Construction method, device and equipment of suspension system model
Through iterative update and sub-model superposition methods, the target model of the suspension system is built, which solves the problem of difficulty in accurately constructing complex suspension system models in the existing technology, and achieves the improvement of the model's accuracy.
Patent Information
- Application Number
- CN202411772247.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-04
- Publication Date
- 2025-03-14
AI Technical Summary
It is difficult to accurately construct complex suspension system models in the prior art, especially in operation in multiple degrees of freedom systems and different frequency ranges.
The target model of the target suspension system is constructed by building the initial model based on the target model type and the initial sample set, and through iterative update methods. In each iteration, a sub-model of the space to be optimized is constructed and the current model is superimposed with the sub-model to optimize the initial model.
The construction accuracy of the suspension system model is greatly improved, and the target suspension system can be accurately described, solving the problem of over-wide or over-fitting of the model in the prior art.
Smart Images
Figure CN119623090A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing, and in particular to a method, device and equipment for constructing a suspension system model. Background Art
[0002] In a vehicle, the suspension system is mainly used to isolate the vibrations of the engine and transmission to reduce the transmission of these vibrations into the vehicle cabin, thereby improving ride comfort and the NVH (Noise, Vibration and Harshness) performance of the entire vehicle. It plays a vital role in vehicle design and performance optimization.
[0003] In the process of designing a vehicle's suspension system, it is usually necessary to construct a suspension system model so as to predict and optimize the performance of the suspension system through the suspension system model.
[0004] However, the suspension system is usually a multi-degree-of-freedom system, that is, vibrations in multiple directions need to be considered simultaneously, there may be interactions between different degrees of freedom, and the suspension system needs to work effectively within different frequency ranges. Therefore, the suspension system is extremely complex, and existing methods make it difficult to accurately construct a suspension system model. Summary of the invention
[0005] In order to solve the above technical problems, the present application provides a method, device and equipment for constructing a suspension system model to solve the problem that the suspension system model cannot be accurately constructed in the prior art.
[0006] To achieve the above technical objectives, the present application provides the following technical solutions:
[0007] In a first aspect, an embodiment of the present specification provides a method for constructing a suspension system model, comprising:
[0008] constructing an initial model of a target suspension system based on the first target model type and the initial sample set;
[0009] The initial model is iteratively updated to obtain a target model of the target suspension system; wherein, in each iteration, a sub-model corresponding to the space to be optimized in the target design space of the target suspension system is constructed, and the current model is superimposed with the sub-model to obtain an optimization result of the initial model in the current iteration; the target design space includes a predetermined design space of the target suspension system or a space to be optimized determined in a previous iteration; the current model includes the initial model or the optimization result of the initial model in the previous iteration.
[0010] In one embodiment, the process of determining the space to be optimized in the target design space of the target suspension system includes:
[0011] Determining a space to be optimized in the target design space based on a change state of the current model in the target design space;
[0012] And / or, in response to an input operation on the space to be optimized, determining the space to be optimized in the target design space.
[0013] In one embodiment, the change state of the current model in the target design space includes the change rate of each output variable of the current model in the target design space;
[0014] Determining the space to be optimized in the target design space based on the change state of the current model in the target design space includes:
[0015] The space to be optimized is determined based on the relationship between the change rate of each output variable in the target design space and a predetermined change rate threshold.
[0016] In one embodiment, the process of determining the change state of the current model in the target design space includes:
[0017] Acquire a designated model corresponding to the target design space; the designated model includes the initial model and target sub-models determined in iterations before the current iteration, and the design space corresponding to the target sub-model includes the target design space;
[0018] When the number of the designated models is multiple, determining the sub-change state of each designated model in the target design space respectively; the sub-change state of the designated model in the target design space includes the sub-change rate of each output variable in the target design space;
[0019] The sub-change rates corresponding to the output variable are superimposed to obtain the change rate of the output variable in the target design space.
[0020] In one embodiment, constructing a sub-model corresponding to the space to be optimized of the target suspension system in the target design space includes:
[0021] Acquire a newly added sample set corresponding to the space to be optimized; the newly added sample set includes a plurality of input sets and an output set corresponding to each of the input sets;
[0022] Based on the second target model type and the newly added sample set, a sub-model corresponding to the space to be optimized is constructed.
[0023] In one implementation, obtaining the input set in the newly added sample set includes:
[0024] Based on a target sampling interval, sampling the target input variables of the current model in the space to be optimized to obtain a point set corresponding to each target input variable; wherein the target sampling interval is smaller than the sampling interval corresponding to the space to be optimized in the previous iteration;
[0025] Based on the point sets corresponding to the target input variables, an input set in the newly added sample set is determined.
[0026] In one embodiment, the process of determining the space to be optimized in the target design space of the target suspension system includes:
[0027] When the current iteration is the first iteration, determining whether the initial model has a discontinuous point on the predetermined design space;
[0028] When the discontinuity point exists, the design space including the discontinuity point is used as the space to be optimized.
[0029] In one embodiment, after obtaining the target model of the target suspension system, the method further includes:
[0030] Constructing a model tree corresponding to the target model, the model tree comprising a plurality of levels, the plurality of levels corresponding to the number of iterations corresponding to each model in the target model, each model in the target model comprising the initial model and the sub-model determined in each iteration;
[0031] The target model is stored via the model tree.
