Vehicle control device

Abstract

This vehicle control device is provided with: a variable damper (relative displacement suppression device) that is provided between a vehicle body and wheels of a vehicle and that changes a force that suppresses relative displacement between the vehicle body and the wheels; and a vehicle state calculation unit that is mounted on the vehicle and that calculates the state of the vehicle. On the basis of a neural network that learns a result obtained by estimating a sprung speed or a relative speed in a predetermined frequency band in which high-frequency noise reduction or low-frequency phase shift reduction is performed on an input signal from a vehicle state calculation unit. A state estimation unit that outputs a sprung speed or a relative speed to an input signal from the vehicle state calculation unit; and a control value calculation unit (control unit) that controls the variable damper on the basis of the output from the state estimation unit.

Description

Technical Field

[0001] This disclosure relates to vehicle control devices, for example, suitable for use in four-wheeled vehicles. Background Technology

[0002] Patent Document 1 discloses a control device that appropriately reduces noise in sensor signals and suppresses performance degradation of control based on sensor signals. The control device acquires sensor signals that are output from a sensor detecting time-series data and include noise, and reduces the noise in the sensor signals using a recurrent neural network trained to learn the correspondence between a first signal including the noise corresponding to the sensor signals and a second signal representing the first signal after the noise has been removed.

[0003] Prior technology documents

[0004] Patent documents

[0005] Patent Document 1: Japanese Patent Application Publication No. 2021-77030 Summary of the Invention

[0006] The problem that the invention aims to solve

[0007] However, in existing technologies, filters for noise reduction are constructed using neural networks. This allows existing control devices to more effectively reduce noise in sensor signals and prevent degradation of control performance based on sensor signals. However, in existing technologies, differentiation / integration processing is performed as a pre-stage of noise reduction, and the characteristic of phase deviation from the ideal becomes a problem during this stage.

[0008] One embodiment of the present invention aims to provide a vehicle control device capable of making appropriate state estimations.

[0009] Methods for solving problems

[0010] A vehicle control device according to one embodiment of the present invention includes: a relative displacement suppression device disposed between the vehicle body and the wheels of a vehicle, which suppresses force changes that inhibit relative displacement between the vehicle body and the wheels; a vehicle state calculation unit mounted on the vehicle, which calculates the state of the vehicle; a state estimation unit that outputs sprung speed or relative speed based on an input signal from the vehicle state calculation unit using a neural network, the neural network learning to estimate the sprung speed or relative speed in a predetermined frequency band where high-frequency noise or low-frequency phase shift of the input signal from the vehicle state calculation unit is reduced; and a control unit that controls the relative displacement suppression device based on the output from the state estimation unit.

[0011] According to one embodiment of the present invention, appropriate state estimation can be performed. Attached Figure Description

[0012] Figure 1 This is a block diagram showing the vehicle control device according to the first embodiment.

[0013] Figure 2 It means Figure 1 The module diagram of the state estimation section.

[0014] Figure 3 It means Figure 2 A diagram illustrating the learning (training) method of neural networks.

[0015] Figure 4 This is an explanatory diagram showing an example of a neural network for state estimation.

[0016] Figure 5 It is a characteristic curve diagram showing the frequency characteristics of gain and phase when integrating the upper and lower accelerations along the frequency axis.

[0017] Figure 6 It is a characteristic curve diagram showing the frequency characteristics of the gain and phase of the transfer function between the vertical acceleration and the estimated speed on the spring.

[0018] Figure 7 It is a characteristic curve diagram representing the time-varying velocity of the estimated spring.

[0019] Figure 8 It is a characteristic curve diagram representing the time-varying displacement and jerk of the spring.

[0020] Figure 9 This is a block diagram illustrating the vehicle control device according to the second embodiment.

[0021] Figure 10 It means Figure 9 The module diagram of the state estimation section.

[0022] Figure 11 It means Figure 10 A diagram illustrating the learning method of neural networks in the diagram.

[0023] Figure 12 This is a block diagram showing the vehicle control device according to the third embodiment.

[0024] Figure 13 It means Figure 12 The module diagram of the state estimation section.

[0025] Figure 14 It means Figure 13 A diagram illustrating the learning method of neural networks in the diagram.

[0026] Figure 15 This is a block diagram showing the vehicle control device according to the fourth embodiment. Detailed Implementation

[0027] Hereinafter, a detailed description will be given with reference to the accompanying drawings, taking the application of the vehicle control device according to the embodiments of the present invention to a four-wheeled vehicle as an example.

[0028] Figure 1 This refers to vehicle control device 1. Vehicle control device 1 consists of suspension device 5 and controller 11. Here, in... Figure 1 In this vehicle, for example, left and right front wheels and left and right rear wheels (hereinafter collectively referred to as wheels 3) are provided on the lower side of the vehicle body 2, which constitutes the vehicle body. The wheel 3 is configured to include a tire 4, which functions as an elastic element (spring) to absorb the small bumps and depressions of the road surface.

[0029] The suspension device 5 is installed between the vehicle body 2 and the vehicle wheels 3. The suspension device 5 consists of a suspension spring 6 (hereinafter referred to as spring 6) and a damping force adjustable damper (hereinafter referred to as variable damper 7), which is arranged in parallel with spring 6 between the vehicle body 2 and the wheels 3.

[0030] In addition, Figure 1 The diagram shows a configuration where one set of suspension devices 5 is positioned between the vehicle body 2 and the wheels 3. However, for example, four sets of suspension devices 5 may be independently positioned between each of the four wheels 3 and the vehicle body 2. Figure 1 Only one set is shown schematically.

[0031] Here, the variable damper 7 of the suspension device 5 is constructed using a damping force adjustable hydraulic buffer that is installed between the vehicle body 2 and the wheel 3. The variable damper 7 is installed between the vehicle body 2 and the wheel 3, constituting a relative displacement suppression device that suppresses force changes in the relative displacement between the vehicle body 2 and the wheel 3.

