A novel lightweight optimization design system for vehicle chassis truss

Abstract

The application discloses a novel light-weight optimization design system for a vehicle chassis truss, and relates to the technical field of vehicle chassis truss optimization design; the design steps of the optimization design system are as follows: firstly, an adaptive spiral search strategy is introduced on the basis of a multi-objective hybrid meta-heuristic algorithm based on adaptive differential evolution of successful history, and then a novel SHAMODE-IWOA algorithm is proposed; and then, in order to estimate the reliability level of the chassis truss structure under different design parameter combinations, the proposed SHAMODE-IWOA algorithm is used to learn an ANFIS model and a novel ANFIS-SHAMODE-IWOA model is constructed. The proposed ANFIS-SHAMODE-IWOA multi-objective optimization design system is a novel intelligent model, can be used for evaluating the reliability of the chassis truss structure, improving the development and design efficiency, and facilitating the obtaining of the optimal design parameter combination.

Description

Technical Field

[0001] This invention relates to the field of vehicle chassis truss optimization design technology, and in particular to a novel vehicle chassis truss lightweight optimization design system. Background Technology

[0002] The vehicle chassis is a crucial component ensuring the vehicle's power output and stable, reliable operation. Chassis optimization is a key issue in vehicle lightweighting. During lightweight optimization design, simultaneously optimizing the shape, dimensions, and topology of the chassis truss structure while considering quality and reliability indicators presents a complex and challenging task for designers. It often requires extensive finite element simulation calculations and cross-departmental, multi-disciplinary collaboration to successfully complete and achieve satisfactory results. Currently, intelligent design and efficient, high-quality design have become new trends in the manufacturing industry. Simultaneously, design systems that can facilitate interdisciplinary collaboration while improving work efficiency and reducing cross-departmental coordination have become a highly sought-after goal for design departments. In chassis design, maintaining the reliability and durability of the chassis truss structure while reducing weight is a key factor in improving the design quality of the chassis truss structure, reducing vehicle energy consumption, and improving subsequent processing efficiency. Therefore, establishing a lightweight auxiliary optimization design system that is computationally accurate, reliable, and efficient is particularly crucial.

[0003] Furthermore, optimizing the shape, dimensions, and topology of the vehicle chassis truss structure while considering both quality and reliability indicators is a difficult and challenging task for designers. However, the research on the joint model of multi-objective optimization and reliability estimation of the truss in traditional lightweight design models is still inadequate.

[0004] In existing technologies, optimization algorithms can usually search for multiple Pareto frontiers simultaneously in a single solution process, making it easier to solve multi-objective problems. However, their disadvantages are that they are less efficient and prone to getting trapped in local optima when dealing with high-dimensional problems or non-differentiable constraints.

[0005] Traditional ANFIS models can only be optimized using backpropagation or hybrid learning algorithms, which limits their further development. Furthermore, while ANFIS models have advantages in many aspects, they also have some drawbacks, including:

[0006] 1. High complexity: The structure of the ANFIS model is relatively complex, including multiple parts such as fuzzification, rule base, and inference mechanism, which requires a lot of computing resources and time to train and optimize the model.

[0007] 2. Overfitting: Due to the large flexibility and number of parameters of the ANFIS model, overfitting is prone to occur when the amount of training data is small or there is a lot of noise, resulting in poor generalization ability of the model on new data.

[0008] 3. Difficulty in setting rules: Fuzzy rules in the ANFIS model need to be set manually. Choosing the appropriate number and form of rules is a challenge, and the setting of rules usually depends on the experience of domain experts.

[0009] 4. Difficulty in training convergence: Training an ANFIS model requires updating model parameters through iterative optimization algorithms, but it may sometimes get stuck in local optima, requiring careful adjustment of the learning rate and other hyperparameters.

[0010] 5. Not suitable for large-scale datasets: Due to the complexity of the ANFIS model and the high computational requirements of the training process, it is not suitable for processing large-scale datasets, as the training time is long and consumes a lot of resources. Summary of the Invention

[0011] The purpose of this invention is to address the shortcomings of existing technologies by proposing a novel lightweight optimization design system for vehicle chassis trusses.

