Vehicular control device

Abstract

To make it possible to more appropriately suppress vibration.SOLUTION: A vehicular control device is used for a vehicle including a motor for outputting power to a drive shaft connected to an axle and controls the motor. The vehicular control device uses: a machine learning model which has, as explanatory variables, torque of the motor, rotation speeds of the motor, and wheel speeds at the present time and during a period from the present time to a first time ago, and which has, as an objective variable, future shaft torque of the drive shaft or a future reflection value reflecting torque of the drive shaft after a second time from the present time; and the torque of the motor, the rotation speeds of the motor, and the wheel speeds at the present time and in the past from the present time to the first time ago. Thereby, the vehicular control device estimates the future shaft torque or the future reflection value, and executes vibration damping control to suppress vibration of the motor on the basis of the future shaft torque or the future reflection value which has been estimated.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a vehicle control device. [Background technology]

[0002] Conventionally, a vehicle control device of this type has been proposed for use in a vehicle equipped with a motor that outputs power to a drive shaft connected to an axle (see, for example, Patent Document 1). This device calculates compensation torque using a detection value of the motor's rotation speed and a model that takes into account a predetermined torsional rigidity of the drive system and a friction coefficient between the wheel and the road surface, and controls the motor so that a torque command for the motor minus the compensation torque is output. This is said to perform vibration suppression control with high precision. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2022-58067 Summary of the Invention [Problem to be solved by the invention]

[0004] Incidentally, as a method for suppressing vibration, a method has been proposed in which vibration is determined to have occurred when an integrated value of the fluctuation in the rotation speed of the motor is equal to or greater than a judgment threshold, and vibration suppression control is executed to suppress the vibration. In order to further reduce the vibration, it is necessary to quickly determine the occurrence of vibration and execute vibration suppression control at the initial stage when the vibration is small. As a method for quickly determining the occurrence of vibration, there is a method for lowering the judgment threshold. However, if the judgment threshold is lowered, there are more opportunities to execute vibration suppression control even on road surfaces where vibration suppression control is not required, which may result in a decrease in drivability. Therefore, it is desired to more accurately determine the occurrence of vibration and execute vibration suppression control to more appropriately suppress the vibration.

[0005] The main object of the vehicle control device disclosed herein is to more appropriately suppress vibrations. [Means for solving the problem]

[0006] The vehicle control device of the present disclosure employs the following means to achieve the above-mentioned main object.

[0007] The vehicle control device of the present disclosure includes: A vehicle control device for use in a vehicle having a motor that outputs power to a drive shaft connected to an axle, the vehicle control device controlling the motor, A machine learning model is used in which the torque of the motor, the rotation speed of the motor, and the wheel speed from the present time and from the present time until the first hour are explanatory variables, and a future shaft torque as the torque of the drive shaft from the present time until the second hour or a future reflection value reflecting the future shaft torque is used as a target variable, and the future shaft torque or the future reflection value is estimated using the torque of the motor, the rotation speed of the motor, and the wheel speed from the present time and from the present time until the first hour, and vibration suppression control is performed to suppress vibration of the motor based on the estimated future shaft torque or the future reflection value. The gist of the present invention is as follows.

[0008] In the vehicle control device disclosed herein, a machine learning model is used in which the torque of the motor at present and from the present until the first hour ago are used as explanatory variables, and a future torque as the torque of the drive shaft at the second hour ago from the present or a future reflected value reflecting the future torque is used as a target variable, and a future torque or a future reflected value is estimated using the torque of the motor at present and from the present until the first hour ago, and vibration suppression control is performed to suppress vibration of the motor based on the estimated future torque or future reflected value. Since the future torque or the future reflected value is estimated using the machine learning model and the torque of the motor at present and from the present until the first hour ago, the future torque or the future reflected value can be estimated more accurately. Since vibration suppression control is performed to suppress vibration of the motor based on the accurately estimated future torque or the future reflected value, vibration suppression control with higher accuracy can be performed. As a result, vibration can be suppressed more appropriately. Here, the "first time" can be a time equivalent to a half cycle to one cycle of the expected vibration fluctuation of the drive shaft. The "second time period" may be, for example, a time period corresponding to 1 to 1.5 periods of the expected vibration fluctuation of the drive shaft. [Brief description of the drawings]

