Vehicle control system

Abstract

To provide a control device of a vehicle that can execute driving-system protection control in accordance with a state of a road surface on which a vehicle is running and a habit and a character of each driver to make both reliability on a driving system and power performance compatible.SOLUTION: A control device of a vehicle inputs running data Ddat to a machine-learnt model selected in accordance with a running road surface section Rcd reflecting a state of a running road surface to calculate a maximum torque predicted value Tpr, and also estimates a driver section Lv by clustering, and performs control of setting a limit value constant Lmtv for driving-system protection control in accordance with the running road surface section Rcd and the driver section Lv. When a maximum value of generated torque Tmax exceeds the maximum torque predicted value Tpr, the control device renews the maximum torque predicted value Tpr to a maximum value of generated torque Tmax, and further renews the driver section Lv by clustering, and performs control of setting the limit value constant Lmtv for the driving-system protection control.SELECTED DRAWING: Figure 4

Description

Technical Field

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

Background Art

[0002] In vehicles, generally, in order to avoid damage to the drive system and deterioration of reliability that may occur due to excessive torque being applied to the drive system, drive system protection control such as engine torque limitation, engine speed limitation, and suppression of brake hydraulic pressure gradient is implemented.

[0003] In the drive system protection control, it has been proposed to use a learning model of a neural network as control that predicts a driving operation that applies an excessive load to the vehicle and executes avoidance processing in advance, and to improve the prediction accuracy of the driving operation by reflecting the habits of the driver. For example, the vehicle control device described in Patent Document 1 is like this.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] In the above conventional vehicle control device, although it is said that the prediction accuracy is improved by learning while reflecting the habits of the driver, for situations that occur rarely, such as wheel spin grounding during off-road driving or sudden braking on snowy or muddy roads, which may lead to damage to the drive system if they occur, there is a problem that learning is difficult. If these situations occur before or during learning, they cannot be avoided and there is a risk of damage.

[0006] Furthermore, the aforementioned avoidance process applied the same control to all drivers, resulting in a problem where the control was not optimized for each individual driver. While the control ensured the reliability of the drivetrain by avoiding events that could only occur for a very small number of drivers, it was over-engineered for the majority of other drivers, and power performance was sacrificed.

[0007] The present invention was made against the above circumstances, and its objective is to provide a vehicle control device that implements drivetrain protection control according to the road surface conditions and the habits and characteristics of each driver, thereby achieving an optimal balance between drivetrain reliability and power performance. [Means for solving the problem]

[0008] The gist of the present invention is a vehicle control device that (a) performs drive system protection control to suppress the generation of excessive torque in the drive system, and (b) an estimation unit that selects one of a plurality of pre-stored machine learning models based on the driving road surface conditions, inputs driving data for a predetermined period to the selected machine learning model to calculate a maximum torque prediction value generated during a specific period, and estimates the driver's category of the driver by machine learning clustering based on the maximum torque prediction value, the driving data for the predetermined period, and a pre-prepared driver category, and (c) a determination unit that controls the setting of the limit value constant of the drive system protection control according to the driving road surface conditions and the estimated driver category, and updates the maximum torque prediction value and the driver category and controls the setting of the limit value constant of the drive system protection control if the maximum generated torque during a specific period exceeds the maximum torque prediction value.

[0009] Preferably, the accuracy of the machine learning model's prediction estimation can be improved by retraining the machine learning model using the driving data when the maximum torque prediction value and the driver classification are updated as training data.

[0010] The machine learning models used here (including pre-trained models) refer to models trained using machine learning algorithms. A specific example of a machine learning algorithm is deep learning, which utilizes neural networks to generate its own features and connection weights for training. [Effects of the Invention]

[0011] According to the present invention, in addition to controlling the setting of the limit constant of the drivetrain protection control according to the road surface conditions and the estimated driver classification, if the maximum generated torque for the specified period exceeds the predicted maximum torque value, the predicted maximum torque value and the driver classification are updated and the setting of the limit constant of the drivetrain protection control is controlled. As a result, the drivetrain protection control can be implemented according to the road surface conditions and the driver classification which reflects the habits and characteristics of each driver, and a vehicle control device can be provided that can optimally balance the reliability and power performance of the drivetrain. [Brief explanation of the drawing]

