Control device for vehicle
The vehicle control device uses machine learning on rotational speed data to predict torque and output opposing torque, addressing complexity and stability issues in existing systems by simplifying configuration and maintaining vehicle stability.
Patent Information
- Application Number
- JP2023081818
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-17
- Publication Date
- 2024-11-28
AI Technical Summary
Existing vehicle control devices require multiple types of input data and sensors to estimate torque, leading to increased complexity and cost, and decelerating the vehicle to reduce torque can compromise stability.
A vehicle control device that predicts future torque values using machine learning on rotational speed data without direct torque detection, and performs vibration damping control by outputting opposing torque to mitigate torsional resonance.
Simplifies the system configuration and maintains vehicle stability by predicting and damping torque vibrations without reducing driving performance.
Smart Images

Figure 2024165540000001_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to a vehicle control device that controls a vehicle equipped with a rotating electric machine and a drive transmission mechanism that transmits the driving force of the rotating electric machine to wheels. [Background technology]
[0002] For example, when a vehicle travels on a road surface that is uneven or wavy, such as a corrugated road, the wheels (drive wheels) driven by the vehicle drive device repeatedly touch the ground and lift off the ground, causing periodic rotational fluctuations in the wheels, which can cause mechanical loads on the drive transmission mechanism that transmits the driving force between the vehicle drive device and the wheels. This type of load is known as "torsional resonance" caused by periodic rotational fluctuations. When the occurrence of resonance is detected by a sensor or the like, the amplitude of the torsional resonance becomes large, and the load on the drive transmission mechanism also becomes large. Therefore, it has been proposed to estimate the torque caused by the torsional resonance in the drive transmission mechanism and perform avoidance processing to reduce the torque.
[0003] JP 2023-18375 A discloses a control device (system control unit (100)) that estimates torque caused by torsional resonance using a long-term short-term memory neural network that can handle time-series data while retaining information on the transition along the time axis as a learned model (symbols in parentheses in the background art are those of the referenced document). When this control device determines that an excessive load acts on the drive transmission mechanism, it executes an avoidance process to reduce the torque. The avoidance process is realized by reducing the torque generated in the vehicle drive device by reducing the torque of the wheel drive source (the second motor generator in the document). In addition, the input data (X) for estimating the torque value includes the rotation speed of each of the four wheels detected by the wheel speed sensor, the rotation speed of the drive source, the acceleration in the longitudinal direction of the vehicle, the torque command value of the drive source, the accelerator operation amount, and the vehicle speed. [Prior art documents] [Patent documents]
[0004] [Patent Document 1] JP 2023-18375 A Summary of the Invention [Problem to be solved by the invention]
[0005] The control device disclosed in the above document requires many types of input data to estimate the torque value of the drive transmission mechanism. In order to obtain these many types of data, various sensors are also required. In addition, a mechanism for obtaining the accelerator operation amount may also need to be added. This increases the cost of sensors, etc., and tends to complicate the system. In addition, the control device disclosed in the above document performs the avoidance process by reducing the torque of the driving force source, that is, by limiting the torque, but limiting the torque of the driving force source when the vehicle is decelerating may cause the vehicle to lose stability, which is not preferable.
[0006] In view of the above background, it is desirable to provide a technique for suppressing torque vibrations that occur in a drive transmission mechanism when a vehicle travels on an uneven or undulating road surface, while preventing the system from becoming larger and more complicated. [Means for solving the problem]
[0007] In view of the above, a vehicle control device is a vehicle control device that controls a vehicle equipped with a rotating electric machine and a drive transmission mechanism that transmits the driving force of the rotating electric machine to wheels, and includes a rotational speed detection unit that detects the rotational speed of the rotating electric machine, a torque prediction unit that predicts a future value of the torque transmitted through the drive transmission mechanism by machine learning based on a rotational speed detection value group, which is a collection of detection values of the rotational speed detected by the rotational speed detection unit over a predetermined learning period, and a rotating electric machine control unit that performs vibration suppression control to output a torque to the rotating electric machine that is in the opposite phase to the torque vibration indicated in the torque prediction value, based on the torque prediction value predicted by the torque prediction unit.
