A shock absorber valve system tuning method, product, device and storage medium based on artificial intelligence technology

Abstract

The application relates to the technical field of image processing, in particular to a shock absorber valve system tuning method, product, equipment and storage medium based on artificial intelligence technology. The method comprises the following steps: collecting parameter combinations of various valve systems of a shock absorber; performing data quantization on the parameter combinations to obtain feature combinations; inputting the feature combinations into a neural network model to obtain a change curve of a predicted damping force of the neural network with respect to piston speed. The application predicts the external characteristic function of a passive hydraulic shock absorber under different valve system structures through an artificial intelligence algorithm, so that the number of tests in the chassis tuning process is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a shock absorber valve system tuning method, product, equipment and storage medium based on artificial intelligence technology. BACKGROUND

[0002] The shock absorber is an important part of the vehicle suspension system, and its main function is to alleviate and attenuate the impact vibration caused by road excitation during vehicle driving. In the chassis tuning process, the shock absorber tuning work is carried out after the tire selection, spring and stabilizer bar matching, and is mainly used to balance the handling stability and ride comfort of the vehicle. As the most critical and largest workload part in the chassis tuning, the shock absorber tuning has a crucial influence on improving the quality of the entire chassis.

[0003] The working principle of the shock absorber is based on the oil and gas filled inside. When the car moves up and down, the shock absorber piston reciprocates in the cylinder, driving the internal oil to flow repeatedly, and the damping force output is controlled through the compression valve and the recovery valve. The shock absorber has two working states of stretching and compression, and mainly controls the damping force curve at different speeds by adjusting the valve system. According to the speed, the working of the shock absorber can be divided into three levels: the piston speed is low (less than 0.1 m / s), the piston speed is medium (0.1 m / s to 0.6 m / s), and the piston speed is high (greater than 0.6 m / s). The low-speed stage mainly corresponds to the initial roll control of the vehicle, providing softness under small excitation; the medium-speed stage corresponds to the attenuation of the vehicle under medium impact, affecting the vehicle response in the non-central area; the high-speed stage corresponds to the isolation of the vehicle under large impact, controlling the vehicle body posture in extreme control.

[0004] The traditional shock absorber valve system tuning mainly relies on experience tuning and test verification method, and this method has problems such as long tuning cycle, high cost, and dependence on expert experience. Therefore, the present application is proposed. SUMMARY

[0005] The purpose of the present application is to provide a shock absorber valve system tuning method, product, equipment and storage medium based on artificial intelligence technology, which predicts the external characteristic function of the passive hydraulic shock absorber under different valve system structures through an artificial intelligence algorithm, and reduces the number of tests in the chassis tuning process.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme:

[0007] In a first aspect, the present application provides a shock absorber valve system tuning method based on artificial intelligence technology, comprising:

[0008] Collecting parameter combinations of each valve system of the shock absorber;

[0009] quantize the parameter combination to obtain a feature combination;

[0010] input the feature combination into a neural network model to obtain a damping force change curve predicted by the neural network;

[0011] wherein a loss function Loss of the neural network model during training is as follows:

[0012] ;

[0013] wherein, is a number of training samples, are loss weights of the physical model and the neural network model respectively, is a change curve predicted by the physical model corresponding to the i-th training sample, is a change curve predicted by the neural network model corresponding to the i-th training sample, is a change curve obtained by testing a single training sample;

[0014] the change curve predicted by the physical model satisfies a physical constraint of the shock absorber.

[0015] In a second aspect, the present application provides a computer program product, when running on a computer, causes the computer to execute the shock absorber valve system tuning method based on artificial intelligence technology in the first aspect.

[0016] In a third aspect, the present application provides an electronic device, comprising:

[0017] at least one processor, and a memory communicatively connected with the at least one processor;

[0018] wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the shock absorber valve system tuning method based on artificial intelligence technology.

[0019] In a fourth aspect, the present application provides a computer readable storage medium, and the medium stores computer instructions for causing a computer to execute the shock absorber valve system tuning method based on artificial intelligence technology.