[0032] To continuously adjust the damping force characteristic (i.e., the damping force characteristic) from a hard characteristic to a soft characteristic, a variable damping force actuator 8, consisting of a damping force adjustment valve, is attached to the variable damper 7. Furthermore, the variable damping force actuator 8 may not necessarily be a structure that requires continuous adjustment of the damping force characteristic; for example, it may be a structure capable of adjusting the damping force in multiple stages (two or more stages). Additionally, the variable damper 7 can be a pressure-controlled type or a flow-controlled type. The variable damper 7 may also be a type that controls viscosity, such as in magnetoviscous fluids or electrorheological fluids.

[0033] Controller 11 constitutes a control device. Controller 11, as a control device for controlling the damping characteristics of the variable damper 7, is, for example, composed of a microcomputer. Controller 11 is connected, for example, to a CAN (Controller Area Network) network required for data communication. Controller 11 acquires data related to vehicle behavior (hereinafter referred to as behavior information) via CAN 10. This behavior information includes, for example, front-rear acceleration (front-rear G), lateral acceleration (lateral G), steering angle, yaw rate, wheel speed, etc. Therefore, this behavior information is input to controller 11. The output side of controller 11 is connected to the variable damping force actuator 8 of the variable damper 7.

[0034] In addition, the controller 11 has a storage unit 12 composed of ROM (Read-Only Memory), RAM (Random Access Memory), non-volatile memory, etc. The storage unit 12 of the controller 11 stores various programs, information (vehicle information), data, etc., for controlling the variable damper 7. Based on the behavioral information, the controller 11 estimates the sprung speed and relative speed as vehicle state quantities. At this time, the relative speed is the relative speed between the sprung mass and the unsprung mass, which is the piston speed of the variable damper 7. Based on the estimated vehicle state quantities, the controller 11 calculates the force that should be generated by the variable damper 7 (force generating mechanism) of the suspension device 5, and outputs this control signal (command current) to the damping force variable actuator 8 of the suspension device 5.

[0035] like Figure 1 As shown, the controller 11 includes a damper control unit 13, a weight parameter storage unit 14, and a data readout unit 15. The damper control unit 13 includes a vehicle state calculation unit 16, a state estimation unit 17, and a control value calculation unit 20. The weight parameter storage unit 14 may be part of the storage unit 12, or it may be different from the storage unit 12.

[0036] The vehicle state calculation unit 16 is mounted on the vehicle and calculates the vehicle's state. Specifically, the vehicle state calculation unit 16 acquires data related to the vehicle's state during driving from communication data transmitted within the vehicle via the CAN 10, and calculates behavioral information (vehicle state) such as wheel speed, front-rear acceleration (front-rear G), lateral acceleration (lateral G), steering angle, and yaw rate based on the acquired data. Furthermore, the behavioral information calculated by the vehicle state calculation unit 16 based on the CAN signal includes wheel speed, front-rear acceleration (front-rear G), lateral acceleration (lateral G), steering angle, and yaw rate, but the present invention is not limited to this. The behavioral information is only the information required by the state estimation unit 17 to estimate the sprung speed or relative speed; unnecessary information may be omitted. In addition, the behavioral information is not limited to the above-described behavioral information; for example, various other information may be added to improve the estimation accuracy of sprung speed, etc.

[0037] The state estimation unit 17 receives various behavioral information calculated based on CAN signals as input signals from the vehicle state calculation unit 16. The state estimation unit 17 includes a neural network that learns to estimate sprung speed or relative speed within a predetermined frequency band where high-frequency noise reduction or low-frequency phase shift reduction is applied to the input signals from the vehicle state calculation unit 16. The weight parameters of the learned neural network are stored in the weight parameter storage unit 14. Therefore, the neural network of the state estimation unit 17 is constructed using the weight parameters stored in the weight parameter storage unit 14. The neural network of the state estimation unit 17 outputs sprung speed and relative speed in response to the input signals from the vehicle state calculation unit 16.

[0038] Furthermore, in this embodiment, CAN is used as an example of an in-vehicle network for explanation, but other in-vehicle networks can also be used. Examples of in-vehicle networks include CAN FD (CAN with Flexible Data rate), FlexRay (vehicle communication protocol), and in-vehicle Ethernet.

[0039] The state estimation unit 17 reads the weight parameters stored in the weight parameter storage unit 14 via the data reading unit 15. Based on the behavior information and weight parameters calculated by the vehicle state calculation unit 16, the state estimation unit 17 estimates the data related to suspension control required by the control value calculation unit 20, specifically, it estimates the instantaneous values ​​(sprung speed, relative speed) related to suspension control.

[0040] The control value calculation unit 20 is a control unit that controls the variable damper 7 based on the output from the state estimation unit 17. Based on the instantaneous values ​​input from the state estimation unit 17, the control value calculation unit 20 calculates suspension control values ​​(e.g., target damping force) for controlling the damping force of the variable damper 7 in the suspension device 5. Specifically, the control value calculation unit 20 calculates suspension control values ​​for improving vehicle ride comfort, for example, based on bilinear optimal control, sky-hook control, H∞ control, etc. Based on the suspension control values, the control value calculation unit 20 outputs a command current as a control signal to the variable damping force actuator 8 of the variable damper 7. Furthermore, the control value calculation unit 20 is not limited to calculating vehicle ride comfort; it can also calculate suspension control values ​​for improving handling stability.