[0032] In a second aspect, the embodiments of this specification provide a device for constructing a suspension system model, including:
[0033] A first processing module, configured to construct an initial model of a target suspension system based on a first target model type and an initial sample set;
[0034] The second processing module is used to iteratively update the initial model to obtain the target model of the target suspension system; wherein, in each iteration, a sub-model corresponding to the space to be optimized in the target design space of the target suspension system is constructed, and the current model is superimposed with the sub-model to obtain the optimization result of the initial model in the current iteration; the target design space includes the predetermined design space of the target suspension system or the space to be optimized determined in the previous iteration; the current model includes the initial model or the optimization result of the initial model in the previous iteration.
[0035] In a third aspect, an embodiment of the present specification provides an electronic device, the electronic device comprising at least one processor and at least one memory, the memory storing a computer program, and the computer program, when executed by the processor, implements a method for constructing a suspension system model as described in any one of the above items.
[0036] In a fourth aspect, an embodiment of the present specification provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for constructing a suspension system model as described in any one of the above items is implemented.
[0037] In a fifth aspect, an embodiment of the present specification provides a computer program product or a computer program, wherein the computer program product includes a computer program, and the computer program is stored in a computer-readable storage medium; the processor of the computer device reads the computer program from the computer-readable storage medium, and when the processor executes the computer program, it implements the method for constructing a suspension system model as described in any one of the above items.
[0038] It can be seen from the above technical scheme that the embodiment of the present application provides a method, device and equipment for constructing a suspension system model. The method constructs an initial model of a target suspension system based on a first target model type and an initial sample set, so as to obtain a relatively broad suspension system model; and iteratively updates the initial model to obtain a target model of the target suspension system, wherein in each iteration, a sub-model corresponding to the space to be optimized in the target design space of the target suspension system is constructed, and the current model is superimposed with the sub-model to obtain the optimization result of the initial model in the current iteration, the target design space includes a predetermined design space of the target suspension system or a space to be optimized determined in the previous iteration, and the current model includes the initial model or the optimization result of the initial model in the previous iteration, so that on the basis of a relatively broad suspension system model, local detail features can be gradually added, so that the suspension system model finally optimized can accurately describe the target suspension system, thereby greatly improving the construction accuracy of the suspension system model. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.
[0040] Figure 1 A schematic flow chart of a method for constructing a suspension system model provided for an implementation of this specification.
[0041] Figure 2 A schematic diagram of the structure of a device for constructing a suspension system model provided in an embodiment of this specification. DETAILED DESCRIPTION
[0042] Unless otherwise defined, the technical terms or scientific terms used in the embodiments of this specification shall have the common meanings understood by persons with ordinary skills in the field to which this specification belongs. The words "first", "second" and similar words used in the embodiments of this specification do not indicate any order, quantity or importance, but are only used to avoid confusion of constituent elements.
[0043] Unless the context requires otherwise, throughout the specification, "plurality" means "at least two", and "including" is interpreted as an open, inclusive meaning, that is, "including, but not limited to". In the description of the specification, the terms "one embodiment", "some embodiments", "exemplary embodiments", "examples", "specific examples" or "some examples" are intended to indicate that a particular feature, structure, material or characteristic associated with the embodiment or example is included in at least one embodiment or example of the specification. The schematic representation of the above terms does not necessarily refer to the same embodiment or example.
[0044] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.
[0045] Overview
[0046] As described in the background technology, in a vehicle, the suspension system is mainly used to isolate the vibration of the engine and transmission to reduce the transmission of these vibrations into the vehicle cabin, improve ride comfort and the NVH performance of the vehicle, and plays a vital role in vehicle design and performance optimization.
[0047] In the process of designing a vehicle's suspension system, it is usually necessary to construct a suspension system model so as to predict and optimize the performance of the suspension system through the suspension system model.
[0048] However, the suspension system is usually a multi-degree-of-freedom system, that is, vibrations in multiple directions need to be considered simultaneously, there may be interactions between different degrees of freedom, and the suspension system needs to work effectively within different frequency ranges. Therefore, the suspension system is extremely complex.
[0049] In the process of constructing a suspension system, the existing methods usually select a certain number of sample points by sampling and test these sample points to obtain a series of combinations of system inputs and outputs, and use these input and output combinations to fit the important coefficients of a certain analytical expression to obtain an approximate model. The inventors of the present application have found through research that due to the complexity of the suspension system, when a low-order model is used to construct a suspension system model, the constructed model is difficult to describe the detailed characteristics of the suspension system. When a high-order model is used to construct a suspension system model, the constructed model is prone to overfitting oscillation and is difficult to reliably describe the characteristics of the suspension system. Therefore, it is difficult to accurately construct a suspension system model using existing methods.
[0050] In order to solve the problem that the suspension system model cannot be accurately constructed in the traditional method, in the technical solution of the present application, based on the first target model type and the initial sample set, an initial model of the target suspension system is constructed, so that a relatively broad suspension system model can be obtained; and the initial model is iteratively updated to obtain the target model of the target suspension system, wherein in each iteration, a sub-model corresponding to the space to be optimized in the target design space of the target suspension system is constructed, and the current model is superimposed with the sub-model to obtain the optimization result of the initial model in the current iteration, the target design space includes the predetermined design space of the target suspension system or the space to be optimized determined in the previous iteration, and the current model includes the initial model or the optimization result of the initial model in the previous iteration, so that on the basis of the relatively broad suspension system model, local detail features can be gradually added, so that the suspension system model finally optimized can accurately describe the target suspension system, and the construction accuracy of the suspension system model is greatly improved.