[0008] According to this configuration, the future value of the torque transmitted through the drive transmission mechanism can be predicted by machine learning using the detected value of the rotation speed of the rotating electric machine, without using information on the torque of the driving force source. Therefore, it is easy to simplify the configuration of the vehicle control device. Furthermore, according to this configuration, vibration suppression control is performed to cause the rotating electric machine to output a torque in the opposite phase to the torque vibration transmitted to the drive transmission mechanism based on the torque prediction value of the drive transmission mechanism, so that even when the rotating electric machine outputs a deceleration torque (negative torque), vibration suppression control can be performed without reducing the running stability of the vehicle. In other words, according to this configuration, it is possible to suppress the torque vibration generated in the drive transmission mechanism when the vehicle runs on an uneven or wavy road surface while suppressing the system from becoming larger and more complicated.
[0009] Further features and advantages of the vehicle control device will become apparent from the following description of exemplary and non-limiting embodiments which are given with reference to the drawings. [Brief description of the drawings]
[0010] [Figure 1] FIG. 1 is a skeleton diagram illustrating an example of a vehicle drive device; [Diagram 2] FIG. 1 is a block diagram showing an example of a rotating electrical machine control system; [Diagram 3] Conceptual diagram of vibration suppression control [Figure 4] Conceptual diagram of predicting future torque of drivetrain mechanism from past rotation speed [Diagram 5] A diagram showing a schematic diagram of the procedure for predicting resonance torque. [Figure 6] Schematic block diagram of a vehicle control device DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0011] Hereinafter, an embodiment of the vehicle control device will be described with reference to the drawings. FIG. 1 is a schematic diagram of a vehicle drive device 90 to be controlled by a vehicle control device 50 (see FIG. 2, FIG. 6, etc.). The vehicle control device 50 controls a vehicle 100 (see FIG. 4) including a rotating electric machine 10 including a rotor 11 and a stator 12, and a drive transmission mechanism 40 that transmits the drive force of the rotating electric machine 10 to wheels W. In this embodiment, a form in which the drive transmission mechanism 40 includes a planetary gear mechanism 20 that functions as a speed reducer, and a differential gear mechanism 30 that distributes the drive force from the rotating electric machine 10 to a pair of wheels W via a pair of drive shafts DS is illustrated. However, the drive transmission mechanism 40 may be configured to transmit the drive force of the rotating electric machine 10 to one wheel W without including the differential gear mechanism 30. In addition, the drive transmission mechanism 40 may be configured to directly connect the rotating electric machine 10 and the wheels W by a transmission member such as a drive shaft DS without including a gear mechanism.
[0012] In this embodiment, the planetary gear mechanism 20 includes a sun gear 21 that rotates integrally with the rotor 11, a ring gear 23 fixed to a non-rotating member such as a case, a carrier 22 that rotates integrally with an input element (for example, a differential input gear not shown) of the differential gear mechanism 30, and a pinion gear 24 that is rotatably supported by the carrier 22 and meshes with the sun gear 21 and the ring gear 23. Here, a single-pinion type mechanism is exemplified in order to simply show a configuration example, but the planetary gear mechanism 20 may be configured by a double-pinion type or other composite mechanism. In addition, here, a configuration in which the planetary gear mechanism 20 functions as a reducer is exemplified, but it may also be a speed-up gear. In addition, here, a reducer having fixed speed stages is exemplified, but it may be a reducer or speed-up gear having multiple speed stages equipped with an engagement device such as a clutch or brake, or it may be a transmission that can reduce and increase speed.
[0013] Further, while the vehicle drive device 90 shown here has a single-axis configuration in which the rotating electric machine 10, planetary gear mechanism 20, differential gear mechanism 30, and drive shaft DS are arranged on the same rotation axis, for example, if the drive transmission mechanism 40 is equipped with a parallel axis gear mechanism such as a counter gear mechanism, the vehicle drive device 90 may have a two-axis or three-axis configuration.
