Suspension system
The suspension system uses a trained model to estimate stroke displacement and velocity, calculating sprung mass acceleration and detecting anomalies, enhancing reliability by correcting errors and preventing inappropriate control.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- AISIN CORP
- Filing Date
- 2023-03-24
- Publication Date
- 2026-06-23
AI Technical Summary
Existing suspension systems using learned models for damping force adjustment lack effective mechanisms for detecting abnormalities in their operation.
A suspension system with an estimation unit that calculates stroke displacement and velocity using a trained model, a calculation unit that determines sprung mass acceleration, and an abnormality determination unit that detects anomalies based on errors between measured and estimated values, allowing for high-precision anomaly detection and system reliability.
Enables efficient and accurate detection of abnormalities in suspension systems, improving reliability by preventing control based on inappropriate estimates and resetting model parameters when necessary.
Smart Images
Figure 0007878114000001 
Figure 0007878114000002 
Figure 0007878114000003
Abstract
Description
Technical Field
[0001] Embodiments of the present invention relate to a suspension system.
Background Art
[0002] In a suspension system capable of adjusting a damping force for converging the vibration (expansion and contraction movement) of a spring that alleviates the impact from the road surface to the vehicle body according to the situation, a learned model (artificial intelligence) may be used for controlling the damping force.
Prior Art Documents
Patent Documents
[0003]
Patent Document 1
Patent Document 2
Summary of the Invention
Problems to be Solved by the Invention
[0004] In such a system, considering the possibility that the processing by the learned model loses its appropriateness, a function for detecting the occurrence of an abnormality is required.
[0005] One of the problems to be solved by the embodiments of the present invention is to enable detection of an abnormality in a suspension system that adjusts a damping force using a learned model.
Means for Solving the Problems
[0006] One embodiment of the present invention is a suspension system with adjustable damping force, comprising a shock absorber that generates a damping force to dampen the vibration of a spring interposed between the wheel and the vehicle body, and including an estimation unit that estimates at least one of a stroke displacement amount indicating the displacement amount of the shock absorber and a stroke velocity indicating the displacement velocity of the shock absorber from input information including measured values of sprung mass acceleration using a learned model; a calculation unit that calculates an estimated value of sprung mass acceleration based on the estimation result by the estimation unit; and an abnormality determination unit that determines whether or not there is an abnormality based on the error between the measured value of sprung mass acceleration and the estimated value of sprung mass acceleration.
[0007] According to the above configuration, an estimated value of the sprung mass acceleration is calculated based on at least one of the stroke displacement and stroke velocity estimated by a trained model from input information including the measured value of the sprung mass acceleration, and an anomaly is determined based on the error between the measured value and the estimated value of the sprung mass acceleration. This enables high-precision detection of anomalies in systems that use trained models for control, thereby improving the reliability of the system.
[0008] Furthermore, in the above configuration, when the estimation unit estimates the stroke displacement and stroke velocity, the calculation unit may calculate an estimated value of the spring force, which is the reaction force generated by the spring, based on the estimated value of the stroke displacement estimated by the estimation unit, calculate an estimated value of the damping force based on the estimated value of the stroke velocity estimated by the estimation unit, and calculate an estimated value of the sprung mass acceleration based on the estimated value of the spring force and the estimated value of the damping force.
[0009] With the above configuration, an estimated value of the sprung mass acceleration can be efficiently calculated based on the stroke displacement and stroke velocity estimated by the trained model.
[0010] Furthermore, in the above configuration, when the estimation unit estimates the stroke displacement, the calculation unit may calculate an estimated value of the spring force, which is the reaction force generated by the spring, and an estimated value of the stroke velocity, which indicates the displacement velocity of the shock absorber, based on the estimated value of the stroke displacement estimated by the estimation unit. Based on the estimated value of the stroke velocity, it may calculate an estimated value of the damping force, and based on the estimated value of the spring force and the estimated value of the damping force, it may calculate an estimated value of the sprung mass acceleration.
