, when the least squares method is used, the above equation is applied to unknown parameters [a ₁ a ₂ ] or [a ₁ a ₂ b
_[0 b ₁ b ₂ ] is transformed into a linear function,
The unknown parameter can be estimated by applying the method of least squares to each data to which the detected Δω ₁ is sequentially applied.

[0054] Here, as in the invention of claim 4, if the vibration input detecting means can detect the vibration input [Delta] T ₁ applied to the wheel resonance system by vibration input means, unknown parameters [a ₁ a ₂ b ₀ b ₁ b ₂ ] can all be estimated. On the other hand, when ΔT ₁ is not detected,
[A ₁ a ₂ ] can be estimated.

When the vibration input is applied to the wheel resonance system as described above, the unknown parameter can be estimated with high accuracy even when the vehicle is traveling on a good road surface with few irregularities.

On the other hand, when the excitation torque ΔT ₁ is not given,
With ΔT ₁ = 0, Δω ₂ = ΔG ₂ (s) ΔT _d obtained from the equation (4) is transformed into an equation in which the unknown parameter [a ₁ a ₂ ] is transformed into a linear function, and the detected Δω ₁ is sequentially obtained. By applying the least squares method to each fitted data,
Unknown parameters can be estimated. In this case, there is an advantage that the unknown parameter can be estimated even when the excitation torque ΔT ₁ cannot be applied.

Various modified least squares methods may be used to suppress the identification error. As the modified least squares method,
The well-known auxiliary variable method, extended least squares method, and generalized least squares method can be used.

The estimating means of the present invention uses the correspondence relationship between the transfer function (11) and the expression (11-2) of the vibration model or (12).
From the correspondence between the equation and the equation (12-2), the physical quantity related to the road surface μ gradient D ₀ is calculated using the estimated parameters a ₁ and a ₂ .

[0059]

[Equation 8]

Then, the physical quantity relating to the twisting constant of the tire can be estimated as a ₂ = K / J ₁ . In addition, the resonance frequency of the wheel resonance system can be estimated from this.

When the physical quantity related to the road surface μ gradient D ₀ can be calculated in this way, when the physical quantity is small, it is possible to determine that the frictional characteristics between the tire and the road surface are in a saturated state, and the slipperiness of the road surface is immediately determined. it can. Further, since the μ gradient D ₀ is obtained without assuming that the twisting constant of the tire is known, the μ gradient D ₀ can be obtained with high accuracy regardless of whether or not the resonance frequency changes due to tire replacement or changes in tire air pressure. It is possible to obtain a physical quantity related to the slipperiness between the tire and the road surface. Further, the tire pressure can be diagnosed based on the estimated value of the physical quantity related to the twisting constant of the tire.

Further, in the invention of claim 2, the vibration model is corrected based on the detected response output. For example, in order to improve the identification accuracy, when a preprocessing unit having a frequency characteristic corresponding to the resonance frequency of the wheel resonance system is provided in the preceding stage of the transfer function of the vibration model, the parameter of this preprocessing unit is detected. It is adaptively changed with the operation of the least squares method based on the response output. In this case, the parameters of the preprocessing means can be represented by the parameters of the transfer function.
Then, the parameter of the transfer function is updated again by the response output that has passed through the changed preprocessing means and transfer function. In this way, the frequency characteristic of the pretreatment means approaches the frequency characteristic adapted to the resonance frequency according to the variation of the twisting constant of the tire, and therefore, compared to the case where the parameter of the pretreatment means is fixed. Highly accurate estimation is possible. Further, the calculating means according to the invention of claim 7 calculates a judgment value for judging whether or not the response output is a periodic output, based on the response output detected by the detecting means. Here, the calculating means converts the response output detected by the detecting means and the response output so that the periodic part of the response output changes corresponding to the change of the response output, as the judgment value. You may make it calculate the correlation coefficient with. In addition, the above conversion is an even number differential, an even number differential,
Either even-numbered integration or even-numbered increment. here,
Even if a response output whose magnitude is suddenly large is converted as described above, the converted value includes many components other than the periodic part. That is, since the value obtained by converting the sudden response output as described above does not change in response to the change in the response output, the correlation between the sudden response output and the value obtained by converting the response output as described above is small. Further, in the value obtained by converting the output in which the periodical change does not significantly appear as described above, there are few components converted so as to change corresponding to the change in the response output.
Therefore, the correlation between the output in which the periodic change does not significantly appear and the value obtained by converting the response output as described above is small. In addition,
The calculation means may calculate the determination value each time a predetermined number (for example, one) of response outputs are detected by the detection means. Then, the selecting means selects the output to be input to the estimating means from the response outputs detected by the detecting means based on the judgment value calculated by the calculating means. Then, the estimating means uses the output selected by the selecting means when estimating the unknown element. In this way, the judgment value for judging whether or not the response output is a periodic output is calculated, and based on the calculated judgment value, the output to be input to the estimation means is selected from the detected response outputs. , The unknown element is estimated using the selected output, and therefore the unknown element is estimated based on the response output from which the response output whose magnitude is suddenly large or the output in which the periodic change does not significantly appear is removed. be able to.

BEST MODE FOR CARRYING OUT THE INVENTION Each embodiment of a wheel state estimating device of the present invention will be described in detail below with reference to the drawings. (1st Embodiment) The wheel condition estimation apparatus of 1st Embodiment is based on road surface disturbance (DELTA) T.
_The μ gradient is calculated when only _d is input to the wheel resonance system as a vibration input, and there are a first mode and a second mode.

FIG. 8 shows the configuration of the first aspect of the wheel state estimating device according to the first embodiment. As shown in the drawing, the wheel condition estimating apparatus includes a wheel speed detecting means 1 for detecting a wheel speed omega ₁ of each wheel, wheel from the wheel speed omega ₁ of each detected wheel undergoing road surface disturbance [Delta] T _d Based on the preprocessing filter 2 that detects the wheel speed vibration Δω ₁ of each wheel as the response output of the resonance system and the vibration model of FIG.
Transfer function identification means 3 for identifying the transfer function of each wheel that satisfies the detected wheel speed vibration Δω ₁ by using the least square method, and the friction between the tire and the road surface based on the identified transfer function. And a gradient calculation means 4 for calculating the gradient of the coefficient μ for each wheel.