[0041] like Figure 2As shown, in this embodiment, the state estimation unit 17 estimates the data (sprung speed, relative speed) related to suspension control for each of the four wheels by referring to behavioral information such as wheel speed, front-rear acceleration, lateral acceleration, steering angle, and yaw rate in the data constantly transmitted to the CAN 10. Then, this estimation result is forwarded to the subsequent control value calculation unit 20. At this time, the state estimation unit 17 estimates the sensor data (vehicle state quantities) related to suspension control for each of the four wheels. The state estimation unit 17 is, for example, composed of a neural network. The weight parameters used in the neural network are weight parameters pre-determined for the vehicle through machine learning. Therefore, the state estimation unit 17 can also estimate the sprung speed and relative speed as vehicle state quantities, reflecting the overall rigidity characteristics of the vehicle. As a result, the control value calculation unit 20 controls the variable damper 7 based on the vehicle state quantities appropriately estimated by the state estimation unit 17. Therefore, the suspension device 5 can be controlled according to the vehicle's behavior, improving the accuracy of suspension control.

[0042] Next, refer to Figures 2 to 5 The structure and learning method of the neural network of the state estimation unit 17 will be explained.

[0043] The state estimation unit 17 is an artificial intelligence (AI) system with a fully learned neural network. Weight parameters and behavioral information are input to the state estimation unit 17. The neural network of the state estimation unit 17 is constructed using the weight parameters. For example... Figure 2 As shown, the state estimation unit 17 estimates sensor data such as sprung speed and relative speed by inputting behavioral information into a neural network, and then outputs the data. The state estimation unit 17 outputs the sprung speed and relative speed, which are vehicle state quantities, to the subsequent control value calculation unit 20. The neural network of the state estimation unit 17 is configured in the same way as the neural network of the vehicle behavior estimation unit disclosed in Japanese Patent Application Publication No. 2022-191913.

[0044] As an example, refer to Figure 4 The specific structure of the state estimation unit 17 will be explained. For example... Figure 4As shown, the state estimation unit 17 is, for example, a neural network. The neural network is a hierarchical neural network with a three-layer structure, consisting of input layer (number of elements i) 101, hidden layer (number of elements j) 102, and output layer (number of elements k) 103, which are layered together. Each element of the input layer 101 and each element of the hidden layer 102 are combined with weights W1ij (i=1~I, j=1~J), and each element of the hidden layer 102 and each element of the output layer 103 are combined with weights W2jk (j=1~J, k=1~K). The information about these weights (hereinafter referred to as weight parameters) is represented by the determinant of weights W1ij and W2jk. The weight parameters are pre-calculated through machine learning and stored in the weight parameter storage unit 14. Furthermore, this example shows a simple, fully connected neural network with a single hidden layer 102, but it is not limited to this. For example, a neural network with two or more hidden layers 102 may also be shown.

[0045] like Figure 4 As shown, time-series data, such as the CAN signal (wheel speed, front-rear acceleration, lateral acceleration, yaw rate, etc.) of the left front wheel, is input to the input layer 101 of the neural network. Instantaneous values ​​of, for example, the sprung speed and relative speed of the suspension device 5 installed on the left front wheel of the vehicle are output to the output layer 103. The number of components in the hidden layer 102 is generally determined based on the number of components in the input layer 101 and the output layer 103, and is set to maximize the accuracy of the state estimation based on the neural network. The number of components in the output layer 103 is determined by the output specifications of the state estimation.

[0046] Figure 3 The learning method of the neural network in state estimation unit 17 is shown. In machine learning of neural networks, the learning method will be achieved through... Figure 3 The data generated by the method shown is provided as teacher data. First, CAN signals (wheel speed, front-to-rear acceleration, lateral acceleration, steering angle, yaw rate, etc.) and vertical acceleration of the four wheels are measured when the vehicle body vibrates in a real vehicle. Then, the vertical acceleration is integrated on the frequency axis to calculate the sprung speed and relative speed. The calculated sprung speed and relative speed are used together with the measured CAN signals to create teacher data. Using this teacher data, machine learning of a neural network is performed. In the machine learning of the neural network, the weight parameters are adjusted to learn the correlation between the calculated sprung speed and relative speed data and the CAN signal data. The weight parameters obtained as a learning result are stored in the weight parameter storage unit 14.

[0047] Here, the characteristics of the frequency axis integral are shown. Figure 5 .like Figure 5As shown by the dashed line, in existing real-time integral filters, to reduce the effects of gradients and other factors, the gain at frequencies lower than the control band is reduced relative to the ideal integral characteristic. Furthermore, it is known that due to this effect, the phase near the spring resonant frequency, which was originally intended to be controlled, is advanced.

[0048] In contrast, in the characteristics of frequency-axis integration, the control band becomes the frequency-axis integration range. Here, the control band is a defined band where high-frequency noise reduction or low-frequency phase shift reduction is applied. The control band is, for example, around 0.8~10Hz, including the spring resonant frequency around 1~2Hz. Figure 5 As shown by the double-dotted line, it can be confirmed that in the characteristics of frequency axis integration, compared with the characteristics of an integral filter, the gain is significantly reduced outside the control band, and there is no phase shift. By using such a neural network that has been learned using teacher data, it is possible to take into account both the tilt of sensors such as gradients and the ride comfort on a flat road.

[0049] The neural network of the state estimation unit 17 can improve the estimation accuracy for time series data by taking the measurement results of a certain amount of past time as input. At this time, the time width of the certain amount of past time can vary according to the frequency of the controlled object (the cutoff frequency of the frequency axis integral), thereby balancing computational resources and performance.

[0050] Furthermore, in this embodiment, the state estimation unit 17 performs machine learning on the correlation between the CAN signal acquired by the data acquisition vehicle and the data of the sprung speed and relative speed, but the present invention is not limited thereto. For example, a vehicle model corresponding to the vehicle equipped with the state estimation unit 17 may be constructed, and the state estimation unit 17 may perform machine learning based on the data obtained by simulating the vehicle model.

