Suspension control device and suspension control method

Abstract

The suspension control device of the present invention is connected to a sensor that acquires information about a vehicle or information about the surroundings of the vehicle, and calculates a suspension control value for controlling the suspension of the vehicle based on the information acquired by the sensor. The device includes: an arrival vibration generation unit that generates an arrival vibration arriving at the vehicle based on the information acquired by the sensor; a sampling processing unit that changes the sampling time width of the arrival vibration waveform based on the frequency of the arrival vibration generated by the arrival vibration generation unit and samples the arrival vibration waveform; and a control value calculation unit that calculates the suspension control value based on the sampled value of the vibration waveform sampled by the sampling processing unit.

Description

Technical Field

[0001] This invention relates to a suspension control device and a suspension control method. Background Technology

[0002] A suspension control device is known that controls the vehicle's attitude by changing the stiffness and characteristics of the suspension according to road conditions and driving conditions.

[0003] Patent document 1 describes a device that continuously acquires high-density road displacement information of the road surface in front of a vehicle, reduces the probability of road height detection omissions, thereby obtaining high-precision road displacement information, and provides the acquired road displacement information to the vehicle's suspension system for active suspension pre-aiming control.

[0004] Existing technical documents

[0005] Patent documents

[0006] Patent Document 1: International Publication No. WO2017 / 169365 Summary of the Invention

[0007] The technical problem that the invention aims to solve

[0008] The device disclosed in Patent Document 1 cannot accurately grasp the vibration waveform based on the frequency of the arrival vibration of the vehicle, and may not be able to optimally control the suspension.

[0009] Technical solutions to solve technical problems

[0010] The suspension control device of the present invention is connected to a sensor for acquiring information about a vehicle or the area surrounding the vehicle, and calculates a suspension control value for controlling the suspension of the vehicle based on the information acquired by the sensor. The device includes: an arrival vibration generation unit that generates an arrival vibration of the vehicle based on the information acquired by the sensor; a sampling processing unit that changes the sampling time width of the vibration waveform of the arrival vibration generated by the arrival vibration generation unit and samples the vibration waveform of the arrival vibration according to the frequency of the arrival vibration generated by the arrival vibration generation unit; and a control value calculation unit that calculates the suspension control value based on the sampled value of the vibration waveform sampled by the sampling processing unit.

[0011] The suspension control method of the present invention uses a sensor for acquiring information about the vehicle or the area surrounding the vehicle to control the vehicle's suspension. Based on the information acquired by the sensor, an arrival vibration is generated for the vehicle. According to the frequency of the generated arrival vibration, the sampling time width of the vibration waveform of the arrival vibration is changed and the vibration waveform of the arrival vibration is sampled. A suspension control value is calculated based on the sampled value of the vibration waveform, and the suspension is controlled based on the calculated suspension control value.

[0012] Invention Effects

[0013] According to the present invention, the vibration waveform can be grasped regardless of the frequency of the arrival vibration of the vehicle, thereby improving the controllability of the suspension. Attached Figure Description

[0014] Figure 1 This is a diagram showing the suspension control model.

[0015] Figure 2 This is a block diagram of the electronic control device.

[0016] Figure 3 (A) Figure 3 (B) Figure 3 (C) is a graph showing information obtained from vehicle sensors, etc., and the frequency analysis of the vibrations.

[0017] Figure 4 (A) Figure 4 (B) Figure 4 (C) is a diagram showing the road surface and vibration amplitude.

[0018] Figure 5 (A) Figure 5 (B) Figure 5 (C) is a graph showing the relationship between the vibration waveform value and the suspension control value.

[0019] Figure 6 (A) Figure 6 (B) Figure 6 (C) is a graph showing the relationship between the vibration waveform and the time width.

[0020] Figure 7 (A) Figure 7 (B) is a flowchart illustrating the processing of the electronic control device of Embodiment 1 and a diagram illustrating the input specifications.

[0021] Figure 8 (A) Figure 8 (B) is a diagram showing an example of the control value calculation unit and the storage unit of Embodiment 2.

[0022] Figure 9 This is a structural diagram of the learning system for Implementation Method 2, which uses teacher data to generate neural networks.

[0023] Figure 10 (A) Figure 10 (B) is a flowchart illustrating the processing of the electronic control device of Embodiment 2 and a diagram illustrating the input specifications.

[0024] Figure 11 This is a diagram showing an example of the control value calculation unit in Embodiment 3.

[0025] Figure 12 (A) Figure 12 (B) is a modified example of the control value calculation unit of Embodiment 3, and a graph showing the vibration waveform predicted based on the measured value.

[0026] Figure 13 This is a structural diagram of the learning system for implementation method 3, which uses teacher data to generate neural networks.

[0027] Figure 14 (A) Figure 14 (B) is a flowchart illustrating the processing of the electronic control device of Embodiment 3 and a diagram illustrating the input specifications.

[0028] Figure 15 (A) Figure 15 (B) Figure 15 (C) is a diagram showing the suspension control device of embodiment 4. Detailed Implementation

[0029] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. The following description and drawings are examples for illustrating the present invention, and appropriate omissions and simplifications have been made for clarity. The present invention may also be implemented in various other forms. Unless otherwise specified, the structural elements may be singular or plural.

[0030] To facilitate understanding of the present invention, the positions, dimensions, shapes, and extents of the constituent elements shown in the accompanying drawings may not represent their actual positions, dimensions, shapes, or extents. Therefore, the present invention is not necessarily limited to the positions, dimensions, shapes, and extents disclosed in the accompanying drawings.

[0031] When multiple structural elements with the same or identical functions exist, different subscripts are sometimes added to the same label for description. However, when it is not necessary to distinguish between these multiple structural elements, the subscripts are sometimes omitted for description.

[0032] Furthermore, in the following description, processing is sometimes described as being performed by executing a program. However, since the program is executed by a processor (e.g., CPU, GPU), the processor can be the subject of the processing, performing the specified processing while appropriately utilizing storage resources (e.g., memory) and / or interface devices (e.g., communication ports). Similarly, the subject of processing performed by executing a program can also be a controller, device, system, computer, or node that has a processor. The subject of processing performed by executing a program can be any arithmetic unit, but may also include dedicated circuitry (e.g., FPGA and ASIC) that performs specific processing.

[0033] The program can be installed from a program source onto a device such as a computer. The program source can be, for example, a program distribution server or a computer-readable storage medium. If the program source is a program distribution server, the program distribution server contains a processor and storage resources for storing the programs to be distributed. The processor of the program distribution server can also distribute the programs to other computers. Furthermore, in the following description, two or more programs can be implemented as one program, or one program can be implemented as two or more programs.