[0051] Based on the above-mentioned inventive concept, the method for constructing the suspension system model provided in the embodiment of this specification is exemplarily described below.
[0052] Exemplary Methods
[0053] The embodiments of this specification provide a method for constructing a suspension system model, such as Figure 1 As shown, including:
[0054] S101: construct an initial model of a target suspension system based on a first target model type and an initial sample set.
[0055] Specifically, the target suspension system may be a suspension system to be modeled. The first target model type may be the type of an initial model of the target suspension system, the first target model type may be a default type, or may be determined in response to an input operation of a relevant person on the model type, for example, the first target model type may be a response surface model, a Kriging model, etc.
[0056] The initial sample set may include multiple input sets and an output set corresponding to each input set. For any input set in the initial sample set, the input set may include sampled values of each input variable of the initial model of the target suspension system. The output set corresponding to the input set may include the response values of each output variable of the initial model of the target suspension system. Taking the four-point suspension system as an example, the input variables of the initial model may include the static stiffness of each suspension point of the target suspension system in the vertical direction of the local coordinate system (the direction perpendicular to the plane where the four-point suspension system is located). For example, the input variable x∈{x 1 ,x 2 ,x 3 ,x 4}, x 1 ~x 4 are the static stiffness of the front left suspension point, front right suspension point, rear left suspension point and rear right suspension point in the vertical direction of their local coordinate systems. The output variables of the initial model may include the energy decoupling rate and vibration frequency of the target suspension system in each direction. For example, the output variable y∈{y 1 ,y 2 ,…,y 12},y 1 ~y 12 They are respectively the forward energy decoupling rate and vibration frequency, the backward energy decoupling rate and vibration frequency, the lateral energy decoupling rate and vibration frequency, the vertical energy decoupling rate and vibration frequency, the roll energy decoupling rate and vibration frequency, the pitch energy decoupling rate and vibration frequency, and the yaw energy decoupling rate and vibration frequency.
[0057] In implementation, sampling can be performed in a predetermined design space based on a specified sampling method and sampling interval. The predetermined design space may include a value interval for each input variable, which may be pre-configured or determined in response to an input operation of a person on the design space. The sampling method and sampling interval may be default values or may be determined in response to an input operation of the relevant person on the sampling method and sampling interval. For example, the sampling method may be a two-level sampling method, a three-level sampling method, a Latin hypercube sampling method, etc. For example, each input variable may be sampled based on a specified sampling method and sampling interval to obtain a point set corresponding to each input variable. Thus, the sampling points in the point set corresponding to each input variable may be arranged and combined to obtain multiple input sets in the initial sample set. For any input set in the initial sample set, the response value of each output variable under the input set may be obtained by experiment and used as the output set corresponding to the input set.
[0058] Therefore, an input set and an output set corresponding to the input set can be used as a sample, and the relevant parameters in the analytical expression corresponding to the first target model type can be fitted through each sample in the initial sample set to obtain the initial model of the target suspension system.
[0059] It can be understood that the model type of the initial model can adopt a lower-order model, and the sampling interval corresponding to the initial sample set can adopt a larger sampling interval to reduce the amount of calculation and the number of experiments, and can effectively avoid the phenomenon of overfitting oscillation in the initial model.
[0060] S102. Iteratively update the initial model to obtain the target model of the target suspension system; wherein, in each iteration, a sub-model corresponding to the space to be optimized in the target design space of the target suspension system is constructed, and the current model is superimposed with the sub-model to obtain the optimization result of the initial model in the current iteration; the target design space includes the predetermined design space of the target suspension system or the space to be optimized determined in the previous iteration; the current model includes the initial model or the optimization result of the initial model in the previous iteration.
[0061] Specifically, after the initial model is determined, the initial model may be iteratively updated, and the update result of the initial model in the last iteration is used as the target model of the target suspension system.
[0062] For any iteration, in the current iteration, a sub-model corresponding to the space to be optimized of the target suspension system can be constructed. The space to be optimized of the target suspension system can be a sub-space in the target design space. Among them, when the current iteration is the first iteration, the target design space can be a predetermined design space of the target suspension system, that is, the area where the initial model needs to be optimized is determined in the predetermined design space. When the current iteration is any iteration after the first iteration, the target design space can be the space to be optimized determined in the previous iteration, that is, the model is further optimized on the basis of the previous optimization.
[0063] It can be understood that the number of spaces to be optimized in the current iteration can be one or more. In addition, in the current iteration, if the space to be optimized does not exist, that is, there is no area to be optimized, it can be determined that the iteration termination condition is met, and the target model of the target suspension system is output. In addition, it can also be determined that the iteration termination condition is met when a predetermined number of iterations is reached, which can be specifically set according to actual needs.