[0012] [Figure 1] This is a diagram illustrating the schematic configuration of a vehicle according to one embodiment of the present invention. [Figure 2] This figure shows the main components of the torque suppression control of the electronic control unit shown in Figure 1. [Figure 3] Figures 1 and 2 are block diagrams illustrating the flow of processing data for torque suppression control of the electronic control unit and processing on the server. [Figure 4] Figures 1 and 2 show flowcharts illustrating the key aspects of the torque suppression control of the electronic control unit, specifically the control operation examples. [Modes for carrying out the invention]

[0013] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that in the following embodiments, the drawings have been simplified or modified as appropriate, and the dimensional ratios and shapes of each part are not necessarily depicted accurately. [Examples]

[0014] Figure 1 is a diagram illustrating the schematic configuration of an engine-driven, four-wheel-drive vehicle 10 to which the present invention is applied. In Figure 1, the vehicle 10 includes an engine 12, wheels 14, brakes 44, and power transmission devices 18 that transmit power from the engine 12 and other components to the wheels 14.

[0015] The power transmission system 18 includes a transmission 26 having a power distribution device and an automatic transmission, a transfer case 28, a propeller shaft 30, a differential 34, and a drive shaft 38.

[0016] Vehicle 10 is equipped with an electronic control unit 80, which acts as a controller and includes a control device for controlling the engine 12 and the power transmission system 18, etc. The electronic control unit 80 is composed of a so-called microcomputer, and performs various controls of vehicle 10 by having the CPU perform signal processing according to a stored program. The electronic control unit 80 functionally includes a drive control unit 82 and a torque suppression control unit 84.

[0017] The drive control unit 82 controls all functions related to driving. It includes functions as an engine control unit that controls the operation of the engine 12 by controlling the engine control device 20, which includes the throttle actuator, fuel injector, and ignition device; functions as a gear shift control unit that controls the shifting of the transmission 26 by controlling the transmission control mechanism 42 based on the actual vehicle speed and requested output from a pre-stored AT gear shift map; and functions as a brake control unit that controls the braking action of the brake 44 by controlling the brake control mechanism 48.

[0018] The torque suppression control unit 84 is a control unit that oversees the control related to the present invention, and performs drive system protection control to suppress the generation of excessive torque in the drive system through the control described later.

[0019] In the vehicle 10, taking as an example the situation where excessive torque is applied during off-road driving, on a rocky road or mogul road, one of the wheels 14 may float, and when it is installed on the road surface during subsequent driving, excessive torque may be generated due to the engine torque Te and the inertia torque due to the inertia Itm of the input system of the engine 12 and the transmission 26, and there is a concern that the drive system such as the transmission 26, transfer 28, propeller shaft 30, differential 34, etc. may be damaged. In addition, on a snowy road or muddy road, when the wheel 14 spins and suddenly brakes, excessive torque may be generated due to the inertia torque of the input system. Since the load on the drive system from the output shaft of the transmission 26 to the axle side is more severe when the gear ratio yat is lower, the drive control unit 82 sets the upper limit value of the engine torque Te and the upper limit value of the engine speed Ne when the gear ratios yat of the transmission 26 and transfer 28 are in the low gear, and performs control so that they do not exceed those values. The upper limit value of the engine torque Te, the upper limit value of the engine speed Ne, etc. are stored in the electronic control unit 80 as a map, and the torque suppression control unit 84 performs control to switch according to the driving road surface classification Rcd and driver classification Lv described later.

[0020] The torque Ta generated in the drive system is calculated by the following equation (1). Ta = [Te + (Itm × engine angular acceleration ωe)] × yat ···(1)

[0021] FIG. 2 shows an overview of the torque suppression control unit 84. The torque suppression control unit 84 includes an input unit 86 to which data from various sensors and other control devices in the vehicle, such as driving data Dda, road surface image Iard, and weather information Winfo, are input, and a storage unit 88 that stores a part of the data input to the input unit 86. Further, the torque suppression control unit 84 includes an estimation unit 90 described later, a determination unit 92 described later, and a storage / output unit 94 that stores the data determined by the determination unit 92 in the storage unit 88 or outputs it to other control flows.