[0020] Compared with the prior art, the present application has the following beneficial effects:

[0021] The method for adjusting a shock absorber valve system based on artificial intelligence technology provided in the embodiments of the present application can guide chassis adjustment work in the early stage of vehicle development through parameter combination construction, data quantization, neural network model construction, and physical constraint construction. The method is used for predicting the external characteristic curve of a shock absorber under different valve system combinations, reducing the number of tests in the chassis adjustment process, and greatly improving the efficiency of chassis adjustment. The method is mainly applied to the performance development field of suspension development and chassis adjustment, greatly reduces the labor intensity of shock absorber adjustment, and significantly shortens the development cycle. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the specific embodiments or prior art of the present application, the drawings needed in the description of the specific embodiments or prior art will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings without creative labor on the basis of these drawings.

[0023] Figure 1 is a flowchart of a method for adjusting a shock absorber valve system based on artificial intelligence technology provided in the embodiments of the present application;

[0024] Figure 2 is a structural schematic diagram of an electronic device provided in the present application. DETAILED DESCRIPTION

[0025] The exemplary embodiments of the present application are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present application to help understanding. They should be considered only as exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Also, for the sake of clarity and conciseness, the description below omits the description of well-known functions and structures.

[0026] The present application is further described in detail below in conjunction with the embodiments.

[0027] Figure 1 is a flowchart of a method for adjusting a shock absorber valve system based on artificial intelligence technology provided in the embodiments of the present application, which can be executed by a computer program and integrated in an electronic device. The method for adjusting a shock absorber valve system based on artificial intelligence technology integrated in the electronic device in the present embodiment predicts the external characteristic function of a passive hydraulic shock absorber under different valve system structures through an artificial intelligence algorithm, and obtains the damping force change curve of the shock absorber with various valve system structures with respect to piston speed. As shown in Figure 1 The method for adjusting a shock absorber valve system based on artificial intelligence technology provided in the embodiments of the present application includes the following steps:

[0028] S110, collect parameter combinations of each valve system of the shock absorber.

[0029] The specification parameters and quantity parameters of the piston valve system and the bottom valve system of the shock absorber are sorted to form parameter combinations. The parameter combinations include but are not limited to the following: the number, length, width and diameter of the orifice of the throttle valve of the piston valve system, the diameter and number of the orifice of the piston valve, the inner diameter, outer diameter, thickness and number of various restoring valve plates, and the outer diameter and thickness of various gaskets; the number, length, width and diameter of the orifice of the throttle valve of the bottom valve system, the diameter and number of the orifice of the bottom valve body, the outer diameter, thickness and number of various valve plates, and the outer diameter and thickness of various gaskets.

[0030] S120, data quantization is performed on the parameter combinations to obtain feature combinations.

[0031] In actual applications, there are more than 50 kinds of data including the piston and the bottom valve system for the same shock absorber. If the data is directly input into the neural network model, it will cause problems such as difficulty in capturing effective features, increase in complexity, and performance decline. To solve this problem, a part of the parameter combinations is quantized in this embodiment to reduce the number of features, and the obtained feature combinations have a significant impact on the change curve. The feature combinations include the quantized features and the original parameter combinations.

[0032] Optionally, since the throttle valve plate causes the damping force to change through the throttle flow change, the orifice area of the throttle valve plate is taken as the feature of the throttle valve plate, and the parameters such as the outer diameter, thickness and number of the throttle valve plate are omitted.

[0033] Optionally, the restoring valve plate, the flow-through valve plate and the compression valve plate are installed in a stacked manner by using multiple valve plates of different specifications, the oil flow of the piston hole is affected by the deformation of the valve plate, the deformation of the valve plate is affected by the stiffness, and the stiffness is related to the equivalent thickness of the valve plate. Therefore, the equivalent thicknesses of the restoring valve plate, the flow-through valve plate and the compression valve plate are calculated respectively, and the equivalent thicknesses are taken as the features of the restoring valve plate, the flow-through valve plate and the compression valve plate respectively, and the parameters such as the outer diameter, thickness and number of the valve plate are omitted. The calculation of the equivalent thickness of the shock absorber valve plate is essentially equivalent to a group of stacked valve plates, which is equivalent to a single valve plate with the same bending stiffness. A single valve plate is too thin to provide sufficient damping force. Therefore, the shock absorber valve system is usually composed of multiple valve plates with different thicknesses and diameters. It is very complex to directly analyze the deformation and force relationship of the entire valve plate group. By calculating the “equivalent thickness”, it can be modeled, analyzed and calculated as a single, thicker valve plate to calculate the opening pressure, which greatly simplifies the design process.