[0064] When constructing the sub-model corresponding to the space to be optimized of the target suspension system, the space to be optimized can be re-sampled, and the sub-model of the target suspension system in the space to be optimized can be constructed based on the sampling results. The model type of the sub-model can be the same as the first target model type, or different from the first target model type, and the method for sampling the space to be optimized can be the same as the sampling method of the target design space, or different from the sampling method of the target design space. The sampling interval for sampling the space to be optimized can be smaller than the sampling interval of the target design space, so that the sub-model corresponding to the space to be optimized can describe more detailed characteristics of the target suspension system.
[0065] It can be understood that in the current iteration, a sub-model can be constructed for any space to be optimized, and the value range of the input variables of each sub-model constructed in the current iteration can be the value range of the input variable in the corresponding space to be optimized, that is, each sub-model constructed in the same iteration is a parallel model located in different design spaces.
[0066] After constructing the sub-models corresponding to each space to be optimized, the current model and each sub-model can be superimposed to obtain the optimization result of the initial model in the current iteration. When the current iteration is the first iteration, the current model can be the initial model, and when the current iteration is any iteration after the first iteration, the current model can be the optimization result of the initial model in the previous iteration.
[0067] In implementation, the initial model and the sub-models determined in each iteration may be linearly superimposed based on the predetermined model coefficients. It can be shown as formula (1):
[0068]
[0069] In the formula, is the initial model, Including the sub-models constructed in the rth iteration, c 0 and c r are the predetermined model coefficients corresponding to the initial model and each sub-model constructed in the rth iteration, is an input vector including various input variables, and r is a positive integer.
[0070] Therefore, in the process of constructing the target model of the target suspension system, local detail features can be gradually added on the basis of a relatively broad suspension system model, so that the final optimized suspension system model can accurately describe the target suspension system, thereby greatly improving the construction accuracy of the suspension system model.
[0071] In a feasible implementation manner, the process of determining the space to be optimized in the target design space of the target suspension system includes:
[0072] Determining a space to be optimized in the target design space based on a change state of the current model in the target design space;
[0073] And / or, in response to an input operation on the space to be optimized, determining the space to be optimized in the target design space.
[0074] Specifically, in the process of determining the space to be optimized in the target design space, the space to be optimized in the target design space can be determined based on the change state of the current model in the target design space. The change state of the current model in the target design space may include the change rate of each output variable of the current model in the target design space, wherein the greater the change rate, the more detailed features of the part, and thus, the space to be optimized in the target design space can be determined according to the size of the change rate of each output variable of the current model in the target design space. It can be understood that when the number of target design spaces is multiple in the current iteration, the space to be optimized in each target design space can be determined separately, so that the space to be optimized in each target design space can be quickly and accurately determined.
[0075] At the same time, in the current iteration, the space to be optimized in the target design space can also be determined in response to the input operation of the relevant personnel on the space to be optimized. For example, the relevant personnel can customize the space to be optimized in each target design space according to experience or needs, thereby improving the flexibility in the construction process of the suspension system model.
[0076] In a feasible implementation manner, the change state of the current model in the target design space includes the change rate of each output variable of the current model in the target design space;
[0077] Determining the space to be optimized in the target design space based on the change state of the current model in the target design space includes:
[0078] The space to be optimized is determined based on the relationship between the change rate of each output variable in the target design space and a predetermined change rate threshold.
[0079] Specifically, for any output variable among the output variables of the current model, the change rate of the output variable in the target design space may include the change rate of the output variable in each subspace of the target design space. For any subspace, the subspace may be such that the value interval of the current input variable is the interval between two adjacent sampling points in the previous iteration, and each input variable other than the current input variable is a fixed value, that is, the change rate of the output variable in the subspace is the change rate of the output variable in the corresponding value interval of the current input variable when each input variable other than the current input variable is a corresponding fixed value. It can be understood that the current input variable may be any input variable, and the values of each input variable other than the current input variable may be the permutation and combination of the sampling points of the corresponding input variable in the design space during the previous iteration.
[0080] For example, taking the four-point suspension system as an example, in the previous iteration, x 1 The sampling points are 50, 100, 150, 200, 250, 300, x 2 The sampling points are 50, 100, 150, 200, 250, 300, x 3 The sampling points are 50, 100, 150, 200, 250, 300, x 4 The sampling points of x are 50, 100, 150, 200, 250, and 300 respectively. The subspace can include: 2 、x 3 and x 4 Perform permutations and combinations to obtain 6*6*6=216 combinations. For any of the 216 combinations, x 1 The value intervals of are 50-100, 100-150, 150-200, 200-250, and 250-300. That is, 216*5=1080 subspaces can be obtained. Therefore, in x 1 、x 3 and x 4 When a fixed value is taken, 1080 subspaces can be obtained. 1 、x 2 and x 4 When a fixed value is taken, 1080 subspaces can be obtained, as well as 1 、x 2 and x 3 When a fixed value is taken, 1080 subspaces can be obtained, that is, a total of 1080*4=4320 subspaces can be obtained.
[0081] For any subspace, the rate of change of each output variable in the subspace can be obtained separately. For example, for x 2 =50, x 3=100, x 4 =50 and x 1 The value range of is 50 to 100 in the subspace, and the output variable y 1 The rate of change in this subspace is in, For x 1 =100 when y 1 The value of For x 1 =50 when y 1 The value of .
