JP7768764B2 - Motor control device, motor control method, motor module, and electric power steering device - Google Patents

Description

This disclosure relates to a motor control device, a motor control method, a motor module, and an electric power steering device.

General automobiles are equipped with an electric power steering system (EPS) that includes an electric motor (hereafter simply referred to as "motor") and a motor control device. An electric power steering system is a device that assists the driver's operation of the steering wheel by driving the motor. Conventionally, torque control is used to achieve motor output corresponding to steering torque, thereby assisting steering operation.

Patent Documents 1 and 2 each disclose technology related to disturbance observer control. Patent Document 1 uses a robust controller to reduce the impact of disturbances or parameter fluctuations of the controlled object on steering control. Patent Document 2 uses a resonance point disturbance controller composed of a disturbance observer to suppress resonance point disturbances excited at the longitudinal resonance point of the suspension. Patent Document 3 discloses technology that eliminates friction torque generated by internal friction in the steering mechanism and generates an appropriate steering reaction force that does not feel unnatural in response to road reaction force.

Japanese Patent Application Publication No. 06-219310 International Publication No. 2016/208665 Japanese Patent Application Laid-Open No. 2005-88610

It is desirable to improve the steering feel that the driver feels when assisting the driver's steering.

In recent years, market demands for noise, vibration, and harshness (NVH), one of the standards used to evaluate vehicle comfort, have become increasingly stringent. However, conventional torque control is particularly susceptible to the effects of high-frequency disturbances and is unable to suppress high-frequency torque fluctuations, making it difficult to meet market demands.

Conventionally, a friction model was constructed as a function of the motor's angular velocity ω, and friction compensation control was performed using this model. However, with typical friction characteristics, there was an issue of chattering occurring easily due to the sign of the friction torque suddenly reversing when the motor's angular velocity ω approaches zero.

The present invention has been made to solve at least one of the above problems, and aims to provide a motor control device that can improve the steering feel perceived by the driver by applying model following control and/or friction compensation control to torque control, a motor module equipped with such a control device, an electric power steering device equipped with such a motor module, and a motor control method.

In a non-limiting exemplary embodiment, the control device of the present disclosure is a control device for controlling an electric power steering device equipped with a motor, and includes a model-following controller that generates a correction torque based on the output from the motor, which is a controlled object, and corrects the input to the controlled object using the correction torque. The model-following controller includes a high-pass filter having a first cutoff frequency and a low-pass filter having a second cutoff frequency greater than the first cutoff frequency. When the transfer functions of the low-pass filter and the high-pass filter are Q(s) and HPF(s), respectively, the transfer function of the controlled object is configured to be constrained to a nominal model in a frequency band where the gain in the gain characteristic of Q(s)·HPF(s) is 1.

In a non-limiting exemplary embodiment, the motor module of the present disclosure includes a motor and the above-described control device.

In a non-limiting exemplary embodiment, the electric power steering device of the present disclosure includes the motor module described above.

In a non-limiting exemplary embodiment, the control method of the present disclosure is a computer-implemented method for controlling a motor of an electric power steering device equipped with a motor, the method comprising: a model-following controller including a high-pass filter having a first cutoff frequency and a low-pass filter having a second cutoff frequency greater than the first cutoff frequency; where the transfer functions of the low-pass filter and the high-pass filter are Q(s) and HPF(s), respectively; the model-following controller constrains the transfer function of the motor, a controlled object, to a nominal model in a frequency band where the gain in the gain characteristic of Q(s)·HPF(s) is 1; and the computer is caused to generate a correction torque based on the output from the controlled object and correct the input to the controlled object with the correction torque.

According to an exemplary embodiment of the present disclosure, there are provided a motor control device that can improve the steering feel perceived by the driver by applying model following control and/or friction compensation control to torque control, a motor module including the control device, an electric power steering device including the motor module, and a motor control method.

FIG. 1 is a diagram schematically illustrating an example of the configuration of an electric power steering device according to an embodiment of the present disclosure. FIG. 2 is a block diagram illustrating a typical example of the configuration of a control device according to an embodiment of the present disclosure. FIG. 3 is a functional block diagram illustrating the functions of a processor for controlling a motor according to an embodiment of the present disclosure. FIG. 4 is a functional block diagram showing an example of the configuration of a model following controller in the first implementation example. FIG. 5 is a graph illustrating the gain characteristic T(s) of Q(s)·HPF(s) and the gain characteristic of the reciprocal of the modeling error Δ(s) between the plant and the nominal model P _n (s). FIG. 6 is a graph illustrating a gain diagram of the transfer function C(s) of the phase compensator in the torque controller. FIG. 7 is a graph illustrating a gain diagram of the transfer function HPF(s) of the high-pass filter. FIG. 8 is a graph illustrating a gain diagram of the nominal model P _n (s). FIG. 9 is a graph showing the measurement results of the steering angle and the torsion torque when the model following control is not applied. FIG. 10 is a graph showing the measurement results of the steering angle and the torsion torque when the model following control is applied. FIG. 11 is a graph showing the measurement results of the change in steering angle over time when model following control is not applied and when model following control is applied. FIG. 12 is a functional block diagram showing an example of the configuration of a model following controller in the second implementation example. FIG. 13 is a graph showing simulation results of steering angle and steering torque when friction compensation control is not applied and when friction compensation control is applied. FIG. 14 is a graph showing simulation results of steering angle and steering torque when conventional friction compensation control is applied and when friction compensation control according to the second implementation example is applied.