[0011] With the above configuration, an estimated value of the sprung mass acceleration can be efficiently calculated based on the stroke displacement estimated by the trained model.
[0012] Furthermore, in the above configuration, if the estimation unit estimates the stroke velocity, the calculation unit may calculate an estimated value of the stroke displacement and an estimated value of the damping force based on the estimated stroke velocity estimated by the estimation unit, calculate an estimated value of the spring force, which is the reaction force generated by the spring, based on the estimated stroke displacement, and calculate an estimated value of the sprung acceleration based on the estimated spring force and the estimated damping force.
[0013] With the above configuration, an estimated value of the sprung acceleration can be efficiently calculated based on the stroke velocity estimated by the trained model.
[0014] Furthermore, in the above configuration, the anomaly detection unit may reset the parameters of the trained model if it determines that an anomaly exists.
[0015] The above configuration allows for the normalization of the functionality of the trained model.
[0016] Furthermore, in the above configuration, the abnormality detection unit may prohibit the use of the estimation result by the estimation unit for controlling the damping force if it determines that an abnormality exists.
[0017] The above configuration makes it possible to avoid executing control based on inappropriate estimation results. [Brief explanation of the drawing]
[0018] [Figure 1] Figure 1 is a diagram showing an example of the configuration of a suspension system mounted on a vehicle according to the first embodiment. [Figure 2] Figure 2 is a diagram showing an example of the configuration of a suspension device according to the first embodiment. [Figure 3] Figure 3 is a diagram showing an example of the functional configuration of an ECU according to the first embodiment. [Figure 4] Figure 4 is a diagram for explaining an example of a method for calculating an estimated acceleration above the spring in the first embodiment. [Figure 5] Figure 5 is a diagram showing an example of the relationship between the acceleration above the spring, the error, and the abnormality determination flag when an abnormality occurs in the first embodiment. [Figure 6] Figure 6 is a flowchart showing an example of processing by the ECU according to the first embodiment. [Figure 7] Figure 7 is a diagram showing an example of the functional configuration of an ECU according to the second embodiment. [Figure 8] Figure 8 is a diagram for explaining an example of a method for calculating an estimated acceleration above the spring in the second embodiment. [Figure 9] Figure 9 is a flowchart showing an example of processing by the ECU according to the second embodiment. [Figure 10] Figure 10 is a diagram showing an example of the functional configuration of an ECU according to the third embodiment. [Figure 11] Figure 11 is a diagram for explaining an example of a method for calculating an estimated acceleration above the spring in the third embodiment. [Figure 12] Figure 12 is a flowchart showing an example of processing by the ECU according to the third embodiment.
Mode for Carrying Out the Invention
[0019] Illustrative embodiments of the present invention are disclosed below. The configurations of the embodiments shown below, as well as the functions, results, and effects brought about by such configurations, are examples only. The present invention can be realized by configurations other than those disclosed in the following embodiments, and it is possible to obtain at least one of the various effects based on the basic configuration or derived effects.
[0020] (First Embodiment) Figure 1 shows an example of the configuration of a suspension system S mounted on a vehicle 1 of the first embodiment. The vehicle 1 illustrated here is a four-wheeled automobile capable of traveling on a road surface, and has a body 2 and four wheels 3.
[0021] The suspension system S includes a suspension device 11, an acceleration sensor 12, and an ECU (Electronic Control Unit) 13.
[0022] The suspension system 11 is installed between each of the four wheels 3 and the vehicle body 2, and is a device that reduces the impact from the road surface to the vehicle body 2.
[0023] The acceleration sensors 12 are installed on top of each of the four suspension devices 11 and are sensors that detect the sprung mass acceleration corresponding to each installation position. Sprung mass acceleration is the acceleration of the vertical displacement of the part above the suspension device 11 (mainly the vehicle body 2). Note that the number and installation positions of the acceleration sensors 12 are not limited to these.