[0082] During implementation, for any subspace, the change rate of each output variable in the subspace can be compared with a predetermined change rate threshold. If there is at least one output variable in the subspace whose change rate is greater than the predetermined change rate threshold, the subspace is taken as a candidate subspace, and the candidate subspaces are merged to obtain the space to be optimized, thereby effectively ensuring the reliability of the determination result of the space to be optimized. Furthermore, by constructing a sub-model corresponding to the space to be optimized, the characteristics of the target suspension system in the space to be optimized can be effectively described, thereby further improving the accuracy of the suspension system model.
[0083] In a feasible implementation manner, the process of determining the change state of the current model in the target design space includes:
[0084] Acquire a designated model corresponding to the target design space; the designated model includes the initial model and target sub-models determined in iterations before the current iteration, and the design space corresponding to the target sub-model includes the target design space;
[0085] When the number of the designated models is multiple, determining the sub-change state of each designated model in the target design space respectively; the sub-change state of the designated model in the target design space includes the sub-change rate of each output variable in the target design space;
[0086] The sub-change rates corresponding to the output variable are superimposed to obtain the change rate of the output variable in the target design space.
[0087] Specifically, the designated model corresponding to the target design space may include all models in the current model, and the corresponding design space includes the target design space, for example, it may include the initial model and the target sub-model in each sub-model determined in each iteration before the current iteration, and the target sub-model is the sub-model of the corresponding design space including the target design space. That is, when the current iteration is the first iteration, the designated model corresponding to the target design space is the initial model. When the current iteration is any iteration after the first iteration, the designated model corresponding to the target design space may include the initial model, and may also include the target sub-model in each sub-model determined in each iteration before the current iteration.
[0088] For any specified model, the sub-change state of the specified model in the target design space may include the sub-change rate of each output variable in the target design space. Wherein, for any output variable, the sub-change rate of the output variable in the target design space may include the sub-change rate of each subspace of the output variable in the target design space.
[0089] In implementation, for any output variable, when the number of specified models is multiple, a sub-change rate of the output variable in the target design space can be obtained for each specified model, thereby, multiple sub-change rates of the output variable in the target design space can be obtained, and the multiple sub-change rates can be superimposed to obtain the change rate of the output variable in the target design space. For example, for any subspace in the target design space, the sub-change rates of the output variable in the subspace can be superimposed to obtain the change rate of the output variable in the subspace, so that the change rate of the output variable in each subspace of the target design space can be obtained, and then the change rate of the output variable in each subspace of the target design space is used as the change rate of the output variable in the target design space.
[0090] In the process of superimposing multiple sub-rates of change, linear superposition can be performed according to the predetermined model coefficients corresponding to each specified model. For example, when the number of specified models is two, the output variable y 1 The rate of change in this subspace k = c 0 k 0 +c 1 k 1 , k 0 is the output variable y determined according to the initial model 1 The sub-rate of change in this subspace, k 1 is the output variable y determined according to the specified model built in the first iteration 1 The sub-rate of change in this subspace, c 0 and c 1 are the predetermined model coefficients corresponding to the initial model and the sub-model constructed in the first iteration, respectively.
[0091] Thus, the change rate of each output variable in the target design space can be obtained, so that the change state of the current model in the target design space can be determined quickly and accurately.
[0092] In a feasible implementation manner, the step of constructing a sub-model corresponding to the space to be optimized of the target suspension system in the target design space includes:
[0093] Acquire a newly added sample set corresponding to the space to be optimized; the newly added sample set includes a plurality of input sets and an output set corresponding to each of the input sets;
[0094] Based on the second target model type and the newly added sample set, a sub-model corresponding to the space to be optimized is constructed.
[0095] Specifically, in the process of constructing the sub-model corresponding to the target suspension system in the space to be optimized, a newly added sample set corresponding to the space to be optimized can be obtained. The newly added sample set may include multiple input sets and an output set corresponding to each input set, wherein the newly added sample set may include input sets that are different from the input sets in the initial sample set located in the space to be optimized.
[0096] Optionally, obtaining an input set in the newly added sample set includes:
[0097] Based on a target sampling interval, sampling the target input variables of the current model in the space to be optimized to obtain a point set corresponding to each target input variable; wherein the target sampling interval is smaller than the sampling interval corresponding to the space to be optimized in the previous iteration;
[0098] Based on the point sets corresponding to the target input variables, an input set in the newly added sample set is determined.
[0099] Specifically, in the process of obtaining the input set in the newly added sample set, the target input variable in each input variable can be sampled based on the target sampling interval. The target input variable may include an input variable whose value interval is not a fixed value in the space to be optimized. For example, taking the four-point suspension system as an example, if the space to be optimized is x 2 =50, x 3 =100, x 4 =50 and x 1 The value range of x is 200 to 300. 1As the target input variable, sampling is performed based on the target sampling interval in the value interval of 200 to 300 to obtain the point set corresponding to the target input variable. The method for sampling each target input variable can be a default value, or can be determined in response to the input operation of the sampling method by the relevant personnel in the current iteration. For example, the sampling method can be a two-level sampling method, a three-level sampling method, a Latin hypercube sampling method, etc.