Below, with reference to the accompanying drawings, embodiments of a motor control device mounted in an electric power steering device of the present disclosure, a motor control method, a motor module including the control device, and an electric power steering device including the motor module will be described in detail. However, more detailed explanation than necessary may be omitted. For example, detailed explanations of already well-known matters and redundant explanations of substantially identical configurations may be omitted. This is to avoid unnecessary redundancy in the following explanation and to make it easier for those skilled in the art to understand.

The following embodiments are merely illustrative, and the motor control device and motor control method according to the present disclosure are not limited to the following embodiments. For example, the numerical values, steps, and order of steps shown in the following embodiments are merely examples, and various modifications are possible as long as no technical inconsistencies are created. The embodiments and examples described below are merely illustrative, and various combinations are possible as long as no technical inconsistencies are created.

1. Configuration of the electric power steering device 1000
FIG. 1 schematically shows an example of the configuration of an electric power steering device 1000 according to an embodiment of the present disclosure.

The electric power steering device 1000 (hereinafter referred to as "EPS") has a steering system 520 and an assist torque mechanism 540 that generates assist torque. The EPS 1000 generates assist torque that assists the steering torque of the steering system generated when the driver operates the steering wheel. The assist torque reduces the operating burden on the driver.

The steering system 520 includes, for example, a steering wheel 521, a steering shaft 522, universal joints 523A, 523B, a rotating shaft 524, a rack and pinion mechanism 525, a rack shaft 526, left and right ball joints 552A, 552B, tie rods 527A, 527B, knuckles 528A, 528B, and left and right steering wheels 529A, 529B.

The assist torque mechanism 540 includes, for example, a steering torque sensor 541, a steering angle sensor 542, an automotive electronic control unit (ECU) 100, a motor 543, a reduction gear 544, an inverter 545, and a torsion bar 546. The steering torque sensor 541 detects the amount of torsion of the torsion bar 546 to detect the steering torque in the steering system 520. The steering angle sensor 542 detects the steering angle of the steering wheel. Note that the steering torque may be an estimated value derived by calculation, rather than a value from the steering torque sensor. The steering angle can also be calculated based on the output value of an angle sensor.

ECU 100 generates a motor drive signal based on detection signals from steering torque sensor 541, steering angle sensor 542, a vehicle speed sensor (not shown), and the like, and outputs the signal to inverter 545. For example, inverter 545 converts DC power into three-phase AC power, which is a pseudo-sine wave of U, V, and W phases, in accordance with the motor drive signal, and supplies the power to motor 543. Motor 543 is, for example, a surface permanent magnet synchronous motor (SPMSM) or a switched reluctance motor (SRM), and receives the three-phase AC power supply to generate an assist torque corresponding to the steering torque. Motor 543 transmits the generated assist torque to steering system 520 via reduction gear 544. Hereinafter, ECU 100 will be referred to as EPS control device 100.

The control device 100 and motor are modularized and manufactured and sold as a motor module. The motor module includes a motor and control device 100 and is suitable for use in EPS. Alternatively, the control device 100 can be manufactured and sold as a control device for controlling EPS, independent of the motor.

2. Configuration example of control device 100
2 shows a typical example of the configuration of a control device 100 according to an embodiment of the present disclosure. The control device 100 includes, for example, a power supply circuit 111, an angle sensor 112, an input circuit 113, a communication I/F 114, a drive circuit 115, a ROM 116, and a processor 200. The control device 100 can be realized as a printed circuit board (PCB) on which these electronic components are mounted. The control device 100 is used to control a motor of an electric power steering device that includes a motor.

The vehicle speed sensor 300, steering torque sensor 541, and steering angle sensor 542 mounted on the vehicle are communicatively connected to the processor 200, and the vehicle speed, steering torque, and steering angle are transmitted from the vehicle speed sensor 300, steering torque sensor 541, and steering angle sensor 542 to the processor 200, respectively.

The control device 100 is electrically connected to the inverter 545 (see Figure 1). The control device 100 controls the switching operations of multiple switch elements (e.g., MOSFETs) included in the inverter 545. Specifically, the control device 100 generates control signals (hereinafter referred to as "gate control signals") that control the switching operations of each switch element and outputs them to the inverter 545.

The control device 100 generates a torque command value based on the steering torque, etc., and controls the torque and rotational speed of the motor 543, for example, by vector control. The control device 100 is not limited to vector control and can perform other closed-loop control. The rotational speed is expressed as the number of rotations (rpm) at which the rotor rotates per unit time (e.g., 1 minute) or the number of rotations (rps) at which the rotor rotates per unit time (e.g., 1 second). Vector control is a method of decomposing the current flowing through the motor into a current component that contributes to torque generation and a current component that contributes to magnetic flux generation, and independently controlling each of the mutually orthogonal current components.

The power supply circuit 111 is connected to an external power supply (not shown) and generates the DC voltage required for each block in the circuit. The generated DC voltage is, for example, 3V or 5V.

The angle sensor 112 is, for example, a resolver or a Hall IC. Alternatively, the angle sensor 112 can be realized by combining a magnetoresistive (MR) sensor having an MR element with a sensor magnet. The angle sensor 112 detects the rotation angle of the rotor and outputs the detected value to the processor 200. The control device 100 may include a speed sensor and an acceleration sensor that detect the rotation speed and acceleration of the motor, instead of the angle sensor 112. The processor 200 can calculate the angular velocity ω [rad/s] based on the electrical angle _θm of the motor.

