JP3207570B2 - Vehicle speed estimation method and vehicle speed detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for estimating a vehicle speed (speed during traveling) required for detecting a dangerous state in an apparatus for assisting safe driving of an automobile, and more particularly, to a method for coupling to a drive system of the automobile. The present invention relates to a vehicle speed estimating method that does not require the above means, and a vehicle speed detecting device using the method.

[0002]

2. Description of the Related Art In a safe driving support system which detects a danger state before an accident occurs and generates, for example, an alarm while an automobile is running, a vehicle speed signal is required to detect the danger state. Sometimes.

[0003] For example, during traveling, the distance from the preceding vehicle (inter-vehicle distance) is always measured, and when the inter-vehicle distance becomes shorter than the dangerous distance (distance set as a distance that may collide with the preceding vehicle). In a collision warning radar that emits a warning, it is necessary to change the setting of the dangerous distance in real time in accordance with the vehicle speed, and a vehicle speed signal is required for automatically changing the setting of the dangerous distance.

Conventionally, there are mechanical and electric methods for extracting a vehicle speed signal. In each of the methods, the vehicle speed signal is extracted using means directly or indirectly coupled to a drive system of an automobile.

[0005]

According to the above-mentioned conventional method, there are the following problems, especially when a completed automobile is additionally equipped with a safe driving support device.

(1) Since the vehicle speed detection method differs depending on the manufacturer and the type of the vehicle, a safe driving support device is required for each vehicle speed detection method and lacks versatility.

(2) At the time of additional installation of the safe driving support device,
Equipment that requires remodeling on the vehicle body side, especially for mechanically extracting a vehicle speed signal, requires additional construction of a harness, and the installation work is not easy and the equipment cost is high.

[0008] The present invention proposes to solve the above problems, and can be installed in any vehicle model of any manufacturer, and when additional equipment is installed, safe driving that does not require modification of the vehicle body side It is an object of the present invention to provide a vehicle speed estimation method for providing a support device and a vehicle speed detection device using the method.

[0009]

The inventor of the present invention analyzes the frequency of a vibration occurring in a vehicle body during the running of an automobile by frequency spectrum analysis, and integrates the vibration level obtained by a plurality of analyzes for each frequency component. In a relatively low frequency range of 10HZ to 100HZ, the integrated data of the vibration level changes with a certain correlation with a change in the vehicle speed, and the correlation varies depending on the type of vehicle, but the state of the traveling road surface ( Paved, unpaved, wet, dry surfaces, etc.).

The present invention estimates the vehicle speed by utilizing the above phenomena. That is, while the vehicle is running, a process of continuously detecting vibrations occurring in the vehicle body of the vehicle, and performing a frequency spectrum analysis of a vibration signal obtained by the detection is repeatedly performed at a set cycle.
Vibration levels for a certain range of frequency components set within a range of 100 Hz to 100 HZ are integrated for each of the above-described frequency components for the set number of times, and the vibration level integrated values of singular points and the like appearing in the obtained vibration level integrated data Alternatively, the vehicle speed is estimated based on the frequency, and the singular point is the next singular point where the correlation with the vehicle speed appears relatively remarkably, that is, (1) the vibration level appearing in the vibration level integrated data And (2) the minimum point of the vibration level appearing in a frequency range lower than the maximum point, and in order to increase the accuracy of the estimated vehicle speed, as another specific point, (3) the minimum point A point separated by a certain frequency exceeding the frequency of the above-mentioned maximum point based on the frequency is selected as a point for determining the correlation, and the vibration level of each point is integrated. Or it is obtained so as to estimate the vehicle speed from the frequency correlation and the vehicle speed.

Further, since the singular point or the specific point in the above (1) to (3) has a correlation with the vehicle speed, the vehicle speed can be estimated from the vibration level integrated value or the frequency at any one point. In order to increase the accuracy of the estimated vehicle speed, it is preferable to estimate the vehicle speed using a plurality of points. In this case, each estimated vehicle speed obtained from the plurality of points is processed by fuzzy inference and finally processed. If a more accurate vehicle speed estimation value is obtained, a more accurate vehicle speed estimation becomes possible.

[0012]

[Function] Vibration generated in the body of an automobile is continuously detected, frequency spectrum is analyzed every set time, and the analysis data is accumulated (integrated). The noise component is canceled by the integration of the analysis data, and in particular, a component having regularity in a relatively low frequency range of 10 Hz to 100 Hz appears as shown in FIGS. 4A and 4B.