[0024] The ECU 13 is an information processing device that performs information processing to control each suspension device 11, and is connected to in-vehicle devices such as the suspension devices 11 via a network such as CAN (Controller Area Network) for communication. The ECU 13 also acquires signals (analog signals) output from the acceleration sensor 12 via an appropriate line. The ECU 13 may be configured using, for example, a CPU (Central Processing Unit), memory, FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit), etc. In addition to the suspension devices 11 and the acceleration sensor 12, the in-vehicle devices connected to the ECU 13 may include, for example, a wheel speed sensor that detects the rotational speed (wheel speed) of the wheels 3, a 3-axis acceleration sensor that detects the tilt of the vehicle body 2, a steering angle sensor that detects the steering angle, and other ECUs that perform predetermined control.
[0025] Figure 2 shows an example of the configuration of the suspension system 11 of the first embodiment. In Figure 2, the configuration of one suspension system 11 interposed between one wheel 3 and the vehicle body 2 is schematically shown. The suspension system 11 has a spring 21 and a shock absorber 22. Note that all four suspension systems 11 have the same configuration.
[0026] The spring 21 is interposed between the wheel 3 and the vehicle body 2 and is a component that expands and contracts to absorb the impact from the road surface R to the vehicle body 2, generating a reaction force (restoring force) according to its expansion and contraction state. Hereinafter, the reaction force corresponding to the expansion and contraction state of the spring 21 will be referred to as the spring force.
[0027] The shock absorber 22 is a unit that generates a damping force to dampen the vibration (extension and contraction motion) of the spring 21. The damping force of the shock absorber 22 can be changed according to the control signal (control current) from the ECU 13. The specific configuration of the shock absorber 22 is not particularly limited, but for example, a configuration using a hydraulic damper whose damping force changes according to fluid pressure, or an actuator that operates to change the fluid pressure of the hydraulic damper according to the control current from the ECU 13 can be adopted.
[0028] The ECU 13 in this embodiment performs processing to optimize the damping force of the shock absorber 22 based on various information about the vehicle 1 obtained from the acceleration sensor 12, CAN, etc. The various information may include sprung mass acceleration obtained from the acceleration sensor 12, wheel speed obtained from the wheel speed sensor, etc., longitudinal acceleration, lateral acceleration, yaw rate, etc. obtained from the 3-axis acceleration sensor mounted on the vehicle body 2, etc. The specific control method of the damping force by the ECU 13 is not particularly limited, but for example, control based on the skyhook theory may be used.
[0029] Figure 3 shows an example of the functional configuration of the ECU 13 of the first embodiment. The ECU 13 of this embodiment has an estimation unit 101, a drive control unit 102, a calculation unit 103, and an abnormality determination unit 104. These functional units 101 to 104 can be configured by the cooperation of hardware elements and software elements (programs, etc.) that constitute the ECU 13. In addition, at least one of these functional units 101 to 104 may be configured by dedicated hardware (circuits, etc.).
[0030] The estimation unit 101 uses a trained model to estimate the stroke displacement and stroke velocity of the shock absorber 22 from predetermined input information. The stroke displacement is the vertical displacement of the shock absorber 22. The stroke velocity is the vertical displacement velocity of the shock absorber 22. The input information includes at least the measured sprung mass, which is the measured value of the spring bearing acceleration detected by the acceleration sensor 12. The input information may further include wheel speed, longitudinal acceleration, lateral acceleration, yaw rate, and other information.
[0031] The trained model is a model generated by machine learning (deep learning) performed in advance on a neural network using predetermined training data, and may be a predetermined algorithm to which parameters (weights) determined by machine learning are applied. The parameters of the trained model may include parameters that are updated according to usage (driving of vehicle 1). The trained model of this embodiment takes input information including measured sprung acceleration and outputs an estimated stroke displacement, which is an estimated value of the stroke displacement, and an estimated stroke velocity, which is an estimated value of the stroke velocity. The specific form of the trained model is not particularly limited, but for example, a model having an RNN (Recurrent Neural Network) structure or an LSTM (Long Short Term Memory) structure may be adopted.