In FIG. 8, the wheel speed detecting means 1 is a so-called wheel speed sensor which outputs a sensor output signal corresponding to the wheel speed, and a calculating means which calculates an actual rotation speed signal of each wheel from the sensor output signal. , Can be configured from.

Further, the preprocessing filter 2 is a bandpass filter which passes only a frequency component of a certain band centered on the frequency expected to be the resonance frequency of the present wheel resonance system, and a high band filter which contains the resonance frequency component. It can be configured by a high-pass filter or the like that passes only frequency components. In the first mode, the parameter defining the frequency characteristic of the bandpass filter or the highpass filter is fixed to a constant value, and in the second mode described later, this parameter is identified by the transfer function identification means 3. The parameters are adapted and changed.

The output of the preprocessing filter 2 is assumed to have the DC component removed. That is, the wheel speed ω ₁
Only the wheel speed vibration Δω ₁ around is extracted.

Now, the transfer function F (s) of the preprocessing filter 2 is

[0068]

[Equation 9]

It is assumed that However, c _i is the coefficient of the filter transfer function, and s is the Laplace operator. Next, an arithmetic expression on which the transfer function identification means 3 depends will be derived. In the present embodiment, the calculation of the preprocessing filter 2 is included in the calculation of the transfer function identification means 3 to be performed.

First, the transfer function to be identified in the first embodiment uses the road surface disturbance ΔT _d as a vibration input and the wheel speed vibration Δω ₁ detected by the preprocessing filter 2 at this time as a response output 2 The following model is used. That is,

[0071]

[Equation 10]

Assume the vibration model of Here, v is an observation noise included when observing the wheel speed signal. (1
By transforming equation (5), the following equation is obtained.

[0073]

[Equation 11]

First, the equation obtained by applying the preprocessing filter of equation (14) to equation (16) is discretized. At this time, Δω ₁ , Δ
T _d and v are the discretized data Δω ₁ (k), ΔT _d (k), and v that are sampled for each sampling period T _s.
(K) (k is a sampling number: k = 1,2,3, ...). The Laplace operator s can be discretized using a predetermined discretization method. In this embodiment, as an example, the discretization is performed by the following bilinear transformation. Note that d is a 1-sample delay operator.

[0075]

[Equation 12]

Since the order m of the preprocessing filter is preferably 2 or more, in the present embodiment, m = 2 in consideration of the calculation time, and the following equation is obtained.

[0077]

[Equation 13]

Further, in order to identify the transfer function from each data of the wheel speed vibration Δω ₁ based on the least square method, (17)
The equation is transformed into the following equation so as to be in the form of a linear function with respect to the parameter to be identified. Note that " ^T " is the transpose of the matrix.

[0079]

[Equation 14]

It is In the above equation, θ is the parameter of the transfer function to be identified. Next, the operation of this embodiment will be described.

In the transfer function identifying means 3, the least squares method is applied to each data obtained by sequentially fitting the discretized data of the detected wheel speed vibration Δω _{1 to} the equation (22),
The unknown parameter θ is estimated and the transfer function is identified thereby.

Specifically, the detected wheel speed vibration Δω ₁
To discrete data Δω (k) (k = 1,2,3, ...),
The data are sampled at N points, and the parameter θ of the transfer function is estimated by using the arithmetic expression of the least square method shown below.

[0083]

[Equation 15]

Here, the amount with the symbol "^" is defined as its estimated value. Further, the least squares method may be calculated as a recursive least squares method for obtaining the parameter θ by the following recurrence formula.

[0085]

[Equation 16]

Here, ρ is a so-called forgetting coefficient, which is normally set to a value of 0.95 to 0.99. At this time, the initial value is

[0087]

[Equation 17]

It suffices to set Various modified least squares methods may be used as a method for reducing the estimation error of the least squares method. In the present embodiment, an example will be described in which the auxiliary variable method, which is a least squares method in which an auxiliary variable is introduced, is used. According to this method, m (k)
Using as an auxiliary variable, the parameters of the transfer function are estimated using the following equation.

[0089]

[Equation 18]

Further, the sequential calculation is as follows.

[0091]

[Formula 19]

The principle of the auxiliary variable method is as follows.
Substituting equation (22) into equation (28),

[0093]

[Equation 20]

Therefore, if the auxiliary variable is selected so that the second term on the right side of the equation (32) becomes zero, the estimated value of θ agrees with the true value of θ. Therefore, in the present embodiment, as an auxiliary variable,
ζ (k) = [− ξ _y1 (k) −ξ _y2 (k)] ^T
The one delayed so as not to correlate with (k) is used. That is, m (k) = [− ξ _y1 (k−L) −ξ _y2 (k−L)] ^T (33). However, L is a delay time.

After identifying the transfer function as described above,
In the μ gradient calculating means 4, the physical quantity related to the road surface μ gradient D ₀ is

[0096]

[Equation 21]

Is calculated. Thus, if the physical quantity related to the road surface μ gradient D ₀ can be calculated by the equation (34), for example,
When the physical quantity is small, it can be easily determined that the frictional characteristics between the tire and the road surface are saturated.

Next, the second mode will be described with reference to FIG. As shown in FIG. 9, in the second mode, an adapting unit 5 that changes the characteristic of the preprocessing filter 6 according to the parameter identified by the transfer function identifying unit 3 is further provided.

In the second mode, the transfer function of the preprocessing filter 6 preferably has the same characteristics as the denominator polynomial of the transfer function to be identified, but the transfer function to be identified is unknown, so the preprocessing is performed. The transfer function of the filter 6 is constructed by the following equation using the parameters of the transfer function identified by the transfer function identifying means 3.

[0100]

[Equation 22]

At that time, the transfer function identifying unit 3 executes the following calculation as in the case of the recursive least squares method of the first mode.

[0102]

[Equation 23]

Then, the adaptive means 5 constructs the coefficient of the transfer function of the equation (35) of the preprocessing filter 6 from the respective elements a ₁ and a ₂ of the parameter θ of the estimated transfer function. And
From the wheel speed vibration Δω ₁ detected by the pre-processing filter 2 to which the coefficient is adapted, the transfer function identifying means 3 estimates the parameter of the transfer function of the wheel resonance system again.