That is, FIG. 4A shows the frequency versus vibration level characteristics before starting the vehicle engine (when the vehicle is stopped, that is, during idling).
Three maximum points appear between Z.

FIG. 4 (B) shows frequency versus vibration level characteristics during driving of an automobile.
Between Z, the lowest point B of the vibration level appears first from the lower frequency, and then the maximum point P of the vibration level appears. In the frequency range higher than the maximum point P, the vibration level exhibits relatively complicated characteristics. Decreases. The higher the vehicle speed, the lower the vibration level in a frequency range lower than the frequency Fp of the maximum point P, and the higher the vibration level in a high frequency range. Is the same regardless. However, although the actual measurement value differs depending on the vehicle type, the actual measurement value is irrelevant to the state of the traveling road surface for the same vehicle type.

Therefore, focusing on the singular points appearing on the above characteristics, that is, the maximum point P and the minimum point B, the correlation between the vibration levels Lp and Lb and the frequencies Fp and Fb of the two points P and B and the vehicle speed is obtained. The vibration levels Lp and Lb decrease with a certain function as the vehicle speed increases as shown in FIG. 5A, and the frequencies Fp and Fb increase with a certain function as the vehicle speed increases as shown in FIG. 5B. It was confirmed.

Further, as a result of obtaining a correlation in a frequency range higher than the frequency Fp of the maximum point P, the vibration level at the point (B + α) appearing at a frequency F (b + α) higher than the frequency Fb of the minimum point B by a certain frequency α. L (b + α) is found to have the best correlation with the vehicle speed.
It has been confirmed that (b + α) increases by a certain function as the vehicle speed increases, as shown in FIG. FIG.
In (B), the point (B + α) appears near the minimum point. However, depending on the vehicle speed, the point (B + α) is not always near the minimum point. Further, in FIGS. 5A to 5C, the discontinuous portion appearing in the low vehicle speed portion is due to the characteristic shown in FIG. 4A, and the vehicle speed in this portion is determined to be zero (stop). It will be.

As described above, since the vibration of the vehicle body includes a component having a correlation with the vehicle speed, the vehicle speed can be estimated by detecting this component, and the vibration is directly or indirectly coupled to the drive system of the vehicle. A vehicle speed detection device that does not require any means can be realized.

Further, if the vehicle speed detecting device according to the present invention is used in the vehicle speed detecting portion of the safe driving assist device requiring vehicle speed information, a versatile safe driving assist device that can be mounted on any vehicle type is realized. Since the modification of the vehicle body is not required for mounting the safe driving support device, it is very easy to add the equipment to an existing automobile.

[0019]

1 is a block diagram, FIG. 2 is a flowchart of a vehicle speed estimating process, FIG. 3 is a flowchart of a correlation data initial setting process, and FIG. 4 is a frequency diagram. FIG. 6 is a diagram showing vibration level characteristics,
(A) is a characteristic example when the vehicle body is stopped (idling),
FIG. 5B is a diagram showing a correlation between the vibration signal and the vehicle speed, and FIG. 5A is a diagram showing a correlation between the level and the vehicle speed at the maximum and minimum points of the vibration level.
6B shows an example of the correlation between the same frequency and the vehicle speed, and FIG. 6C shows an example of the correlation of the vibration level vehicle speed at a point higher than the frequency of the vibration level minimum point by a certain frequency (α).
(A), (B) is a figure explaining the vehicle speed estimation process by fuzzy inference.

In FIG. 1, reference numeral 1 denotes a vehicle speed detecting device according to an embodiment of the present invention, and 2 denotes a safe driving support device using data (an estimated vehicle speed) from the vehicle speed detecting device 1, such as a collision warning radar. It is.

In the vehicle speed detecting device 1, 11 is a vibration sensor, 12 is an A / D converter, 13 is an initial setting operation unit, 14
Is a CPU and 15 is a memory.

The vibration sensor 11 serves as vibration detecting means for detecting low-frequency micro-vibration of the vehicle body of the vehicle, and is mounted at a position where the vibration of the vehicle body can be easily detected. Further, since the vibration phenomenon of the object is an event that can be grasped by acceleration or angular velocity, an acceleration sensor or angular velocity sensor can be used as the vibration sensor 11 (an acceleration sensor is used in the embodiment).