[0014] The rotating electric machine 10, which functions as a driving force source for the wheels W, is electrically connected to a DC power supply 70 constituted by an electricity storage device such as a secondary battery or a capacitor, via an inverter circuit 60, as shown in Fig. 2. The rotating electric machine 10 has a function as a motor (electric motor) that receives power from the DC power supply 70 to generate power, and a function as a generator (electric power generator) that receives power from the wheels W to generate power. The rotating electric machine 10 generates driving force by running using power stored in the DC power supply 70, and also generates power using driving force transmitted from the wheels W to charge the DC power supply 70.
[0015] The rotating electric machine 10 is driven and controlled by a rotating electric machine control unit 3 (see FIG. 6 ) based on the target torque of the rotating electric machine 10. The rotating electric machine control unit 3 controls the switching of an inverter circuit 60 composed of a plurality of switching elements, and causes the inverter circuit 60 to convert power between DC and multiple-phase (three-phase in this embodiment) AC. The inverter circuit 60 is composed of a plurality of switching elements. Although not shown because it is publicly known, the inverter circuit 60 has a plurality of sets (three sets here) of arms for one AC phase, each of which is composed of a series circuit of an upper-stage switching element on the positive pole side of DC and a lower-stage switching element on the negative pole side. Each switching element is provided with a freewheel diode with the forward direction being the direction from the negative pole to the positive pole (the direction from the lower stage to the upper stage). As the switching element, it is preferable to use a power semiconductor element such as an IGBT (Insulated Gate Bipolar Transistor), a power MOSFET (Metal Oxide Semiconductor Field Effect Transistor), a SiC-MOSFET (Silicon Carbide - Metal Oxide Semiconductor FET), a SiC-SIT (SiC - Static Induction Transistor), or a GaN-MOSFET (Gallium Nitride - MOSFET).
[0016] 2 and 6, the rotating electric machine control unit 3 includes a feedback calculation unit 7, and performs current feedback control, for example, by a known vector control method, based on the rotational position of the rotor 11 (magnetic pole position of the permanent magnet), the rotational speed of the rotor 11, and the current flowing through the stator coil 13 of each of the three phases, to drive and control the rotating electric machine 10 via the inverter circuit 60. The rotational position of the rotor 11 is detected by a rotation sensor 51 such as a resolver or an inductive position sensor. The rotation sensor 51 also detects the rotational speed of the rotor 11. The current flowing through the stator coil 13 is detected by a current sensor 52. The current sensor 52 is preferably a non-contact type current sensor installed near a power line such as a bus bar connecting the inverter circuit 60 and the stator coil 13.
[0017] Incidentally, as shown in FIG. 4, when the vehicle 100 runs on a road surface R having unevenness or undulations such as a wavy road, the wheels W driven by the vehicle drive device 90 repeatedly touch the ground and lift off the ground, causing periodic rotational fluctuations in the wheels W, which may cause mechanical loads on the drive transmission mechanism 40 that transmits driving force between the vehicle drive device 90 and the wheels W. Such loads are known as "torsional resonance" caused by periodic rotational fluctuations. Such resonance can also be detected using a sensor. However, at the time when the occurrence of resonance is detected by a sensor or the like, the amplitude of the torsional resonance becomes large, and the load on the drive transmission mechanism 40 also becomes large. Therefore, it is preferable that the vehicle control device 50 estimates the torque caused by the torsional resonance in the drive transmission mechanism 40 and performs avoidance processing to reduce the torque. The vehicle control device 50 of this embodiment is configured to suppress torque vibrations that occur when the vehicle 100 runs on a road surface R having unevenness or undulations while suppressing the system from becoming larger and more complicated. In the following description, the torque that is generated due to resonance in the drive transmission mechanism 40 and that should be avoided as described above may be referred to as "resonance torque".
[0018] The vehicle control device 50 includes a rotation speed detection unit (rotation sensor 51) that detects the rotation speed ωmg (see FIGS. 3, 4, and 6) of the rotating electric machine 10, and a torque prediction unit 1 that predicts a future value of the torque transmitted through the drive transmission mechanism 40 by machine learning based on a rotation speed detection value group that is a collection of detection values of the rotation speed ωmg detected by the rotation sensor 51 during a predetermined learning period T1. As described above, the vehicle control device 50 includes the rotating electric machine control unit 3, and the rotating electric machine control unit 3 performs vibration suppression control that causes the rotating electric machine 10 to output a torque in the opposite phase to the torque vibration indicated by the torque prediction value Tds (see FIGS. 3, 4, and 6) based on the torque prediction value Tds predicted by the torque prediction unit 1.