In the second mode, the characteristics of the preprocessing filter are appropriately changed according to the estimated parameters, so that a better estimated value can be obtained as compared with the first mode in which the parameters of the preprocessing filter are fixed. Is obtained.

FIG. 10 shows the estimation result when the second mode is carried out. In the figure, the horizontal axis is the true value of the physical quantity D ₀ related to the slipperiness between the tire and the road surface, and the vertical axis is the estimated value of D ₀ according to the second mode.
As shown in the figure, it can be seen that the physical quantity D ₀ relating to the slipperiness between the tire and the road surface is well estimated by this mode.

The present invention is applied to an antilock brake control device.
When applied, the estimated physical quantity D ₀Match the reference value
The braking force is controlled so that This physical quantity D₀Is
It is calculated without assuming that the resonance frequency is known.
Tires, tire replacement and tire pressure drop, etc.
Irrespective of whether the resonance frequency is changed by
Anti-lock braking operation becomes possible. Also,
Similar effect when applied to traction control
BR> can be played. (Second Embodiment) It can be understood from the equations (6) and (12).
As described above, the vibration model of the present invention includes unknown elements of the wheel state.
Then, the twisting constant K of the tire is included. this
Focusing on the point, in the second embodiment, in the first embodiment
Similarly to the road surface disturbance ΔT_dOnly as a vibration input for both wheels
If the transfer function of the vibration model is applied to the vibration system
The physical quantity related to the twisting constant K of the tire
It is something.

The second embodiment will be described with reference to the block diagram of FIG. The same components as those in the first embodiment are designated by the same reference numerals and detailed description thereof will be omitted.

As shown in FIG. 11, in the second embodiment, the spring constant calculating means 7 for calculating the physical quantity relating to the twisting constant of the tire based on the parameter identified by the transfer function identifying means 3 is used as the transfer function. It is connected to the output terminal of the identification means 3.

Next, the flow of calculation by the spring constant calculating means 7 will be described. When the parameters of the transfer function are identified by the transfer function identifying means 3 of FIG. 11, the identified transfer function is

[0110]

[Equation 24]

Since it can be expressed as follows, from the correspondence with the vibration model expressed by equation (12)

[0112]

[Equation 25]

It becomes: Therefore, in the spring constant calculating means 7, from the equation (43), the twist of the tire is calculated from the estimated value (with ^) of the identification parameter a ₂ by the transfer function identifying means 3 and the rim-side inertia moment J ₁ given in advance. The spring constant K is calculated.

Furthermore,

[0115]

[Equation 26]

Is the resonance frequency of the wheel resonance system. Therefore, in the spring constant calculating means 7, from the equation (44), as one of the physical quantities relating to the twisting constant K of the tire,
The resonance frequency of the wheel resonance system can also be calculated.

Further, since the twisting constant K and the resonance frequency of the tire obtained by this embodiment depend on the tire air pressure, they can be used as the information of the tire air pressure diagnostic device, which enables safe driving. Become.

In this tire pressure diagnostic device, for example,
The estimated twisting constant K of the tire is compared with a predetermined reference value, and when the constant is equal to or larger than the reference value, it is diagnosed as "tire pressure normal", and when the constant is smaller than the reference value, Diagnose "abnormal tire pressure" and warn the driver. In addition, the second reference value corresponding to the tire pressure suitable for traveling on a highway is also facilitated, and the spring constant K becomes the second reference value.
If it is smaller than the reference value of, the driver may be warned to replenish the tire pressure. In addition,
By preparing these reference values for each type of tire, it is possible to maintain the same judgment criteria for tire pressure diagnosis even when tires are replaced.

Further, there is an advantage that the excitation frequency can be provided to the antilock brake control device based on the excitation of the braking force. For example, even if the resonance frequency changes due to a change in tire air pressure or tire replacement, if the braking force is slightly excited at the changed resonance frequency, the resonance gains calculated in the same frictional state show substantially the same value. The detection accuracy of the peak μ can be maintained. Further, by detecting the resonance frequency from the peak value of the frequency spectrum of the wheel speed signal, by comparing the resonance frequency with the resonance frequency at the time of tire grip, in the antilock brake control device for detecting the state immediately before the peak μ, By using the resonance frequency estimated according to the present embodiment as a reference for comparison, the state immediately before the peak μ can always be detected accurately regardless of variations in tire air pressure.

Although the parameters of the preprocessing filter 2 are fixed in the second embodiment, the coefficients of the preprocessing filter 2 are estimated as in the second mode of the first embodiment. It may be adaptively changed according to the parameters of the transfer function.

FIG. 12 shows the estimation result of the resonance frequency when the second embodiment is carried out. In the figure, the horizontal axis is the true value of the resonance frequency, and the vertical axis is the resonance frequency calculated using the equation (44) according to the present embodiment. As shown in the figure, it can be seen that the resonance frequency can be estimated very well according to the present embodiment. (Third Embodiment) A wheel state estimating apparatus according to a third embodiment identifies a transfer function of a wheel resonance system when an excitation torque ΔT ₁ is input to the wheel resonance system as a vibration input. There is a first mode for calculating the μ gradient and a second mode for calculating the twisting constant of the tire.

FIG. 13 shows the configuration of the first mode of the wheel state estimating device according to the third embodiment. The same components as those in the first embodiment shown in FIG. 8 are designated by the same reference numerals, and detailed description thereof will be omitted.

As shown in FIG. 13, the present wheel state estimating device is further provided with a vibrating means 8 for superposing the exciting torque ΔT ₁ as a vibrating input around the average braking / driving torque T _1. Has been. The vibrating means 8 vibrates the wheel around its central axis by exciting the brake pressure and the engine output. For example, in the case of exciting the braking torque, in the control command of the pressure increasing / decreasing valve of the hydraulic actuator for the brake pressure, by superimposing the pressure increasing / decreasing command of a predetermined frequency on the average brake pressure command,
Braking torque can be applied.

Next, an arithmetic expression on which the transfer function identifying means 3 according to the present embodiment depends will be derived. Note that, also in the present embodiment, the calculation of the pre-processing filter 2 is included in the calculation of the transfer function identification means 3 to be executed.