[0100] In implementation, the target sampling interval may be smaller than the sampling interval corresponding to the space to be optimized in the previous iteration. It is understandable that when the current iteration is the first iteration, the sampling interval corresponding to the space to be optimized in the previous iteration may be the sampling interval used when determining each input set in the initial sample set. Thus, by sampling in the space to be optimized with a smaller sampling interval to construct a new sample set, more detailed features of the target suspension system can be obtained, and the accuracy of the suspension system model is effectively improved. At the same time, only a small amount of sampling data is required in the area outside the space to be optimized, which greatly reduces the number of sampling points, thereby reducing the number of experiments in the process of determining the output set.
[0101] After obtaining the point set corresponding to the target input variable, the point set corresponding to each target input variable and the numerical value corresponding to each input variable other than the target input variable can be arranged and combined to obtain the input set in the newly added sample set. In addition, for any input set in the newly added sample set, the response value of each output variable under the input set can be obtained by experiment, and used as the output set corresponding to the input set.
[0102] After obtaining the newly added sample set, a sub-model corresponding to the space to be optimized can be constructed based on the second target model type and the newly added sample set. The second target model type can be the same as the first target model type or different from the first target model type. The second target model type can be a default value, for example, the model type used in different iterations can be pre-configured. In addition, the second target model type can also be determined in response to the input operation of the model type by the relevant personnel in the current iteration.
[0103] During implementation, an input set and the output set corresponding to the input set can be taken as a sample. By adding each sample in the sample set, the relevant parameters in the analytical expression corresponding to the second target model type are fitted to obtain a sub-model corresponding to the space to be optimized, so that the sub-model in each iteration can be constructed flexibly and accurately, and then through the sub-model, the more detailed characteristics of the target suspension system can be more accurately described.
[0104] In a feasible implementation manner, the process of determining the space to be optimized in the target design space of the target suspension system includes:
[0105] When the current iteration is the first iteration, determining whether the initial model has a discontinuous point on the predetermined design space;
[0106] When the discontinuity point exists, the design space including the discontinuity point is used as the space to be optimized.
[0107] Specifically, considering the complexity of the suspension system, when the current iteration is the first iteration, it is also possible to determine whether the initial model has a discontinuous point in the predetermined design space. For example, it is possible to determine whether there is a point in the predetermined design space that makes the denominator of the initial model equal to 0, and it is also possible to generate a graph corresponding to the initial model, and determine whether there is a discontinuous point based on the graph.
[0108] When there is a discontinuity, the design space including the discontinuity can be used as the space to be optimized. For example, when the discontinuity has an intersection with the space to be optimized determined based on other methods, the space to be optimized determined based on other methods can be used as the design space including the discontinuity. When the discontinuity does not have an intersection with the space to be optimized determined based on other methods, the design space of a predetermined size including the discontinuity can be used as the space to be optimized. The predetermined size can be a default value or can be determined based on the sampling interval corresponding to the target design space. For example, the predetermined size can be the size of a sampling interval on the design space in the last iteration. The space to be optimized determined based on other methods can be a space to be optimized determined based on the changing state of the current model in the target design space and / or in response to an input operation on the space to be optimized.
[0109] In implementation, in the process of constructing a sub-model corresponding to the space to be optimized including discontinuous points, the sub-model corresponding to the space to be optimized can be constructed by a step function.
[0110] It can be understood that when there are no discontinuities, the space to be optimized in the target design space can be further determined based on other methods. For example, the space to be optimized in the target design space can be determined based on the changing state of the current model in the target design space and / or in response to input operations on the space to be optimized.
[0111] In a feasible implementation manner, after obtaining the target model of the target suspension system, the method further includes:
[0112] Constructing a model tree corresponding to the target model, the model tree comprising a plurality of levels, the plurality of levels corresponding to the number of iterations corresponding to each model in the target model, each model in the target model comprising the initial model and the sub-model determined in each iteration;
[0113] The target model is stored via the model tree.
[0114] Specifically, after obtaining the target model of the target suspension system, a model tree corresponding to the target model can also be constructed. The model tree can include multiple levels, and the multiple levels can correspond to the number of iterations corresponding to each model in the target model. Each model in the target model can include an initial model and a sub-model determined in each iteration, that is, the level of the model tree can be n+1, n is the number of iterations, and each sub-model constructed in the same iteration can be stored in the same level of the model tree. In implementation, the model tree can be constructed in order from small to large number of iterations.
[0115] The target model can be stored through the model tree. Any level in the model tree can include at least one branch. In implementation, the branch where the sub-model is located can be determined according to the design space corresponding to the sub-model. The design space corresponding to each sub-model on the same branch can include the same area, so as to quickly load the corresponding sub-model according to the design space, so as to quickly and effectively load each model in the target model, and then facilitate related calculations and analysis according to the target model.
[0116] As an optional implementation, the target model can be stored in DOM (Document Object Model Tree), which can describe the target model based on text in XML (Extensible Markup Language) format. For any layer in DOM, it can include level nodes, and the attributes of the level nodes can include "level" and "name". "Level" can be used to indicate the level of the layer in the model tree, and "name" can be used to indicate the order name of the corresponding model in the target model, for example, 0th order, 1st order, 2nd order, etc. The order of the corresponding model in the target model can be determined by the number of iterations corresponding to the model. For example, the order of the initial model is 0th order, and the order of the sub-model constructed in the jth iteration is jth order, where j is a positive integer. The child nodes of the hierarchical nodes can nest one or more model nodes. The model nodes are used to describe the corresponding model information. For example, the attributes of the model nodes may include "category" and "name". "category" can be used to indicate the type of the corresponding model, such as RSM (Response Surface Methodology), and "name" can be used to indicate the name of the corresponding model, such as the second-order response surface model.