The A / D converter 12 converts an analog signal into a digital signal. The vibration sensor 11 outputs a vibration signal in the form of an analog signal, whereas the processing in the CPU 14 converts the vibration signal in the form of a digital signal into a digital signal. It is provided as needed. The vibration sensor 11
Output a vibration signal in the form of a digital signal, the A / D converter 12 is not required.

The initial setting operation section 13 is an operation section for activating a measuring means (included in the processing of the CPU 14) for previously measuring the correlation between the vibration and the vehicle speed, and is constituted by, for example, a push button switch.

The CPU 14 collectively performs the processing in the vehicle speed detecting device 1. The main function of the present invention is to convert a vibration signal into, for example, a known FFT (Fast F) signal.
frequency transformation and processing from 10 Hz to 100 Hz.
A certain range in the low frequency range of HZ (for example, 20HZ to 70HZ)
FFT processing function 141 (frequency spectrum analysis means) for outputting vibration level data for the frequency component of HZ), vehicle speed estimation function 142 (vehicle speed estimation means) for estimating the vehicle speed based on the frequency spectrum analysis processing data, Fuzzy processing function 143 (fuzzy inference means) for making the vehicle speed estimation value estimated by the function 142 highly accurate by fuzzy inference, and the vehicle speed detection device 1
Before the operation, there is a correlation data setting function 144 (correlation data measuring means) for performing a process of measuring the correlation between the vehicle speed and the vibration signal in advance and creating correlation data matching the vehicle type.
All of these functions are achieved by software of the CPU 14. Note that the fuzzy processing function 143 often uses a CPU dedicated to fuzzy inference separately from the CPU 14 because the number of processed data is enormous (for convenience of explanation, the fuzzy processing function 143 is replaced with the function of the CPU 14). I do.)

The memory 15 includes a first memory 151 (first storage means) for temporarily storing a vibration signal input from the vibration sensor 11 via the A / D converter 12 for a set time, and 1 stores the analysis data obtained by frequency spectrum analysis of the vibration signal stored in the memory 151;
It has a memory 152 (second storage means) and a third memory 153 (third storage means) in which correlation data between the vibration signal and the vehicle speed is stored in advance, and is omitted in the drawing. However, there is a program memory of the CPU 14, a temporary storage memory for intermediate data during processing, and the like.

The collision warning radar 2 includes a vehicle speed detection device 1
This is an example of a safe driving support device that uses vehicle speed information (vehicle speed signal) output from a vehicle. With this collision warning radar, it is necessary to change the dangerous distance at which a warning is required according to the vehicle speed in real time in order to smooth the flow of vehicles on the road, and therefore, vehicle speed information is required However, very high accuracy is not required as vehicle speed information for changing the setting of the dangerous distance, and a vehicle speed estimated value obtained by the present invention can be sufficiently used.

The vehicle speed detecting operation will be described below in accordance with the processing flow shown in FIG.

The vibration sensor 11 constantly detects vibration occurring in the vehicle body when the vehicle is idling and running, and a vibration signal output from the vibration sensor 11 is converted into a digital value by the A / D converter 12. After being converted to CP
It is continuously sent to U14.

The CPU 14 reads the vibration signal output from the A / D converter 12 and stores the vibration signal in the first memory 151 (step S1).

Next, the CPU 14 determines whether or not a vibration signal for a predetermined time (set time) has been stored in the first memory 151 (step S2). Before the storage for the set time, steps S1 and S2 are performed. The vibration signal is stored in the first memory 151 one after another by repeating the processing described above, and when the storage of the vibration signal for the set time is confirmed, the processing shifts to the processing after step S3.

The monitoring of the storage of the vibration signal for the set time is performed as follows. That is, the CPU 14 reads the vibration signal from the A / D converter 12 based on a fixed time reference (clock signal).
By counting the number of times the vibration signal is read in 4 and monitoring whether the count value has reached a set value, it is possible to confirm the end of storing the vibration signal in the first memory 151 for a set time.

When the vibration signal for the set time is stored in the first memory 151 as described above, the CPU 14 then performs the frequency spectrum analysis processing of the vibration signal stored in the first memory 151 to the FFT processing function 141. (Step S3), for example, the vibration level analysis processing data (first data) appearing in the frequency components in a relatively low range of 20 Hz to 70 Hz is stored in the second memory 152 for each frequency component (Step S4). ).