First, the transfer function to be identified in the third embodiment uses the road surface torque ΔT ₁ as a vibration input and the wheel speed vibration Δω ₁ detected by the preprocessing filter 2 as a response output 2 The following model is used. That is,

[0126]

[Equation 27]

Assume the vibration model of. Here, v is an observation noise included when observing the wheel speed signal. (Four
By transforming equation (5), the following equation is obtained.

[0128]

[Equation 28]

First, the equation obtained by applying the preprocessing filter of equation (14) to equation (46) is discretized. At this time, Δω ₁ , Δ
T _d and v are the discretized data Δω ₁ (k), ΔT _d (k), and v that are sampled for each sampling period T _s.
(K) (k is a sampling number: k = 1,2,3, ...). The Laplace operator s can be discretized by the above-described bilinear transformation using the delay calculation element d.

Since the order m of the preprocessing filter is preferably 2 or more, in the present embodiment, m = 2 in consideration of the calculation time, and the following expression is obtained.

[0131]

[Equation 29]

Further, in order to identify the transfer function from each discretized data of the wheel speed vibration Δω ₁ based on the least squares method, the equation (47) should be in the form of a linear function with respect to the parameter to be identified. , Is transformed into the following equation. In addition, "
Let ^T "be the transpose of the matrix.

[0133]

[Equation 30]

It is In the above equation, θ is the parameter of the transfer function to be identified. Next, the operation of this embodiment will be described.

In the transfer function identifying means 3, the least squares method is applied to each data obtained by successively applying the discretized data of the detected wheel speed vibration Δω _{1 to} the equation (52).
The unknown parameter θ is estimated and the transfer function is identified thereby.

Specifically, the detected wheel speed vibration Δω₁
To discrete data Δω (k) (k = 1,2,3, ...),
The data is sampled at N points. Then, the above formula ζ (k) and
And ξ _y0Using (k), (24) in the first embodiment
The parameter a of the transfer function is calculated by the same calculation as the equation and thereafter.₁
And a₂The estimated value (with ^) of is calculated. In addition, this implementation
In the form, the recursive least squares method and the auxiliary variable method may be used.
As in the second aspect of the first embodiment, the pretreatment process
The coefficient of filter 2 is used as the parameter of the identified transfer function.
It can be adapted and changed appropriately.

Here, the excitation torque ΔT by the vibrating means 8
Examples of the signal waveform of ₁ are shown in FIGS. 14 (A), 14 (B) and 14 (C).

The signal shown in FIG. 14A is a pseudo random signal and can be generated based on, for example, an M-sequence signal well known as one of pseudo random signals. In this case, since the vibration input has many frequency components, there is an advantage that the transfer function estimation accuracy is improved.

The signal shown in FIG. 14 (B) is a stepwise signal in which the exciting torque rises abruptly at a certain time from the state where the exciting torque is 0 and thereafter becomes a constant exciting torque.
When this stepwise signal is used, it is not necessary to change the torque as frequently as the pseudo-random signal, so that there are advantages that vibration and noise are small and the load on the actuator is small.

Further, the signal shown in FIG. 14C is an impulse-like signal which rapidly rises from the state where the excitation torque is 0 at a certain time and returns to the state where the excitation torque is 0 again after a lapse of a certain time. When this impulse-like signal is used, the torque has no offset component and can be instantaneously given in various running states such as braking, driving, and steady running, so that the transfer function can be estimated in any running state. .

Next, as a second mode, a wheel condition estimating apparatus in the first mode will be described in which the vibration model further includes a physical quantity related to the twisting constant of the tire.

As can be seen from the expressions (5) and (11), the vibration model of the present invention includes the twisting constant K of the tire as an unknown element of the wheel state. Focusing on this point, in the second mode, when the excitation torque ΔT ₁ is input to the wheel resonance system as a vibration input, the physical quantity related to the twisting constant K of the tire is estimated from the transfer function of the vibration model. is there.

The second mode will be described with reference to the block diagram of FIG. As shown in FIG. 15, in the second mode, as in the second embodiment, the spring constant calculating means 7 for calculating a physical quantity relating to the twisting constant of the tire based on the parameters identified by the transfer function identifying means 3. Is connected to the output end of the transfer function identification means 3.

Next, the flow of calculation by the spring constant calculating means 7 will be described. When the parameters of the transfer function are identified by the transfer function identifying means 3 of FIG. 15, the identified transfer function is

[0145]

[Equation 31]

Since it can be expressed as follows, from the correspondence with the vibration model expressed by equation (11),

[0147]

[Equation 32]

It becomes: Therefore, in the spring constant calculation means 7, from Equation (55), the twist of the tire is calculated from the estimated value (with ^) of the identification parameter a ₂ by the transfer function identification means 3 and the rim-side inertia moment J ₁ given in advance. The spring constant K is calculated.

Furthermore,

[0150]

[Expression 33]

Is the resonance frequency of the wheel resonance system. Therefore, in the spring constant calculating means 7, from the equation (44), as one of the physical quantities relating to the twisting constant K of the tire,
The resonance frequency of the wheel resonance system can also be calculated.

Further, since the twisting constant K and the resonance frequency of the tire obtained in this embodiment depend on the tire air pressure, they can be used as information of the tire air pressure diagnostic device. Further, there is an advantage that the excitation frequency can be provided to the anti-lock brake control device based on the excitation of the braking force.

In this mode, as in the first mode, the signal waveform of the vibration input may be any of the waveforms of FIGS. 14 (A), 14 (B) and 14 (C). it can.

As described above, in the third embodiment, even if the road surface has few irregularities and the amplitude of the wheel speed vibration Δω ₁ is small, the excitation torque ΔT ₁ is applied to the wheel resonance system to a certain level or more. Since the response output of 1 can be obtained at all times, it is possible to accurately estimate the μ gradient and the twisting constant of the tire regardless of the road surface condition. (Fourth Embodiment) The wheel state estimating apparatus according to the fourth embodiment detects a vibration input and a response output when the excitation torque ΔT ₁ is input to the wheel resonance system as a vibration input. Is used to identify the transfer function of the wheel resonance system, and there are a first mode for calculating the μ gradient and a second mode for calculating the twisting constant of the tire.