[0051] Next, refer to Figures 6 to 8 Specific examples will be given regarding the state estimation results of the state estimation unit 17 and the improvement effect of the vehicle control device 1 equipped with the state estimation unit 17 on the ride comfort.

[0052] First, refer to Figure 6 The state estimation results obtained using the neural network of state estimation unit 17 will be explained. Figure 6 The solid line in the figure represents the frequency characteristics of the gain and phase of the transfer function when the transfer function between the sprung speed and the vertical acceleration estimated by the neural network (state estimation unit 17) of the first embodiment is calculated. At this time, the neural network of the state estimation unit 17 learns the sprung speed calculated by integrating the frequency axis as teacher data.

[0053] on the other hand, Figure 6The double-dotted lines in the diagram represent the frequency characteristics of the gain and phase of the transfer function when calculating the transfer function between the spring velocity and the vertical acceleration derived using a neural network based on existing technology. Here, the spring velocity, processed by integral filtering and learned by the neural network based on existing technology, is used as the teacher data.

[0054] like Figure 6 As shown, in existing neural networks, like the characteristics of an integral filter, the phase leads by -90° (ideal integral characteristic) near the spring resonant frequency. In contrast, the neural network of the state estimation unit 17 learns the spring speed calculated by integrating the frequency axis, thus confirming that the phase does not lead near the spring resonant frequency.

[0055] Next, refer to Figure 7 The state estimation results of the neural network using the state estimation unit 17 when driving on a gradient road surface are explained. Figure 7 The solid line in the figure represents the time variation of the estimated spring speed based on the neural network (state estimation unit 17) of the present invention. Figure 7 The double-dotted line in the figure represents the time-varying estimated spring speed based on a neural network using existing technology.

[0056] like Figure 7 As shown, in the prior art, neural networks learn the sprung speed processed by integral filtering as teacher data. Therefore, in the prior art, the estimated result varies greatly with the timing of gradient (along) changes. In contrast, the neural network (state estimation unit 17) of the present invention learns the sprung speed calculated by integrating along the frequency axis. Therefore, in the neural network of the present invention, it can be confirmed that even with the timing of gradient changes, the estimated value does not change significantly.

[0057] Next, refer to Figure 8 The effect of improving ride comfort when using the estimated sprung speed of the state estimation unit 17 to control the vehicle will be explained. Figure 8 The solid lines in the figure represent the time variations of sprung displacement and sprung jerk when suspension control is performed using the estimated sprung speed based on the neural network of the state estimation unit 17. Figure 8 The double-dotted lines in the figure represent the time-varying sprung displacement and jerk when using estimated sprung speeds based on existing neural networks for suspension control.

[0058] like Figure 8As shown, in existing technologies, even low-frequency road surfaces can generate significant jerk. In contrast, the neural network of this invention significantly reduces high-frequency gain through frequency axis integration. Therefore, in control using this invention, by further reducing high-frequency gain, unnecessary high-frequency control caused by noise and other factors can be reduced. Consequently, in control using this invention, a reduction in jerk can be confirmed.

[0059] Thus, the vehicle control device 1 of this embodiment includes: a variable damper 7 (relative displacement suppression device) disposed between the vehicle body 2 and the wheels 3 to suppress force changes in relative displacement between the vehicle body 2 and the wheels 3; a vehicle state calculation unit 16 mounted on the vehicle to calculate the vehicle's state (behavior information); a state estimation unit 17 that, based on a neural network that has learned the estimated sprung speed or relative speed in a predetermined frequency band where high-frequency noise reduction or low-frequency phase shift reduction has been implemented on the input signal from the vehicle state calculation unit 16, outputs the sprung speed or relative speed for the input signal from the vehicle state calculation unit 16; and a control value calculation unit 20 (control unit) that controls the variable damper 7 based on the output from the state driving unit 17.

[0060] At this time, the state estimation unit 17 can perform appropriate state estimation even when the sensor is tilted, such as on a gradient road surface (slope, incline, etc.). As a result, the vehicle control device 1 can reduce unnecessary control caused by errors in state estimation, thus improving vehicle ride comfort. Furthermore, the phase shift of the sprung speed and relative speed estimated by the state estimation unit 17 near the sprung resonant frequency is reduced. Therefore, the vehicle control device 1 is not limited to gradient roads, but can also improve ride comfort on flat roads.

[0061] then, Figures 9 to 11 This indicates the second embodiment. The second embodiment is characterized in that the vehicle status calculation unit has a first detection unit mounted on the vehicle using CAN signals and a second detection unit different from the first detection unit described above. Furthermore, in the second embodiment, the same reference numerals are used to label the same components as in the first embodiment, and their descriptions are omitted.

[0062] Figure 9 This refers to the vehicle control device 31 according to the second embodiment. The vehicle control device 31 consists of a suspension device 5 and a controller 32.

[0063] The controller 32 in the second embodiment constitutes a control device. The controller 32, as a control device for controlling the damping characteristics of the variable damper 7, is, for example, composed of a microcomputer. The controller 32 in the second embodiment is configured similarly to the controller 11 in the first embodiment. The controller 32 acquires behavioral information data from the CAN 10 and the vertical acceleration sensor 30. This behavioral information includes, for example, front-rear acceleration (front-rear G), lateral acceleration (lateral G), steering angle, yaw rate, wheel speed, and vertical acceleration. Therefore, this behavioral information is input to the controller 32. The output side of the controller 32 is connected to the variable damping force actuator 8 of the variable damper 7.

[0064] The controller 32 has a storage unit 33 composed of ROM, RAM, non-volatile memory, etc. The storage unit 33 of the controller 32 stores various programs, information (vehicle information), data, etc., used to control the variable damper 7. Based on the behavioral information, the controller 32 estimates the sprung speed and relative speed, which are vehicle state quantities. Based on the estimated vehicle state quantities, the controller 32 calculates the force that should be generated by the variable damper 7 (force generating mechanism) of the suspension device 5, and outputs this control signal (command current) to the damping force variable actuator 8 of the suspension device 5.