[0034] [Implementation Method 1]

[0035] Figure 1 This is a diagram illustrating a suspension control model 1 of the suspension control device according to Embodiment 1 of the present invention. This suspension control model 1 is a simplified model focusing on one wheel of the vehicle around the suspension.

[0036] Vibrations from the road surface S are transmitted to the tire 104, from the wheel 102 to the damping force control suspension 103, and then to the vehicle body 101. The vehicle body 101 is equipped with a vehicle sensor 105 for acquiring vehicle information such as vehicle speed, and a road surface measurement sensor 106 for acquiring information about the vehicle's surroundings. The road surface measurement sensor 106 is, for example, a stereo camera.

[0037] Information acquired by vehicle sensor 105 and road surface measurement sensor 106 is input to electronic control unit 107. Electronic control unit 107 is equivalent to the suspension control unit of this embodiment. For example, it uses parallax information from a stereo camera image of the road surface in front of the vehicle to acquire the road surface displacement in front of the vehicle. Then, based on the road surface displacement and vehicle speed, it predicts the arrival vibration of the approaching vehicle. Details will be described later. Electronic control unit 107 changes the sampling time width of the vibration waveform of the approaching vibration according to the predicted frequency of the arrival vibration, calculates the suspension control value based on the sampled values ​​of the vibration waveform, and controls the suspension 103 based on the calculated suspension control value.

[0038] exist Figure 1The model is shown with a focus on one wheel of the vehicle, but the electronic control unit 107 also controls the suspension 103 for the other wheels with suspension 103 based on the calculated suspension control values.

[0039] Figure 2 This is a block diagram of the electronic control device 107.

[0040] The electronic control device 107 includes a vibration generation unit 109, a frequency determination unit 110, a sampling processing unit 111, a storage unit 112, and a control value calculation unit 113.

[0041] The arrival vibration generation unit 109 calculates the road surface displacement in front of the vehicle based on the data received from the road surface measurement sensor 106. Then, based on the calculated road surface displacement and the vehicle speed detected from the vehicle sensor 105, it generates the predicted arrival vibration of the arriving vehicle as time series data.

[0042] The frequency determination unit 110 determines the frequency of the arriving vibration generated by the arriving vibration generation unit 109. For example, it performs frequency analysis on the arriving vibration and determines whether the peak frequency is higher or lower than a reference nHz.

[0043] The sampling processing unit 111 samples the vibration waveform of the arriving vibration generated by the arriving vibration generation unit 109. At this time, the sampling time width of the vibration waveform is changed according to the frequency of the arriving vibration determined by the frequency determination unit 110. For example, when the arriving vibration is low-frequency, the time width is set wider and the sampling interval is set sparsely; when the arriving vibration is high-frequency, the time width is set narrower and the sampling interval is set more closely. The sampled vibration waveform values ​​are input to the control value calculation unit 113.

[0044] The control value calculation unit 113 applies parameters pre-stored in the storage unit 112 as needed and calculates suspension control values ​​based on sampled values ​​of the vibration waveform transmitted from the sampling processing unit 111. Furthermore, although the parameters are described using the case where they are stored in the storage unit 112 as an example, parameters can also be obtained from outside the electronic control device 107 via a network or the like and applied. In addition, the parameters could be, for example, the weight parameters of a neural network, which will be described later.

[0045] Figure 3 (A) Figure 3 (B) Figure 3 (C) is a graph showing the information obtained from vehicle sensor 105 and the frequency analysis of the vibration. Figure 3 (A) is a diagram showing the information obtained from vehicle sensor 105 and road surface measurement sensor 106. Figure 3 (B) Figure 3(C) is a graph representing the frequency analysis of the arriving vibration.

[0046] exist Figure 3 In the example shown in (A), an acceleration sensor 204 is mounted on vehicle 201 as vehicle sensor 105, and a camera 205 is mounted on vehicle 201 as road surface measurement sensor 106. The suspension 103 includes front wheel suspension 202 and rear wheel suspension 203.

[0047] Examples include a camera image 206 observed by camera 205 when vehicle 201 is traveling on a road surface S with protrusions, and time-series data, i.e., sensor data 207, output by an acceleration sensor 204, for example, mounted on a spring of the front wheel suspension 202.

[0048] In the electronic control device 107, if the camera 205 is a stereo camera, distance information is derived from the parallax information of each camera image. Based on the vehicle speed, the timing and magnitude of the arrival vibration of the approaching vehicle 201 can be estimated and predicted, and this arrival vibration is generated as time-series data. Alternatively, if the camera 205 is a monocular camera, the timing and magnitude of the arrival vibration of the approaching vehicle can be estimated and predicted based on images from different frames in the time direction, and this arrival vibration is generated as time-series data. Furthermore, in Figure 3 In (A), although the case of a straight path in front of the vehicle is illustrated, the direction of travel can also be calculated based on the vehicle's steering angle information, and the arriving vibration can be generated as time series data in various driving paths such as curved paths.

[0049] The frequency of the vibration is determined by frequency analysis performed by the frequency determination unit 110. Figure 3 (B) and Figure 3 (C) is a graph illustrating this frequency analysis. The horizontal axis represents frequency, and the vertical axis represents the magnitude of the frequency components. Figure 3 In (B), the peak frequency is below the reference nHz. In this case, the frequency determination unit 110 determines that the arriving vibration is low-frequency. Figure 3 In (C), the peak frequency is higher than the reference nHz. In this case, the frequency determination unit 110 determines that the arriving vibration is high frequency.

[0050] Figure 4 (A) Figure 4 (B) Figure 4 (C) is a diagram showing the road surface and vibration amplitude.

[0051] Figure 4 (A) shows an example of a vehicle 201 traveling at 50 km / h on a road surface S1 with a convex shape having a depth W1. Figure 4(B) shows an example of a vehicle 201 traveling at 100 km / h on a road surface S2 with a convex shape having a depth W2. The convex shape with depth W2 has twice the depth (width) of the convex shape with depth W1, and the same height.

[0052] Figure 4 (C) is a graph showing the vibration amplitude received by vehicle 201. Figure 4 (A) Figure 4 In case (B), if Figure 4 If the vibration amplitude 401 experienced by vehicle 201 is represented by a time axis as shown in (C), then the vibration amplitude 401 experienced by vehicle 201 is the same. In other words, from the perspective of vibration, even if the displacement of the road surface ahead can be known, the vibration reaching vehicle 201 will vary depending on the speed of vehicle 201.