FIG. 16 shows the configuration of the first mode of the wheel condition estimating device according to the fourth embodiment. The same components as those in the third embodiment shown in FIG. 13 are designated by the same reference numerals and detailed description thereof will be omitted.

As shown in FIG. 16, the present wheel state estimating device includes a vibration input detecting means 9 for detecting the actual value of the excitation torque ΔT ₁ as a vibration input given to the wheel by the vibration means 8. Is further equipped. Then, the pre-processing filter 2 performs a predetermined filtering process on the detected vibration input together with the wheel speed ω ₁ , and the transfer function identification means 3 performs the filtering process on the vibration input component and the wheel speed vibration Δω _1. The parameters of the transfer function are estimated from and.

The method of detecting the vibration input by the vibration input detecting means 9 is, for example, when exciting the braking torque, the brake pressure (wheel cylinder pressure) corresponding to the braking torque.
Is detected by a pressure sensor, etc., and is multiplied by a predetermined constant to convert it to a braking torque that includes an excitation component, and by subtracting the average braking torque value from this braking torque, only the excitation torque component is excited. Extract as input.

Next, an arithmetic expression on which the transfer function identifying means 3 according to the present embodiment depends will be derived. Note that, also in the present embodiment, the calculation of the pre-processing filter 2 is included in the calculation of the transfer function identification means 3 to be executed.

First, the transfer function to be identified in the fourth embodiment has the excitation torque ΔT ₁ as a vibration input and the wheel speed vibration Δω ₁ detected by the preprocessing filter 2 at this time as a response output 2 The following model is used. That is,

[0160]

[Equation 34]

The vibration model of is assumed. Where ΔT ₁
Is the actual excitation torque detected by the vibration input detection means 9, and v is the observation noise included when observing the wheel speed signal. By modifying the equation (57), the following equation is obtained.

[0162]

[Equation 35]

First, the equation obtained by applying the preprocessing filter of equation (14) to equation (58) is discretized. At this time, Δω ₁ , Δ
T _d and v are the discretized data Δω ₁ (k), ΔT _d (k), and v that are sampled for each sampling period T _s.
(K) (k is a sampling number: k = 1,2,3, ...). The Laplace operator s can be discretized by the above-described bilinear transformation using the delay calculation element d.

Further, it is desirable that the order m of the preprocessing filter is 2 or more. Therefore, in the present embodiment, the calculation time is taken into consideration and m = 2, and the following equation is obtained.

[0165]

[Equation 36]

Further, in order to identify the transfer function from each discretized data of the wheel speed vibration Δω ₁ based on the least squares method, the equation (47) should be in the form of a linear function with respect to the parameter to be identified. , Is transformed into the following equation. In addition, "
Let ^T "be the transpose of the matrix.

[0167]

[Equation 37]

It is In the above equation, θ is the parameter of the transfer function to be identified, but in the present embodiment, ΔT ₁
Therefore, unlike the above-described embodiment, the coefficient b ₀ ,
It can be seen that b ₁ and b ₂ can also be estimated.

Next, the operation of the first mode according to the fourth embodiment will be described. The transfer function identifying means 3 estimates the unknown parameter θ by applying the least squares method to each data obtained by sequentially fitting the discretized data of the detected wheel speed vibration Δω _{1 to} the equation (64), and The transfer function is identified by.

Specifically, the detected wheel speed vibration Δω₁
To discrete data Δω (k) (k = 1,2,3, ...),
The data is sampled at N points. Then, the above formula ζ (k) and
And ξ _y0Using (k), (24) in the first embodiment
The parameter a of the transfer function is calculated by the same calculation as the equation and thereafter.₁
And a₂The estimated value (with ^) of is calculated. In addition, this implementation
In the form, the recursive least squares method and the auxiliary variable method may be used.
Yes. When the auxiliary variable method is used, as in this embodiment
If an excitation input can be detected at, the identified transfer function
With parameters

[0171]

[Equation 38]

From the above, the estimated value of the true output x (k) (with ^)
Can be obtained sequentially, and using this estimated value, the auxiliary variable can be constructed as in the following equation.

[0173]

[Formula 39]

[0174] Further, ζ (k) = [- ξ y1 (k) -ξ y2 (k) ξ u0 (k) ξ u1 (k) ξ u2 (k)] T (69) a formula error r (k) Use something that is delayed so that it has no correlation with. That is,

[0175]

[Formula 40]

It is assumed that However, L is the number of delay steps. Further, in the present aspect, the coefficient of the preprocessing filter 2 can be appropriately changed by adapting to the parameter of the identified transfer function, as in the second aspect of the first embodiment.

Also, the waveform of the vibration input is shown in FIG.
Although it may be as shown in FIGS. 14B and 14C,
In this aspect, when the resonance frequency when the tire grips the road surface (hereinafter referred to as “tire resonance frequency”) is clear, the vibration can be applied at the resonance frequency.
According to the experimental studies by the inventors, the vibration model expressed by the equation (11) is

[0178]

[Formula 41]

Since it is known that (11), (7
When the transfer characteristic from the vibration input to the response output when the vibration is applied at the tire resonance frequency is calculated based on equation (1),

[0180]

[Equation 42]

The transfer function is the zeroth order transfer function. Therefore, the transfer function to be identified is set as Δω ₁ = G _d ΔT ₁ (73), and the transfer function identification means 3 identifies the transfer function using the least square method as in the following equation.

[0182]

[Equation 43]

That is. The least squares method can also be calculated as the recursive least squares method as shown in the following equation.

[0184]

[Equation 44]

Here, ρ is a so-called forgetting coefficient, which is normally set to a value of 0.95 to 0.99. At this time, the initial value is

[0186]

[Equation 45]

It may be set as follows. Furthermore, by applying a modified least squares method such as the auxiliary variable method to the above least squares method, a more accurate estimated value can be obtained.

Then, the estimated value of the parameter G _d of the identified transfer function is calculated from the correspondence between the equations (72) and (73).

[0189]

[Equation 46] Therefore, the μ-gradient calculating means 4 calculates the physical quantity relating to the road slipperiness D ₀ from the equation (81) using the estimated value of G _d .