[0032] The drive control unit 102 determines a target value for the damping force of the shock absorber 22 based on information obtained from the acceleration sensor 12 and CAN, as well as estimation results from the estimation unit 101, and outputs a control signal (control current) to the shock absorber 22 so that the damping force of the shock absorber 22 becomes the target value. The method for determining the target value of the damping force is not particularly limited, but for example, a method based on the skyhook theory may be adopted.
[0033] The calculation unit 103 calculates the estimated sprung acceleration, which is an estimated value of the sprung acceleration, based on the estimation results from the estimation unit 101, i.e., the estimated stroke displacement and estimated stroke velocity output from the trained model. The method for calculating the estimated sprung acceleration from the estimated stroke displacement and estimated stroke velocity can be implemented by appropriately utilizing known calculation methods. For example, the estimated spring force, which is an estimated value of the spring force of the spring 21, can be calculated from the estimated stroke displacement, and the estimated damping force, which is an estimated value of the damping force of the shock absorber 22, can be calculated from the estimated spring force and estimated damping force based on the equation of motion, etc.
[0034] The abnormality determination unit 104 determines whether or not there is an abnormality based on the error between the measured sprung mass acceleration obtained from the acceleration sensor 12 and the estimated sprung mass acceleration calculated by the calculation unit 103. For example, the abnormality determination unit 104 determines that there is an abnormality if the error is greater than or equal to a threshold.
[0035] Furthermore, the abnormality detection unit 104 may execute a predetermined abnormality response process if it determines that an abnormality exists. The abnormality response process may, for example, reset parameters in the trained model. The parameters to be reset may, for example, be parameters that are updated by use (driving of the vehicle 1), and may be stored values in RNNs or LSTMs, etc. This may allow the operation of the trained model to be normalized. The abnormality response process may also be, for example, a process that prohibits the use of the estimation result by the estimation unit 101 for controlling the damping force of the shock absorber 22. This makes it possible to avoid controlling the damping force based on inappropriate estimation results.
[0036] Figure 4 is a diagram illustrating an example of a method for calculating estimated sprung mass acceleration in the first embodiment. As shown in Figure 4, the trained model of this embodiment outputs estimated stroke displacement and estimated stroke velocity when input information, including measured sprung mass acceleration obtained from the acceleration sensor 12, is input. The input information may include, in addition to measured sprung mass acceleration, wheel speed, longitudinal acceleration, lateral acceleration, yaw rate, feedback value of estimated spring force, feedback value of estimated damping force, etc. By including information other than measured sprung mass acceleration in the input information in this way, the estimation accuracy of the trained model can be improved.
[0037] The estimated spring force is calculated by multiplying the estimated stroke displacement output from the trained model by the spring constant of spring 21. Furthermore, the estimated damping force, which corresponds to the estimated stroke velocity output from the trained model, is calculated using the FV (Forth-Velocity) relation, which shows the correspondence between stroke velocity and damping force. The FV relation may be generated, for example, based on previously conducted experiments or simulations, or based on information acquired during driving. The estimated spring force and estimated damping force thus calculated may be fed back as input information to the trained model.
[0038] The estimated sprung mass acceleration is calculated by adding the estimated spring force and estimated damping force calculated as described above and dividing the result by the sprung mass (for example, the weight of vehicle body 2). Then, based on the error between the measured sprung mass acceleration and the estimated sprung mass acceleration, it is determined whether or not there is an abnormality.
[0039] The above calculation method is illustrative, and the method for calculating the estimated sprung mass acceleration in this embodiment is not limited to this.
[0040] Figure 5 shows an example of the relationship between sprung mass acceleration, error, and abnormality determination flag when an abnormality occurs in the first embodiment. In Figure 5, the correspondence between sprung mass acceleration change information 201, error change information 202, and flag change information 203 is illustrated.