[0022] The estimation unit 90 performs the following three processes based on the input data from the input unit 86 and the storage unit 88. First, it selects a road surface classification Rcd based on the road surface conditions. Second, it selects one of several pre-stored machine learning models according to the selected road surface classification Rcd, inputs driving data Ddat for a predetermined period to the selected machine learning model, and calculates the maximum torque prediction value Tpr that will occur during the specific period. Third, it estimates the driver's driver classification Lv using machine learning clustering based on the calculated maximum torque prediction value Tpr, the driving data Ddat for the predetermined period, and a pre-prepared driver classification Lv.

[0023] The determination unit 92 controls the setting of the limit value constant Lmtv for drivetrain protection control according to the road surface classification Rcd and estimated driver classification Lv output from the estimation unit 90. In addition, if the maximum generated torque Tmax for a specific period exceeds the maximum torque prediction value Tpr, it updates the maximum torque prediction value Tpr and driver classification Lv and controls the setting of the limit value constant Lmtv for drivetrain protection control. Machine learning models, maps, and calculation formulas created in advance through experiments or simulations, which are necessary for the processing and calculations of the estimation unit 90 and the determination unit 92, are stored in ROM.

[0024] Furthermore, the torque suppression control unit 84 (electronic control unit 80) includes a transmitting unit 96 for transmitting stored data to a server 100 located outside the vehicle 10, and a receiving unit 98 for receiving data from the server.

[0025] Figure 3 is a block diagram illustrating the flow of processing data for torque suppression control by the torque suppression control unit 84. In Figure 3, the areas within the parallelograms represent the output results or values ​​of each process, and the symbols beginning with S in parentheses indicate the corresponding steps in the flowchart of Figure 4.

[0026] Figure 4 is a flowchart illustrating the main parts of the control operation of the torque suppression control unit 84. Hereafter, the control operation of the torque suppression control unit 84 will be explained in accordance with the processing steps in Figure 4, with reference to Figure 3.

[0027] In step S1 of Figure 4 (the step will be omitted hereafter), driving data and weather data are acquired. As shown in Figure 3, driving data Dda such as accelerator opening θacc, engine torque Te, engine speed Ne, transmission gear Sat, right front wheel rotation speed Nfr, left front wheel rotation speed Nfl, right rear wheel rotation speed Nrr, right rear wheel rotation speed Nrl, brake hydraulic pressure Pbcr, longitudinal acceleration Gx and lateral acceleration Gy of the vehicle 10, and shift lever operating position POSsh are acquired via the input unit 86, and the time-series driving data Ddat for a predetermined period is stored in the storage unit 88. In addition, road surface images Iard taken by the onboard camera and weather information Winfo received through communication with the server 100 or other external network are also acquired via the input unit 86.

[0028] In S2, which corresponds to the estimation unit 90, a road surface classification Rcd is selected based on the road surface conditions. Based on the camera image Iard and weather information Winfo acquired via the input unit 86, a road surface classification Rcd1 is selected from the M types of pre-set road surface classifications Rcd, depending on the road surface (e.g., rocky road, mogul road, snowy road, muddy road, etc.) and the road surface condition (e.g., dry or wet). (Rcd = Rcd1 is assumed.) In addition, machine learning may be used to estimate the selection of the road surface classification Rcd, with the driving data Ddat also being input.

[0029] In S3, which corresponds to the estimation unit 90, a machine learning model is selected from N types of pre-created machine learning models for estimating maximum torque prediction value Tpr, according to the road surface category Rcd1 selected in S2. Preferably, the machine learning model is a model created using LSTM (Long Short-Term Memory).

[0030] In S4, which corresponds to the estimation unit 90, the maximum torque prediction value Tpr1 is calculated from the driving data Ddat using the machine learning model selected in S3 (Tpr = Tpr1).

[0031] In S5, which corresponds to the estimation unit 90, the maximum torque prediction value Tpr1 calculated in S4 and the driving data Ddat are input to a pre-prepared machine learning model that performs clustering, and driver category Lv1 is estimated from the pre-prepared driver category Lv of type P (Lv = Lv1). The machine learning model that performs clustering is preferably a model created by TKSM (Time Series k-means). Furthermore, weighting may be performed when inputting the maximum torque prediction value Tpr1.