In this case, the frequency of the vibration input is the resonance frequency obtained from the twisting constant of the tire calculated by the second mode according to the second embodiment or the third embodiment of the present invention. Can be used. In this way, when the vibration is applied only at the tire resonance frequency, the power of the vibration input can be concentrated and applied to the frequency component, so that the SN of the output from the vibration input detection means 9 and the response output detection means 1 can be obtained. There is an advantage that the ratio is improved and the calculation accuracy is improved.

Next, as a second mode, a wheel state estimating apparatus in the first mode further including a physical quantity relating to a tire torsional constant will be described with reference to the block diagram of FIG.

As shown in FIG. 17, in the second mode, the transfer function identifying means 3 is used instead of the μ gradient calculating means 4 of the first mode.
Spring constant calculating means 7 for calculating a physical quantity relating to a twisting constant of a tire based on the parameter identified by
Is connected to the output end of the transfer function identification means 3.

Next, the flow of calculation by the spring constant calculating means 7 will be described. When the parameters of the transfer function are identified by the transfer function identifying means 3 in FIG. 17, the identified transfer function is

[0194]

[Equation 47]

Since it can be expressed as follows, from the correspondence with the vibration model expressed by equation (11),

[0196]

[Equation 48]

It becomes: Therefore, the spring constant computing means 7 uses the estimated value (with ^) of the identification parameter a ₂ by the transfer function identifying means 3 and the rim-side inertia moment J ₁ given in advance to calculate the twist of the tire from the equation (83). The spring constant K is calculated.

Furthermore,

[0199]

[Equation 49]

Is the resonance frequency of the wheel resonance system. Therefore, in the spring constant calculating means 7, from the equation (44), as one of the physical quantities relating to the twisting constant K of the tire,
The resonance frequency of the wheel resonance system can also be calculated.

Further, since the twisting constant K and the resonance frequency of the tire obtained in this embodiment depend on the tire air pressure, they can be used as information of the tire air pressure diagnostic device. Further, there is an advantage that the excitation frequency can be provided to the anti-lock brake control device based on the excitation of the braking force.

Also in this embodiment, the signal waveform of the vibration input is as shown in FIG. 14 except when the vibration is performed at the tire resonance frequency.
The waveform can be any one of (A), FIG. 14 (B), and FIG. 14 (C).

As described above, in the fourth embodiment,
Since the detected excitation input is used, more transfer functions
The parameters can be estimated. (Fifth Embodiment) Wheel state estimation of the fifth embodiment
The device has a response output that is a periodic signal among the response outputs.
Wheel resonance system based on the selected response output
The transfer function of is identified and the μ gradient is calculated. First
The configuration of the wheel state estimating device of the fifth embodiment is shown in FIG.
This will be described using a lock diagram. As shown in FIG. 18, pretreatment
A preprocessing filter is provided between the filter 2 and the transfer function identification means 3.
Wheel speed vibration Δω output from the₁Of which is periodic
Select only the signal and output it to the transfer function identification means 3.
The only difference is that another means 10 is provided.
The other configuration is the same as that of the first embodiment. Wheel speed vibration
Motion Δω₁If is a periodic signal, input it within the specified period.
Wheel speed vibration Δω₁Is a frequency-amplitude (power) graph
When plotted on, the graph shown by reference numeral 100 is obtained.
Should be. However, the wheel speed vibration Δω₁But shake
Outputs with a suddenly large width (for example, alias (ali
noise due to aliasing), etc.) 102 and periodical variations.
Output that does not show significant change (output with small S / N ratio) 10
4 is used to identify the transfer function of the wheel resonance system and the μ gradient is calculated.
When calculated, the variation is large and the estimation accuracy is poor. Yo
Therefore, it is necessary to select only the response output that is a periodic signal.
is there. FIG. 2 is a time series graph of the periodic output 100.
Compared with 0 (B), the amplitude plotted in time series is
The suddenly large output 102 is as shown in FIG.
Moreover, the periodicity is bad. In addition, the time-series graphed cycle
The output 104 in which the dynamic change does not significantly appear is shown in FIG.
As shown in, the periodicity is very bad. On the other hand, FIG.
The periodic output 100 shown by the solid line in FIG.
The output 100 is the periodic portion of the periodic output 100
It is changed to change corresponding to the periodic change of the output 100.
The converted value (shown by the dotted line in FIG. 20 (B)) is constant.
Have a correlation. That is, in the above-described embodiment
As explained, the periodic output 100 _y0And
Then convert the periodic output 100 as described above, eg
For example, even (for example, twice) times differentiation (other, even times difference)
Minute, even number integration, or even number summation)
Value is ξ_y2Therefore, ξ_y0(= X) and ξ_y2(= Y)
As shown in FIG. 21, the point defined by
Located within a line and on a given straight line 112 if it is perfectly periodic
The wider the range 11
It comes to be located in 4. Therefore, in the present embodiment,
Wheel speed vibration Δω₁Is the response output ξ_y0And the response output ξ
_y0Ξ converted from the above_y2Correlation coefficient σ (equation (8
5)) is calculated, and the periodic signal is calculated based on the calculated correlation coefficient σ.
Only the response output that is the No. is selected.