[0009] [Figure 1] 1 is a schematic configuration diagram of an electric vehicle 20 according to an embodiment of the present disclosure. [Diagram 2] 4 is a flowchart showing an example of a control routine. [Diagram 3] FIG. 4 is an explanatory diagram for explaining how a future shaft torque Tes is estimated. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] An embodiment of the present disclosure will be described with reference to the drawings. Fig. 1 is a schematic configuration diagram of an electric vehicle 20 according to an embodiment of the present disclosure. As shown in the figure, the electric vehicle 20 according to the embodiment includes a motor 32, an inverter 34, a battery 36, and an electronic control unit (hereinafter referred to as "ECU") 50.

[0011] The motor 32 is configured as a three-phase AC motor, and includes a rotor in which a permanent magnet is embedded in a rotor core, and a stator in which a three-phase coil is wound around a stator core. The rotor of the motor 32 is connected to a drive shaft 26 that is connected to the drive wheels 22a, 22b via a differential gear 24. The inverter 34 is used to drive the motor 32, and is connected to a power line 38 together with a battery 36. The inverter 34 includes six transistors T11-T16 as switching elements, and six diodes D11-D16 that are connected in parallel to the six transistors T11-T16, respectively. The transistors T11-T16 are arranged in pairs, two at a time, so as to be on the source side and the sink side with respect to the positive and negative pole lines of the power line 38, respectively. Each of the connection points of the paired transistors of the transistors T11-T16 is connected to each of the three-phase (U-phase, V-phase, W-phase) coils of the motor 32. Therefore, when a voltage is applied to the inverter 34, the ECU 50 adjusts the ratio of the on-time of the paired transistors T11 to T16, thereby forming a rotating magnetic field in the three-phase coil and rotating the motor 32. A smoothing capacitor 39 is attached to the power line 38. The battery 36 is configured as, for example, a lithium ion secondary battery or a nickel-metal hydride secondary battery having a rated voltage of several hundred volts, and is connected to the power line 38 together with the inverter 34 as described above.

[0012] The ECU 50 includes a microcomputer, and the microcomputer has a CPU, a ROM, a RAM, a flash memory, an input / output port, and a communication port. The ECU 50 inputs signals from various sensors via the input ports. For example, the ECU 50 inputs a rotational position θm from a rotational position sensor (e.g., a resolver) 32a that detects the rotational position of the rotor of the motor 32, and phase currents Iv and Iw from current sensors 32v and 32w that detect the V-phase and W-phase currents of the motor 32. The ECU 50 also inputs a voltage Vb from a voltage sensor 36a attached between the terminals of the battery 36, a current Ib from a current sensor 36b attached to the output terminal of the battery 36, and a voltage VH of the capacitor 39 (the voltage of the power line 38) from a voltage sensor 39a attached between the terminals of the capacitor 39. The ECU 50 also receives a start signal from a start switch 60, a shift position SP from a shift sensor 62 that detects the operation position of a shift lever 61, an accelerator opening Acc from an accelerator pedal position sensor 64 that detects the depression amount of an accelerator pedal 63, a brake pedal position BP from a brake pedal position sensor 66 that detects the depression amount of a brake pedal 65, and a vehicle speed V from a vehicle speed sensor 67. The ECU 50 also receives a longitudinal acceleration A from an acceleration sensor 68 that detects the longitudinal acceleration of the vehicle, and wheel speeds Vwa to Vwd from wheel speed sensors 69a to 69d that are attached to the four wheels on the front, rear, left and right and detect the wheel speeds. The ECU 50 outputs various control signals, such as control signals to the transistors T11 to T16 of the inverter 34, via an output port. The ECU 50 calculates the electrical angle θe and the rotation speed Nm of the motor 32 based on the rotation position θm of the rotor of the motor 32 from the rotation position sensor 32a, and calculates a rotation speed change rate ΔNm, which is the amount of change in the rotation speed Nm per unit time, based on the rotation speed Nm. The ECU 50 calculates the motor torque Tm, which is the torque output from the motor 32, based on the phase currents Iv, Iw from the current sensors 32v, 32w, and calculates the torque change rate ΔTm, which is the change in the motor torque Tm per unit time, based on the motor torque Tm.The ECU 50 calculates the wheel speed change rates ΔVwa-ΔVwd, which are the amounts of change in the wheel speeds Vwa-Vwd per unit time, based on the wheel speeds Vwa-Vwd from the wheel speed sensors 69a-69d. The ECU 50 inputs and calculates the above-mentioned data every time titr (for example, every few msec). The ECU 50 stores the wheel speeds Vwa-Vwd detected by the wheel speed sensors 69a-69d during a data collection period P1 from the present time until a first time t1 ago, and the motor torque Tm and rotation speed Nm calculated during the data collection period P1, as past wheel speeds Vwapre-Vwdpre, past torque Tmpre, and past rotation speed Nmpre in a RAM or flash memory. The first time t1 is set to a time equivalent to a half cycle to one cycle of the expected vibration fluctuation of the drive shaft 26, for example, 50 msec, 75 msec, 100 msec, etc. During the data collection period P1, the past wheel speeds Vwapre-Vwdpre, the past torque Tmpre, and the past rotation speed Nmpre may be detected and calculated multiple times, so the RAM or flash memory may store multiple values ​​of each of the past wheel speeds Vwapre-Vwdpre, the past torque Tmpre, and the past rotation speed Nmpre. Note that data from the RAM or flash memory prior to the data collection period P1 may be erased.