[0019] That is, the vehicle control device 50 predicts the future value of the torque (torque prediction value Tds) transmitted through the drive transmission mechanism 40 by machine learning using the detection value of the rotation speed ωmg of the rotating electric machine 10, without detecting the torque or using a torque command value. Therefore, it is easy to simplify the configuration of the vehicle control device 50. In addition, when the rotating electric machine 10 outputs a deceleration torque (negative torque), for example, if the output of the torque is limited for vibration suppression, the running stability of the vehicle 100 may be reduced. In this embodiment, however, based on the torque prediction value Tds, vibration suppression control is performed so that the rotating electric machine 10 outputs a torque in the opposite phase to the torque vibration transmitted to the drive transmission mechanism 40. Therefore, in this embodiment, the vehicle control device 50 can perform vibration suppression control without reducing the running stability of the vehicle 100.
[0020] The rotation speed ωmg of the rotating electric machine 10 detected by the rotation speed detection unit may be in a form in which the rotation of the rotor 11 is directly detected as in this embodiment, but may also be the rotation speed of a rotating member that rotates in synchronization with the rotation of the rotor 11 and at a rotation speed proportional to the rotation speed of the rotor 11, such as the sun gear 21, the carrier 22, or a differential input member of the differential gear mechanism 30. However, in feedback control of the rotating electric machine 10, the rotation of the rotor 11 or a rotor shaft (no sign) that is connected to the rotor 11 and rotates integrally with the rotor 11 is generally detected by the rotation sensor 51. Therefore, by using such a rotation sensor 51 as the rotation speed detection unit, a rotation speed detection unit for vibration suppression control can be provided without adding an additional sensor.
[0021] Incidentally, corresponding to the group of rotation speed detection values, the torque prediction value Tds may also be a set of multiple values, such as the torque prediction value Tds in the prediction period T2 shown in Fig. 4. Alternatively, the torque prediction value Tds estimated as a single value may be repeatedly calculated multiple times over time to form a torque prediction value group including multiple torque prediction values Tds. In other words, the term "torque prediction value Tds" includes both a single value and a set (group).
[0022] In addition, when calculating the torque prediction value Tds from the rotation speed ωmg, as one mode, the future rotation speed ωmg is predicted from the past rotation speed ωmg, and the future torque prediction value Tds can be calculated from the future rotation speed ωmg. For example, the torque prediction unit 1 includes a rotation speed prediction unit (which can also be called an arrival vibration prediction unit) that predicts a future value of the rotation speed ωmg of the rotating electric machine 10 based on a group of rotation speed detection values. In addition, the torque prediction unit 1 includes a conversion unit that converts the rotation speed prediction value, which is the future value of the rotation speed ωmg predicted by the rotation speed prediction unit, into a torque prediction value Tds, which is the future value of the torque transmitted through the drive transmission mechanism 40. Since the conversion unit converts the dimension (domain) of the physical quantity from the rotation speed to the torque, the conversion unit may be called a domain conversion unit.
[0023] The vehicle control device 50 further includes a resonance prediction unit 2 that predicts future occurrence of resonance in the drive transmission mechanism 40 based on the torque prediction value Tds before the vibration of the torque transmitted through the drive transmission mechanism 40 reaches a resonance peak. The rotating electric machine control unit 3 preferably performs vibration damping control when the resonance prediction unit 2 predicts future occurrence of resonance. In this case, vibration damping control can be performed based on the torque prediction value Tds before the vibration of the torque transmitted through the drive transmission mechanism 40 reaches a resonance peak. This makes it easy to avoid excessive torque acting on the drive transmission mechanism 40.