The input circuit 113 receives the motor current value (hereinafter referred to as the "actual current value") detected by a current sensor (not shown), converts the level of the actual current value to an input level for the processor 200 as necessary, and outputs the actual current value to the processor 200. A typical example of the input circuit 113 is an analog-to-digital conversion circuit.

Processor 200 is a semiconductor integrated circuit, also referred to as a central processing unit (CPU) or microprocessor. Processor 200 sequentially executes a computer program stored in ROM 116, which contains instructions for controlling motor drive, to achieve the desired processing. In addition to or instead of processor 200, control device 100 may include a field programmable gate array (FPGA) equipped with a CPU, a graphics processing unit (GPU), an application specific integrated circuit (ASIC), an application specific standard product (ASSP), or a combination of two or more circuits selected from these circuits. Processor 200 sets a current command value according to the actual current value and rotor rotation angle, etc., generates a pulse width modulation (PWM) signal, and outputs it to drive circuit 115.

The communication I/F 114 is, for example, an input/output interface for transmitting and receiving data in accordance with an in-vehicle control area network (CAN).

Driver circuit 115 is typically a gate driver (or pre-driver). Drive circuit 115 generates gate control signals in accordance with the PWM signal and applies the gate control signals to the gates of multiple switch elements in inverter 545. When the drive target is a motor that can be driven at a low voltage, a gate driver may not necessarily be required. In that case, the gate driver function may be implemented in processor 200.

ROM 116 is electrically connected to processor 200. ROM 116 is, for example, a writable memory (e.g., PROM), a rewritable memory (e.g., flash memory, EEPROM), or a read-only memory. ROM 116 stores a control program including instructions for causing processor 200 to control motor drive. For example, the control program is temporarily loaded into RAM (not shown) at boot time.

Figure 3 shows functional blocks of a processor 200 for controlling a motor according to an embodiment of the present disclosure. In the illustrated implementation example, the processor 200, which is a computer, sequentially executes the processes (or tasks) required to control the motor using a torque controller, a model-following controller, a subtractor, and an adder.

Each functional block is implemented in the processor 200 as software (or firmware) and/or hardware. The processing of each functional block is typically written in a computer program in software module units and stored in ROM 116. However, when using an FPGA or the like, all or part of these functional blocks may be implemented as a hardware accelerator. Furthermore, the motor control method according to the embodiment of the present disclosure may be implemented in a computer and performed by causing the computer to perform the desired operations.

The control device 100 includes a torque controller 210, a model following controller 230, a subtractor AD1, and an adder AD2. In other words, the processor 200 implements functions corresponding to the torque controller 210, the model following controller 230, the subtractor AD1, and the adder AD2.

Torque controller 210 operates based on steering torque T _h and provides an input to controlled object 220, which is a motor. For example, steering torque T _h detected by steering torque sensor 541 is input to torque controller 210. Torque controller 210 generates target motor torque (or torque command value) T _ref by applying phase compensation to steering torque T _h when the steering frequency or steering speed is within a predetermined range, and inputs the target motor torque (or torque command value) T ref to controlled object 220.

The torque controller 210 illustrated in Figure 3 has a base assist calculation unit 211 and a phase compensator 212.

The base assist calculation unit 211 acquires the steering torque T _h and the vehicle speed. The base assist calculation unit 211 generates a base assist torque based on the steering torque T _h and the vehicle speed. For example, the base assist calculation unit 211 may have a look-up table (LUT) that defines the correspondence between the steering torque T h , the vehicle speed, and the base assist torque. The base assist calculation unit 211 can determine the corresponding base assist torque based on the steering _torque T _h and the vehicle speed by referring to the LUT. Furthermore, the base assist calculation unit 211 can determine a base assist gain based on a slope defined by the ratio of the amount of change in the base assist torque to the amount of fluctuation in the steering torque _{T h} .

The phase compensator 212 in the embodiment of the present disclosure adjusts the assist gain within a range of possible steering frequencies when the driver operates the steering wheel, thereby compensating for the stiffness of the torsion bar. In the embodiment of the present disclosure, an example of the predetermined range is 5 Hz or less. The phase compensator 212 may apply, for example, first-order phase compensation to the steering torque (torsion torque) when the steering frequency is 5 Hz or less. The first-order phase compensation is expressed, for example, by a transfer function in the mathematical expression (1).
Here, s is the Laplace transformer, _f1 is the frequency (Hz) of the zero point of the transfer function, and _f2 is the frequency (Hz) of the pole of the transfer function. A graph with gain (or loop gain) on the vertical axis and the logarithm of frequency on the horizontal axis is called a gain diagram. In a gain diagram, the zero point represents the intersection of the gain curve with the horizontal axis representing 0 dB, and the pole represents the maximum point of the gain curve. For example, phase lead compensation can be applied by making the pole frequency higher than the zero point. The larger the frequency interval, the greater the amount of phase lead.

The phase compensator 212 generates a target motor torque T _ref based on the base assist torque and base assist gain output from the base assist calculation unit 211. For example, the phase compensator 212 is a stabilization compensator and can apply stability phase compensation to the base assist torque. The phase compensator 212 can have a second-order or higher transfer function whose frequency characteristics vary depending on the base assist gain. The second-order or higher transfer function is expressed using a responsiveness parameter ω and a damping parameter ζ. The second-order or higher transfer function can be expressed by, for example, Equation 2. By making the order of the transfer function second-order, damping can be applied to the characteristics of the transfer function. Changing the damping makes it possible to adjust the phase characteristics.
where s is the Laplace transformer, _ω1 is the frequency of the zero, _ω2 is the frequency of the pole, _ζ1 is the damping of the zero, and _ζ2 is the damping of the pole. The pole frequency _ω2 is smaller than the zero frequency _ω1 .