[0053] Figure 5 (A) Figure 5 (B) Figure 5 (C) is a graph showing the relationship between the sampled values ​​of the vibration waveform and the suspension control values. Figure 5 (A) shows the control value calculation unit 113. Figure 5 (B) shows the sampled values ​​of the vibration waveform that arrived at the vibration. Figure 5 (C) shows the suspension control values.

[0054] like Figure 5 As shown in (A), the control value calculation unit 113 calculates the suspension control value based on the sampled values ​​x1 to xi of the input vibration waveform.

[0055] like Figure 5 As shown in (B), the vibration waveform of the arriving vibration is sampled by the sampling processing unit 111, thereby converting the vibration waveform of the arriving vibration into discrete time series data, which is then input to the control value calculation unit 113. The sampling processing unit 111 defines a window width 503 (hereinafter, sometimes referred to as the time width) with a certain time width on the time axis, and samples i sampled values ​​(x1 to xi) from the vibration waveform within the window width 503. Here, without changing the number i of sampled values ​​in the window width 503, a time width that can well capture the characteristics of the time series data is set. As described later, the sampling processing unit 111 changes the time width for sampling the vibration waveform of the arriving vibration according to the frequency of the arriving vibration.

[0056] Then, the control value calculation unit 113 calculates the suspension control value based on the sampled values ​​of the vibration waveform sampled by the sampling processing unit 111. Then, it outputs... Figure 5 (C) shows the suspension control value of 504.

[0057] Figure 6 (A) Figure 6(B) Figure 6 (C) is a graph showing the relationship between the vibration waveform and the time width. Figure 6 (A) shows the case of a specified time width. Figure 6 (B) shows the case where the time width is narrowed. Figure 6 (C) is Figure 6 (B) is a magnified view of a portion of the image.

[0058] like Figure 5 As shown in (B), when the frequency components of the vibration waveform are in a lower frequency band, i sampled values ​​(x1 to xi) within a window width of 503 can be sampled to capture the vibration waveform representing the convex shape of the road surface. However, as Figure 6 As shown in (A), when the frequency component of the vibration waveform is in a higher frequency band, it is sometimes impossible to capture the vibration waveform representing the convex shape of the road surface between sampling points x4 and x5 within the window width 503.

[0059] Therefore, before sampling is performed by the sampling processing unit 111, the frequency determination unit 110 performs frequency analysis on the vibration waveform of the arriving vibration generated by the arriving vibration generation unit 109. If it is determined that the frequency is high, the sampling processing unit 111... Figure 6 (B) shows how to narrow the time width, that is, to set a smaller window width of 603, as shown. Figure 6 As shown in (C), i sampled values ​​(x1 to xi) are acquired. Furthermore, regardless of the time width (window width), the number of samples taken from the vibration waveform is a fixed number i. Since the number of sampled vibration waveform values ​​is fixed, the calculation and processing of the suspension control values ​​in the control value calculation unit 113 can be simplified.

[0060] Figure 7 (A) Figure 7 (B) is a flowchart illustrating the processing of the electronic control device 107 of Embodiment 1 and a diagram illustrating the input specifications. Figure 7 (A) is a flowchart illustrating the processing of the electronic control device 107. Figure 7 (B) is a diagram illustrating the input specifications during the processing of the electronic control unit 107. This flowchart illustrates the processing performed by the electronic control unit 107 executing a program. The flowcharts shown in other embodiments are similar.

[0061] In step S701, it is determined whether the vehicle is in a moving state or a stationary state. If it is in a stationary state, proceed to step S702 to determine whether the ignition switch is on or off. In step S702, if the ignition switch is off, the process ends. If the ignition switch is on, return to step S701. If it is determined in step S701 that the vehicle is in a moving state, proceed to step S703.

[0062] In step S703, the arrival vibration generation unit 109 of the electronic control device 107 calculates the road surface displacement in front of the vehicle based on data received from the road surface measurement sensor 106. Then, in the next step S704, the arrival vibration generation unit 109 generates arrival vibration of the arriving vehicle based on the calculated road surface displacement and the vehicle speed detected from the vehicle sensor 105. This arrival vibration is a time series data predicting the arrival of the vehicle. Furthermore, the arrival time of the vehicle is calculated based on the vehicle speed, and as described below, at the arrival time, the suspension is controlled based on the calculated suspension control value.

[0063] Next, in step S705, the frequency determination unit 110 determines the frequency of the arriving vibration generated by the arriving vibration generation unit 109. Specifically, frequency analysis is performed on the arriving vibration to determine whether the peak frequency is higher or lower than a reference nHz. If it is lower than the reference nHz, the process proceeds to step S706; if it is higher than the reference nHz, the process proceeds to step S707.

[0064] In step S706, the sampling processing unit 111 uses Figure 7 (B) shows the input specifications for low frequencies for sampling. Figure 7 (B) shows an example of input specifications for low frequencies, with a window width of 10 seconds, a sampling interval of 20ms, and 50 samples. That is, the vibration waveform is sampled by setting a wider window width and making the sampling interval sparser. Thus, as... Figure 5 As shown in (A), i sampled values ​​(x1 to xi) are obtained from the vibration waveform in a window width of 503. In this example, i = 50.

[0065] In step S707, the sampling processing unit 111 uses Figure 7 (B) shows the input specifications for high frequency sampling. Figure 7 (B) shows an example of input specifications for high frequencies, with a window width of 0.5 seconds, a sampling interval of 10 m seconds, and a sample count of 50. That is, the vibration waveform is sampled by setting a narrower time width and a denser sampling interval. The sample count is the same as in the low-frequency case. Thus, as... Figure 6 As shown in (C), i sampled values ​​(x1 to xi) are obtained from the vibration waveform within a window width of 603. In this example, i = 50.

[0066] After the processing in step S706 or S707, the process proceeds to step S708. In step S708, the control value calculation unit 113 uses the sampled values ​​(x1 to xi) of the sampled vibration waveform to calculate the suspension control value. The electronic control device 107 controls the suspension based on the calculated suspension control value according to the time when the corresponding arrival vibration arrives at the vehicle.

[0067] After the processing in step S708, the process returns to step S701. By repeating steps S703 to S708, the vibration waveform can be accurately determined, and the suspension can be optimally controlled, regardless of the frequency of the vibration arriving at the vehicle. As a result, because the characteristics of the intended road surface can be accurately captured and the suspension controlled, good ride comfort can be achieved.