[0024] In addition, it is preferable that the rotating electrical machine control unit 3 starts vibration suppression control when the amplitude of the torque vibration indicated by the torque prediction value Tds exceeds a predetermined first threshold value, and ends vibration suppression control when the amplitude of the torque vibration indicated by the torque prediction value Tds becomes equal to or less than a predetermined second threshold value that is lower than the first threshold value. In this case, by not performing unnecessary vibration suppression control, it is possible to reduce the calculation processing load and interference with other controls.
[0025] In addition, when the amplitude of the torque vibration does not exceed the first threshold value while the vibration damping control is not being executed, the vibration damping control is not started as a matter of course. In other words, the rotating electrical machine control unit 3 does not execute the vibration damping control while it is determined that there will be no torque resonance in the future.
[0026] A specific configuration for determining the torque prediction value Tds from the rotation speed ωmg of the rotor 11 will be described below. Fig. 3 shows the basic technical idea and scheme of vibration suppression control in this embodiment. That is, the rotating electric machine control unit 3 predicts the torque prediction value Tds, which is the torque waveform of the drive shaft DS, from the rotation speed ωmg of the rotor 11 of the rotating electric machine 10, and realizes vibration suppression control by controlling the rotating electric machine 10 using a resonance control signal generated based on the torque prediction value Tds.
[0027] In this embodiment, the vehicle control device 50 calculates the value "x tWhen "t∈[0,T-1]" is input at time step "t∈[0,T max Train a model that outputs ".
[0028]
number
[0029] Furthermore, “T” and “T max " respectively indicate the prediction start point (time t1 in Figure 4) and the prediction end point. When torsional resonance occurs, extremely large amplitudes are observed in specific time intervals, so here we adopt "quantile regression" for torsional resonance prediction. The prediction of each quantile is expressed as the following equation (2).
[0030]
number
[0031] The left side of equation (2) indicates that the qth quantile estimated τ steps ahead is observed at time t. q (·) indicates the prediction model.
[0032] The structure of the prediction model preferably includes, for example, a recurrent neural network (RNN), a temporal convolutional neural network (TCN), a multi-head attention mechanism, etc. For example, in one embodiment, the vehicle control device 50 preferably has a function of extracting features of an input sequence by extracting recurrent features and features by temporal convolution to create feature pairs, calculating attention between them, and further extracting important features by providing a gate structure (a mechanism for feature extraction) in the network output.
[0033] FIG. 5 shows a schematic diagram of a procedure for predicting resonance torque, which is a torque that occurs in the drive transmission mechanism 40 due to resonance and should be avoided. The first step #1 is an input step, in which the rotation speed ωmg of the rotating electric machine 10 is acquired. The second step #2 is a point prediction step, in which future prediction is performed using a recurrent neural network and a time convolution neural network. The recurrent neural network has a problem in improving the prediction performance in cross-domain (between the rotation speed ωmg and the torque prediction value Tds). Meanwhile, the time convolution neural network has a problem in improving the long-term prediction performance in the same domain. By using the recurrent neural network and the time convolution neural network together, it is easy to realize appropriate point prediction.
[0034] The third step #3 is a resonance detection step. As shown in Fig. 6, in this embodiment, the vehicle control device 50 includes a resonance prediction unit 2 that predicts future occurrence of resonance of the drive transmission mechanism 40 based on the torque prediction value Tds before the torque vibration transmitted through the drive transmission mechanism 40 reaches a resonance peak. In this embodiment, the resonance prediction unit 2 predicts future occurrence of resonance when the amplitude of the torque vibration indicated by the torque prediction value Tds exceeds a first threshold value.
[0035] The fourth step #4 is an output step of outputting a signal (e.g., a command value, a resonance torque suppression signal) to the feedback calculation unit 7 for outputting a torque in the opposite phase to the resonance torque to the rotating electric machine 10. In the vibration suppression control, a torque in the opposite phase to the torque vibration is output to the rotating electric machine 10, but when the amplitude of the resonance is small, the vibration suppression control is not particularly necessary. Therefore, it is preferable for the rotating electric machine control unit 3 to perform the vibration suppression control when the resonance prediction unit 2 predicts the occurrence of future resonance. In this case, a control value for determining whether or not resonance control is required is required. According to experiments and simulations by the inventors, it has been confirmed that the robustness of the characteristic change of the vehicle 100 is higher in the control value based on the torque prediction value Tds than in the control value based on the rotation speed ωmg. That is, the inventors have confirmed that the case where the control value is set based on the torque prediction value Tds of the drive shaft DS is more appropriate in determining whether or not resonance control is required than the case where the control value is set based on the rotation speed ωmg of the rotating electric machine 10.