[0065] like Figure 9 As shown, the controller 32 includes a damper control unit 34, a weight parameter storage unit 14, and a data readout unit 15. The damper control unit 34 includes a vehicle state calculation unit 35, a state estimation unit 36, and a control value calculation unit 20.

[0066] The vehicle state calculation unit 35 includes a first detection unit 35A mounted on the vehicle and using CAN signals, and a second detection unit 35B, which is different from the first detection unit 35A. The first detection unit 35A acquires data related to the vehicle state during driving from communication data transmitted within the vehicle via CAN 10, and calculates, for example, front-rear acceleration (front-rear G), lateral acceleration (lateral G), steering angle, yaw rate, wheel speed, etc., based on the acquired data. The second detection unit 35B calculates the sprung (body 2) vertical acceleration based on the vertical acceleration sensor 30. The vehicle state calculation unit 35 outputs vehicle behavior information, including information acquired from the CAN signals (front-rear acceleration, lateral acceleration, steering angle, yaw rate, wheel speed, etc.) and information acquired from the vertical acceleration sensor 30 (sprung vertical acceleration), to the state estimation unit 36.

[0067] like Figure 10As shown, the state estimation unit 36 ​​receives various behavioral information calculated based on the CAN signal and the detection signal from the vertical acceleration sensor 30 as input signals from the vehicle state calculation unit 35. The state estimation unit 36 ​​includes a neural network that learns to estimate sprung speed or relative speed within a predetermined frequency band where high-frequency noise reduction or low-frequency phase shift reduction is applied to the input signal from the vehicle state calculation unit 35. The weight parameters of the learned neural network are stored in the weight parameter storage unit 14. Therefore, the neural network of the state estimation unit 36 ​​is constructed using the weight parameters stored in the weight parameter storage unit 14. The neural network of the state estimation unit 36 ​​outputs sprung speed and relative speed in response to the input signal from the vehicle state calculation unit 35.

[0068] The neural network of the state estimation unit 36 ​​is configured in the same way as the neural network of the state estimation unit 17 in the first embodiment. However, in the machine learning of the neural network, the data shown below is provided as teacher data. Figure 11 The learning method of the neural network in the state estimation unit 36 ​​is shown. In the machine learning of the neural network in the state estimation unit 36, firstly, CAN signals (wheel speed, front-rear acceleration, lateral acceleration, steering angle, yaw rate, etc.) are measured when the vehicle body vibrates in a real vehicle, and the vertical acceleration of the four wheels is measured using the vertical acceleration sensor 30. Then, the vertical acceleration is integrated on the frequency axis to calculate the sprung speed and relative speed. The calculated sprung speed and relative speed, together with the measured CAN signals and the vertical acceleration measured by the vertical acceleration sensor 30, are used to create teacher data. Using this teacher data, machine learning of the neural network is performed. In the machine learning of the neural network, the weight parameters are adjusted to learn the correlation between the calculated sprung speed and relative speed data and the CAN signals and vertical acceleration data. The weight parameters obtained as a learning result are stored in the weight parameter storage unit 14.

[0069] Thus, in the second embodiment, the same effect as in the first embodiment can be obtained. In the second embodiment, the vehicle state calculation unit 35 has a first detection unit 35A mounted on the vehicle and using CAN signals, and a second detection unit 35B, which is different from the first detection unit 35A. At this time, the state estimation unit 36 ​​learns the relationship between sprung speed and relative speed based on the input signals (wheel speed, front-rear G, lateral G, steering angle, yaw rate, etc.) from the first detection unit 35A and the vertical acceleration from the second detection unit 35B during neural network learning, and calculates the future sprung speed and relative speed based on the vertical acceleration from the vehicle state calculation unit 35.

[0070] In the state estimation unit 17 of the first embodiment, CAN signals (wheel speed, front and rear acceleration, lateral acceleration, steering angle, yaw rate, etc.) commonly used in sensorless semi-active damping systems are learned as input signals for the neural network. In contrast, the second embodiment is applied to a system with sensors including an up-and-down acceleration sensor 30. In this case, differentiation and integration processing are not used, so the state estimation unit 36 ​​of the second embodiment, like the state estimation unit 17 of the first embodiment, reduces the phase shift of sprung speed and relative speed near the sprung resonant frequency. Furthermore, the state estimation unit 36 ​​can perform appropriate state estimation even when the sensor is tilted, such as on a gradient road surface (slope, incline, etc.). Therefore, in the vehicle control device 31 of the second embodiment, ride comfort on gradient roads, flat roads, etc., can also be improved.

[0071] Furthermore, in the second embodiment, the vehicle state calculation unit 35 calculates the data (vertical acceleration) required for estimating sprung speed and relative speed based on the detection signal from the vertical acceleration sensor 30, but the present invention is not limited thereto. For example, the vehicle state calculation unit may also calculate the data required for estimating sprung speed and relative speed based on the detection signal from the vehicle height sensor. Alternatively, the vehicle state calculation unit may calculate the data required for estimating sprung speed and relative speed based on the detection signals from both the vertical acceleration sensor and the vehicle height sensor.

[0072] In the second embodiment, the vehicle state calculation unit 35 calculates behavioral information data based on the CAN signal in addition to the detection signal from the vertical acceleration sensor 30, but the present invention is not limited thereto. When the state estimation unit 36 ​​estimates the sprung speed and relative speed based on the detection signal from the vertical acceleration sensor 30, the vehicle state calculation unit 35 only needs to input the detection signal from the vertical acceleration sensor 30, and the input of the CAN signal can also be omitted.