[0068] Furthermore, the arrival vibration is not limited to being acquired from camera 205, but can also be acquired from the accelerometer 204, etc. In this case, the vehicle actions acquired by vehicle sensor 105 are pre-stored as a history. Then, based on the actual vehicle actions that make vehicle 201 move and are detected by vehicle sensor 105, electronic control unit 107 predicts the arrival vibration that will arrive at vehicle 201 immediately after detection by referring to the vehicle actions pre-stored as history. When accelerometer 204 is used as vehicle sensor 105, the arrival vibration that will arrive immediately thereafter is predicted based on the history of acceleration changes up to date. In addition, as vehicle actions, accelerator or brake pedal operation information, steering wheel operation information, etc., can be used.

[0069] [Implementation Method 2]

[0070] Figure 8 (A) and Figure 8 (B) is a diagram showing an example of the control value calculation unit 123 and the storage unit 112 of Embodiment 2. Figure 8 (A) shows the control value calculation unit 123. Figure 8 (B) Storage unit 112 is shown. In this embodiment, the control value calculation unit 113 shown in Embodiment 1 is replaced with a control value calculation unit 123 composed of a neural network, and the weight parameters 804 and 805 are stored in the storage unit 112 shown in Embodiment 1. Other structures are the same as those shown in Embodiment 1. Figures 1-6 same.

[0071] like Figure 8 As shown in (A), the neural network in the control value calculation unit 123 of this embodiment is a hierarchical neural network with a three-layer structure obtained by layering the elements in the input layer (number of elements i+1) 801, the hidden layer (number of elements j+1) 802, and the output layer (number of elements K) 803. An element representing a bias term is set in both the input layer 801 and the hidden layer 802. The elements of the input layer 801 and the hidden layer 802 are combined using weights W1ij (i = 1 to I+1, j = 1 to J+1), and the elements of the hidden layer 802 and the output layer 803 are combined using weights W2jk (j = 1 to J+1, k = 1 to K). In this embodiment, as... Figure 8As shown in (B), the information of these weights (hereinafter referred to as weight parameters) is divided into weight parameters 804 for low frequencies and weight parameters 805 for high frequencies to be used in relation to the vibrations. The weight parameters 804 for low frequencies and weight parameters 805 for high frequencies are represented by the determinants of weights W1ij and W2jk, respectively. In this example, the simplest hidden layer 802 shows a single fully connected neural network, but it is not limited to this.

[0072] The neural network learns the correlation between the sampled values ​​(x1~xi) of the sampled vibration waveform and the suspension control values. The number of elements in the input layer 801 is the same as the number of sampling points in the window width, and the number of elements in the output layer 803 is determined according to the resolution of the suspension control values. For example, if the resolution of the suspension control values ​​is a 256-level digital value, then the number of elements in the output layer 803 is 256. That is, the number of suspension control values ​​and the number of elements are the same. Then, the suspension control values ​​and elements are set to a one-to-one relationship, and are configured such that only one element out of the 256 elements is 1 (High), and the other elements are 0 (Low). The number of elements in the hidden layer 802 is determined by considering the balance between the number of elements in the input layer 801 and the number of elements in the output layer 803.

[0073] Figure 9 This is a structural diagram of the learning system for generating teacher data for a neural network in the control value calculation unit 123 of this embodiment. The learning system generates data such as... Figure 8 The weight parameters of the neural network are shown.

[0074] The learning system includes a vehicle motion simulator 901, a control value / parameter setting unit 902, a vehicle motion evaluation unit 903, and a neural network learning unit 904.

[0075] The vehicle motion simulator 901 takes as input the road surface profile representing road displacement as the road surface setting and the vehicle speed as the vehicle speed setting. It then calculates the vehicle motion of the target vehicle at each speed, such as longitudinal acceleration, vertical acceleration, lateral acceleration, yaw rate, pitch rate, and yaw rate, based on the vehicle model. Additionally, it calculates vibration waveforms based on the road surface profile and vehicle speed.

[0076] For example, the control value / parameter setting unit 902 has various parameters, such as those relating to the structure from the road surface to the vehicle chassis, and can arbitrarily set the control value of the suspension.

[0077] The vehicle motion evaluation unit 903 compares vehicle motion data output from the vehicle motion simulator 901, such as the acceleration of the seat where the passenger is sitting, with desired values ​​or existing simulation results to determine whether improvement is needed. Then, if the seat acceleration exceeds the desired value, or if further improvement is expected, the process switches to the control value / parameter setting unit 902, sets a different control value, and executes the vehicle motion simulator 901. Conversely, if the seat acceleration is below the desired value, or if further improvement is not expected, the optimal control value obtained by the vehicle motion simulator 901 is defined as the optimal suspension control value and transmitted to the neural network learning unit 904.

[0078] The neural network learning unit 904 sets sampled values ​​of the vibration waveform time-series data derived from the vehicle motion simulator 901 as the input side of the neural network, and sets the optimal suspension control values ​​derived from the vehicle motion evaluation unit 903 as the output side of the neural network, and learns the relationship between them through the neural network. Furthermore, the neural network learns by applying the error backpropagation method, a commonly known learning method.

[0079] Traditional suspension control is based on the "skyhook" theory, which assumes the vehicle is suspended in mid-air, to calculate optimal control values ​​using equations of motion. However, there are unrealistic scenarios where the vehicle is suspended in mid-air, which may not yield optimal suspension control values. In contrast, this implementation takes the approach of learning through a neural network to determine the relationship between road surface displacement and optimal suspension control values. Using a simulator, even in a virtual environment, allows for the observation of vehicle actions at various suspension control values. By investigating vehicle actions with a wider range of parameters, it is possible to derive optimal control values ​​that surpass the skyhook theory.

[0080] Furthermore, during the search conducted by this learning system, the evaluation results from the vehicle motion evaluation unit 903 are used to perform reinforcement learning using a vehicle motion simulator by rewarding good results and penalizing bad results. Ideally, the acceleration felt by the passenger should be zero. The acceleration on the passenger's seat is set as the evaluation function to solve the optimization problem of minimizing acceleration.

[0081] In this learning process, the frequency of the vibration waveform based on the arrival vibration of the road surface displacement is divided into low-frequency and high-frequency cases, and the weight parameters of the neural network that outputs the optimal control value of the suspension are determined. Then, the determined weight parameters 804 for low-frequency and 805 for high-frequency are stored in the storage unit 112 (see reference). Figure 2 )middle.

[0082] Figure 10 (A) Figure 10 (B) is a flowchart illustrating the processing of the electronic control device 107 in Embodiment 2 and a diagram illustrating the input specifications. Figure 10 (A) is a diagram showing the processing flow of the electronic control device 107. Figure 10 (B) is a diagram showing the input specifications during processing by the electronic control unit 107. Figure 7 Processes shown that are the same as those processed by the electronic control device 107 in Embodiment 1 are given the same reference numerals to simplify their description.