[0012] To achieve the above objectives, the present invention adopts the following technical solution:

[0013] A novel lightweight optimization design system for vehicle chassis trusses, the design steps of which are as follows:

[0014] S1: First, based on the multi-objective hybrid heuristic algorithm of adaptive differential evolution based on success history, an adaptive spiral search strategy is introduced, and then a new SHAMODE-IWOA algorithm is proposed.

[0015] S2: Then, in order to estimate the reliability level of the chassis truss structure under different combinations of design parameters, the proposed SHAMODE-IWOA algorithm was used to learn the ANFIS model and a new ANFIS-SHAMODE-IWOA model was constructed.

[0016] S3: Finally, to obtain the optimal combination of design parameters, the SHAMODE-IWOA algorithm is used to study multi-objective optimization based on minimum design quality and optimal reliability metric function.

[0017] Preferably, the specific steps of S1 are as follows:

[0018] S11: Parameter initialization; randomly generate NP initialization solution sets;

[0019] x i,G =[x 1,G x 2,G x 3,G， …, x NP,G ]

[0020] Where i is the index, NP is the index extreme value, which is equal to the number of design variables, and G is the number of iterations;

[0021] The initial Pareto archive Pareto1 is populated with non-dominated solutions; an empty external archive (A1) is created for the regeneration process; during this process, the initial values ​​of all adaptive parameters are initialized.

[0022] S12: Mutation. During the mutation process, a random strategy is used to generate a mutation amount.

[0023]

[0024] Among them, V i，G ∈[0,1] is the scaling factor that controls the influence of the differential change;

[0025] x i，G This is a feasible solution for the Gth generation;

[0026] x pbest It is a solution randomly selected from an external Pareto archive;

[0027] x r1，G It is a solution randomly selected from the current population (xG). It is the union of the current population and the external archive (x) G ∪A G A solution randomly selected from ( )

[0028] S13: Cross;

[0029] S14: Select; Extract the current individual x i，G and test individual h i，G The non-dominated ranking scheme of NSGA-II was used to analyze the joint population (x) G ∪u G Sort the solutions; then, in the next iteration, retain the NP solutions with the highest non-dominated level; if the number of solutions with the highest non-dominated level exceeds NP, some of them will be randomly removed to maintain a constant population size; finally, the NP survivors of the current iteration will be stored in x. G+1 In the middle; after the selection process, from h G ∪Pareto G All non-dominated solutions sorted in Pareto will be stored in Pareto. G+1 In the process, if the number of non-dominant solutions exceeds the maximum Pareto archive size, some of them will be randomly removed from the archive;

[0030] S15: Parameter adaptive adjustment strategy; all adaptive parameters, including the external archive (A), the historical memory of the scaling factor (MF), and the cross ratio (MCR), are updated at the end of each iteration; at the end of each iteration, the indices of the successfully updated descendants that survived the selection process are stored in a vector named sind; then, the successful offspring (x) are generated. sind The parent vector of G is stored in the external file A. G+1 If the number of solutions stored in the external archive exceeds a specified value, some of these solutions will be randomly deleted to maintain a constant archive size.

[0031] S16: Improve the whale adaptive spiral strategy by integrating the spiral motion of WOA into the SHAMODE algorithm; modify it during the mutation process, in which each mutant vector has the opportunity to be further updated with the spiral motion of WOA, and then activate the crossover process.

[0032] Preferably, in S13, the crossover phase is performed as shown in the following formula:

[0033]

[0034] Among them, R i，G The crossover ratio, its range R i，G ∈[0,1]; rand([0,1)) represents a simple random number; j rand The x index is randomly generated from [1, 2, ..., n], where n is the number of design variables.

[0035] Preferably, in step S15, VC and RG are updated according to the formula, as shown below:

[0036] V i，G =randc i (μ v ,0.1)

[0037] R i，G =randn i (μ R ,0.1)

[0038] Where, μ v and μ R The mean is 0.5, and its initial value is 0.5. ramdct(μ, 0.1) and randnt(μR, 0.1) are random numbers generated based on the Cauchy and normal distributions, respectively, where the variance is... The value is (0.1, 0.1);

[0039] Furthermore, let M V and M R Scaling factor V i，G and cross ratio Ri，G Historical memory, μ of each individual v and μ R They can be randomly selected from these; and at the end of each iteration, one element of the memory (MF and MCR) will be updated with the Lehmer mean of the parameters of the successfully updated offspring, as follows:

[0040]

[0041]

[0042] Among them, L mean (V sind，G ) and L mean (R sind，G ) are respectively V i，G and R i，G The Lehmer mean is given by the subscript sind, which indicates the individual with a successful offspring. That is, if there is at least one successful offspring, the k-th memory position will be updated; otherwise, such an element will remain unchanged. The index k is a memory element index that is initially equal to 1 and increases linearly as the process continues. If k > H, then k is reset to 1.