[0036] The rotating electrical machine control unit 3 starts vibration suppression control when the amplitude of the torque vibration indicated in the torque prediction value Tds exceeds a first threshold value, and ends vibration suppression control when the amplitude of the torque vibration indicated in the torque prediction value Tds becomes equal to or smaller than a second threshold value. Here, "the first threshold value>the second threshold value." As shown in FIG. 6, in this embodiment, the rotating electrical machine control unit 3 includes a control value calculation unit 4. The control value calculation unit 4 determines whether or not vibration suppression control is necessary, and calculates a control value for generating an antiphase torque for vibration suppression control based on the amplitude of the torque vibration.
[0037] The rotating electric machine control unit 3 also includes a suppression signal generation unit 5 that generates a suppression signal (torque in opposite phase to the torque vibration) for suppressing vibration based on the torque prediction value Tds. In a fourth step #4, the torque in opposite phase generated by the suppression signal generation unit 5 is output to the feedback calculation unit 7. The rotating electric machine control unit 3 (control value calculation unit 4, suppression signal generation unit 5) starts vibration suppression control when the amplitude of the torque vibration predicted by the resonance prediction unit 2 exceeds a first threshold value. After starting the vibration suppression control, the rotating electric machine control unit 3 ends the vibration suppression control when the amplitude of the torque vibration indicated by the torque prediction value Tds becomes equal to or less than a second threshold value.
[0038] The feedback calculation unit 7 drives and controls the rotating electric machine 10 so as to output a torque obtained by superimposing a torque of an opposite phase on a torque set by current feedback control based on the rotational position of the rotor 11, the rotational speed of the rotor 11, and the current flowing through the stator coils 13 of each of the three phases. When vibration damping control is not executed, the feedback calculation unit 7 drives and controls the rotating electric machine 10 so as to output a torque set based on the rotational position, the rotational speed, and the current without superimposing the torque of the opposite phase. [Explanation of symbols]
[0039] 1: torque prediction unit, 2: resonance prediction unit, 3: rotating electric machine control unit, 10: rotating electric machine, 40: drive transmission mechanism, 50: vehicle control device, 51: rotation sensor (rotation speed detection unit), 100: vehicle, T1: learning period, Tds: torque prediction value, W: wheel, ωmg: rotation speed
Claims
1. A vehicle control device that controls a vehicle including a rotating electric machine and a drive transmission mechanism that transmits a drive force of the rotating electric machine to wheels, a rotation speed detection unit that detects a rotation speed of the rotating electric machine; a torque prediction unit that predicts a future value of a torque transmitted through the drive transmission mechanism by machine learning based on a rotational speed detection value group, which is a set of detection values of the rotational speed detected by the rotational speed detection unit during a predetermined learning period; A control device for a vehicle comprising: a rotating electric machine control unit that performs vibration suppression control to cause the rotating electric machine to output a torque in opposite phase to the torque vibration indicated by the torque prediction value based on the torque prediction value predicted by the torque prediction unit.
2. a resonance prediction unit that predicts future occurrence of resonance of the drive transmission mechanism before a vibration of torque transmitted through the drive transmission mechanism reaches a resonance peak based on the torque prediction value, The vehicle control device according to claim 1 , wherein the rotating electrical machine control unit performs the vibration suppression control when the resonance prediction unit predicts the occurrence of resonance in the future.
3. 3. The vehicle control device according to claim 1, wherein the rotating motor control unit starts the vibration damping control when the amplitude of the torque vibration indicated in the torque prediction value exceeds a predetermined first threshold value, and terminates the vibration damping control when the amplitude of the torque vibration indicated in the torque prediction value becomes equal to or lower than a predetermined second threshold value that is lower than the first threshold value.
Citation Information
Patent Citations
Vehicle control device
JP2023018375A