[0073] then, Figures 12 to 14 This indicates the third embodiment. The third embodiment is characterized in that the state estimation unit predicts future input signals based on input signals from the vehicle state calculation unit, and outputs sprung speed or relative speed for these future input signals, thereby performing feedforward control. Furthermore, in the third embodiment, the same reference numerals are used to denote the same components as in the first embodiment described above, and their descriptions are omitted.

[0074] Camera 40 constitutes a road condition measurement unit installed at the front of the vehicle body 2. Camera 40 measures and detects the road condition in front of the vehicle (specifically, including the distance and angle to the road surface of the object being detected, and the image position and distance). Camera 40 is, for example, a stereo camera. Camera 40 detects the road condition, including the distance and angle to the object being photographed (the road surface in front of the vehicle), by capturing a pair of left and right images. Therefore, the preview image of the front of the vehicle captured by camera 40 (i.e., road surface preview information) is output to controller 42 as the detection result of the road condition measurement unit. In addition, the road condition measurement unit is not limited to camera 40 composed of a stereo camera; for example, it can be a camera composed of a combination of millimeter-wave radar and a monocular camera, or it can be composed of multiple millimeter-wave radars, etc.

[0075] Figure 12 This refers to the vehicle control device 41 according to the third embodiment. The vehicle control device 41 consists of a suspension device 5 and a controller 42.

[0076] The controller 42 in the third embodiment constitutes a control device. The controller 42, as a control device for controlling the damping characteristics of the variable damper 7, is, for example, composed of a microcomputer. The controller 42 in the third embodiment is configured similarly to the controller 11 in the first embodiment. The controller 42 acquires behavioral information data from the CAN 10 and the camera 40. This behavioral information includes, for example, front-rear acceleration (front-rear G), lateral acceleration (lateral G), steering angle, yaw rate, wheel speed, road surface information, etc. Therefore, this behavioral information is input to the controller 42. The output side of the controller 42 is connected to the variable damping force actuator 8 of the variable damper 7.

[0077] The controller 42 has a storage unit 43 composed of ROM, RAM, non-volatile memory, etc. The storage unit 43 of the controller 42 stores various programs, information (vehicle information), data, etc., for controlling the variable damper 7. Based on the behavioral information, the controller 42 estimates the sprung speed and relative speed, which are vehicle state quantities. Based on the estimated vehicle state quantities, the controller 42 calculates the force that should be generated by the variable damper 7 (force generating mechanism) of the suspension device 5, and outputs this control signal (command current) to the damping force variable actuator 8 of the suspension device 5.

[0078] like Figure 12 As shown, the controller 42 includes a damper control unit 44, a weight parameter storage unit 14, and a data readout unit 15. The damper control unit 44 includes a vehicle state calculation unit 45, a state estimation unit 46, and a control value calculation unit 20.

[0079] The vehicle state calculation unit 45 includes a first detection unit 45A mounted on the vehicle and using CAN signals, and a second detection unit 45B, which is different from the first detection unit 45A. The first detection unit 45A acquires data related to the vehicle state during driving from communication data transmitted within the vehicle via CAN 10, and calculates, based on the acquired data, parameters such as front-rear acceleration (front-rear G), lateral acceleration (lateral G), steering angle, yaw rate, and wheel speed. The second detection unit 45B calculates the road surface displacement in front of the vehicle (in the direction of travel) from the camera 40. The vehicle state calculation unit 45 outputs vehicle behavior information, including information acquired from the CAN signals (front-rear acceleration, lateral acceleration, steering angle, yaw rate, wheel speed, etc.) and information acquired from the camera 40 (road surface displacement), to the state estimation unit 46.

[0080] like Figure 13 As shown, the state estimation unit 46 receives various behavioral information calculated based on the CAN signal and the detection signal from the camera 40 as input signals from the vehicle state calculation unit 45. The state estimation unit 46 includes a neural network that learns to estimate sprung speed or relative speed within a predetermined frequency band where high-frequency noise reduction or low-frequency phase shift reduction is applied to the input signals from the vehicle state calculation unit 45. The weight parameters of the learned neural network are stored in the weight parameter storage unit 14. Therefore, the neural network of the state estimation unit 46 is constructed using the weight parameters stored in the weight parameter storage unit 14. The neural network of the state estimation unit 46 outputs sprung speed and relative speed for the input signals from the vehicle state calculation unit 45.

[0081] The neural network of the state estimation unit 46 is configured in the same way as the neural network of the state estimation unit 17 in the first embodiment. However, in the machine learning of the neural network, the data shown below is provided as teacher data. Figure 14 The learning method of the neural network in the state estimation unit 46 is shown. In the machine learning of the neural network in the state estimation unit 46, firstly, CAN signals (wheel speed, front-to-rear acceleration, lateral acceleration, steering angle, yaw rate, etc.) are measured when the vehicle body vibrates in the actual vehicle, and road surface displacement and vertical acceleration are measured using various sensors. Then, the vertical acceleration when the vehicle reaches the measured road surface displacement position is integrated on the frequency axis to calculate the sprung speed and relative speed. The calculated sprung speed and relative speed are combined with the measured CAN signals and the road surface displacement measured by the camera 40 to generate teacher data. Using this teacher data, machine learning of the neural network is performed. In the machine learning of the neural network, the weight parameters are adjusted to learn the correlation between the calculated sprung speed and relative speed data and the CAN signals and road surface displacement data. The weight parameters obtained as a learning result are stored in the weight parameter storage unit 14.

[0082] Thus, in the third embodiment, the same effect as in the first embodiment can be achieved. In the third embodiment, the state estimation unit 46 predicts future input signals based on input signals from the vehicle state calculation unit 45, and outputs sprung speed or relative speed for these future input signals, thereby enabling the vehicle control device 41 to perform feedforward control. In the first and second embodiments, although vehicle body state quantities (wheel speed, vertical acceleration, etc.) are used as inputs to the neural network, the inputs to the neural network in this invention are sufficient as long as the input signals that can be calculated numerically to become the sprung speed and relative speed output are not insufficient. Therefore, as shown in the third embodiment, even a preview control system that controls the suspension based on the actual road displacement of the vehicle before it passes through can be applied.