[0043] Preferably, in step S16, the specific formula for activating the crossover process is as follows:

[0044]

[0045] D i =|x pbest2 -u i，G |

[0046] in, It is a new type of mutation with a spiral renewal process, while x pbest2 Another feasible solution (with x) is randomly selected from the current Pareto file. pbest1 (Different); Variable 1 and rand are random numbers in the intervals [-1, 1] and [0, 1], respectively.

[0047] Preferably, in step S16, the spiral shape parameter l is set to a dynamic value that changes with the number of iterations. This allows the individual whale to dynamically adjust the spiral shape during the search phase after entering the spiral search stage, enhancing the algorithm's global search capability and improving its convergence accuracy. Specifically, as follows:

[0048]

[0049]

[0050] D i =|x pbest2 -ui，G |

[0051] Where γ is the spiral shape adjustment coefficient, G max G represents the maximum number of iterations, and G represents the current number of iterations.

[0052] Preferred method: The specific method for constructing the model in S2 is as follows:

[0053] Layer 1: This adaptive node is the membership function layer of the input variables, transforming the input into a fuzzy set, and the output function is:

[0054]

[0055]

[0056] Where x1 and x2 are inputs; A i and B i-2 This represents a fuzzy set, which is a linguistic variable obtained through calculation; and For the membership function, a Gaussian function is chosen:

[0057]

[0058] Where parameter c ij and σ ij These are the initial parameters that need to be adjusted through a learning algorithm;

[0059] Layer 2: This layer releases the strength of the rules. The node function is multiplied by the input, and the result is represented by fuzzy rules.

[0060]

[0061] Layer 3: The number of nodes in Layer 3 is the same as in Layer 2. The results from Layer 2 are normalized.

[0062]

[0063] Layer 4: Layer 4 is the consequent network, which obtains fuzzy if-then rules and calculates the output of the fuzzy rules.

[0064]

[0065] Layer 5: Layer 5 is the output layer, which calculates the sum of the input signals to obtain the overall output.

[0066]

[0067] In step S2, after the ANFIS model structure is established, the SHAMODE-iWOA algorithm is used to train the ANFIS model; the root mean square error is used as the fitness function to train ANFIS, as shown in the following formula:

[0068]

[0069] Where n is the number of training set data; r is the r-th data point; dr is the actual value of the r-th training set data; and pr is the r-th predicted value.

[0070] Preferably, the method for obtaining the dataset in the optimization design system includes the following steps:

[0071] S31: Database required for constructing chassis truss structure research;

[0072] S32: Construct the SFE model of the chassis truss structure;

[0073] S33: After solving, principal component analysis is used to extract the design variables in the SFE model based on their contribution.

[0074] S34: Select the design variables with the largest contribution as design parameters;

[0075] S35: After generating the DOE matrix for the design variables, solve it. The Bending, Torion, and Mass solutions are calculated using a finite element solver.

[0076] S36: After solving the problem, the dataset is obtained by eliminating outlier data;

[0077] S37: Divide the experimental dataset obtained after solving the model into two parts: training dataset and test dataset; the training dataset is used to train the ANFIS-SHAMODE-IWOA model, and the test dataset is used to test the performance of the ANFIS-SHAMODE-IWOA model.