[0034] After data quantization, the adjustable parameters of a single throttle valve plate originally have 5 parameters, which are changed to 1 parameter after quantization. The adjustable parameters of the restoring valve system and the compression valve system can reach 9-18 according to the number of stacked valve plates, which are changed to 1 after quantization.

[0035] S130, input the feature combination into the neural network model to obtain a damping force change curve with piston speed predicted by the neural network.

[0036] The feature combination is written in vector form and input into a neural network model, which has the mapping ability to obtain the change curve according to the feature combination. The horizontal axis of the change curve is the piston speed, and the vertical axis is the damping force. In an embodiment, the change curve is divided into multiple speed segments, and the trend of the change curve of each speed segment is the same, then the neural network model can output the damping force value of each speed point, and further calculate the linear equation parameters (including slope and intercept) of different speed segments.

[0037] Before using the neural network model to predict the change curve, the neural network model needs to be selected and trained.

[0038] This embodiment adopts a physics informed neural network (PINN) model as the model for predicting the change curve of the shock absorber. The model uses a deep learning method that combines basic neural networks and physical equations. Its main feature is to incorporate the physical constraints of the physical system into the training process of the neural network, so that the neural network model can not only learn the input feature combination, but also meet the physical laws. By modifying the traditional loss function and adding physical constraints, the output of the neural network model can not only fit the training samples, but also meet the dynamic behavior of the physical system, reducing the need for training set data. Traditional deep learning methods often require a large amount of experimental data, while this method can be trained with a small amount of data plus physical equation constraints.

[0039] The neural network model in this embodiment uses a multi-layer perceptron neural network as the main network, and uses the regularization technique of Dropout to prevent network overfitting, and updates the parameters of the neural network model through the backpropagation algorithm. Table 1 is the hyperparameters of the multi-layer perceptron neural network.

[0040] Table 1 Hyperparameters of the neural network model

[0041]

[0042] In Table 1, the Adam (Adaptive Moment Estimation) optimizer is a widely used adaptive optimization algorithm in the field of deep learning, combining the advantages of momentum mechanism and adaptive learning rate adjustment. Leaky ReLU is an improved version of the ReLU activation function, aiming to solve the "neuron death" problem caused by the zero gradient of ReLU in the negative input interval. The input is a 13-dimensional feature combination, and the output is the damping force value of 10 speed points.

[0043] In a multi-layer perceptron neural network, the neurons of each layer are connected to all the neurons of the next layer. Information starts from the input layer, passes through one or more hidden layers, and finally reaches the output layer, propagating in one direction.

[0044] Dropout refers to randomly discarding (setting its activation value to zero) a portion of neurons in the network during the training process of the neural network with a certain probability p. Dropout forces the network not to rely on any specific neuron or feature combination through this random "discarding" method. Each iteration trains a slightly different "sub-network". This can be seen as a kind of model averaging, effectively reducing the complex co-adaptation between neurons, making the network more generalizable and less likely to overfit to noise in the training data. Throughout the training process, Dropout randomly masks a portion of neurons in each iteration, as if training multiple sub-networks, thereby suppressing overfitting and improving the prediction ability and robustness of the model for unseen positions.

[0045] The PINN in this embodiment combines neural networks and physical constraints, and the output of the neural network model is subjected to physical constraints (i.e., obtained from the tuning experience of the tuning engineer) as the output of the physical model. The physical model is equivalent to a correction layer connected after the neural network model, and it can be seen that the change curve predicted by the physical model satisfies the physical constraints of the shock absorber. The change curve predicted by the physical model is obtained by correcting the change curve predicted by the neural network. The physical model and the neural network model respectively obtain the predicted change curves, and the prediction error of the two models is weighted and fused as the final error to update the weights of the neural network model. The final error calculation method, i.e., the loss function Loss, is as follows:

[0046] Equation 1

[0047] wherein, N is the number of training samples, are the loss weights of the physical model and the neural network model, respectively, is the change curve predicted by the physical model corresponding to the i-th training sample, a change curve predicted by the neural network model corresponding to the ith training sample, a change curve obtained by testing a single training sample.