[0083] Furthermore, in the third embodiment, a case is illustrated where the road condition measurement unit is a camera 40 composed of a stereo camera. The present invention is not limited to this; the road condition measurement unit may be, for example, a LiDAR (Light Detection and Ranging) system, a structure combining a millimeter-wave radar and a monocular camera, or it may be composed of multiple millimeter-wave radars. Furthermore, the road condition measurement unit is not limited to camera 40; for example, it may obtain map information from a GPS-based server, or it may obtain information from other vehicles via vehicle-to-vehicle communication.

[0084] Furthermore, in the third embodiment, the vehicle state calculation unit 45 calculates behavioral information data based on the CAN signal in addition to the detection signal from the camera 40, but the present invention is not limited thereto. For example, if a camera and a vehicle speed sensor are connected to the vehicle state calculation unit, the vehicle state calculation unit does not need to input a CAN signal to the vehicle state calculation unit as long as it can calculate the information (road displacement and vehicle speed) required to estimate the sprung speed and relative speed.

[0085] then, Figure 15 This refers to the fourth embodiment. The fourth embodiment is characterized in that the state estimation unit obtains vehicle information from the position information recorded in the recording unit based on the current position information obtained from the position information acquisition unit. Furthermore, in the fourth embodiment, the same reference numerals are used to label the same components as in the second embodiment described above, and their descriptions are omitted.

[0086] The GPS receiver 50 is a location information acquisition unit that acquires the vehicle's location information. The GPS receiver 50 is installed on the vehicle body 2 and receives signals from GPS (Global Positioning System) satellites (hereinafter referred to as GPS signals). The GPS receiver 50 calculates the vehicle's current location information based on the GPS signals. The GPS receiver 50 outputs the current location information to the controller 52. Furthermore, the location information acquisition unit is not limited to using GPS signals. For example, the location information acquisition unit may also estimate the vehicle's current location information using technologies such as vehicle speed sensors, gyroscopes, and map mapping.

[0087] Figure 15 This refers to the vehicle control device 51 according to the fourth embodiment. The vehicle control device 51 consists of a suspension device 5 and a controller 52.

[0088] The controller 52 of the fourth embodiment constitutes a control device. The controller 52, as a control device for controlling the attenuation characteristics of the variable damper 7, is, for example, composed of a microcomputer. The controller 52 of the fourth embodiment is configured similarly to the controller 32 of the second embodiment. The controller 52 acquires behavioral information data from the CAN 10 and the vertical acceleration sensor 30. The controller 52 is connected to the GPS receiver 50. The output side of the controller 32 is connected to the variable actuator 8 for the attenuation force of the variable damper 7.

[0089] The controller 52 has a storage unit 53 composed of ROM, RAM, non-volatile memory, etc. The storage unit 53 of the controller 52 stores various programs, information (vehicle information), data, etc., used to control the variable damper 7. In addition, the storage unit 53 constitutes a recording unit for recording location information, etc. Therefore, the storage unit 53 records location information from the GPS receiver 50 and vehicle information related to that location information from the vehicle status calculation unit 35.

[0090] Based on behavioral information, controller 52 estimates the sprung speed and relative speed as vehicle state quantities. In addition, based on location information from GPS receiver 50, controller 52 estimates the sprung speed and relative speed as vehicle state quantities. Based on the estimated vehicle state quantities, controller 52 calculates the force that should be generated by the variable damper 7 (force generating mechanism) of suspension device 5, and outputs this control signal (command current) to the damping force variable actuator 8 of suspension device 5.

[0091] like Figure 15 As shown, the controller 52 includes a storage unit 53, a damper control unit 54, a weight parameter storage unit 14, and a data readout unit 15. The damper control unit 54 includes a vehicle state calculation unit 35, a state estimation unit 55, and a control value calculation unit 20.

[0092] The state estimation unit 55 is configured similarly to the state estimation unit 36 ​​in the second embodiment. The state estimation unit 55 receives various behavioral information calculated based on the CAN signal and the detection signal from the vertical acceleration sensor 30 as input signals from the vehicle state calculation unit 35. The state estimation unit 55 includes a neural network that learns to estimate sprung speed or relative speed within a predetermined frequency band where high-frequency noise reduction or low-frequency phase shift reduction is applied to the input signal from the vehicle state calculation unit 35. The weight parameters of the learned neural network are stored in the weight parameter storage unit 14. Therefore, the neural network of the state estimation unit 55 is configured using the weight parameters stored in the weight parameter storage unit 14. The neural network of the state estimation unit 55 outputs sprung speed and relative speed in response to the input signal from the vehicle state calculation unit 35.

[0093] Furthermore, the state estimation unit 55 obtains vehicle information from the position information recorded in the storage unit 53 (recording unit) based on the current position information from the GPS receiver 50. Specifically, the state estimation unit 55 determines the position information corresponding to the vehicle's current position in the position information recorded in the storage unit 53, and obtains behavioral information (rear acceleration, lateral acceleration, steering angle, yaw rate, wheel speed, sprung acceleration, etc.) as vehicle information corresponding to the current position. Therefore, even when a behavioral state based on the current position information is input, the neural network of the state estimation unit 55 can output sprung speed and relative speed for that input signal.

[0094] The control value calculation unit 20 calculates the suspension control value (e.g., target damping force) for controlling the damping force of the variable damper 7 of the suspension device 5 based on the instantaneous values ​​of the sprung speed and relative speed input from the state estimation unit 55.