[0078] Preferably, in the optimized design system, the multi-objective optimization of the chassis truss structure includes the following steps:

[0079] S41: Establishment of multi-objective optimization equations;

[0080] S42: Solving and analyzing multi-objective optimization;

[0081] Specifically, the establishment of the multi-objective optimization equation is as follows:

[0082] The multi-objective equations for truss reliability and quality are established as follows:

[0083]

[0084] Where Pr is the failure probability; x is the design variable vector; y is the physical vector, which includes yield strength, torsional stiffness and applied load; f1 is the structural mass, which is the sum of the products of mass and density of topological units; f2 is the reliability metric function: f2 = 1 / β, where β is the reliability coefficient;

[0085] The reliability coefficient (β) refers to the shortest distance between the limit state line and the origin of the transformation space; a larger β indicates higher reliability (higher safety), as expressed by the following formula:

[0086]

[0087] Where μ,m can be characterized as the average value of the force under the limit failure function state, σ m It is characterized as the variance of the rate of change of mechanical properties under the limit failure function state.

[0088] Preferably, in the solution and analysis of the multi-objective optimization, the SHAMODE-IWOA algorithm is used to optimize the multi-objective function constructed above; based on the iterative solution results, the "compromise method" is used to obtain the optimized value of the design index; and the finite element model is further simulated and calculated.

[0089] The beneficial effects of this invention are as follows:

[0090] 1. The ANFIS-SHAMODE-IWOA multi-objective optimization design system proposed in this invention, as a novel intelligent model, can be used to evaluate the reliability of chassis truss structures, improve development and design efficiency, facilitate the acquisition of the optimal combination of design parameters, and help improve the level of green and intelligent manufacturing design. Attached Figure Description

[0091] Figure 1 This is a general structural diagram of the overall working process and model of a novel lightweight vehicle chassis truss optimization design system proposed in this invention;

[0092] Figure 2 This is a schematic diagram of the database construction in a novel lightweight optimization design system for vehicle chassis trusses proposed in this invention;

[0093] Figure 3 This is a comparison chart of design calculation optimization indicators for a novel lightweight vehicle chassis truss optimization design system proposed in this invention.

[0094] Figure 4 This is a graph showing the stress acquisition test results of the present invention. Detailed Implementation

[0095] The technical solution of the present invention will be further described in detail below with reference to specific embodiments.

[0096] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0097] Example 1:

[0098] A novel lightweight optimization design system for vehicle chassis trusses, the design steps of which are as follows:

[0099] S1: First, based on the adaptive differential evolution multi-objective hybrid heuristic algorithm (SHAMODE) based on success history, an adaptive spiral search strategy is introduced, and then a new SHAMODE-IWOA algorithm is proposed.

[0100] S2: Then, in order to estimate the reliability level of the chassis truss structure under different combinations of design parameters, a new ANFIS-SHAMODE-IWOA model was constructed by learning the ANFIS model using the proposed SHAMODE-IWOA algorithm.

[0101] S3: Finally, to obtain the optimal combination of design parameters, the SHAMODE-IWOA algorithm was used to study multi-objective optimization based on minimum design quality and optimal reliability metric function.

[0102] The specific steps of S1 are as follows:

[0103] S11: Parameter initialization; randomly generate NP initialization solution sets;

[0104] x i，G =[x 1，G x 2，G x 3，G , ..., x NP，G ]

[0105] Where i is the index, NP is the index extreme value, which is equal to the number of design variables, and G is the number of iterations;

[0106] The initial Pareto archive Pareto1 is populated with non-dominated solutions; an empty external archive (A1) is created for the regeneration process; during this process, the initial values ​​of all adaptive parameters are initialized.

[0107] S12: Mutation. During the mutation process, a random strategy is used to generate a mutation amount.

[0108]

[0109] Among them, Vi，G ∈[0,1] is the scaling factor that controls the influence of the differential change;

[0110] x i，G This is a feasible solution for the Gth generation;

[0111] x pbest It is a solution randomly selected from an external Pareto archive;

[0112] x r1，G It is a solution randomly selected from the current population (xG). It is the union of the current population and the external archive (x) G ∪A G A solution randomly selected from ( )

[0113] S13: Crossover, the crossover phase is performed as shown in the following formula:

[0114]

[0115] Among them, R i，G The crossover ratio, its range R i，G ∈[0,1]; rand([0,1)) represents a simple random number; j rand x is a randomly generated index from [1, 2, ..., n], where n is the number of design variables;