[0048] First, training samples are collected. The training samples are characteristic combinations of each valve system of the shock absorber obtained by manual debugging during the shock absorber tuning process. The damping force of the shock absorber under different motion strokes (stretching, compression) and different speeds (0.05 m / s, 0.1 m / s, 0.3 m / s, 0.6 m / s, 1 m / s) is recorded by testing to form a corresponding point pair of speed-damping force. The corresponding point pair is the true value .

[0049] The training sample is input into the neural network model to be trained to obtain the output of the neural network model Then After being processed by the correction layer, the output of the neural network model is obtained The , and are brought into the Loss function to obtain the Loss value. The parameters in the neural network model are optimized in the direction of minimizing the Loss value until the difference between the Loss values of two adjacent optimizations is less than a set threshold, that is, the Loss reaches stability, and the neural network model training is completed.

[0050] The shock absorber valve system tuning method based on artificial intelligence technology provided by the embodiments can guide chassis tuning work in the early stage of vehicle development through parameter combination construction, data quantization, neural network model construction and physical constraint construction. It is used to predict the external characteristic curve of the shock absorber under different valve system combinations, reduce the number of tests in the chassis tuning process, and greatly improve the efficiency of chassis tuning. The method is mainly applied to the performance development field of suspension development, chassis tuning, etc., greatly reduces the labor intensity of shock absorber tuning, and significantly shortens the development cycle.

[0051] The operation of the physical model (i.e., the correction layer) and how to obtain the change curve predicted by the physical model will be described in detail below. When the neural network model is trained, the change curve predicted by the physical model is obtained in the following manner:

[0052] First, the constraint rules are obtained by statistically analyzing the training samples.

[0053] The constraint rules include the relationship between the restoring gasket thickness and the speed inflection point in the change curve, the relationship between the throttling flow of the throttle valve and the damping force in the low-speed section of the change curve, the relationship between the number of restoring valve and the damping force in the high-speed section of the change curve, the relationship between the restoring valve thickness and the damping force in the medium-speed section and the high-speed section of the change curve, and the relationship between the equivalent thickness of the restoring valve and the slope of the change curve.

[0054] Secondly, the change curve predicted by the neural network model is corrected according to the constraint rule, and a change curve predicted by a physical model is obtained.

[0055] For the relationship between the thickness of the restoring gasket and the speed inflection point in the change curve, the speed inflection point is the horizontal coordinate value of the point where the slope of the change curve changes obviously, that is, the piston speed value. The thinner the thickness of the restoring gasket is, the greater the damping force at the speed inflection point is; on the contrary, the thicker the thickness of the restoring gasket is, the smaller the damping force at the speed inflection point is. Based on this, in the parameter combination, the relationship between the thickness of the restoring gasket and the damping force at the speed inflection point is fitted, for example, a linear relationship. Then, for the input scene of each training sample, the damping force at the speed inflection point under the current training sample (the thickness of the current restoring gasket) is calculated according to the linear relationship. The calculated damping force at the speed inflection point is replaced with the damping force at the same speed inflection point in the change curve output by the neural network model, and a change curve predicted by a physical model is formed.

[0056] For the relationship between the throttling flow of the throttle valve and the low-speed section damping force in the change curve. The low-speed section damping force is the damping force when the piston speed is less than 0.1 m / s. The smaller the throttling flow is, the greater the low-speed damping force is, and vice versa. Based on this, in the parameter combination, the relationship between the throttling flow of the throttle valve and the low-speed section damping force in the change curve is fitted, for example, a linear relationship. Then, for the input scene of each training sample, the low-speed section damping force under the current training sample (the current throttling flow) is calculated according to the linear relationship. The calculated low-speed section damping force is replaced with the low-speed section damping force in the change curve output by the neural network model, and a change curve predicted by a physical model is formed.