The model following controller 230 is configured to estimate disturbance torque based on the motor angular velocity ω (MR angular velocity), which is the output from the controlled object 220, calculate the estimated disturbance torque, and feed it back to the input of the controlled object 220. The torque fed back from the model following controller 230 to the input of the controlled object 220 corresponds to a "correction torque" that corrects the input of the controlled object 220. The model following controller 230 generates this correction torque based on the output of the controlled object 220. An example of the model following controller 230 is a model following controller configured to perform model following control. The specific configuration of the model following controller 230 will be described in detail later.

The subtractor AD1 subtracts the estimated disturbance torque output from the model following controller 230 from the target motor torque T _ref . The output from the subtractor AD1 is input to the adder AD2 and the model following controller 230. The adder AD2 adds the disturbance torque T _d to the output from the subtractor AD1 and outputs the result to the control object 220. Here, examples of disturbances in the embodiment of the present disclosure include friction, torque ripple or rattle caused by mechanisms such as the motor and reduction gear, self-aligning torque, or disturbances that may occur when traveling on an unpaved, bumpy road or gravel road. Here, the self-aligning torque refers to the torque that acts in the direction of returning the steering wheel due to the elasticity of the tires that twist when the steering wheel is turned.

[First Implementation Example]
The model-following controller has an inverse plant model, a high-pass filter, and a low-pass filter (or Q filter). When the transfer functions of the low-pass filter and the high-pass filter are Q(s) and HPF(s), respectively, the model-following controller is configured so that the transfer function P(s) of the controlled object is constrained to a nominal model _Pn (s) in a frequency band where the gain in the gain characteristics of Q(s) and HPF(s) is 1. Note that in this specification, "the transfer function of the controlled object is constrained to the nominal model" means, for example, that the controlled object is controlled so that the transfer function of the controlled object appears to be the transfer function of the nominal model when the input-output relationship is viewed.

Figure 4 shows an example configuration of the model-following controller 230A in the first implementation example. The model-following controller 230A has a controlled object inverse model 231, a low-pass filter 232, a high-pass filter 233, and a subtractor SU1. The high-pass filter 233 has a first cutoff frequency, and the low-pass filter 232 has a second cutoff frequency.

The angular velocity ω of the motor is input to the controlled plant inverse model 231. The subtractor SU1 subtracts the output of the subtractor AD1 from the output of the controlled plant inverse model 231 to generate an estimated disturbance torque ^ _Td . The estimated disturbance torque ^ _Td is filtered by a series-connected low-pass filter 232 and a high-pass filter 233 in this order, and is input to the subtractor AD1. In this way, the model-following controller 230A feeds back the estimated disturbance torque ^ _Td to the input of the controlled plant 220. Note that "^ _Td " means _Td with a hat shown in FIGS. 4 and 12.

The model-following controller 230A performs model-following control, which refers to a feedback loop that uses the angular velocity ω of the motor, which is the controlled object 220, as the outer loop of current control. In the first implementation example, the feedback loop formed by the model-following controller 230A makes it possible to compensate for torque ripple that depends on the angular velocity ω. The angular velocity ω signal used for control can be corrected for each type of motor, making it possible to increase the accuracy of the angular velocity ω signal compared to current signals, etc. As a result, highly accurate torque ripple compensation can be applied to torque control.

The model-following controller 230A is similar in configuration to conventional disturbance estimators (or disturbance observers), but has different intended functions and effects. Conventional disturbance estimators estimate disturbance torque by selecting an inverse plant model close to the plant model, and reduce the effects of disturbance by adjusting the disturbance torque in advance. The frequency band that is the target of compensation is the low frequencies below 4 Hz that can occur due to vehicle behavior.

Model-following control according to an embodiment of the present disclosure utilizes the effect of a feedback loop that constrains the plant to a nominal model defined by an inverse plant model. The frequency band that can be compensated for is approximately 4 Hz to 150 Hz, which is different from the frequency band of conventional disturbance estimators. For example, if an inverse plant model is defined so that there is no torque ripple, model-following control constrains the plant model to characteristics that do not have torque ripple. As a result, torque ripple can be reduced by applying torque ripple compensation. Alternatively, by constructing an inertia or viscosity model and constraining the plant model to that model, it is possible to reduce the inertia or viscosity of the plant model. By performing model-following control, in addition to compensating for motor torque ripple, loss torque compensation or motor inertia compensation, for example, can be performed.

In this specification, the controlled object 220, the nominal model (or plant model) used to constrain the controlled object 220, the controlled object inverse model 231 defined by the inverse plant model of the plant model, the transfer function of the low-pass filter 232, and the transfer function of the high-pass filter 233 are referred to as P(s), P _n (s), P _n ^-1 (s), Q(s), and HPF(s), respectively.

The plant model (nominal model) is expressed by Expression 3, and the inverse plant model is expressed by Expression 4. By appropriately setting J _mn and B _mn , it is possible to impart desired frequency characteristics to P(s) of the controlled object 220. In this embodiment, the plant model (nominal model) is a model of a one-inertia system.