[0095] Thus, in the fourth embodiment, the same effects as in the first and second embodiments can be achieved. In the fourth embodiment, the vehicle includes a GPS receiver 50 (position information acquisition unit) that acquires the vehicle's position information; and a storage unit 53 (recording unit) that records the position information from the GPS receiver 50 and vehicle information from a vehicle state calculation unit 35 that is related to the position information. Furthermore, the state estimation unit 55, based on the current position information from the GPS receiver 50, obtains the spring acceleration, etc., as vehicle information from the position information recorded in the storage unit 53. Therefore, even when a behavioral state based on the current position information is input, the neural network of the state estimation unit 55 can output the spring speed and relative speed for that input signal. Therefore, in the fourth embodiment, by using the position information from the GPS receiver 50, vehicle control can continue even if there is an anomaly in the vehicle state calculation unit 35. In addition, in the fourth embodiment, since the vehicle is controlled using the position information from the GPS receiver 50, control that predicts the vehicle state in advance can be achieved through feedforward control.

[0096] Furthermore, in the fourth embodiment, a case where vehicle control using location information is applied in the second embodiment is illustrated, but the present invention is not limited thereto. Vehicle control using location information can also be applied in the first and third embodiments.

[0097] In the fourth embodiment, location information and vehicle information are stored in the storage unit 53 of the controller 52, but the present invention is not limited thereto. The location information and vehicle information are stored in an external database such as the cloud, and can be downloaded from a cloud service (cloud) as needed via a communication network.

[0098] In the first embodiment, an example is given of the controller 11 acquiring vehicle operating status information, including wheel speed, via CAN 10, but the present invention is not limited thereto. For example, the controller 11 may also directly acquire detection values ​​from various sensors. Additionally, the controller 11 may also acquire behavioral information from other controllers, etc. This structure can be applied to the second to fourth embodiments.

[0099] In the above embodiments, the case of a variable damper 7 composed of a semi-active damper as the force generating mechanism has been described as an example. However, the present invention is not limited to this; an active damper (either an electric actuator or a hydraulic actuator) can also be used as the force generating mechanism. In the above embodiments, the case of a force generating mechanism that generates a force adjustable between the vehicle body 2 sides and the wheel 3 sides has been described as an example, using a variable damper 7 composed of a damping force adjustable hydraulic buffer. However, the present invention is not limited to this; for example, in addition to a hydraulic buffer, an air suspension, a stabilizer, an electromagnetic suspension, etc., can also be used as the force generating mechanism.

[0100] In the above embodiments, a vehicle control device for a four-wheeled vehicle has been described as an example. However, the present invention is not limited thereto, and can also be applied to two-wheeled vehicles, three-wheeled vehicles, and trucks, buses, etc., which are used as work vehicles or transport vehicles.

[0101] Furthermore, the present invention is not limited to the embodiments described above, and includes various modifications. For example, the embodiments described above are detailed for ease of understanding and explanation of the present invention, and are not limited to having all the structures described. In addition, a part of the structure of a certain embodiment can be replaced with the structure of another embodiment, and it is also possible to add the structure of another embodiment to the structure of a certain embodiment. Furthermore, for a part of the structure of each embodiment, other structures can be added, deleted, or replaced.

[0102] This application claims priority based on Japanese Patent Application No. 2023-142300, filed on September 1, 2023. The entire disclosure of Japanese Patent Application No. 2023-142300, filed on September 1, 2023, including the description, claims, drawings and abstract, is incorporated herein by reference in its entirety.

[0103] Explanation of reference numerals in the attached figures

[0104] 1, 31, 41, 51: Vehicle control device; 2: Vehicle body; 3: Wheel; 5: Suspension device; 7: Variable damper (relative displacement suppression device); 8: Variable damping force actuator; 11, 32, 42, 52: Controller; 12, 33, 43, 53: Storage unit (recording unit); 16, 35, 45: Vehicle status calculation unit; 17, 36, 46, 55: Status estimation unit; 20: Control value calculation unit (control unit); 35A, 45A: First detection unit; 35B, 45B: Second detection unit; 50: GPS receiver (location information acquisition unit).

Claims

1. A vehicle control device, comprising: A relative displacement suppression device is installed between the vehicle body and the wheels to suppress changes in force that cause relative displacement between the vehicle body and the wheels. A vehicle status calculation unit, mounted on the vehicle, calculates the status of the vehicle; The state estimation unit outputs sprung speed or relative speed based on the input signal from the vehicle state calculation unit using a neural network. The neural network learns to estimate the sprung speed or relative speed in a specified frequency band where high-frequency noise is reduced or low-frequency phase shift is reduced in the input signal from the vehicle state calculation unit. as well as The control unit controls the relative displacement suppression device based on the output from the state estimation unit.

2. The vehicle control device according to claim 1, wherein, The state estimation unit predicts future input signals based on input signals from the vehicle state calculation unit, and outputs sprung speed or relative speed for the future input signals, thereby performing feedforward control.

3. The vehicle control device according to claim 1, wherein, The vehicle state calculation unit has: The first detection unit, mounted on the vehicle, uses CAN signals; The second testing department is different from the first testing department.

4. The vehicle control device according to claim 3, wherein, In the learning of the neural network, the state estimation unit learns the relationship between the input signal from the first detection unit and the sprung speed or relative speed calculated based on the vertical acceleration or vehicle height value from the second detection unit, and calculates the future sprung speed or relative speed based on the vertical acceleration or vehicle height value from the vehicle state calculation unit.

5. The vehicle control device according to claim 4, wherein, The vehicle has the following features: The location information acquisition unit acquires the location information of the vehicle; and The recording unit records location information from the location information acquisition unit and vehicle information related to the location information from the vehicle state calculation unit. The state estimation unit obtains the vehicle information from the location information recorded in the recording unit based on the current location information from the location information acquisition unit.

Images