[0013] In the electric vehicle 20 of this embodiment configured as described above, the ECU 50 sets a required torque Td* required for the drive shaft 26 based on the accelerator opening Acc and the vehicle speed V, sets the set required torque Td* as a torque command Tm* for the motor 32, and performs switching control of the transistors T11-T16 of the inverter 34 so that the motor 32 is driven by the torque command Tm*. In this embodiment, this control is sometimes referred to as "normal control".

[0014] Next, a description will be given of the operation of the electric vehicle 20 of this embodiment configured as described above, in particular the operation when suppressing vibrations. Fig. 2 is a flowchart showing an example of a control routine executed by the ECU 50. This routine is executed at every time itr.

[0015] When this routine is executed, the CPU of the ECU 50 executes a process of inputting input values ​​necessary for control, such as the motor torque Tm, the rotation speed Nm, the longitudinal acceleration A, the wheel speed Vwa to Vwd, the torque change rate ΔTm, the rotation speed change rate ΔNm, the wheel speed change rate ΔVwa to ΔVwd, the past torque Tmpre, the past rotation speed Nmpre, and the past wheel speed Vwapre to Vwbpre (S100). The longitudinal acceleration A and the wheel speed Vwa to Vwd are inputted with values ​​detected by the acceleration sensor 68 and the wheel speed sensors 69a to 69d, respectively. The motor torque Tm and the rotation speed Nm are inputted with values ​​calculated based on the phase currents Iv and Iw from the current sensors 32v and 32w and the rotation position θm of the rotor of the motor 32 from the rotation position sensor 32a, respectively. The torque change rate ΔTm, the rotation speed change rate ΔNm, and the wheel speed change rate ΔVwa to ΔVwd are inputted with values ​​calculated based on the motor torque Tm, the rotation speed Nm, and the wheel speed Vwa to Vwd. The past torque Tmpre, past rotation speed Nmpre, and past wheel speeds Vwapre to Vwbpre are input from the RAM or flash memory.