The storage of the analysis processing data in the second memory 152 is performed by adding new data to the already stored data. That is, at the same time as the processing of storing the analysis processing data in the second memory 152, the integration processing of the data is performed.

Next, the CPU 14 determines whether or not the analysis processing data has been stored in the second memory 152 a predetermined number of times, that is, whether or not the number of times of integration of the data has reached a predetermined number (step S1). S5) If the set number has not been reached, the processing from steps S1 to S5 is repeated to store the analysis processing data in the second memory 152.

When a large number of data subjected to the FFT operation are integrated as described above, irregular noise components included in the vibration signal are canceled by the integration, and only the vibration level due to the regular component is analyzed. Appears in

When the integration of the set number of times of the first data is completed, the analysis processing integrated data (second data) shown in FIG. 4B is stored in the second memory 152.

Next, the CPU 14 reads the integrated data of the analysis processing from the second memory 152 (step S6), and the pattern of the data is the pattern at the time of stop (the pattern of FIG. 4A) or the pattern at the time of running (the pattern at the time of running). It is determined whether the pattern is the pattern at the time of stop (step S7). If the pattern is the stop pattern, the process proceeds to step S11 to output the estimated vehicle speed value 0 (zero) to the collision prevention radar 2.

If it is determined in step S7 that the integrated data of the analysis processing is a pattern during traveling, then the CPU 14
Are the vibration level maximum point P and the maximum point P of the data.
The vibration level minimum point B appearing in the lower frequency range is detected, and the vibration level Lp of the maximum point P and the frequency Fp,
The vibration level Lb of the minimum point B, the frequency Fb, and a point (B +
α) of vibration level L (b + α) (step S)
8).

Next, the CPU 14 confirms that the correlation data has been set in the third memory 153 (step S).
9), the vibration level Lp,
Lb, L (b + α) or frequencies Fp, Fb are stored in the third memory 153 using the collation data (FIG.
(Data shown in (C)) (Step S10)
The estimated vehicle speed is obtained and output to the collision prevention radar 2 (step S11). In step S9, when it is detected that the correlation data is not set in the third memory 153, the process proceeds to the correlation data initial setting process. This processing will be described later with reference to FIG.

The data used in the vehicle speed estimation processing does not require all of the vibration levels Lp, Lb, L (b + α) or the frequencies Fp, Fb.
The vehicle speed can also be estimated from the two data. However, in order to increase the accuracy, it is better to estimate the vehicle speed using a plurality or all of the data. In this case, if the fuzzy inference is further performed by using the fuzzy processing function 143, the vehicle speed can be increased with higher accuracy. Can be estimated. The fuzzy inference processing will be described later. When there is data that is not used for vehicle speed estimation in the above data, it goes without saying that the detection and collation of the unused data is not required, and the correlation data for the unused data is not required.

As described above, since the vibration characteristics shown in FIG. 4B slightly differ depending on the vehicle type, the correlation data shown in FIGS. 5A to 5C also differ depending on the vehicle type. Therefore, when operating the vehicle speed detecting device 1,
It is necessary to perform an initial setting operation for presetting correlation data matching the vehicle type in the third memory 153. Therefore, FIG.
The initialization process of the correlation data will be described below according to the processing flow shown in FIG.

The initial setting operation of the correlation data is performed as follows. That is, when it is confirmed that the vehicle speed has reached a predetermined first vehicle speed of, for example, 20 km / H by the speedometer by driving the automobile, the push button of the initial setting operation unit 13 is pressed, and Similarly, the vehicle speed is predetermined as the second vehicle speed, for example, 40 km /
When it is confirmed that the state has become H, the push button is pressed again. By the above two push button operations, the CPU 1
4 performs the following processing.

Steps S1 to S9 in FIG. 2 are similarly performed in the initial setting process.

In the step S9 in FIG.
After confirming that no correlation data is set in 3,
The CPU 14 starts the processing by the correlation data setting function 144.

The CPU 14 monitors whether or not a push button has been pressed on the initial setting operation unit 13 (step S101).
While the push button is not pressed, step S in FIG.
The processing from 1 to S9 is repeated.