[0041] Sprung mass acceleration change information 201 illustrates the time-series changes of the measured value (measured sprung mass acceleration) and the estimated value (estimated sprung mass acceleration) of the sprung mass acceleration. Error change information 202 illustrates the time-series changes of the error (estimated value - measured value) between the measured value and the estimated value of the sprung mass acceleration. Th1 indicates the positive threshold, and Th2 indicates the negative threshold. Flag change information 203 illustrates the time-series changes of the value (true / false value) of a flag indicating the presence or absence of an abnormality. Here, the flag value is "0" when normal and "1" when an abnormality occurs as an example. The time axes (horizontal axis) of sprung mass acceleration change information 201, error change information 202, and flag change information 203 are assumed to be the same.
[0042] Figure 5 illustrates a case where the measured and estimated values of the sprung mass acceleration gradually diverge over time, and the error reaches a threshold Th1 at time t1. In such a case, the flag value becomes 1 at time t1, and the abnormality handling process is executed. As a result, as shown in Figure 5, the divergence (error) between the measured and estimated values of the sprung mass acceleration almost disappears after time t1, and the flag value becomes 0.
[0043] Figure 6 is a flowchart showing an example of processing by the ECU 13 of the first embodiment. The estimation unit 101 inputs input information to the trained model and obtains the estimated stroke displacement and estimated stroke velocity (S101). The calculation unit 103 calculates the estimated spring force from the estimated stroke displacement (S102), calculates the estimated damping force from the estimated stroke velocity (S103), and calculates the estimated sprung mass acceleration from the estimated spring force and estimated damping force (S104).
[0044] The anomaly detection unit 104 calculates the error between the measured sprung mass acceleration and the estimated sprung mass acceleration (S105) and determines whether the error is greater than or equal to a threshold (S106). If the error is not greater than or equal to the threshold (S106: No), the processing from step S101 onwards is executed again. If the error is greater than or equal to the threshold (S106: Yes), the anomaly detection unit 104 performs anomaly-counter processing such as resetting the parameters of the trained model (S107).
[0045] As described above, according to this embodiment, an estimated value of sprung mass acceleration is calculated based on the stroke displacement and stroke velocity estimated by a trained model from input information including the measured value of sprung mass acceleration, and an anomaly is determined based on the error between the measured value and the estimated value of sprung mass acceleration. This makes it possible to detect anomalies in the suspension system S that uses a trained model for control with high accuracy, and improves the reliability of the system.
[0046] Other embodiments will be described below, but explanations of parts that have the same or similar effects as the first embodiment will be omitted as appropriate.
[0047] (Second Embodiment) The trained model in the first embodiment outputs an estimated stroke displacement and an estimated stroke velocity in response to input information, whereas the trained model in the second embodiment outputs an estimated stroke displacement in response to input information, but does not output an estimated stroke velocity.
[0048] Figure 7 shows an example of the functional configuration of the ECU13 in the second embodiment. The estimation unit 101 in this embodiment uses the learned model described above to estimate the stroke displacement amount (estimated stroke displacement amount) from input information including the measured sprung mass acceleration. The calculation unit 103 in this embodiment calculates the estimated sprung mass acceleration (estimated sprung mass acceleration) based on the estimated stroke displacement amount estimated by the estimation unit 101.
[0049] Figure 8 is a diagram illustrating an example of a method for calculating the estimated sprung mass acceleration in the second embodiment. As shown in Figure 8, the trained model of this embodiment outputs an estimated stroke displacement when it receives input information including the measured sprung mass acceleration obtained from the acceleration sensor 12. In this embodiment, the estimated stroke velocity is calculated by differentiating the estimated stroke displacement output from the trained model.