[Equation 50] However, ρ _c is a forgetting factor, and N is the number of response outputs. That is, the selection means 10 executes the selection processing routine shown in FIG. 22 every time the wheel speed vibration Δω ₁ is input from the preprocessing filter 2. In step 22, pre-processing filter (band pass) processing is performed. In step 24, the response output ξ _y0 obtained from the wheel speed vibration Δω ₁ after the pre-processing filter processing
And a correlation coefficient σ (equation (85)) with the value ξ _{y2 obtained} by converting the response output ξ _y0 as described above. In step 26, it is determined whether or not the calculated correlation coefficient σ is within a predetermined range (σ ₁ <σ <σ ₂ ).
It is determined whether ω ₁ is a periodic signal. If the calculated correlation coefficient σ is not within the predetermined range, the wheel speed vibration Δω ₁ input this time is not a periodic signal, and this routine is ended. As a result, since the wheel speed vibration Δω ₁ input this time is not input to the transfer function identification means 3, the transfer function of the wheel resonance system is identified and the μ gradient is calculated based on a signal that is not a periodic signal. Can be prevented. On the other hand, when the calculated correlation coefficient σ is within the predetermined range, since the wheel speed vibration Δω ₁ input this time is a periodic signal, the wheel speed vibration Δω ₁ input this time is transmitted in step 28. Input to the function identification means 3. This allows
The transfer function of the wheel resonance system is identified based on the periodic signal, and the μ gradient is calculated. Therefore, it is possible to improve the identification accuracy of the transfer function and the estimation accuracy of the μ gradient. Here, even if the output (noise) having a suddenly large amplitude is converted as described above, the converted value includes many components other than the periodic part. That is, since the value obtained by converting the output with the suddenly large amplitude as described above does not change in response to the change in the response output, the output with the suddenly large amplitude and the value converted as described above The correlation coefficient σ with and does not fall within the above predetermined range. For example, FIGS. 24A and 24B
As shown in, the noise K appears in a predetermined range in which the frequency includes the frequency F ₁ and the correlation coefficient σ is outside the predetermined range.
(See FIG. 23A) is removed by the sorting means 10. Further, as shown in FIG. 25, the value obtained by converting the output C in which no periodical change does not significantly appear as described above has few components converted so as to change corresponding to the change in the response output. . Therefore, the output C where the periodic change does not significantly appear
And the correlation coefficient σ between the response output and the value converted as described above
Does not fall within the above range. For example, in FIG.
As shown in FIG. 26B, the signal group C (see FIG. 26A) whose correlation coefficient σ is out of the predetermined range is removed by the selecting means 10. The output C in which the periodical change does not significantly appear is obtained when the vehicle travels on the dirt road. On the other hand, as shown in FIGS. 24 (C) and 24 (D), the signal group B (see FIG. 23 (B)) having no noise component and having the correlation coefficient σ within the predetermined range is selected by the selecting means 10. Are input to the transfer function identification means 3. Also, FIG.
As shown in FIG. 5, the value obtained by converting the periodic output D as described above has many components converted so as to change corresponding to the change in the response output. Therefore, the correlation coefficient σ between the periodic output D and the value obtained by converting the response output as described above falls within the above predetermined range. For example, FIG.
As shown in (B), the signal group D (see FIG. 26A) in which the correlation coefficient σ is within the predetermined range is not removed by the selecting means 10. Therefore, it is input to the transfer function identification means 3. The periodic output D shown in FIG. 25 is obtained when the vehicle travels on an asphalt road. Therefore,
The unknown element can be estimated based on the response output from which the response output having a suddenly large magnitude or the output in which the periodical change does not significantly appear is removed. In the fifth embodiment described above, the μ gradient is calculated, but the present invention is not limited to this, and the twisting constant of the tire may be calculated. That is, the present embodiment is applicable not only to the above-described first embodiment (first aspect) but also to all the above-described embodiments.

In each of the embodiments of the present invention described above, the transfer function of the continuous time model is identified and the parameters of the transfer function are directly estimated.
It is possible to avoid the above problems associated with the identification of the discrete time model.

Further, since the wheel states included in the vibration model established in various running states are identified, the wheel states in various running states such as braking, driving and steady running are estimated without being limited to braking. be able to.

The embodiments of the present invention have been described above. However, the present invention is not limited to the above-mentioned examples, and can be arbitrarily changed without departing from the scope of the present invention.

For example, in the above-described first to fifth embodiments, the wheel resonance system has been described as the secondary vibration model represented by the equations (11) and (12). It is obvious that the wheel state may be estimated based on the third-order vibration model expressed by the equation (6).

As described above, according to the present invention,
Based on a vibration model that includes a physical quantity related to the slipperiness between the tire and the road surface as an unknown element, at least the detected response output is substantially satisfied without assuming that the twisting constant of the tire is known. Since the unknown element is estimated as described above, it is possible to highly accurately determine the physical quantity related to the slipperiness between the tire and the road surface, and even when the resonance frequency is changed due to tire replacement or tire pressure change. The excellent effect of being able to respond immediately is obtained.

Further, in the invention of claim 1, the vibration model has a model that is established as the frictional characteristic between the tire and the road surface irrespective of the traveling state such as braking, driving and steady traveling. As described in the prior art, there is no problem that the opportunity for estimation is limited to braking, and the wheel state can be equally estimated in various traveling states such as braking, driving, and steady traveling. It should be noted that even in the case of a vehicle model that does not have a vibrating means in the brake unit or a vehicle model that can be excited only during braking, the wheel condition can be estimated by using the road surface disturbance as the vibration input.
Furthermore, the vibration model relates to the twisting constant of the tire.
The physical quantity is included as an unknown element of the wheel state, and the physical quantity is estimated.
Since it is possible to determine the tire pressure,
The effect that diagnosis is possible is also obtained.

Further, according to the second aspect of the invention, the vibration model is adaptively corrected based on the detected response output, so that the estimation accuracy can be further improved.

[0210]

Further, in the invention of claim 3, since the vibration input is applied to the wheel resonance system, highly accurate estimation is possible even when the road surface has few irregularities and the road surface disturbance is small.
The effect is obtained.

Further, the inventions of claims 4 to 6 are:
Since the vibration input applied to the wheel resonance system is detected by the vibration input means, a further effect that more parameters can be estimated is obtained. Further, the invention of claim 7 and claim 8 calculates a judgment value for judging whether or not the response output is a periodic output, and based on the calculated judgment value, The output to be input to the estimation means is selected, and the unknown element is estimated using the selected output. Therefore, the response output is obtained by removing the response output that is suddenly large in size or the output in which no periodic change is noticeable. The effect that the unknown element can be estimated based on

[Brief description of drawings]

FIG. 1 is a diagram showing a dynamic model equivalent to a wheel resonance system according to the present invention.

FIG. 2 is a diagram showing friction characteristics between a tire and a road surface in a wheel resonance system according to the present invention.

FIG. 3 is a conceptual diagram of a vibration model according to the present invention.

4 (A) and (B) show the actual brake pressure P _m.
3A and 3B are experimental results of frequency characteristics in the wheel resonance system when various values are changed, where FIG.