[0116] S14: Select; Extract the current individual x i，G and test individual h i，G The non-dominated ranking scheme of NSGA-II was used to analyze the joint population (x) G ∪u G Sort the solutions; then, in the next iteration, retain the NP solutions with the highest non-dominated level; if the number of solutions with the highest non-dominated level exceeds NP, some of them will be randomly removed to maintain a constant population size; finally, the NP survivors of the current iteration will be stored in x. G+1 In the middle; after the selection process, from h G ∪Pareto G All non-dominated solutions sorted in Pareto will be stored in Pareto. G+1 In the process, if the number of non-dominant solutions exceeds the maximum Pareto archive size, some of them will be randomly removed from the archive;

[0117] S15: Parameter adaptive adjustment strategy; all adaptive parameters, including the external archive (A), the history memory (MF) of the scaling factor, and the crossover ratio (MCR), are updated at the end of each iteration; the maximum number of solutions in the external archive (A) is 1.4 × NP; at the end of each iteration, the indices of the successfully updated descendants that survived the selection process are stored in a vector named sind; then, the successful offspring (x) are generated. sind The parent vector of G is stored in the external file A. G+1 If the number of solutions stored in the external archive exceeds a specified value, some of these solutions will be randomly deleted to maintain a constant archive size.

[0118] The VC and RG are updated according to the formula, as shown below:

[0119] V i，G =randc i (μ v ,0.1)

[0120] R i，G =randn i (μ R ,0.1)

[0121] Where, μ v and μ R The mean is 0.5, and its initial value is 0.5. ramdct(μ, 0.1) and randnt(μR, 0.1) are random numbers generated based on the Cauchy and normal distributions, respectively, where the variance is... The value is (0.1, 0.1);

[0122] Furthermore, let M V and M R Scaling factor V i，G and cross ratio R i，G Historical memory, μ of each individual v and μ R They can be randomly selected from these; and at the end of each iteration, one element of the memory (MF and MCR) will be updated with the Lehmer mean of the parameters of the successfully updated offspring, as follows:

[0123]

[0124]

[0125] Among them, L mean (V sind，G ) and L mean (R sind，G ) are respectively V i，G and R i，GThe Lehmer mean is given, and the subscript sind indicates the individual with a successful offspring. That is, if there is at least one successful offspring, the k-th memory position will be updated; otherwise, such an element will remain unchanged. The index k is a memory element index that is initially equal to 1 and increases linearly as the process continues; if k > H, then k is reset to 1.

[0126] S16: Improve the whale adaptive spiral strategy by integrating the spiral motion of WOA into the SHAMODE algorithm; modify the mutation process so that each mutant vector has the opportunity to be further updated with the spiral motion of WOA, and then activate the crossover process, as follows:

[0127]

[0128] D i =|x pbest2 -u i，G |

[0129] in, It is a new type of mutation with a spiral renewal process, while x pbest2 Another feasible solution (with x) is randomly selected from the current Pareto file. pbest1 (Different); Variable 1 and rand are random numbers in the intervals [-1, 1] and [0, 1], respectively;

[0130] By setting the spiral shape parameter l to a dynamic value that varies with the number of iterations, the spiral shape of the individual whale can be dynamically adjusted after entering the spiral search phase, thereby enhancing the algorithm's global search capability and improving its convergence accuracy. The details are as follows:

[0131]

[0132]

[0133] D i =|x pbest2 -u i，G |

[0134] Where γ is the spiral shape adjustment coefficient, G max G represents the maximum number of iterations, and G represents the current number of iterations.

[0135] The specific method for constructing the model in S2 is as follows:

[0136] Layer 1: This adaptive node is the membership function layer of the input variables, transforming the input into a fuzzy set, and the output function is:

[0137]

[0138]

[0139] Where x1 and x2 are inputs; A i and B i-2 This represents a fuzzy set, which is a linguistic variable obtained through calculation; and For the membership function, a Gaussian function is chosen:

[0140]

[0141] Where parameter c ij and σ ij These are the initial parameters that need to be adjusted through a learning algorithm;

[0142] Layer 2: This layer releases the strength of the rules. The node function is multiplied by the input, and the result is represented by fuzzy rules.

[0143]

[0144] Layer 3: The number of nodes in Layer 3 is the same as in Layer 2. The results from Layer 2 are normalized.