[0050] The subsequent processing, namely the calculation of the estimated spring force, estimated damping force, and estimated sprung mass acceleration, is performed in the same manner as in the first embodiment. That is, the estimated spring force is calculated by multiplying the estimated stroke displacement output from the trained model by the spring constant of the spring 21, and the estimated damping force corresponding to the estimated stroke velocity is calculated using a predetermined FV relation. Then, the estimated sprung mass acceleration is calculated based on the estimated spring force and estimated damping force.
[0051] The above calculation method is illustrative, and the method for calculating the estimated sprung mass acceleration in this embodiment is not limited to this.
[0052] Figure 9 is a flowchart showing an example of processing by the ECU 13 of the second embodiment. The estimation unit 101 inputs input information to the trained model and obtains the estimated stroke displacement (S201). The calculation unit 103 calculates the estimated spring force and estimated stroke velocity from the estimated stroke displacement (S202), calculates the estimated damping force from the estimated stroke velocity (S203), and calculates the estimated sprung mass acceleration from the estimated spring force and estimated damping force (S204).
[0053] The anomaly detection unit 104 calculates the error between the measured sprung mass acceleration and the estimated sprung mass acceleration (S205) and determines whether the error is greater than or equal to a threshold (S206). If the error is not greater than or equal to the threshold (S206: No), the processing from step S201 onwards is executed again. If the error is greater than or equal to the threshold (S206: Yes), the anomaly detection unit 104 performs anomaly-counter processing such as resetting the parameters of the trained model (S207).
[0054] As described above, according to this embodiment, even when using a trained model that outputs an estimated stroke displacement (but does not output an estimated stroke velocity) for input information including measured values of sprung mass acceleration, it is possible to detect abnormalities in the system and improve the reliability of the system, similar to the first embodiment.
[0055] (Third embodiment) The trained model in the first embodiment outputs an estimated stroke displacement and an estimated stroke velocity in response to input information, but the trained model in the third embodiment outputs an estimated stroke velocity in response to input information and does not output an estimated stroke displacement.
[0056] Figure 10 shows an example of the functional configuration of the ECU 13 in the third embodiment. The estimation unit 101 in this embodiment uses the learned model described above to estimate the stroke velocity (estimated stroke velocity) from input information including the measured sprung mass acceleration. The calculation unit 103 in this embodiment calculates the estimated sprung mass acceleration (estimated sprung mass acceleration) based on the estimated stroke velocity estimated by the estimation unit 101.
[0057] Figure 11 is a diagram illustrating an example of a method for calculating estimated sprung mass acceleration in the third embodiment. As shown in Figure 11, the trained model of this embodiment outputs an estimated stroke velocity when it receives input information including the measured sprung mass acceleration obtained from the acceleration sensor 12. In this embodiment, the estimated stroke displacement is calculated by integrating the estimated stroke velocity output from the trained model.
[0058] The subsequent processing, namely the calculation of the estimated spring force, estimated damping force, and estimated sprung mass acceleration, is carried out in the same manner as in the first embodiment. That is, the estimated spring force is calculated by multiplying the estimated stroke displacement calculated as described above by the spring constant of the spring 21, and the estimated damping force corresponding to the estimated stroke velocity is calculated using a predetermined FV relation. Then, the estimated sprung mass acceleration is calculated based on the estimated spring force and estimated damping force.
[0059] The above calculation method is illustrative, and the method for calculating the estimated sprung mass acceleration in this embodiment is not limited to this.
[0060] Figure 12 is a flowchart showing an example of processing by the ECU 13 of the third embodiment. The estimation unit 101 inputs input information to the trained model and obtains the estimated stroke velocity (S301). The calculation unit 103 calculates the estimated stroke displacement and estimated damping force from the estimated stroke velocity (S302), calculates the estimated spring force from the estimated stroke displacement (S303), and calculates the estimated sprung acceleration from the estimated spring force and estimated damping force (S304).