[0083] In step S701, if it is determined that the vehicle is in motion, proceed to step S703. In step S703, the arrival vibration generation unit 109 of the electronic control device 107 calculates the road surface displacement in front of the vehicle based on the data received from the road surface measurement sensor 106. Then, in the next step S704, the arrival vibration generation unit 109 generates the arrival vibration of the approaching vehicle based on the calculated road surface displacement and the vehicle speed detected from the vehicle sensor 105.

[0084] Next, in step S705, the frequency determination unit 110 determines the frequency of the arriving vibration generated by the arriving vibration generation unit 109. Specifically, frequency analysis is performed on the arriving vibration to determine whether the peak frequency is higher or lower than a reference nHz. If it is lower than the reference nHz, the process proceeds to step S706; if it is higher than the reference nHz, the process proceeds to step S707.

[0085] In step S706, the sampling processing unit 111 uses Figure 10 (B) shows the input specifications used for low frequencies for sampling. Therefore, as... Figure 5 As shown in (A), i sampled values ​​(x1 to xi) are obtained from the vibration waveform in the window width 503. Then, proceed to step S716.

[0086] In step S716, the weighting parameter 804 for low frequency is read from the storage unit 112 (refer to...). Figure 8 (B)). The weight parameter 804 is the weight parameter between elements used in the neural network.

[0087] In step S707, the sampling processing unit 111 uses Figure 10 (B) shows the input specifications used for high-frequency sampling. Therefore, as... Figure 6 As shown in (C), i sampled values ​​(x1 to xi) are obtained from the vibration waveform within a window width of 603. Then, proceed to step S717.

[0088] In step S717, the weighting parameter 805 for high frequency is read from the storage unit 112 (refer to...). Figure 8(B)). The weight parameter 805 is the weight parameter between elements used in the neural network.

[0089] After the processing in step S716 or S717, the process proceeds to step S708. In step S708, the control value calculation unit 123 uses the sampled values ​​(x1 to xi) of the sampled vibration waveform to calculate the suspension control value, and controls the suspension based on the suspension control value. The control value calculation unit 123 is composed of a neural network, and calculates the suspension control value based on the weight parameters 804 and 805 for low frequency or high frequency read in the processing in step S716 or S717.

[0090] The window width differs between low-frequency and high-frequency vibrations, but the number of sampled values ​​remains the same, thus matching the number of elements in the input layer 801 of the neural network. Furthermore, the number of elements in the output layer 803 is the same as the number of suspension control values. Moreover, the number of weight parameters used for low-frequency and high-frequency vibrations is also the same. Therefore, the neural network structure does not need to be changed according to the frequency of the arriving vibration, and its processing can be accelerated.

[0091] After the processing in step S708, return to the processing in step S701, and repeat the above steps S703 to S708.

[0092] According to this embodiment, the vibration waveform can be accurately grasped regardless of the frequency of the arrival vibration of the vehicle, and the relationship between road surface characteristics and optimal suspension control values ​​can be learned. As a result, good ride comfort can be obtained because the characteristics of the intended road surface can be accurately captured and the suspension can be optimally controlled.

[0093] [Implementation Method 3]

[0094] Figure 11 This diagram illustrates an example of the control value calculation unit 133 in Embodiment 3. In this embodiment, the control value calculation unit 113 shown in Embodiment 1 is replaced by a control value calculation unit 133 composed of a neural network, which has a different structure than that described in Embodiment 2. Other structures are the same as those shown in Embodiment 1. Figures 1-6 same.

[0095] This embodiment not only uses the arrival vibration generated by the arrival vibration generation unit 109 described in Embodiments 1 and 2, i.e. the arrival vibration predicted based on the image of the camera 205, but also uses the arrival vibration predicted based on the vehicle's movement during driving, such as the speed of the suspension piston and the up-and-down acceleration of the vehicle's springs, to calculate the suspension control value, thereby improving the calculation accuracy.

[0096] like Figure 11As shown, the neural network in the control value calculation unit 133 of this embodiment is a hierarchical neural network with a three-layer structure, in which the elements of the input layers 1001-1003, the hidden layer 1004, and the output layer 1005 are combined in a hierarchical manner. The elements of the input layer 1001 are X11-X1i, the elements of the input layer 1002 are X21-X2i, and the elements of the input layer 1003 are X31-X3i.

[0097] Sampled values ​​(x1 to xi) of the vibration waveform predicted from the image of camera 205 are input to elements of input layer 1001. Sampled values ​​(p1 to pi) of the vibration waveform based on piston velocity are input to elements of input layer 1002. Sampled values ​​(a1 to ai) of the vibration waveform based on the up-and-down acceleration of the spring are input to elements of input layer 1003. Similar to Embodiments 1 and 2, these sampled values ​​input to the neural network are obtained by sampling the vibration waveform of the arriving vibration generated by the arriving vibration generation unit 109 within a window width corresponding to the frequency.

[0098] Figure 12 (A) is a diagram showing a modified example of the control value calculation unit 133 of Embodiment 3. In this modified example, it shows the case where no sampled values ​​(x1 to xi) of the vibration waveform predicted based on the image of the camera 205 are input.

[0099] like Figure 12 As shown in (A), the neural network in the control value calculation unit 133 of this embodiment can be composed of a three-layer hierarchical neural network that combines the elements of the input layers 1002-1003, the hidden layer 1004, and the output layer 1005 in a hierarchical manner. The elements of the input layer 1002 are X21-X2i, and the elements of the input layer 1003 are X31-X3i.

[0100] The piston speed and sampled values ​​(p1 to pi) of the vibration waveform predicted based on the piston speed are input to the elements of input layer 1002. The up-and-down acceleration of the spring and sampled values ​​(a1 to ai) of the vibration waveform predicted based on the up-and-down acceleration of the spring are input to the elements of input layer 1003. Similar to Embodiments 1 and 2, these sampled values ​​input to the neural network are obtained by sampling the vibration waveform of the arriving vibration generated by the arriving vibration generation unit 109 within a window width corresponding to the frequency.

[0101] Figure 12 (B) is a graph illustrating the predicted vibration waveform based on measured values, using piston speed as an example. The horizontal axis represents time, and the vertical axis represents piston speed. White circles represent measured values, and black circles represent predicted values.