Let T(s) be the complementary sensitivity function of the inner loop formed by the model-following controller, and Δ(s) be the modeling error of the plant model. T(s) is expressed as Q(s)HPF(s), and the relationship shown in Equation 5 holds for Δ(s). The robust stability of the model-following controller is guaranteed when the small gain theorem shown in Equation 6 holds between T(s) and Δ(s). For disturbance suppression, it is sufficient for T(s) = 1, but in consideration of robust stability, Equation 6 must be satisfied. As can be seen from this, disturbance suppression and robust stability are incompatible.

Figure 5 shows an example of a gain diagram of the transfer function of the entire steering system. The horizontal axis of the gain diagram represents frequency (Hz), and the vertical axis represents gain (dB). In the first implementation example, to achieve disturbance suppression based on the frequency band, the frequency band is divided into Region I, where T(s) = 1 and disturbance suppression is required, and Region II, where T(s) is lowered to ensure robust stability. In Region II, 1/Δ(s) > T(s) holds.

The gain characteristics of the transfer function of the entire steering system have peaks, for example, near 20 Hz and 50 Hz, and the modeling error appears in the peak near 50 Hz of the two peaks. That is, Δ(s) has a peak near 50 Hz, and 1/Δ(s) shown in FIG. 5 has a bottom near 50 Hz. Adjustment methods for the gain characteristics include adjusting 1/Δ(s) and the break point of T(s). Adjustment of 1/Δ(s) is performed by adjusting J _mn and B _mn of the plant model, and adjustment of the break point of T(s) is performed by adjusting the second cutoff frequency of the low-pass filter 232. Furthermore, sensitivity to disturbances can be adjusted by the steering assist amount, steering speed, or vehicle speed. When the bottom frequency of the modeling error is close to the boundary frequency between Region I and Region II, a method is often adopted in which the order of the low-pass filter 232 is increased to sharply reduce T(s) in Region I, where disturbance suppression is required.

The control device 100 performs torque control for low-frequency torque signals and controls high-frequency disturbances so that the angular velocity ω ≈ 0, thereby stabilizing steering to prevent the steering wheel from being seized. To achieve this goal, the control device 100 uses the torque controller 210 to reduce the high-frequency gain of the torque control, and the model following controller 230A to constrain the control object P(s) to a characteristic that reduces the high-frequency gain. The reason for performing the latter process is to prevent the control object 220 from reacting to a disturbance such as Td shown in Figure 4 when it is input to the control object 220.

FIG. 6 illustrates a gain diagram of the transfer function C(s) of the phase compensator 212 in the torque controller 210. FIG. 7 illustrates a gain diagram of the transfer function HPF(s) of the high-pass filter 233. FIG. 8 illustrates a gain diagram of the nominal model P _n (s). In the gain diagram, the horizontal axis represents frequency (Hz) and the vertical axis represents gain (dB). For example, by applying the phase compensator 212 having the gain characteristics of the transfer function C(s) shown in FIG. 6, the high-frequency gain can be reduced in the gain characteristics of the nominal model P _n (s), as shown in FIG. 8. The cutoff frequency fc in the gain diagram of the transfer function C(s) is, for example, 2 Hz to 10 Hz, and the cutoff frequency fc in the gain diagram of P _n (s) is, for example, 2 Hz to 20 Hz.

The model-following controller 230A is configured so that the transfer function P(s) of the controlled object 220 is constrained to the nominal model P _n (s) in the frequency band where the gain in the gain characteristic of Q(s)·HPF(s) is 1. The inverse plant model P _n ^-1 (s) is designed to provide the inverse characteristic to be constrained and to utilize the gain characteristic of Q(s)·HPF(s). By appropriately designing the plant models J _mn and B _mn , the gain characteristic of the nominal model P _n (s) can be obtained, as shown in FIG. 8, in which the gain decreases in the high-frequency range. The boundary frequency between Region I and Region II (the lower limit of the frequency range defining Region I) is the maximum frequency that can be input by the driver, and is generally approximately 2 Hz to 10 Hz. This frequency depends on the first cutoff frequency of the high-pass filter 233. Therefore, the lower limit frequency of the effective range of the model following control is determined by adjusting the first cutoff frequency of the high-pass filter 233 so as not to impede the torque control.

The low-pass filter 232 and the high-pass filter 233 are connected in series. The low-pass filter 232 may be composed of multiple LPFs. That is, Q(s) may be expressed as the transfer function of an n-stage LPF (n is 1 or greater). The second cutoff frequency is greater than the first cutoff frequency. The first cutoff frequency is, for example, 2 Hz or greater and 10 Hz or less, and preferably, for example, 5 Hz or greater and 7 Hz or less. The second cutoff frequency is, for example, 3 Hz or greater and preferably 50 Hz or less. However, the upper limit of the second cutoff frequency may be set to approximately 140 Hz to 200 Hz. The cutoff frequency fc of the gain characteristic of the nominal model P _n (s) shown in FIG. 8 depends on the first cutoff frequency and the second cutoff frequency and is, for example, 2 Hz or greater and 20 Hz or less.

The inventors have confirmed the effects of applying model following control according to an embodiment of the present disclosure by conducting measurements on an actual vehicle. In the measurements on the actual vehicle, they measured the effect of reducing torque ripple and steering pull by applying model following control to torque control. Here, steering pull refers to the steering wheel swaying from side to side when going over a bump with the hands off the wheel.