[0016] When an input value required for control is input in S100, a future shaft torque Tes to be output to the drive shaft 26 after a second time t2 from the present is estimated using the input value and a machine learning model (S110). The second time t2 can be, for example, a time corresponding to one to 1.5 periods of the expected vibration fluctuation of the drive shaft 26. FIG. 3 is an explanatory diagram for explaining how the future shaft torque Tes is estimated. Time tnow indicates the current time. A black circle indicates an example of the rotation speed Nm and the motor torque Tm among the input values ​​input in S100. A black star indicates an example of the future shaft torque Tes. The machine learning model is a model obtained by machine learning with the motor torque Tm, the rotation speed Nm, the longitudinal acceleration A, the wheel speed Vwa to Vwd, the torque change rate ΔTm, the rotation speed change rate ΔNm, the wheel speed change rate ΔVwa to ΔVwd, the past torque Tmpre, the past rotation speed Nmpre, and the past wheel speed Vwapre to Vwbpre as explanatory variables, and the future shaft torque Tes as an objective function. As the machine learning algorithm, it is recommended to use nonlinear regression such as Random Forest or Light GBM (Light Gradient Boosting Machine). Machine learning is executed in parallel with this routine. This type of machine learning makes it possible to estimate the future shaft torque Tes with greater accuracy.

[0017] After estimating the future shaft torque Tes, it is determined whether the future shaft torque Tes is equal to or greater than a judgment threshold Tref (S120). The judgment threshold Tref is a threshold for determining whether the road surface is undulating and vibration is large, and is a value obtained by adding a torque that is determined in advance as a torque at which vibration can be determined to be large to the above-mentioned required torque Td*.

[0018] When the future axle torque Tes is equal to or greater than the judgment threshold value Tref in S120, it is judged that the road surface is wavy and vibration is large, and the wavy road judgment flag Fb is set to a value of 1 (S130). The wavy road judgment flag Fb is initially set to a value of 0. In this process, a more accurately estimated future axle torque Tes is used to judge whether the road surface is wavy and vibration is large, so that it is possible to more appropriately judge whether vibration is large.

[0019] Then, the vibration suppression control for suppressing the vibration of the motor 32 is executed (S140), and this routine is terminated. In the vibration suppression control, a torque obtained by subtracting a predetermined torque ΔTd from the above-mentioned required torque Td* is set as a torque command Tm*, and switching control of the transistors T11 to T16 of the inverter 34 is performed so that the motor 32 is driven by the torque command Tm*. The predetermined torque ΔTd is a value determined in advance by experiments, analysis, machine learning, etc. as a torque for suppressing vibration by reducing the torque output from the motor 32 to the drive shaft 26. The vibration suppression control is a control for lowering the torque of the motor 32 compared to normal control. By executing the vibration suppression control, the absolute value of the shaft torque Td as the torque output to the drive shaft 26 can be reduced, and the torque fluctuation of the drive shaft 26 can be reduced, thereby suppressing the vibration of the vehicle. In this embodiment, the vibration suppression control for suppressing the vibration of the motor 32 is executed based on the future shaft torque Tes estimated with high accuracy, so that the vibration suppression control with higher accuracy can be executed. This allows the vibration to be suppressed more appropriately.

[0020] When the future axle torque Tes is less than the judgment threshold value Tref in S120, it is judged whether or not the undulating road judgment flag Fb is set to the value 1 (S150). When the undulating road judgment flag Fb is set to the value 1, vibration suppression control is executed, so S150 is a process for judging whether or not vibration suppression control is currently being executed. When the undulating road judgment flag Fb is set to the value 0, vibration suppression control is not currently being executed, so normal control is executed (S180) and this routine is terminated.

[0021] When the undulating road determination flag Fb is set to 1 in S150, i.e., when vibration suppression control is currently being executed, it is determined whether the state in which the future axle torque Tes is less than the determination threshold value Tref continues for a predetermined time ts (S160). The predetermined time ts is a threshold for determining whether the future axle torque Tes is definitely less than the determination threshold value Tref. If the state in which the future axle torque Tes is less than the determination threshold value Tref has not continued for the predetermined time ts, it is determined that the future axle torque Tes is not definitely less than the determination threshold value Tref, the undulating road determination flag Fb is set to 1 (S130), vibration suppression control is executed (S140), and this routine ends.