[0145]

[0146] Layer 4: Layer 4 is the consequent network, which obtains fuzzy if-then rules and calculates the output of the fuzzy rules.

[0147]

[0148] Layer 5: Layer 5 is the output layer, which calculates the sum of the input signals to obtain the overall output.

[0149]

[0150] In step S2, after the ANFIS model structure is established, the SHAMODE-iWOA algorithm is used to train the ANFIS model; the root mean square error is used as the fitness function to train ANFIS, as shown in the following formula:

[0151]

[0152] Where n is the number of training set data; r is the r-th data point; dr is the actual value of the r-th training set data; and pr is the r-th predicted value.

[0153] An ANFIS model was constructed to obtain the design parameter combination with the best reliability and lowest quality. The model takes Torsion, Mass, Bending, Design Parameter, and Reliability Coefficient as input parameters and outputs the reliability prediction results. Furthermore, a multi-objective optimization model was constructed using the ANFIS model results as one of the objective functions, ultimately yielding the optimal design parameters.

[0154] The optimization design system obtains the dataset using the following steps:

[0155] S31: Database required for constructing chassis truss structure research;

[0156] S32: Construct the SFE model of the chassis truss structure;

[0157] S33: After solving, principal component analysis is used to extract the design variables in the SFE model based on their contribution.

[0158] S34: Select the design variables with the largest contribution as design parameters;

[0159] S35: After generating the DOE matrix for the design variables, solve it. The Bending, Torion, and Mass solutions are calculated using a finite element solver.

[0160] S36: After solving the problem, the dataset is obtained by eliminating outlier data;

[0161] S37: The experimental dataset obtained after solving the model is divided into two parts: the training dataset and the test dataset; the training dataset is used to train the ANFIS-SHAMODE-IWOA model, and the test dataset is used to test the performance of the ANFIS-SHAMODE-IWOA model.

[0162] To eliminate the impact of sample selection on model training, a random sampling method was used, with 70% of the dataset selected as the training dataset and 30% as the test dataset.

[0163] The multi-objective optimization of the chassis truss structure in the optimization design system includes the following steps:

[0164] S41: Establishment of multi-objective optimization equations;

[0165] S42: Solving and analyzing multi-objective optimization.

[0166] Specifically, the establishment of the multi-objective optimization equation is as follows:

[0167] The multi-objective equations for truss reliability and quality are established as follows:

[0168]

[0169] Where Pr is the failure probability; x is the design variable vector; y is the physical vector, which includes yield strength, torsional stiffness and applied load; f1 is the structural mass, which is the sum of the products of the mass and density of the topological unit; f2 is the reliability metric function: f2 = 1 / β, where β is the reliability coefficient.

[0170] The reliability coefficient (β) refers to the shortest distance between the limit state line and the origin of the transformation space; a larger β indicates higher reliability (higher safety), as expressed by the following formula:

[0171]

[0172] Where μ,m can be characterized as the average value of the force under the limit failure function state, σ m It is characterized as the variance of the rate of change of mechanical properties under the limit failure function state.

[0173] In the solution and analysis of the multi-objective optimization, the SHAMODE-IWOA algorithm is used to optimize the constructed multi-objective function; based on the iterative solution results, the "compromise method" is used to obtain the optimized value of the design index; and the finite element model is further simulated and calculated.

[0174] test:

[0175] To further verify the optimization results, the study used electrical testing to conduct stress tests on the key locations of the optimized truss structure. The optimized chassis truss was mounted on a vehicle and driven on a typical test road surface. Ten sets of tests were conducted on the relevant harsh road surfaces, with 20 cycles in each set.

[0176] The strain data at each stress level were separated by the rainflow counting method. The separation results are shown in the figure below. The damage coupling values ​​corresponding to each stress level at the four locations shown in the figure above were calculated respectively.

[0177] The proposed SHAMODE-IWOA algorithm was used to optimize the chassis truss structure of a certain brand of dump truck. Based on the optimized calculation results, the relevant design parameters of the truss were adjusted, and a prototype was manufactured and installed. The system underwent three tests: mathematical model calculation, finite element simulation calculation, and actual stress testing. The test results show that the optimized chassis truss structure reduces the total mass of the truss, improves reliability, enhances torsion and bending performance, and also improves mechanical properties. Ultimately, this demonstrates that the algorithm has good engineering application value.