Figure 9 shows the measurement results of steering angle and torsion torque when model following control is not applied. Figure 10 shows the measurement results of steering angle and torsion torque when model following control is applied. In the graph, the horizontal axis represents steering angle (deg) and the vertical axis represents torsion torque (Nm). The graph shows the waveform measured when steering from end to end (from turning the steering wheel fully left to fully right, or vice versa) at an angular velocity of 180 (deg/s).

Zooming in on the area enclosed by the dashed line on the graph shows that torque ripple is suppressed when model following control is applied compared to when model following control is not applied. Specifically, it was found that the amount of fluctuation in torsion torque was reduced by approximately 0.1 [Nm].

Figure 11 shows the measurement results of the change in steering angle over time when model following control is not applied and when model following control is applied. In the graph, the horizontal axis represents time [sec] and the vertical axis represents steering angle [deg]. The dashed rectangle indicates the time region where the vehicle went over the step. It was found that by applying model following control to torque control, fluctuations in steering angle when the vehicle went over the step were suppressed, and steering pull was appropriately reduced.

In the first implementation example, applying model-following control to torque control makes it possible to reduce high-frequency components of disturbances. As a result, it is possible to appropriately reduce torque ripple that can occur when steering and steering wheel pull that can occur when the vehicle goes over a bump.

[Second Implementation Example]
Next, a model following controller according to a second implementation example will be described with reference to Figures 12 to 14. The model following controller according to the second implementation example differs from the model following controller according to the first implementation example in that it includes a friction compensation calculator. Below, the differences from the model following controller according to the first implementation example will be mainly described.

Since the disturbances estimated by the model-following controller include mechanical friction such as that of the motor and reduction gear, the model-following controller in the second implementation example is configured to extract the friction component from the estimated disturbance torque and apply friction compensation to the estimated disturbance torque. Friction compensation is performed on, for example, motor friction, reduction gear friction, or the difference in friction between the left and right sides of the reduction gear.

When applying conventional friction compensation control, when the motor's angular velocity ω is near zero, it is necessary to gradually change the friction compensation torque (Nm) relative to the motor's angular velocity ω to prevent chattering, which can result in high-precision friction compensation control being impossible. According to the inventor's research, the best way to solve this problem is to sequentially estimate and compensate for friction.

A model-following controller according to a second implementation example is configured to feed back disturbance compensation torque to the input of the controlled object. Specifically, the model-following controller has a high-pass filter that removes low-frequency components from the estimated disturbance torque, a friction compensation calculator connected in parallel to the high-pass filter that applies friction compensation to the estimated disturbance torque to calculate an estimate of the mechanical friction torque, and an adder that adds the estimated friction torque to the estimated disturbance torque from which the low-frequency components have been removed by the high-pass filter to generate disturbance compensation torque. In the second implementation example, the estimated disturbance torque corresponds to the "first correction torque" and the disturbance compensation torque corresponds to the "second correction torque."

Figure 12 shows an example configuration of the model following controller 230B in the second implementation example. The model following controller 230B is configured to perform model following control in the same way as the model following controller 230A in the first implementation example. However, the model following control function is not required.

The model following controller 230B has a friction compensation calculator 250. The friction compensation calculator 250 is connected in parallel to the high-pass filter 233 and calculates an estimated value of the friction torque of the mechanism by applying friction compensation to the estimated disturbance torque ^ _Td . The friction compensation calculator 250 has a subtractor 251, a limiter 252, and a gain adjuster 253. The subtractor 251 subtracts the output value from the high-pass filter 233 from the output value from the low-pass filter 232. The limiter 252 imposes a limit on the output value from the subtractor 251. If the input value exceeds the upper or lower threshold, the limiter 252 clips the input value to the upper or lower threshold.

Gain adjuster 253 multiplies the output value from limiter 252 by gain K. The maximum value of gain K of gain adjuster 253 is determined under conditions in which the transfer function of controlled object 220 is constrained to the nominal model. The maximum value of gain K is set, for example, to approximately 1 to 1.2.

The estimated disturbance torque ^ _Td includes mechanical friction. In estimating the disturbance, first, friction is estimated from the transmission path of the motor's output torque, and then torque acting on the motor, such as self-aligning torque, is estimated. Therefore, the friction compensation calculator 250 calculates a value equivalent to the friction torque from the initially estimated disturbance as the estimated value of the friction torque. Generally, an appropriate amount of friction is required for EPS, so by setting the estimated value of the friction torque to a value smaller than the actual friction force, it is possible to achieve highly accurate friction compensation while maintaining an appropriate amount of friction force.

In order to apply friction compensation to the estimated disturbance torque used in model-following control, it is necessary to pay attention to the stability conditions of model-following control. According to the small gain theorem described above, this condition requires that the gain in the gain characteristic of the transfer function of the friction compensation calculator 250, which is constrained to characteristics that take stability into consideration, does not exceed 1. This is derived from the design conditions of the low-pass filter 232. In the second implementation example, the friction compensation gain, i.e., the value of gain K, is set to a maximum of 1 to always satisfy this condition, and a subtractor 251 is provided before the limiter 252 to apply subtraction processing so that the gain in the gain characteristic becomes 1 under this condition. In other words, the friction compensation calculator 250 behaves as a low-pass filter with a transfer function of 1-HPF(s).