[0102] like Figure 12As shown in (B), the vibration waveform value 1009 is input from the vehicle sensor 105. The sampled values ​​of the vibration waveform within the window width 1010 generated by the vibration generation unit 109 (white and black circles in the figure) are input to the control value calculation unit 133, which is composed of a neural network. Here, of the sampled values ​​of the vibration waveform included in the window width 1010, approximately 80% of the first half are measured values ​​(white circles in the figure), while approximately 20% of the second half are predicted values ​​determined based on the conditions of the first half (black circles in the figure). Thus, feedforward control, which derives the optimal suspension control value immediately before actual driving, can be achieved, rather than feedback control, which derives the optimal suspension control value after the measured values ​​are consistent. When there is a prediction deviation, optimal control may not be possible, but since the fault only occurs when the characteristics of road displacement change, the prediction is unlikely to be felt by the occupant; therefore, the predicted value can be included in a portion of the vibration waveform value. Furthermore, the value of the vibration waveform input to the control value calculation unit 133 can be determined based on the time series prediction of an RNN (Recurrent Neural Network). The example of piston speed illustrates this; the same applies to the up-and-down acceleration of the spring. (Refer to...) Figure 12 In the example described in (A), the piston speed and the up-and-down acceleration of the spring include not only measured values ​​but also predicted values.

[0103] In this embodiment, an acceleration sensor for detecting piston speed and an acceleration sensor for detecting vertical acceleration of the spring are respectively installed as vehicle sensors 105 on the vehicle body 101. Based on the sensor values ​​input from these vehicle sensors 105, a reference is also included at the vibration generation unit 109. Figure 12 (B) The predicted value described generates an arrival vibration in the interior. Then, the frequency determination unit 110 determines the frequency of the generated arrival vibration, and the sampling processing unit 111 samples the frequency of the generated arrival vibration with a window width corresponding to the frequency. The number of sampled values ​​from the vibration waveforms of these arrival vibrations is the same regardless of the window width. The sampled values ​​(p1 to pi) of the vibration waveforms of the arrival vibration generated based on the piston speed are input to the elements of the input layer 1002, and the sampled values ​​(a1 to ai) of the vibration waveforms of the arrival vibration generated based on the up-and-down acceleration of the spring are input to the elements of the input layer 1003.

[0104] Figure 13 This is a structural diagram of the learning system for generating teacher data for a neural network in the control value calculation unit 133 of this embodiment. The learning system generates data such as... Figure 11 The weight parameters of the neural network are shown. In relation to... Figure 9 The same parts of the learning system shown are given the same labels to simplify their description.

[0105] Figure 13The vehicle motion simulator 901 shown calculates the vibration waveforms of the camera image based on the road surface contour and vehicle speed, and calculates the vibration waveforms corresponding to the piston speed and the vertical acceleration of the spring, respectively. The neural network learning unit 904 learns by taking into account the sampled values ​​of these vibration waveforms as time-series data derived from the vehicle motion simulator 901.

[0106] For reference Figure 9 As explained, the learning system performs multiple simulations with different suspension control values, and derives the optimal control value for the simulator based on the judgment of the vehicle motion evaluation unit 903. Figure 13 In the example shown, an acceleration sensor model is set on the vehicle in the simulator, and the vehicle motion simulator 901 calculates the time-series data of the sensor output obtained under optimal control conditions, particularly the piston speed and spring acceleration used in the calculation of previous suspension control values. Sampled values ​​of vibration waveforms obtained from the road surface displacement and vehicle speed based on camera 205, sampled values ​​of vibration waveforms based on piston speed obtained from vehicle sensor 105, and sampled values ​​of vibration waveforms based on spring acceleration are input from the vehicle motion simulator 901 to the neural network learning unit 904. Additionally, the optimal suspension control value determined by the vehicle motion evaluation unit 903 is input. In the neural network learning unit 904, sampled values ​​representing the vibration waveforms predicted from the road surface by camera 205, the vibration waveforms of piston speed, and the vibration waveforms of spring acceleration are respectively set on the input side of the neural network, and the optimal suspension control value is set on the output side of the neural network, and the relationship between them is learned.

[0107] In this learning process, the weight parameters A to D of the neural network that outputs the optimal control value for the suspension are determined based on the frequency of the vibration waveform that reaches the vibration (see below). Figure 14 (B)). Then, the weighting parameters corresponding to the determined frequency are stored in the storage unit 112 (see reference). Figure 2 )middle.

[0108] Figure 14 (A) Figure 14 (B) is a flowchart illustrating the processing of the electronic control device 107 in Embodiment 3 and a diagram illustrating the input specifications. Figure 14 (A) is a diagram showing the processing flow of the electronic control device 107. Figure 14 (B) is a diagram showing the input specifications in the processing of the electronic control device 107.

[0109] exist Figure 7 (A) and Figure 10In Embodiments 1 and 2, the cases of low-frequency and high-frequency arriving vibrations are distinguished, and the input specifications of the control value calculation unit and the weight parameters of the neural network are different. In Embodiment 3, the input specifications of the control value calculation unit and the weight parameters of the neural network are different according to the frequency of the arriving vibration. Figure 14 In the process of (A), regarding the handling of... Figure 7 (A) Figure 10 The parts that are treated in the same way are given the same labels and are briefly described.

[0110] Additionally, the following description explains the use of electronic control device 107. Figure 11 The control value calculation unit 133 shown is an example. Similarly, it can also be applied to the use of... Figure 12 (A) shows the control value calculation unit 133.

[0111] exist Figure 14 In step S701 shown in (A), if it is determined that the vehicle is in motion, proceed to step S703. In step S703, the arrival vibration generation unit 109 of the electronic control device 107 calculates the road surface displacement in front of the vehicle based on the data received from the road surface measurement sensor 106. Then, in the next step S704, the arrival vibration generation unit 109 generates the arrival vibration of the approaching vehicle based on the calculated road surface displacement and the vehicle speed detected from the vehicle sensor 105.

[0112] Next, in step S705, the frequency determination unit 110 determines the frequency of the arriving vibration generated by the arriving vibration generation unit 109. Specifically, it performs frequency analysis of the arriving vibration and determines whether the frequency is 0.5 Hz or less, greater than 0.5 Hz and less than 1.0 Hz, greater than 1.0 Hz and less than 2.0 Hz, or greater than 2.0 Hz.

[0113] If the frequency is below 0.5Hz, proceed to step S721 and set. Figure 14 (B) is one of the input specifications shown. Figure 14 The input specifications shown in (B) are an example. In this example, the window width is set to 2.0 seconds, the sampling interval is set to 40 m seconds, the number of samples is set to 50, and the sampling processing unit 111 samples the vibration waveform. Then, proceed to the next step S722, referring to... Figure 14 (B) is one of the input specifications shown to set the weight parameter A.