[0057] The relationship between the number of restoring valve pieces and the medium-speed section damping force and the high-speed section damping force in the change curve. The medium-speed section damping force is the damping force when the piston speed is 0.1 m / s to 0.6 m / s, and the high-speed section damping force is the damping force when the piston speed is greater than 0.6 m / s. The greater the number of restoring valve pieces is, the greater the medium-speed section damping force and the high-speed section damping force are; on the contrary, the smaller the number of restoring valve pieces is, the smaller the medium-speed section damping force and the high-speed section damping force are. Based on this, in the parameter combination, the relationship between the number of restoring valve pieces and the medium-speed section damping force and the high-speed section damping force in the change curve is fitted, for example, a linear relationship. Then, for the input scene of each training sample, the medium-speed section damping force and the high-speed section damping force under the current training sample (the current number of restoring valve pieces) are calculated according to the linear relationship. The calculated medium-speed section damping force and the high-speed section damping force are replaced with the medium-speed section damping force and the high-speed section damping force in the change curve output by the neural network model, and a change curve predicted by a physical model is formed.

[0058] For the relationship between the medium-speed section damping force and the high-speed section damping force of the recovery valve plate thickness and the change curve. The medium-speed section damping force and the high-speed section damping force both increase as the recovery valve plate thickness increases; conversely, the medium-speed section damping force and the high-speed section damping force both decrease as the recovery valve plate thickness decreases. Based on this, in the parameter combination, the relationship between the thickness of the recovery valve plate and the medium / high-speed section damping force of the change curve is fitted, for example, a linear relationship. Then, for the input scene of each training sample, the medium-speed section damping force and the high-speed section damping force under the current training sample (the current recovery valve plate thickness) are calculated according to the linear relationship. The calculated medium-speed section damping force and high-speed section damping force replace the medium-speed section damping force and high-speed section damping force in the change curve output by the neural network model to form the change curve predicted by the physical model.

[0059] For the relationship between the equivalent thickness of the recovery valve plate and the slope of the change curve, including the following three steps:

[0060] The first step is to obtain the equivalent thickness of the recovery valve plate according to the i-th training sample, and to obtain the slope of each speed section on the change curve according to the equivalent thickness.

[0061] Specifically, by statistically analyzing the training samples, a corresponding point pair of the equivalent thickness and the slope of each speed section is obtained; linear fitting is performed according to the corresponding point pair to obtain a plurality of relationship formulas with the equivalent thickness as the independent variable and the slope of each speed section as the dependent variable. Wherein, each speed section includes a low-speed section, a medium-speed section and a high-speed section, or more than four speed sections divided in more detail. The slopes of the damping forces in the same speed section are the same.

[0062] The relationship formula of one speed section is as follows:

[0063] Formula 2

[0064] Formula 3

[0065] Wherein, D is the equivalent thickness of the recovery valve plate, k and b are the coefficients obtained by fitting, K rebound is the slope of the speed section, is the identification of the recovery valve plate type, is the number of the i-th recovery valve plate, d i is the actual thickness of a single recovery valve plate, is the outer diameter of a single recovery valve plate. is the maximum value of the outer diameter of the recovery valve plate.

[0066] The equivalent thickness in the i-th training sample is brought into each relationship to obtain the slope of each speed segment. For example, the equivalent thickness of the i-th training sample is calculated, and the equivalent thickness is brought into D in formula 2 to obtain the slope of the speed segment that satisfies the physical constraint .

[0067] Secondly, the damping force of each speed segment is extracted from the change curve predicted by the neural network model corresponding to the i-th training sample.

[0068] The i-th training sample is input into the neural network model to obtain the change curve predicted by the neural network model. The change curve is divided into multiple speed segments according to the obvious difference in slope, for example, divided into a low speed (less than 0.1 m / s) segment, a medium speed (0.1 m / s to 0.6 m / s) segment and a high speed (greater than 0.6 m / s) segment. For each speed segment, the damping force of the speed segment is extracted, and the damping force of each speed segment should be linear.

[0069] Thirdly, the damping force of each speed segment predicted by the physical model is obtained according to the extracted damping force and slope of each speed segment.

[0070] The output of the physical model and the output of the neural network model satisfy the following formula:

[0071] ; Formula 4

[0072] Wherein, and are different speeds in the same speed segment. is the damping force of each speed segment of the physical model to be solved, is the damping force of each speed segment extracted from the output of the neural network model.