The estimated disturbance torque ^ _Td includes a low-frequency component ^ _Td1 , a medium-frequency component ^ _Td2 , and a high-frequency component ^ _Td3 . The low-pass filter 232 removes the high-frequency component ^ _Td3 from the estimated disturbance torque ^ _Td , and the high-pass filter 233 further removes the low-frequency component ^ _Td1 from the estimated disturbance torque ^ _Td . In this way, only the medium-frequency component ^ _Td2 of the estimated disturbance torque, which is in the range from the first cutoff frequency of the high-pass filter 233 to the second cutoff frequency of the low-pass filter 232, is subject to friction compensation. However, because the assumed friction included in the disturbance is a low-frequency component of the disturbance, the above filter processing results in the low-frequency component ^ _Td1 being excluded from friction compensation. Therefore, by connecting the friction compensation calculator 250 in parallel with the high-pass filter 233, the low-frequency component ^ _Td1 of the disturbance, which has been filtered by the high-pass filter 233 and is no longer compensated for, is added again to the estimated disturbance torque ^ _Td , thereby achieving friction compensation. More specifically, the friction compensation calculator 250 multiplies the low-frequency component ^ _Td1 by a gain K and adds the result to the medium-frequency component ^ _Td2 , thereby generating the disturbance compensation torque. Note that "^ _Td1 " means _Td1 with a hat shown in FIG. 12, "^ _Td2 " means _Td2 with a hat shown in FIG. 12, and "^ _Td3 " means _Td3 with a hat shown in FIG. 12.

Vehicles equipped with EPS can be driven in a driving mode that includes an automatic driving mode and a manual driving mode. In this case, the gain K of the gain adjuster 253 may be switched depending on the driving mode. The larger the gain K, the greater the degree of friction reduction. The gain K set in the automatic driving mode is preferably larger than the gain K set in the manual driving mode. This allows for optimal friction compensation to be applied in the automatic driving mode, where greater friction reduction is required.

The model following controller 230B further includes an adder AD3. The adder AD3 adds the output value from the gain adjuster to the output value from the high-pass filter 233. The output from the adder AD3 is fed back to the input of the controlled object 220 as disturbance compensation torque.

Assistance devices have been developed that recognize lane marks, such as white or yellow lines, when driving on expressways, and assist the vehicle in autonomous driving by following the lane. It is known that in vehicles equipped with EPS and an assistance device, if there is a difference in friction between the left and right reduction gears, this can affect the control of the assistance device, which drives the vehicle straight along the center of the lane. Friction compensation control according to an embodiment of the present disclosure can solve the above problem by sequentially calculating an estimated value of friction torque even when there is a difference in friction between the left and right reduction gears. The motor angular velocity ω, which is the output of the plant model, contains information regarding the difference in friction between the left and right reduction gears.

The inventors conducted simulations to confirm the effects of applying friction compensation control through gain adjustment. The simulations measured the friction reduction effect of friction compensation control.

Figure 13 shows the simulation results of steering angle and steering torque when friction compensation control by gain adjustment is not applied and when friction compensation control by gain adjustment is applied. In the graph, the horizontal axis represents steering angle [deg], and the vertical axis represents steering torque [Nm]. The dashed line represents the waveform when friction compensation control is not applied, and the solid line represents the waveform when friction compensation control is applied. The arrows in the figure represent the range of steering torque, which corresponds to the magnitude of friction. It was found that friction can be appropriately reduced by applying friction compensation control.

Figure 14 shows simulation results for steering angle and steering torque when conventional friction compensation control and friction compensation control with gain adjustment are applied. In the graph, the horizontal axis represents steering angle [deg], and the vertical axis represents steering torque [Nm]. The dashed line represents the waveform when conventional friction compensation control is applied, while the solid line represents the waveform when friction compensation control with gain adjustment is applied. With conventional friction compensation control, as described above, when the motor angular velocity ω is near zero, the change in friction compensation torque (Nm) relative to the motor angular velocity ω must be made gradual to prevent chattering. As a result, a spike in steering torque was observed when the steering wheel was turned (see the area circled by dashed lines in the figure). In contrast, when friction compensation control with gain adjustment was applied, no spike was observed, demonstrating that friction was appropriately reduced.

In the second implementation example, by further applying friction compensation control using gain adjustment to torque control, it is possible to appropriately reduce friction while reducing high-frequency components of disturbances.

[Third Implementation Example]
In the third implementation example, the control object includes a steering wheel 521, universal joints 523A and 523B, a rotating shaft 524, a torsion bar 546, a motor 543, and a reduction gear 544. Because the control object in the third implementation example includes parts that can rotate relatively to each other via the torsion bar 546, the motion of the control object cannot be described by a simple equation of motion for a one-inertia system. The control object in the third implementation example changes between a one-inertia system and a two-inertia system depending on the strength with which the driver of the vehicle grips the steering wheel 521. The stronger the driver grips the steering wheel 521, the closer the control object becomes to a one-inertia system. The weaker the driver grips the steering wheel, the closer the control object becomes to a two-inertia system. In the third implementation example, an angular velocity corresponding to the angular velocity of the reduction gear 544 is input to the model-following controller as the output of the control object.

In the third implementation example, the plant model (nominal model) is a model having frequency characteristics between those of a one-inertia system and a two-inertia system. The transfer function P _n (s) of the plant model (nominal model) in the third implementation example is expressed by Expression 7, and the transfer function P _n ⁻¹ (s) of the inverse plant model is expressed by Expression 8.

In Equation 7 and Equation 8, s is a Laplace transformer, J _STGn is a parameter representing the moment of inertia of the nominal model, B _STGn is a parameter representing the viscous friction coefficient of the nominal model, ω _1n is the frequency of the zero point of the transfer function P _n (s), ω _2n is the frequency of the pole of the transfer function P _n (s), ζ _1n is the damping ratio at the zero point of the transfer function P _n (s), and ζ _2n is the damping ratio at the pole of the transfer function P _n (s).