5 (A) and 5 (B) are vibration models calculated in the vibration model according to the present invention, respectively, when a physical quantity D ₀ relating to slipperiness between a tire and a road surface is variously changed. 3A and 3B are graphs showing the transfer characteristics of FIG.

6 (A) and 6 (B) are frequency analysis results of wheel speed signals actually detected at driving wheels, where (A) relates to an asphalt road and (B) relates to a dirt road.

7 (A) and 7 (B) are vibration models calculated according to various changes of a physical quantity D ₀ relating to the slipperiness between a tire and a road surface in the vibration model according to the present invention. 3A and 3B are graphs showing the transfer characteristics of FIG.

FIG. 8 is a block diagram showing a configuration of a first aspect according to the first exemplary embodiment of the present invention.

FIG. 9 is a block diagram showing a configuration of a second aspect according to the first exemplary embodiment of the present invention.

FIG. 10 is an estimation result of a physical quantity D ₀ related to the slipperiness between a tire and a road surface, which is estimated by the wheel state estimation device according to the first embodiment.

FIG. 11 is a block diagram showing a configuration of a second exemplary embodiment of the present invention.

FIG. 12 is an estimation result of a resonance frequency of a wheel resonance system estimated by the wheel state estimation device according to the second embodiment.

FIG. 13 is a block diagram showing a configuration of a first aspect according to a third exemplary embodiment of the present invention.

14A to 14C are diagrams showing a vibration input waveform of a vibration input means according to the third embodiment, and FIG.
Is a pseudo-random waveform, (B) is a stepwise waveform, (C)
Shows an impulse-like waveform.

FIG. 15 is a block diagram showing a configuration of a second aspect according to the third exemplary embodiment of the present invention.

FIG. 16 is a block diagram showing a configuration of a first aspect according to a fourth exemplary embodiment of the present invention.

FIG. 17 is a block diagram showing a configuration of a second aspect according to the fourth exemplary embodiment of the present invention.

FIG. 18 is a block diagram showing a configuration according to a fifth exemplary embodiment of the present invention.

FIG. 19 is a graph showing a periodic signal and two signals that are not periodic.

20A and 20C are graphs showing signals that are not periodic signals, and FIG. 20B is a graph showing periodic signals.

FIG. 21 is a diagram showing a relationship between a response output x and a value y obtained by converting the response output so that a periodic part changes corresponding to a change in the response output.

FIG. 22 is a flowchart showing a selection processing routine executed by the selection means.

23A is a graph showing the correlation coefficient of a periodic signal and a non-periodic signal (noise) in time series, and FIG. 23 (B) is a graph showing a periodic signal and a non-periodic signal (noise). ) Is a graph showing the vehicle speed in time series.

24 (A) and (B) are graphs showing wheel speed signals with relatively low periodicity, and (C) and (D) are
Is a graph showing a relatively high periodic wheel speed signal,
(A) and (C) are front right wheel speed signals, and (B) and (D) are front left wheel signals.

FIG. 25 is a graph showing a frequency-spectrum relationship between a periodic signal and a signal in which a periodic change does not significantly appear.

FIG. 26A is a graph showing the correlation coefficient of a periodic signal and a signal in which a periodic change does not significantly appear in a time series, and FIG. 26B is a graph showing a periodic signal and a periodic signal. It is a graph which showed the wheel speed corresponding to the signal which does not show a remarkable change in time series.

[Explanation of symbols] 1 Wheel speed detection means 2 Pre-processing filter 3 Transfer function identification means 4 μ gradient calculation means 5 Application means 7 Spring constant calculation means 10 Sorting means

Continuation of the front page (72) Inventor Katsuhiro Asano, Nagachite-cho, Aichi-gun, Aichi-gun, Nagakage, Yoko 41, No. 1 at Toyota Central Research Institute Co., Ltd. No. 1 Toyota Central Research Institute Co., Ltd. (56) Reference JP-A-8-334454 (JP, A) JP-A-9-52572 (JP, A) JP-A-8-324414 (JP, A) JP-A-11 -20649 (JP, A) JP-A-5-79934 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) B60T 8/58 B60R 16/02 661

Claims (8)

(57) [Claims] 1. A wheel state estimating device for estimating a wheel state in a wheel resonance system including a friction characteristic between a tire and a road surface, the detecting means detecting a response output to a vibration input to the wheel resonance system. And the transfer characteristics of the wheel resonance system from the vibration input to the response output, at least the physical quantity related to the slipperiness between the tire and the road surface and the physical quantity related to the twisting constant of the tire.
And an estimation means for estimating the unknown element that substantially satisfies at least the response output detected by the detection means, based on the vibration model. A wheel state estimation device characterized by the above. 2. The wheel state estimating device according to claim 1, further comprising a correcting unit that corrects the vibration model based on a response output detected by the detecting unit. 3. Excitation applying an excitation input to the wheel resonance system
The wheel state estimation device according to claim 1 or 2 , further comprising means . 4. The wheel resonance system is applied to the wheel resonance system by the vibrating means.
The wheel state estimation device according to claim 3 , further comprising a vibration input detection means for detecting the vibration input obtained . 5. The wheel resonance system is applied to the wheel resonance system by the vibrating means.
The vibration input that can be obtained is random input, impulse input,
The wheel state estimation device according to claim 4, wherein the wheel state estimation device is any one of step inputs . 6. The wheel resonance system is applied by the vibrating means.
The frequency of the vibration input obtained is the twist constant of the tire.
5. The resonance frequency obtained from
Wheel condition estimation device. 7. A response output detected by the detecting means.
It is determined whether the response output is a periodic output or not.
A calculating unit for calculating a judgment value for calculating the judgment value, and based on the judgment value calculated by the calculating unit,
The estimated hand is selected from the response outputs detected by the detection means.
Selecting means for selecting an output input to the stage , wherein the estimating means estimates the unknown element by the selecting means.
Characterized by using the output selected by another means
The wheel state estimation device according to any one of claims 1 to 6 . 8. The calculation means uses the previous value as the judgment value.
The response output detected by the detection means and the response output
The periodic part of the response output corresponds to the change of the response output
Request to calculate the correlation coefficient with the converted value
Item 7. A wheel state estimation device according to item 7.