[0178] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A novel lightweight optimization design system for vehicle chassis trusses, characterized in that, The design steps of the optimization design system are as follows: S1: First, based on the multi-objective hybrid heuristic algorithm of adaptive differential evolution based on success history, an adaptive spiral search strategy is introduced, and then a new SHAMODE-IWOA algorithm is proposed. S2: Then, in order to estimate the reliability level of the chassis truss structure under different combinations of design parameters, the proposed SHAMODE-IWOA algorithm was used to learn the ANFIS model to construct a new ANFIS-SHAMODE-IWOA model. S3: Finally, to obtain the optimal combination of design parameters, the SHAMODE-IWOA algorithm is used to study multi-objective optimization based on minimum design mass and optimal reliability metric function; The specific method for constructing the model in S2 is as follows: Layer 1: The adaptive nodes are the membership function layer of the input variables, which transforms the input into a fuzzy set, and the output function is: ； ； in and For input; and This represents a fuzzy set, which is a linguistic variable obtained through calculation; and For the membership function, a Gaussian function is chosen: ； Where parameters and These are the initial parameters that need to be adjusted through a learning algorithm; Layer 2: This layer releases the strength of the rules. The node function is multiplied by the input, and the result is represented by fuzzy rules. ； Layer 3: The number of nodes in Layer 3 is the same as in Layer 2. The results from Layer 2 are normalized. ； Layer 4: Layer 4 is the consequent network, which obtains fuzzy if-then rules and calculates the output of the fuzzy rules. ； Layer 5: Layer 5 is the output layer, which calculates the sum of the input signals to obtain the overall output. ； In step S2, after the ANFIS model structure is established, the SHAMODE-iWOA algorithm is used to train the ANFIS model; the root mean square error is used as the fitness function to train ANFIS, as shown in the following formula: ； Where n is the number of training set data; r is the r-th data point; dr is the actual value of the r-th training set data; and pr is the r-th predicted value. The optimization design system obtains the dataset using the following steps: S31: Database required for constructing chassis truss structure research; S32: Construct the SFE model of the chassis truss structure; S33: After solving, principal component analysis is used to extract the design variables in the SFE model based on their contribution. S34: Select the design variables with the largest contribution as design parameters; S35: After generating the DOE matrix for the design variables, solve it. The Bending, Torion, and Mass are solved by the finite element solver. S36: After solving the problem, the dataset is obtained by eliminating outlier data; S37: The experimental dataset obtained after solving the model is divided into two parts: training dataset and test dataset; the training dataset is used to train the ANFIS-SHAMODE-IWOA model, and the test dataset is used to test the performance of the ANFIS-SHAMODE-IWOA model. The multi-objective optimization of the chassis truss structure in the optimization design system includes the following steps: S41: Establishment of multi-objective optimization equations; S42: Solving and analyzing multi-objective optimization; Specifically, the establishment of the multi-objective optimization equation is as follows: The multi-objective equations for truss reliability and quality are established as follows: ； Where Pr is the failure probability; x is the design variable vector; y is the physical vector, which includes yield strength, torsional stiffness and applied load; f1 is the structural mass, which is the sum of the products of the mass and density of the topological unit; f2 is the reliability metric function: f2=1 / β, where β is the reliability coefficient; The reliability coefficient β refers to the shortest distance between the limit state line and the origin of the transformation space; the larger the β, the higher the reliability and the higher the safety, as expressed by the following formula: ； in, It can be characterized as the average value of the force under the limit failure function state. It is characterized as the variance of the rate of change of mechanical properties under the limit failure function state.