[0061] The anomaly detection unit 104 calculates the error between the measured sprung mass acceleration and the estimated sprung mass acceleration (S305) and determines whether the error is greater than or equal to a threshold (S306). If the error is not greater than or equal to the threshold (S306: No), the processing from step S301 onwards is executed again. If the error is greater than or equal to the threshold (S306: Yes), the anomaly detection unit 104 performs anomaly-counter processing such as resetting the parameters of the trained model (S307).
[0062] As described above, according to this embodiment, even when using a trained model that outputs an estimated stroke velocity (but does not output an estimated stroke displacement) for input information including measured values of sprung mass acceleration, it is possible to detect abnormalities in the system and improve the reliability of the system, similar to the first embodiment.
[0063] The program for implementing the functions of the suspension system S of the above embodiment on a computer (e.g., ECU14) may be provided as an installable or executable file recorded on a computer-readable recording medium such as a CD-ROM, flexible disk (FD), CD-R, or DVD (Digital Versatile Disk).
[0064] The program may also be configured to be stored on a computer connected to a network such as the Internet and provided by being downloaded via the network. Alternatively, the program may be configured to be provided or distributed via a network such as the Internet.
[0065] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These novel embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents. [Explanation of symbols]
[0066] 1...Vehicle, 2...Body, 3...Wheels, 11...Suspension system, 12...Accelerometer, 13...ECU, 21...Spring, 22...Shock absorber, 101...Estimation unit, 102...Drive control unit, 103...Calculation unit, 104...Anomaly detection unit, 201...Sprung mass acceleration change information, 202...Error change information, 203...Flag change information, R...Road surface, S...Suspension system
Claims
1. A suspension system comprising a shock absorber that generates a damping force to dampen the vibration of a spring interposed between the wheel and the vehicle body, wherein the damping force is adjustable, An estimation unit that uses a trained model to estimate at least one of the stroke displacement amount, which indicates the displacement amount of the shock absorber, and the stroke velocity, which indicates the displacement velocity of the shock absorber, from input information including measured values of the sprung mass acceleration. A calculation unit that calculates an estimated value of the sprung acceleration based on the estimation result by the estimation unit, An abnormality determination unit that determines whether or not there is an abnormality based on the error between the measured value of the sprung acceleration and the estimated value of the sprung acceleration, A suspension system equipped with [specific features / features].
2. When the estimation unit estimates the stroke displacement and the stroke velocity, the calculation unit calculates an estimated value of the spring force, which is the reaction force generated by the spring, based on the estimated value of the stroke displacement estimated by the estimation unit, calculates an estimated value of the damping force based on the estimated value of the stroke velocity estimated by the estimation unit, and calculates an estimated value of the sprung mass acceleration based on the estimated value of the spring force and the estimated value of the damping force. The suspension system according to claim 1.
3. When the estimation unit estimates the stroke displacement, the calculation unit calculates, based on the estimated stroke displacement amount estimated by the estimation unit, an estimated spring force which is the reaction force generated by the spring, and an estimated stroke velocity which indicates the displacement velocity of the shock absorber, an estimated damping force based on the estimated stroke velocity, and an estimated sprung acceleration based on the estimated spring force and the estimated damping force. The suspension system according to claim 1.
4. When the estimation unit estimates the stroke velocity, the calculation unit calculates an estimated value of the stroke displacement and an estimated value of the damping force based on the estimated value of the stroke velocity estimated by the estimation unit, calculates an estimated value of the spring force, which is the reaction force generated by the spring, based on the estimated value of the stroke displacement, and calculates an estimated value of the sprung acceleration based on the estimated value of the spring force and the estimated value of the damping force. The suspension system according to claim 1.
5. The anomaly detection unit, if it determines that an anomaly exists, resets the parameters in the trained model. The suspension system according to any one of claims 1 to 4.
6. If the abnormality determination unit determines that an abnormality exists, it prohibits the use of the estimation result from the estimation unit for controlling the damping force. The suspension system according to any one of claims 1 to 4.