[0117] It is understandable that the structure of the target model (i.e., the content of the sub-nodes of the model node) can support expansion. For example, taking the second-order response surface model as an example, the sub-nodes of the model node include "Variables", "Ranges", "Response" and "Coefficients". "Variables" can be used to represent input variables, "Ranges" can be used to represent the design space of the corresponding model, that is, including the value interval of each input variable in the corresponding model, "Response" can be used to represent the output variable, and "Coefficients" can be used to represent the parameter matrix of the corresponding model. Among them, "Variables", "Ranges" and "Response" can be sub-nodes that all model nodes must contain, and "Coefficients" can be a sub-node unique to the model node.
[0118] In the process of loading the target model to calculate the response value of a certain point of the target model in the predetermined design space, the model can be traversed according to the hierarchy. In any hierarchy, all models of the corresponding order can be traversed, and it is determined whether the point falls into the design space corresponding to each model. If so, the model is loaded and the response value is calculated. If not, it is skipped. After the traversal is completed, all the response values are linearly accumulated to obtain the response value of the point.
[0119] Exemplary Devices
[0120] In an exemplary embodiment of the present specification, a device for constructing a suspension system model is also provided, such as Figure 2 As shown, including:
[0121] A first processing module 201 is used to construct an initial model of a target suspension system based on a first target model type and an initial sample set;
[0122] The second processing module 202 is used to iteratively update the initial model to obtain the target model of the target suspension system; wherein, in each iteration, a sub-model corresponding to the space to be optimized in the target design space of the target suspension system is constructed, and the current model is superimposed with the sub-model to obtain the optimization result of the initial model in the current iteration; the target design space includes the predetermined design space of the target suspension system or the space to be optimized determined in the previous iteration; the current model includes the initial model or the optimization result of the initial model in the previous iteration.
[0123] In a feasible implementation manner, the second processing module 202 is specifically configured to:
[0124] Determining a space to be optimized in the target design space based on a change state of the current model in the target design space;
[0125] And / or, in response to an input operation on the space to be optimized, determining the space to be optimized in the target design space.
[0126] In a feasible implementation manner, the change state of the current model in the target design space includes the change rate of each output variable of the current model in the target design space;
[0127] The second processing module 202 is specifically used for:
[0128] The space to be optimized is determined based on the relationship between the change rate of each output variable in the target design space and a predetermined change rate threshold.
[0129] In a feasible implementation manner, the second processing module 202 is specifically configured to:
[0130] Acquire a designated model corresponding to the target design space; the designated model includes the initial model and target sub-models determined in iterations before the current iteration, and the design space corresponding to the target sub-model includes the target design space;
[0131] When the number of the designated models is multiple, determining the sub-change state of each designated model in the target design space respectively; the sub-change state of the designated model in the target design space includes the sub-change rate of each output variable in the target design space;
[0132] The sub-change rates corresponding to the output variable are superimposed to obtain the change rate of the output variable in the target design space.
[0133] In a feasible implementation manner, the second processing module 202 is specifically configured to:
[0134] Acquire a newly added sample set corresponding to the space to be optimized; the newly added sample set includes a plurality of input sets and an output set corresponding to each of the input sets;
[0135] Based on the second target model type and the newly added sample set, a sub-model corresponding to the space to be optimized is constructed.
[0136] In a feasible implementation manner, the second processing module 202 is specifically configured to:
[0137] Based on a target sampling interval, sampling the target input variables of the current model in the space to be optimized to obtain a point set corresponding to each target input variable; wherein the target sampling interval is smaller than the sampling interval corresponding to the space to be optimized in the previous iteration;
[0138] Based on the point sets corresponding to the target input variables, an input set in the newly added sample set is determined.
[0139] In a feasible implementation manner, the second processing module 202 is specifically configured to:
[0140] When the current iteration is the first iteration, determining whether the initial model has a discontinuous point on the predetermined design space;
[0141] When the discontinuity point exists, the design space including the discontinuity point is used as the space to be optimized.
[0142] In a feasible implementation manner, the second processing module 202 is further configured to:
[0143] After obtaining the target model of the target suspension system, constructing a model tree corresponding to the target model, the model tree including multiple levels, the multiple levels corresponding to the iteration numbers corresponding to each model in the target model, each model in the target model including the initial model and the sub-model determined in each iteration;
[0144] The target model is stored via the model tree.
[0145] The device for constructing the suspension system model provided in this embodiment belongs to the same application concept as the method for constructing the suspension system model provided in the above embodiments of this application, and can execute the method for constructing the suspension system model provided in any of the above embodiments of this application, and has the corresponding functional modules and beneficial effects of executing the method for constructing the suspension system model. For technical details not fully described in this embodiment, please refer to the specific processing content of the method for constructing the suspension system model provided in the above embodiments of this application, and will not be repeated here.