[0032] In S6, which corresponds to the determination unit 92, it is determined whether the maximum value Tmax of the generated torque Ta calculated from the driving data Ddat acquired in S1 and equation (1) is less than or equal to the maximum torque prediction value Tpr1 calculated in S4 (Tpr1 ≥ Tmax).

[0033] If the judgment in S6 is affirmed, in S7, which corresponds to the determination unit 92, the driver category Lv is determined to be driver category Lv1 selected in S5 (Lv=Lv1 remains unchanged).

[0034] If the judgment in S6 is rejected, in S8 corresponding to the determination unit 92, the value of the maximum torque prediction value Tpr is changed from Tpr1 to Tmax (Tpr = Tmax).

[0035] Next, in S9, which corresponds to the determination unit 92, the same clustering as in S5 is performed using the maximum torque prediction value Tpr (=Tmax) and the driving data Ddat as input data, and the driver category Lv is updated to the updated value Lv2 (Lv = Lv2).

[0036] Next, in S10, which corresponds to the determination unit 92, the relearning flag Fml is turned ON, and in S11, the relearning driving data Dmlt is stored. The storage time for the driving data Dmlt is set in advance to a suitable value, for example, 60 seconds, so that the latest value is always stored. The driving data Dmlt from 60 seconds prior to the moment the relearning flag Fml was turned ON is stored in the storage / output unit 94.

[0037] Furthermore, in S12, which corresponds to the determination unit 92, a limit value constant Lmtv for drivetrain protection control is selected and determined according to the road surface category Rcd selected in S2 and the driver category Lv estimated in S7 or S9. Preferably, a map of the upper limits for engine torque and engine speed is selected and determined.

[0038] The upper part of Figure 3 shows the flow of processed data in the server 100. The maximum torque prediction value Tpr, the retraining flag Fml, and the driving data Dmlt stored in the memory / output unit 94 are transmitted from the vehicle 10 to the server 100 at an appropriate timing and received by the server input unit 102. In the machine learning execution unit 104, the machine learning model is retrained using these received data as training data, a retrained model Mml is created, and stored in the server memory / output unit 106. By updating the machine learning model in the electronic control unit 80 to the retrained model Mml via OTA (Over The Air) or by a dealer, the accuracy of prediction value estimation can be improved.

[0039] As described above, the electronic control device 80 of this embodiment can implement drivetrain protection control according to the road surface conditions and driver classifications that reflect the habits and characteristics of each driver, and can provide a vehicle control device that optimally balances the reliability and power performance of the drivetrain.

[0040] It should be noted that the above is merely one embodiment of the present invention, and the present invention can be implemented in various modified forms without departing from its spirit.

[0041] For example, although the vehicle in the above-described embodiment was a vehicle that used only the engine 12 as a power source, it may also be a hybrid vehicle equipped with both the engine 12 and a rotating machine as power sources, or an electric vehicle that uses only a rotating machine as a power source. Furthermore, the transmission in the transmission 26 may be any of the following: a stepped transmission, a belt-type continuously variable transmission, an electric continuously variable transmission, etc.

[0042] It should be noted that the above-described embodiment is merely one example, and the present invention can be implemented in various modified and improved forms based on the knowledge of those skilled in the art. [Explanation of Symbols]

[0043] 10: Vehicle, 80: Electronic control unit, 82: Drive control unit, 84: Torque suppression control unit, 86: Input unit, 88: Memory unit, 90: Estimation unit, 92: Decision unit, 94: Memory / Output unit, 96: Transmission unit, 98: Receiving unit, 100: Server, 102: Server input unit, 104: Machine learning execution unit, 106: Server memory / output unit

Claims

[Claim 1] A vehicle control device that performs drivetrain protection control to suppress the generation of excessive torque in the drivetrain, An estimation unit selects one of several pre-stored machine learning models based on the road surface conditions, inputs driving data for a predetermined period into the selected machine learning model, calculates a predicted maximum torque value for a specific period, and estimates the driver's category using machine learning clustering based on the predicted maximum torque value, the driving data for the predetermined period, and a pre-prepared driver category. The system includes a determination unit that controls the setting of the limit value constant of the drive system protection control according to the road surface conditions and the estimated driver category, and that updates the maximum torque value and the driver category and controls the setting of the limit value constant of the drive system protection control if the maximum torque generated during a specific period exceeds the maximum torque prediction value. A vehicle control device characterized by the following features.

Images