In the third implementation example, the nominal model is a model having frequency characteristics between those of a one-inertia system and a two-inertia system. The formula (7) representing the transfer function P _n (s) of the above nominal model is an equation obtained by adding a damping term to the equation representing the two-inertia system. In the formula (7), the damping terms are 2ζ _1n ω _1n s and 2ζ _2n ω _2n s. The equation obtained by removing these damping terms from the formula (7) becomes the equation representing the two-inertia system. In the third implementation example, the order of the transfer function P _n (s) of the nominal model is 3.

In the third implementation example, the nominal model is a model that takes into account mechanical characteristics when a driver (steerer) steers the steering wheel 521. The more tightly the driver grips the steering wheel 521, the closer the controlled object approaches a one-inertia system; and the more loosely the driver grips the steering wheel 521, the closer the controlled object approaches a two-inertia system. Therefore, the transfer function P(s) of the controlled object in the third implementation example changes between a one-inertia system and a two-inertia system depending on how the force is applied to the steering wheel 521 from the driver's arms. In the third implementation example, by using a nominal model that has frequency characteristics between a one-inertia system and a two-inertia system, the modeling error Δ(s) between the transfer function P _n (s) of the nominal model and the transfer function P(s) of the controlled object can be prevented from becoming too large, regardless of whether the state of the controlled object is in a one-inertia system or a two-inertia system. Therefore, the controlled object can be suitably controlled using the nominal model regardless of how the driver steers the steering wheel 521. As described above, in the third implementation example, the nominal model is a model that takes into account the mechanical characteristics that are imparted to the controlled object depending on how the driver grips the steering wheel 521. The control device 100 in the third implementation example has such a nominal model as an internal model, thereby enabling suitable control of the controlled object. Other configurations in the third implementation example can be similar to those of the other implementation examples described above.

Note that while the control device in each of the implementation examples described above is configured with a torque controller that provides input to the controlled object, which is a motor, this is not limited to this. The control device according to the present disclosure does not necessarily have to include a torque controller.

Embodiments of the present disclosure can be used in a motor control device for controlling an EPS mounted on a vehicle.

100: Control unit (ECU), 200: Processor, 210: Torque controller, 211: Response phase compensation unit, 212: Phase compensator, 220: Control target, 230, 230A, 230B: Model following controller, 231: Control target inverse model, 232: Low-pass filter, 233: High-pass filter, 250: Friction compensation calculator, 251: Subtractor, 252: Limiter, 253: Gain adjuster, 1000: Electric power steering device, AD1: Subtractor, AD2, AD3: Adders, SU1: Subtractor

Claims (6)

A control device for controlling an electric power steering device including a motor, the control device comprising:
a model following controller that generates a correction torque based on an output from a controlled object that is the motor, and corrects a target motor torque that is an input to the controlled object with the correction torque;
The model following controller
a high-pass filter having a first cutoff frequency and a low-pass filter having a second cutoff frequency greater than the first cutoff frequency ;
an inverse model of a controlled object to which the output of the controlled object is input;
Including,
the control object inverse model is an inverse model of a nominal model, which is a plant model,
the process of calculating the correction torque by the model following controller includes subtracting the target motor torque corrected by the correction torque from an output of the controlled object inverse model, and subjecting the subtracted value to filtering by the low-pass filter and filtering by the high-pass filter;
the model following controller is configured such that, when transfer functions of the low-pass filter and the high-pass filter are Q(s) and HPF(s), respectively, an output of the controlled object becomes equal to an output of the nominal model when the target motor torque is input to the nominal model in a frequency band where a gain in a gain characteristic of Q(s)·HPF(s) is 1 ;
The control device , wherein the nominal model includes a parameter representing a moment of inertia of the nominal model and a parameter representing a viscous friction coefficient of the nominal model . The control device described in claim 1, wherein the first cutoff frequency is greater than or equal to 2 Hz and less than or equal to 10 Hz. The control device described in claim 2, wherein the second cutoff frequency is 3 Hz or higher. A motor;
The control device according to any one of claims 1 to 3;
A motor module comprising: An electric power steering device equipped with the motor module described in claim 4. 1. A computer-implemented method for controlling a motor of an electric power steering device including a motor, comprising:
a model following controller including a high-pass filter having a first cutoff frequency , a low-pass filter having a second cutoff frequency greater than the first cutoff frequency , and a controlled object inverse model which is an inverse model of a nominal model which is a plant model and to which an output of a controlled object which is the motor is input , wherein when transfer functions of the low-pass filter and the high-pass filter are Q(s) and HPF(s), respectively, the model following controller is configured so that the output of the controlled object becomes equal to the output of the nominal model when a target motor torque is input to the nominal model in a frequency band where a gain in a gain characteristic of Q(s)·HPF(s) is 1; generating a correction torque based on the output from the controlled object and correcting the target motor torque which is an input of the controlled object with the correction torque ;
performing a process of calculating the correction torque by the model following controller, the process including subtracting the target motor torque after being corrected by the correction torque from the output of the controlled object inverse model, and filtering the subtracted value using the low-pass filter and the high-pass filter ;
The method , wherein the nominal model includes a parameter representing a moment of inertia of the nominal model and a parameter representing a viscous friction coefficient of the nominal model .