[0146] Exemplary Devices
[0147] In an exemplary embodiment of the present specification, an electronic device is also provided, the electronic device comprising at least one processor and at least one memory, the memory storing a computer program, and the computer program, when executed by the processor, implements the method for constructing a suspension system model as described in any of the above embodiments.
[0148] Exemplary computer program products and storage media
[0149] In addition to the above-mentioned methods and devices, the method for constructing a suspension system model provided in the embodiments of this specification may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method for constructing a suspension system model according to various embodiments of this specification described in the above "Exemplary Method" section of this specification.
[0150] The computer program product can be written in any combination of one or more programming languages to write program codes for executing the operations of the embodiments of this specification, and the programming languages include object-oriented programming languages, such as Java, C++, etc., and also include conventional procedural programming languages, such as "C" language or similar programming languages.
[0151] In addition, an embodiment of the present specification also provides a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to execute the steps in the method for constructing a suspension system model according to various embodiments of the present specification described in the above "Exemplary Method" section of the present specification.
[0152] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this specification can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0153] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0154] The above-mentioned embodiments only express several implementation methods of this specification, and the descriptions are relatively specific and detailed, but they cannot be understood as limiting the scope of the solutions provided by the embodiments of this specification. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of this specification, which all belong to the protection scope of this specification. Therefore, the protection scope of the patent of this specification shall be based on the attached claims.
Claims
1. A method for constructing a suspension system model, characterized in that: include: constructing an initial model of a target suspension system based on the first target model type and the initial sample set; The initial model is iteratively updated to obtain a target model of the target suspension system; wherein, in each iteration, a sub-model corresponding to the space to be optimized in the target design space of the target suspension system is constructed, and the current model is superimposed with the sub-model to obtain an optimization result of the initial model in the current iteration; the target design space includes a predetermined design space of the target suspension system or a space to be optimized determined in a previous iteration; the current model includes the initial model or the optimization result of the initial model in the previous iteration.
2. The method according to claim 1, characterized in that The process of determining the space to be optimized in the target design space of the target suspension system includes: Determining a space to be optimized in the target design space based on a change state of the current model in the target design space; And / or, in response to an input operation on the space to be optimized, determining the space to be optimized in the target design space.
3. The method according to claim 2, characterized in that The change state of the current model in the target design space includes the change rate of each output variable of the current model in the target design space; Determining the space to be optimized in the target design space based on the change state of the current model in the target design space includes: The space to be optimized is determined based on the relationship between the change rate of each output variable in the target design space and a predetermined change rate threshold.
4. The method according to claim 2, characterized in that: The process of determining the change state of the current model in the target design space includes: Acquire a designated model corresponding to the target design space; the designated model includes the initial model and target sub-models determined in iterations before the current iteration, and the design space corresponding to the target sub-model includes the target design space; When the number of the designated models is multiple, determining the sub-change state of each designated model in the target design space respectively; the sub-change state of the designated model in the target design space includes the sub-change rate of each output variable in the target design space; The sub-change rates corresponding to the output variable are superimposed to obtain the change rate of the output variable in the target design space.
5. The method according to claim 1, characterized in that The step of constructing a sub-model corresponding to the space to be optimized of the target suspension system in the target design space includes: Acquire a newly added sample set corresponding to the space to be optimized; the newly added sample set includes a plurality of input sets and an output set corresponding to each of the input sets; Based on the second target model type and the newly added sample set, a sub-model corresponding to the space to be optimized is constructed.
6. The method according to claim 5, characterized in that Obtaining an input set from the newly added sample set includes: Based on a target sampling interval, sampling the target input variables of the current model in the space to be optimized to obtain a point set corresponding to each target input variable; wherein the target sampling interval is smaller than the sampling interval corresponding to the space to be optimized in the previous iteration; Based on the point sets corresponding to the target input variables, an input set in the newly added sample set is determined.
7. The method according to claim 1, characterized in that The process of determining the space to be optimized in the target design space of the target suspension system includes: When the current iteration is the first iteration, determining whether the initial model has a discontinuous point on the predetermined design space; When the discontinuity point exists, the design space including the discontinuity point is used as the space to be optimized.
8. The method according to any one of claims 1 to 7, characterized in that: After obtaining the target model of the target suspension system, the method further includes: Constructing a model tree corresponding to the target model, the model tree comprising a plurality of levels, the plurality of levels corresponding to the number of iterations corresponding to each model in the target model, each model in the target model comprising the initial model and the sub-model determined in each iteration; The target model is stored via the model tree.
9. A device for constructing a suspension system model, characterized in that: include: A first processing module, configured to construct an initial model of a target suspension system based on a first target model type and an initial sample set; The second processing module is used to iteratively update the initial model to obtain the target model of the target suspension system; wherein, in each iteration, a sub-model corresponding to the space to be optimized in the target design space of the target suspension system is constructed, and the current model is superimposed with the sub-model to obtain the optimization result of the initial model in the current iteration; the target design space includes the predetermined design space of the target suspension system or the space to be optimized determined in the previous iteration; the current model includes the initial model or the optimization result of the initial model in the previous iteration.
10. An electronic device, characterized in that: The electronic device includes at least one processor and at least one memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the method for constructing a suspension system model according to any one of claims 1 to 8 is implemented.