[0022] When the undulating road determination flag Fb is set to 1 in S150 and the state in which the future axle torque Tes is less than the determination threshold value Tref continues for a predetermined time ts in S160, it is determined that the future axle torque Tes is definitely less than the determination threshold value Tref and the torque fluctuation of the drive shaft 26 will not become large even without executing vibration suppression control, the undulating road determination flag Fb is set to 0 (S170), normal control is executed (S180), and this routine is terminated. When the torque fluctuation of the drive shaft 26 becomes sufficiently small due to the vibration suppression control by such processing, it is possible to switch from the vibration suppression control to the normal control.

[0023] According to the electric vehicle 20 of the embodiment described above, a machine learning model is used in which the current motor torque Tm, rotation speed Nm, longitudinal acceleration A, wheel speed Vwa-Vwd, torque change rate ΔTm, rotation speed change rate ΔNm, wheel speed change rate ΔVwa-ΔVwd, and past torque Tmpre, past rotation speed Nmpre, and past wheel speed Vwapre-Vwbpre from the present to a first time t1 are used as explanatory variables, and future axle torque Tes as the torque of the drive shaft 26 from the present to the second time after is used as the objective variable, and the future axle torque Tes is estimated using the current motor torque Tm, rotation speed Nm, longitudinal acceleration A, wheel speed Vwa-Vwd, torque change rate ΔTm, rotation speed change rate ΔNm, wheel speed change rate ΔVwa-ΔVwd, past torque Tmpre, past rotation speed Nmpre, and past wheel speed Vwapre-Vwbpre, and vibration suppression control is executed to suppress vibration of the motor 32 based on the estimated future axle torque Tes, so that vibration can be suppressed more appropriately.

[0024] In the above embodiment, the current motor torque Tm, rotation speed Nm, longitudinal acceleration A, wheel speed Vwa-Vwd, torque change rate ΔTm, rotation speed change rate ΔNm, wheel speed change rate ΔVwa-ΔVwd, past torque Tmpre, past rotation speed Nmpre, and past wheel speed Vwapre-Vwbpre are used as explanatory variables. However, it is sufficient to use at least the motor torque Tm, rotation speed Nm, wheel speed Vwa-Vwd, past torque Tmpre, past rotation speed Nmpre, and past wheel speed Vwapre-Vwbpre as explanatory variables, and the longitudinal acceleration A, torque change rate ΔTm, rotation speed change rate ΔNm, and wheel speed change rate ΔVwa-ΔVwd are not necessarily used as explanatory variables.

[0025] In the above embodiment, the future shaft torque Tes is estimated. However, when the drive shaft 26 is connected to another shaft such as a gear mechanism, a reflection value (future reflection value) that reflects the torque of the drive shaft 26 two hours after the present, such as the torque output to the other shaft, may be estimated instead of the future shaft torque Tes.

[0026] In the above-described embodiment, the electric vehicle 20 is provided with the motor 32 and the inverter 34. However, the electric vehicle 20 may be provided with a hybrid vehicle that includes an engine in addition to the motor 32 and the inverter 34, or a fuel cell vehicle that includes a fuel cell in addition to the motor 32 and the inverter 34.

[0027] The above describes embodiments for implementing the present disclosure, but the present disclosure is not limited to these embodiments, and it goes without saying that the present disclosure can be implemented in various forms without departing from the gist of the present disclosure. [Industrial Applicability]

[0028] The present disclosure is applicable to the vehicle control device manufacturing industry and the like. [Explanation of symbols]

[0029] 20 electric vehicle, 22a, 22b drive wheels, 32 motor, 34 inverter, 50 ECU, T11~T16 transistors.

Claims

[Claim 1] A vehicle control device for use in a vehicle having a motor that outputs power to a drive shaft connected to an axle, the vehicle control device controlling the motor, A machine learning model in which the torque of the motor at present and from the present until a first hour ago, the rotation speed of the motor, and the wheel speed are used as explanatory variables, and a future shaft torque as the torque of the drive shaft from the present until a second hour ago or a future reflection value reflecting the future shaft torque is used as a target variable, and the torque of the motor at present and from the present until the first hour ago, the rotation speed of the motor, and the wheel speed are used to estimate the future shaft torque or the future reflection value, and vibration suppression control is performed to suppress vibration of the motor based on the estimated future shaft torque or the future reflection value. Vehicle control device.

Images