[0073] For example, the slope of the low speed segment calculated in formula 2, the two speeds (called the first speed and the second speed) of the low speed segment, and the damping force of the first speed output by the neural network model are brought into formula 4 to calculate the damping force of the second speed predicted by the physical model. The operation method of other speed segments is the same, so that the damping force of any speed predicted by the physical model can be obtained.

[0074] The aforementioned constraint rules can be used to modify the change curve predicted by the neural network model simultaneously or by considering at least one of them, and it can be seen that the prediction curve of the physical model changes constantly with the model training.

[0075] In this embodiment, the relationship between the specification parameters of the shock absorber valve plate structure and the change curve is converted into an empirical formula, and the physical constraint is constructed by modifying the operation of the layer, so as to realize the fusion of the physical model and the neural network model.

[0076] AsFigure 2 The embodiment shown provides an electronic device, comprising:

[0077] at least one processor; and

[0078] a memory connected with the at least one processor in communication; wherein

[0079] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the above method. The at least one processor in the electronic device can perform the above method, and thus has at least the same advantages as the above method.

[0080] Optionally, the electronic device further comprises an interface for connecting various components, including a high-speed interface and a low-speed interface. Various components are connected with each other by different buses, and can be installed on a common motherboard or in other ways as needed. The processor can process instructions executed within the electronic device, including graphical information stored in the memory or on the memory to display a GUI (Graphical User Interface) on an external input / output device such as a display device coupled to the interface. In other embodiments, multiple processors can be used with multiple memories, and / or multiple buses can be used with multiple memories, if necessary. Similarly, multiple electronic devices can be connected (e.g., as a server array, a group of blade servers, or a multi-processor system), each device providing part of the necessary operations. Figure 2 The processor 301 is taken as an example in the embodiment.

[0081] The memory 302 is a computer readable storage medium, which can be used to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the vibration damper valve system tuning method based on artificial intelligence technology in the embodiment of the present application. The processor 301 performs various functional applications and data processing of the device by running the software programs, instructions and modules stored in the memory 302, that is, implements the above-mentioned vibration damper valve system tuning method based on artificial intelligence technology.

[0082] The memory 302 can include a program storage area and a data storage area, where the program storage area can store an operating system, application programs required by at least one function, and the data storage area can store data created according to the use of the terminal, etc. In addition, the memory 302 can include a high-speed random access memory, and can further include a nonvolatile memory such as at least one of a magnetic disk storage device, a flash memory device, or other non-volatile solid state memory device. In some examples, the memory 302 can further include a memory disposed remotely with respect to the processor 301, which can be connected to the device through a network. Examples of the above network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0083] The electronic device can further include an input device 303 and an output device 304. The processor 301, the memory 302, the input device 303, and the output device 304 can be connected by a bus or other means, Figure 2 The above connection by the bus is taken as an example.

[0084] The input device 303 can receive input digital or character information, and the output device 304 can include a display device, an auxiliary lighting device (e.g., an LED), a tactile feedback device (e.g., a vibration motor), etc. The display device can include, but is not limited to, a liquid crystal display (LCD), a light-emitting diode (LED) display, and a plasma display. In some embodiments, the display device can be a touch screen.

[0085] The embodiment provides a computer readable storage medium, and the medium stores computer instructions. The computer instructions are used to make a computer execute the above method. The computer instructions on the computer readable storage medium are used to make a computer execute the above method, and thus at least have the same advantages as the above method.

[0086] The medium in the present application can adopt any combination of one or more computer readable media. The medium can be a computer readable signal medium or a computer readable storage medium. The medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples (non-exhaustive list) of the medium include: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or component.

[0087] A computer readable signal medium can include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated signal can take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium can be any computer readable medium that can be involved in

[0088] The code can be transmitted in any form, including but not limited to electro-magnetic, optical, or any suitable combination thereof. A computer readable medium can comprise one or both of:

[0089] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0090] In the above embodiments, all or part can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another, for example, the computer instructions can be transferred from one website, computer, server or data center to another via wire, such as coaxial cable, optical fiber, digital subscriber line (DSL) or wireless, such as infrared, wireless, microwave, etc. The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be magnetic media (such as floppy disk, hard disk, magnetic tape), optical media or semiconductor media, etc. It should be noted that the computer-readable storage medium mentioned in the embodiments of the present application can be a non-volatile storage medium, in other words, it can be a non-transitory storage medium.