The vehicle speed becomes the first vehicle speed, and the initial setting operation unit 1
When the push button is depressed in step 3, the CPU 14 detects this and detects the frequencies Fp, Fb, F (b + α) and the vibration levels Lp, Lb, L (b + α) of the integrated data of the analysis process detected in step S8 immediately before. ) Is stored in a temporary storage memory (not shown) (step S102), and it is confirmed whether this initial setting operation is the first operation or the second operation (step S103). In this case, since this is the first operation, the processing of the CPU 14 repeats the processing of steps S1 to S9 in FIG.
At 01, it is monitored whether or not the second initialization operation is performed.

The vehicle speed becomes the second vehicle speed, and the initial setting operation unit 1
When the push button is depressed again in step 3, the CPU 14 determines the frequencies Fp, Fb, F (b + α) and the vibration level L of the analysis integrated data in the same manner as in the first operation.
p, Lb, and L (b + α) are stored (S102).

By the above processing, the CPU 14 determines the two different vehicle speeds and vibration levels Lp, Lb, L
Since the relationship between (b + α) and the frequencies Fp and Fb is known, the correlation is calculated based on the relationship (step S104).

As described above, the vibration level-vehicle speed characteristics and the frequency-vehicle speed characteristics show the same tendency regardless of the vehicle type. By connecting the points with the lines representing the respective characteristics, correlation data can be obtained. That is, for example, the vehicle speed and the vibration level Lp will be specifically described as an example. As shown in FIG. 5A, the point P1 of the vibration level Lp1 at the first vehicle speed S1 and the second vehicle speed Sp
If the point P2 of the vibration level Lp2 at 2 is marked on the map in the memory, and the point P2 is connected by a previously measured characteristic curve of the vibration level Lp versus the vehicle speed, correlation data (a) for the vibration level maximum point P is obtained. Can be

After calculating each correlation data as described above, the CPU 14 stores the correlation data in the third memory 153.
(Step S105), and the process ends.

Next, a method for estimating the vehicle speed by fuzzy inference will be described with reference to FIGS.

The membership function is shown in FIG.
This is set for each of the correlation data (a) to (e) shown in (C). K1 to K5 shown in FIG. 6B are membership functions set corresponding to the correlation data (a) to (e), respectively.

Here, the concept of the membership function K will be described with reference to FIG. 6A. If the vehicle speed value obtained by referring to the correlation data is v, the actual vehicle speed is v
A function that passes through a point where the probability i that the vehicle speed is v is the highest, the probability that the vehicle speeds are va and vb is zero, and the probability i is zero, A membership function K is set. The meaning of the membership function K is to indicate the actual vehicle speed when the vehicle speed value v is detected by the probability, and the probability that the detected vehicle speed value v is, for example, the actual vehicle speed vc is ic. .

The fuzzy inference processing using the membership function described above will be described below.

Assuming that the vehicle speed values obtained based on the correlation data (a) to (e) shown in FIGS. 5A to 5C are v1 to v5, as shown in FIG. , Vehicle speed value v1
Ｖv5, as shown in FIG.
When the membership functions K1 to K5 described above are set for each of the vehicle speed values v1 to v5, and then the membership functions K1 to K5 are combined, they are indicated by solid lines.
A probability curve g is obtained. In this probability curve g, the point PK indicating the highest probability is detected, and the vehicle speed value at this point PK, in the example shown in FIG. 6B, the value with the highest probability that the vehicle speed value v2 is the actual vehicle speed And outputs this vehicle speed value v2 as an estimated vehicle speed value. This processing is executed by the CPU 1
4 is performed by the fuzzy processing function 143 as processing subsequent to step S10 in the processing flow shown in FIG.

Although the membership function is a complicated function (for example, a Gaussian distribution function) depending on the object of the inference, a simple function (a triangular shape) as shown in FIG. Function) may be used, and this embodiment also uses this simple function.

[0058]

As described above, according to the present invention, the vehicle speed correlation component included in the vehicle body vibration is extracted by the frequency spectrum analysis processing of the vibration signal and the integration processing of the analysis data, and the vehicle speed is estimated based on this. Thus, it is possible to obtain a vehicle speed detecting device which does not require a means for coupling to a drive system of an automobile, and it is possible to realize various safe driving support devices which can be mounted on any vehicle type by using this device. In addition, since it is not necessary to remodel the vehicle body when installing the device, it is very easy to add equipment to an existing vehicle (completed vehicle), and additional equipment is available for any vehicle type. is there.

[Brief description of the drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

FIG. 2 is a flowchart of an embodiment of the present invention.

FIG. 3 is a flowchart of the embodiment of the present invention.