[0114] If the frequency is greater than 0.5Hz and less than 1.0Hz, proceed to step S731 and set. Figure 14(B) shows the second input specification. In this example, the window width is set to 1.0 second, the sampling interval is set to 20 m seconds, the number of samples is set to 50, and the sampling processing unit 111 samples the vibration waveform. Then, it proceeds to the next step S732, referring to... Figure 14 (B) shows the second input specification to set the weight parameter B.

[0115] If the frequency is greater than 1.0 Hz and less than 2.0 Hz, proceed to step S741 and set. Figure 14 (B) shows the third input specification. In this example, the window width is set to 0.5 seconds, the sampling interval is set to 10 m seconds, the number of samples is set to 50, and the sampling processing unit 111 samples the vibration waveform. Then, proceed to the next step S742, referring to... Figure 14 (B) shows the third input specification to set the weight parameter C.

[0116] If the frequency is greater than 2.0Hz, proceed to step S751 and set. Figure 14 (B) shows the fourth of the input specifications. In this example, the window width is set to 0.25 seconds, the sampling interval is set to 5 m seconds, the number of samples is set to 50, and the sampling processing unit 111 samples the vibration waveform. Then, proceed to the next step S752, referring to... Figure 14 (B) shows the fourth input specification to set the weight parameter D.

[0117] After processing in steps S722, S732, S742, and S752, the process proceeds to step S708. In step S708, the control value calculation unit 133 is composed of a neural network and calculates the suspension control value based on weight parameters corresponding to the frequency.

[0118] In step S705, an example is described whereby the frequency determination unit 110 uses four divisions to determine the frequency of the arriving vibration generated by the arriving vibration generation unit 109. However, this is just an example, and appropriate settings should be made according to the vehicle's structure and the surrounding environment.

[0119] Furthermore, for the sake of simplicity, the following example will be used: In step S703, the arrival vibration generation unit 109 generates the arrival vibration of the approaching vehicle based on the road surface measurement sensor 106 (e.g., camera 205), and in step S705, the frequency determination unit 110 determines the frequency of the arrival vibration. However, in these processes, not only the arrival vibration representing the road surface prediction from camera 205 can be used, but also the arrival vibration representing piston speed and spring acceleration can be used. In this case, all of the arrival vibrations representing the road surface prediction from camera 205, piston speed, and spring acceleration can be used, or any one of them can be combined, or only one of them can be used. Generally, the arrival vibration of the approaching vehicle is similar in all cases, so the convex shape of the road surface can be captured.

[0120] The window width varies depending on the frequency of the arriving vibration, but the number of sampling points remains the same, thus matching the number of elements in the input layer 1001 of the neural network. Furthermore, the number of elements in the output layer 1005 is the same as the number of suspension control values. Also, the number of weight parameters is the same regardless of frequency. Therefore, the structure of the neural network does not need to be changed according to the frequency of the arriving vibration, and its processing speed can be increased.

[0121] After the processing in step S708, return to the processing in step S701 and repeat the processing in steps S703 to S752.

[0122] According to this embodiment, the vibration waveform can be accurately grasped regardless of the frequency of the arrival vibration of the arriving vehicle, and the relationship between road surface characteristics and optimal suspension control values ​​can be learned. Not only can the arrival vibration from the camera 205 be used, but also the arrival vibration from piston speed and spring acceleration can be used to accurately capture the characteristics of the predetermined road surface and optimally control the suspension, thus achieving good ride comfort.

[0123] [Implementation Method 4]

[0124] Figure 15 (A) Figure 15 (B) and Figure 15 (C) is a diagram showing the suspension control device 100 of this embodiment. Figure 15 (A) is a block diagram of the electronic control devices 107 and 1501 that constitute the suspension control device 100 of this embodiment. Figure 15 (B) is a diagram showing the road outline. Figure 15 (C) is a diagram representing road categories.

[0125] In this embodiment, arrival vibration is generated based on the road profile obtained from the server device instead of the arrival vibration generated based on the camera 205 mounted on the vehicle.

[0126] Figure 15 The block diagram of the electronic control device 107 shown in (A) has the same... Figure 2 The electronic control unit 1501 has the same structure as the electronic control unit 107 shown. The electronic control unit 1501 is installed within the vehicle's suspension control unit 100 in the same manner as the electronic control unit 107. The electronic control unit 1501 includes: a transceiver I / F 1502, a GPS 1503, a control unit 1504, a memory 1505 for storing road contours / road categories, and a data transmission unit 1506. The electronic control unit 1501 is sometimes referred to as a road contour acquisition unit.

[0127] The server device (not shown in the diagram) manages digitized maps and road surface contours that record the road surface shape information of each road. The vehicle obtains its position via its onboard GPS 1503. Based on this position, it queries the server device for road surface contours within a certain range, for example, a radius of 5 km, via I / O transceiver 1502, and downloads the corresponding data. The downloaded road surface contours are stored in memory 1505. Then, the road surface contours in front of the moving vehicle are transmitted to vibration generation unit 109 via data transmission unit 1506. Control unit 1504 manages the sending and receiving of this data.

[0128] Here, the data range obtained from the server device is set to a radius of 5km, but it can be larger or smaller than this range. Furthermore, it can collaborate with a car navigation system to obtain the road outline of the predetermined driving route.

[0129] like Figure 15 As shown in (B), the road outlines define roads A and B using the distance between intersections. Furthermore, the distance between intersections is set to, for example, 10m for road A, and road shape information such as height is defined for each 1cm interval of road A. The arrival vibration generation unit 109 plots the road surface displacement by setting the distance as the horizontal axis based on the road shape information. Additionally, by considering the vehicle speed obtained from the vehicle sensor 105, arrival vibration is generated with time set as the horizontal axis. Figure 15 (C) For each road A, B..., a road category is stored, indicating whether it is a good road, a gravel road, or an uneven road, etc. However, these road categories can also be referenced to generate a predetermined arrival vibration for each road category.

[0130] After generating the arrival vibration, the frequency of the generated arrival vibration is determined in the same manner as in any of Embodiments 1 to 3, sampling processing is performed, and suspension control values ​​are calculated to control the suspension.

[0131] According to this embodiment, in addition to achieving the same effects as embodiments 1 to 3, even when road surface displacement cannot be detected by the road surface measurement sensor 106 or the like due to driving environment or driving conditions, this information can be temporarily obtained from outside the suspension control device 100 to continue suspension control. Alternatively, the suspension can be controlled by obtaining this information from outside the suspension control device 100, instead of detecting road surface displacement from the road surface measurement sensor 106 or the like.