[0091] It should be understood that the above-mentioned various forms of processes can be reordered, added or deleted. For example, the steps described in the present application can be executed in parallel, sequentially or in different order, as long as the desired results of the technical solutions disclosed in the present application can be achieved, which is not limited herein.

[0092] The above specific embodiments do not constitute a limitation on the scope of protection of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for adjusting a valve train of a shock absorber based on artificial intelligence technology, characterized by, The method comprises the following steps: Collecting parameter combinations of each valve system of the shock absorber; Data quantization is performed on the parameter combinations to obtain feature combinations; The feature combinations are input into a neural network model to obtain a curve of the predicted damping force with respect to the piston speed predicted by the neural network; The loss function Loss of the neural network model during training is as follows: ； wherein, is a number of training samples, are loss weights of the physical model and the neural network model, respectively, is a change curve predicted by the physical model corresponding to the i-th training sample, is a change curve predicted by the neural network model corresponding to the i-th training sample, is a change curve obtained by performing an experiment on a single training sample; The curve predicted by the physical model satisfies the physical constraints of the shock absorber; The data quantization of the parameter combinations to obtain the feature combinations comprises the following steps: The opening area of the throttle valve is taken as the feature of the throttle valve; The equivalent thicknesses of the restoring valve, the flow valve, and the compression valve are calculated respectively, and the equivalent thicknesses are taken as the features of the restoring valve, the flow valve, and the compression valve respectively.

2. The method of claim 1, wherein the method is based on an artificial intelligence technique. The curve predicted by the physical model is obtained by correcting the curve predicted by the neural network. 3.The method of claim 1, wherein, The neural network model adopts a multilayer perceptron neural network as the main network, and simultaneously adopts a regularization technique to prevent network overfitting.

4. The method of claim 2, wherein the method further comprises: During the training of the neural network model, the curve predicted by the physical model is obtained in the following manner: The equivalent thickness of the restoring valve is obtained according to the i th training sample, and the slopes of each speed segment on the curve are obtained according to the equivalent thickness; The damping forces of each speed segment are extracted from the curve predicted by the neural network model corresponding to the i th training sample; The damping forces of each speed segment predicted by the physical model are obtained according to the extracted damping forces and slopes of each speed segment.

5. The method of claim 4, wherein the method further comprises: The slopes of each speed segment on the curve are obtained according to the equivalent thickness, which comprises the following steps: Corresponding point pairs of the equivalent thickness and the slopes of each speed segment are obtained by statistical analysis of the training samples; Linear fitting is performed according to the corresponding point pairs to obtain a plurality of relational expressions with the equivalent thickness as the independent variable and the slopes of each speed segment as the dependent variable; The slopes of each speed segment are obtained by respectively bringing the equivalent thickness into each relational expression.

6. The method of claim 4, wherein the method further comprises: During the training of the neural network model, the curve predicted by the physical model is obtained in the following manner: A constraint rule is obtained by statistical analysis of the training samples; The curve predicted by the physical model is obtained by correcting the curve predicted by the neural network model according to the constraint rule; The constraint rule comprises the relationship between the thickness of the restoring gasket and the speed inflection point on the curve, the relationship between the throttling flow of the throttle valve and the damping force of the low-speed segment on the curve, the relationship between the number of the restoring valves and the damping forces of the medium-speed segment and the high-speed segment on the curve, the relationship between the thickness of the restoring valve and the damping forces of the medium-speed segment and the high-speed segment on the curve, and the relationship between the equivalent thickness of the restoring valve and the slope of the curve.

7. A computer program product, characterised in that, The computer program product stores computer instructions, and the computer instructions are executed by a processor to realize the steps of the shock absorber valve system tuning method based on the artificial intelligence technology in any one of claims 1-6. The computer program product comprises:

8. An electronic device, comprising: At least one processor and a memory in communication connection with the at least one processor; ​ The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the shock absorber valve system tuning method based on artificial intelligence technology according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The medium stores computer instructions, and the computer instructions are used to enable a computer to perform the shock absorber valve system tuning method based on artificial intelligence technology according to any one of claims 1-6.

Images