4A is a diagram showing vibration characteristics at the time of a stop, and FIG. 4B is a diagram showing vibration characteristics at the time of traveling.

FIGS. 5A, 5B, and 5C are diagrams showing a correlation between a vibration signal component and a vehicle speed.

FIG. 6A illustrates a membership function,
(B) is a conceptual diagram of vehicle speed estimation by fuzzy inference.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Vehicle speed detecting device 2 ... Collision prevention radar 11 ... Vibration sensor 12 ... A / D converter 13 ... Initial setting operation part 14 ... CPU 15 ... Memory 141 ... FFT processing function 142 ... Vehicle speed estimation function 143 ... Fuzzy processing function 144 ... correlation data setting function 151 ... first memory 152 ... second memory 153 ... third memory

Claims (6)

(57) [Claims] 1. A process in which a vibration occurring in a vehicle body is continuously detected while a vehicle is running, and a process of frequency spectrum analysis of a vibration signal obtained by the detection is repeatedly performed at a set cycle, and the process is performed within a range of 10 Hz to 100 Hz. A vibration level for a set fixed range of frequency components is integrated for each of the frequency components for the set number of times of the above-described processing, and utilizing the fact that the obtained vibration level integrated data has a correlation with the vehicle speed of the automobile, The integrated value or frequency of each vibration level at multiple singularities appearing in the level integrated data
The wave number or each of the plurality of singularities and the plurality of features
The product of the vibration level at a specific point determined based on one of the different points
A vehicle speed estimation method for estimating a vehicle speed using a calculated value or a frequency . 2. The vehicle speed estimation method according to claim 1 , wherein the singular point is a maximum point of the vibration level appearing in the vibration level integrated data and a minimum value of the vibration level appearing in a frequency range lower than the maximum point. Is a point separated by a certain frequency exceeding the frequency of the maximum point on the basis of the frequency of the minimum point, by one or more of the vibration level integrated value or frequency of any of the maximum point, the minimum point and the specific point And a vehicle speed estimation method for determining the correlation with the vehicle speed. 3. The vehicle speed estimating method according to claim 2 , wherein when estimating the vehicle speed using a plurality of correlations, the vehicle speed is estimated by fuzzy inference from respective vehicle speed estimation values obtained using the correlations. A vehicle speed estimating method for obtaining an estimated final value. 4. A vibration detecting means mounted on a vehicle body of an automobile for detecting vibration of the vehicle body, and a first storage means for temporarily storing a vibration signal output by the vibration detecting means during a set time. Analyzing the frequency spectrum analysis of the vibration signal stored in the first storage means and outputting first data indicating a frequency component in a certain range set within a range of 10 Hz to 100 Hz versus the vibration level at every set period. Means for storing the first data output from the analyzing means for each frequency component while sequentially adding the first data until a set number is reached, thereby obtaining a second frequency characteristic versus frequency level integrated value characteristic. Second storage means for storing data; and a vibration level for each of a plurality of singularities appearing in the second data.
Or frequency, or each of the plurality of singularities and the
On the second data based on one of the plurality of singularities
Measuring means for measuring in advance the correlation between the determined vibration level or frequency of the specific point and the vehicle speed; and third storage means for storing the correlation data between the vibration level or frequency measured by the measuring means and the vehicle speed. The correlation data stored in the third storage means and the second data stored in the second storage means.
Each of the above multiple singularities expressed in the data of
Vibration level or frequency, or each of the above singularities
And the second data with reference to one of the plurality of singularities.
A vehicle speed detecting device comprising vehicle speed estimating means for outputting a vehicle speed estimated value using a vibration level or a frequency at a specific point determined on the data. 5. The vehicle speed detecting device according to claim 4 , wherein the third storage means stores at least two of the correlation data indicating the correlation between the vibration level or frequency of the singular point or the specific point and the vehicle speed. A vehicle speed detecting device for storing type of correlation data, wherein the vehicle speed estimating means has fuzzy inference means for performing fuzzy inference using a plurality of types of correlation data stored in the third storage means as a membership function. 6. The vehicle speed detecting device according to claim 4, wherein the singular point is a maximum point of the vibration level appearing in the second data and a minimum point of the vibration level appearing in a frequency range lower than the maximum point. A vehicle speed detecting device in which a specific point is a point separated by a certain frequency exceeding the frequency of the maximum point with reference to the frequency of the minimum point.