[0132] The following effects can be obtained by implementing the methods described above.

[0133] (1) The suspension control device 100 is connected to sensors 105 and 106 for acquiring information about the vehicle 201 or the area surrounding the vehicle 201. Based on the information acquired by the sensors 105 and 106, it calculates suspension control values ​​for controlling the suspension 103 of the vehicle 201. This includes: an arrival vibration generation unit 109 that generates an arrival vibration of the vehicle 201 based on the information acquired by the sensors 105 and 106; a sampling processing unit 111 that changes the sampling time width of the arrival vibration waveform based on the frequency of the arrival vibration generated by the arrival vibration generation unit 109 and samples the arrival vibration waveform; and control value calculation units 113, 123, and 133 that calculate the suspension control values ​​based on the sampled values ​​of the vibration waveform sampled by the sampling processing unit 111. Thus, regardless of the frequency of the arrival vibration of the vehicle, the vibration waveform can be determined, thereby improving the controllability of the suspension.

[0134] (2) The control method of the suspension control device 100 is a suspension control method in which the suspension control device 100 uses sensors 105 and 106 to acquire information about the vehicle 201 or the surrounding environment of the vehicle 201 to control the suspension 103 of the vehicle 201. Specifically, based on the information acquired by the sensors 105 and 106, an arrival vibration of the vehicle 201 is generated; according to the frequency of the generated arrival vibration, the sampling time width of the vibration waveform of the arrival vibration is changed and the vibration waveform of the arrival vibration is sampled; a suspension control value is calculated based on the sampled values ​​of the vibration waveform; and the suspension 103 is controlled based on the calculated suspension control value. Therefore, regardless of the frequency of the arrival vibration of the vehicle, the vibration waveform can be grasped, thereby improving the controllability of the suspension.

[0135] This invention is not limited to the embodiments described above. Other embodiments that can be considered within the scope of the technical concept of this invention, as long as they do not impair the characteristics of this invention, are also included within the scope of this invention. Additionally, it may be a combination of the structures described above.

[0136] Label Explanation

[0137] 1...Suspension control model, 100...Suspension control device, 101...Body, 102...Wheel, 103...Suspension, 104...Tire, 105...Vehicle sensor, 106...Road measurement sensor, 107, 1501...Electronic control device, 109...Vibration generation unit, 110...Frequency determination unit, 111...Sampling and processing unit, 112...Storage unit, 113, 123, 133...Control value calculation unit, 201...Vehicle, 202...Front wheel suspension, 203...Rear wheel suspension, 204...Acceleration sensor, 205...Camera, 503, 603 ...Window width (time width), 801, 1001~1003...Input layer of neural network, 802, 1004...Hidden layer of neural network, 803, 1005...Output layer of neural network, 804, 805...Weight parameters, 901...Vehicle motion simulator, 902...Control value / parameter setting unit, 903...Vehicle motion evaluation unit, 904...Neural network learning unit, 1502...Transmit / receive I / F, 1503...GPS, 1504...Control unit, 1505...Memory for road outline / road category, 1506...Data transmission unit, S, S1, S2...Road surface.

Claims

1. A suspension control device connected to a sensor for acquiring information about a vehicle or the area surrounding the vehicle, and calculating suspension control values ​​for controlling the suspension of the vehicle based on the information acquired by the sensor, characterized in that it comprises: The arrival vibration generating unit generates an arrival vibration of the vehicle based on the information obtained by the sensor. The sampling processing unit changes the sampling time width of the vibration waveform of the arriving vibration according to the frequency of the arriving vibration generated by the arriving vibration generation unit and samples the vibration waveform of the arriving vibration. as well as The control value calculation unit calculates the suspension control value based on the sampled values ​​of the vibration waveform sampled by the sampling processing unit. When the arriving vibration is low frequency, the sampling processing unit sets the time width to be wider and the sampling interval to be sparse; when the arriving vibration is high frequency, the time width to be narrower and the sampling interval to be denser, thereby sampling the vibration waveform of the arriving vibration.

2. The suspension control device as described in claim 1, characterized in that, The sampling processing unit sets the number of sampled values ​​when sampling the vibration waveform of the arriving vibration to a certain number, regardless of the frequency of the arriving vibration.

3. The suspension control device as described in claim 1, characterized in that, The arrival vibration generation unit predicts the arrival vibration of the vehicle based on the road surface displacement and the vehicle speed obtained by the sensor.

4. The suspension control device as described in claim 3, characterized in that, The information acquired by the sensor includes at least one of camera images, piston speed of the suspension, and vertical acceleration of the vehicle's springs.

5. The suspension control device as described in claim 1, characterized in that, It includes a road surface contour acquisition unit, which is used to acquire road shape information of the current position of the vehicle. The arrival vibration generation unit predicts the arrival vibration of the vehicle based on the road surface displacement generated according to the road shape information and the vehicle speed.

6. The suspension control device as described in claim 5, characterized in that, The road surface contour acquisition unit downloads the road shape information from the server device.

7. The suspension control device as described in claim 1, characterized in that, The control value calculation unit includes a neural network that uses weight parameters corresponding to the frequency of the arriving vibration to calculate the suspension control value.

8. The suspension control device as described in claim 7, characterized in that, The number of sampled values ​​of the vibration waveform sampled by the sampling processing unit is a fixed number, and is independent of the frequency of the arriving vibration. The input layer of the neural network has the same number of elements as the number of sampled values ​​of the vibration waveform.

9. A suspension control method, which uses sensors for acquiring information about the vehicle or information about the surroundings of the vehicle to control the suspension of the vehicle, characterized in that... Based on the information acquired by the sensor, an arrival vibration is generated for the vehicle. Based on the frequency of the generated arriving vibration, the sampling time width of the vibration waveform of the arriving vibration is changed, and the vibration waveform of the arriving vibration is sampled again. The suspension control value is calculated based on the sampled values ​​of the vibration waveform. The suspension is controlled based on the calculated suspension control values. When the arriving vibration is low frequency, the time width is set to be wider and the sampling interval is set sparsely. When the arriving vibration is high frequency, the time width is set to be narrower and the sampling interval is set more closely, thereby sampling the vibration waveform of the arriving vibration.

10. The suspension control method as described in claim 9, characterized in that, The number of sampled values ​​of the vibration waveform is a fixed number and is independent of the frequency of the vibration. The suspension control value is calculated by applying weight parameters corresponding to the frequency of the arriving vibration to a neural network having the same number of elements in the input layer as the number of sampled values ​​of the vibration waveform.

Images