2. The novel lightweight optimization design system for vehicle chassis trusses according to claim 1, characterized in that, The specific steps of S1 are as follows: S11: Parameter initialization; randomly generate NP initialization solution sets; ； Where i is the index, NP is the index extreme value, which is equal to the number of design variables, and G is the number of iterations; The initial Pareto archive Pareto1 is populated with non-dominated solutions; an empty external archive A1 is created for the regeneration process; during this process, the initial values ​​of all adaptive parameters are initialized. S12: Mutation. During the mutation process, a random strategy is used to generate a mutation amount. ； in, It is a proportionality factor that controls the influence of differential changes; This is a feasible solution for the Gth generation; It is a solution randomly selected from an external Pareto archive; From the current population A solution randomly selected from the pool. It is the union of the current population and the external archive (x) G ∪A G A solution randomly selected from ) S13: Cross; S14: Select; Extract the current individual and test individuals The non-dominated ranking scheme of NSGA-II was used to analyze the joint population (x) G ∪u G Sort the solutions; then, in the next iteration, retain the NP solutions with the highest non-dominated level; if the number of solutions with the highest non-dominated level exceeds NP, some of them will be randomly removed to maintain a constant population size; finally, the NP survivors of the current iteration will be stored in x. G+1 In the middle; after the selection process, from h G ∪Pareto G All non-dominated solutions sorted in Pareto will be stored in Pareto. G+1 In the process, if the number of non-dominant solutions exceeds the maximum Pareto archive size, some of them will be randomly removed from the archive; S15: Parameter adaptive adjustment strategy; all adaptive parameters, including external archive A, historical memory of scaling factor MF, and cross ratio MCR, are updated at the end of each iteration; at the end of each iteration, the indices of the successfully updated offspring that survived the selection process are stored in a vector named sind; then, the successful offspring (x sind The parent vector of G is stored in the external file A. G+1 If the number of solutions stored in the external archive exceeds a specified value, some of these solutions will be randomly deleted to maintain a constant archive size. S16: Improve the whale adaptive spiral strategy by integrating the spiral motion of WOA into the SHAMODE algorithm; modify it during the mutation process, in which each mutant vector has the opportunity to be further updated with the spiral motion of WOA, and then activate the crossover process.

3. The novel lightweight optimization design system for vehicle chassis trusses according to claim 2, characterized in that, In S13, the crossover phase is performed as shown in the following formula: ； in, Cross ratio, its range rand([0,1)) represents a simple random number; The x index is randomly generated from [1,2,...,n], where n is the number of design variables.

4. The novel lightweight optimization design system for vehicle chassis trusses according to claim 3, characterized in that, In step S15, VC and RG are updated according to the formula, as shown below: ； ； in, and The mean is 0.5, and its initial value is 0.

5. `ramdct(μ,0.1)` and `randnt(μR,0.1)` are random numbers generated based on the Cauchy and normal distributions, respectively, where the variance is... The value is (0.1, 0.1); Furthermore, let M V and M R Scaling factor and cross ratio Historical memory, each individual and They can be randomly selected from these; and at the end of each iteration, one element of the memory (MF and MCR) will be updated with the Lehmer mean of the parameters of the successfully updated offspring, as follows: ； ； in, and They are and The Lehmer mean is given by the subscript sind, which indicates the individual with a successful offspring. That is, if there is at least one successful offspring, the k-th memory position will be updated; otherwise, such an element will remain unchanged. The index k is a memory element index that is initially equal to 1 and increases linearly as the process continues. If k > H, then k is reset to 1.

5. A novel lightweight vehicle chassis truss optimization design system according to claim 4, characterized in that, In S16, the specific formula for activating the crossover process is as follows: ； ； in, It is a new variant with a spiral renewal process, and It is another feasible solution randomly selected from the current Pareto archive, and... They are different; variables l and rand are random numbers in the intervals [-1, 1] and [0, 1], respectively.

6. The novel lightweight optimization design system for vehicle chassis trusses according to claim 5, characterized in that, In step S16, the spiral shape parameter l is set to a dynamic value that changes with the number of iterations. This allows the individual whale to dynamically adjust the spiral shape during the search phase after entering the spiral search stage, enhancing the algorithm's global search capability and improving its convergence accuracy. Specifically, the following steps are taken: ； ； ； in, This is the adjustment factor for the spiral shape. G represents the maximum number of iterations, and G represents the current number of iterations.

7. The novel lightweight optimization design system for vehicle chassis trusses according to claim 1, characterized in that, In the solution and analysis of the multi-objective optimization, the SHAMODE-IWOA algorithm is used to optimize the constructed multi-objective function; based on the iterative solution results, the "compromise method" is used to obtain the optimized values ​​of the design index; and the finite element model is further simulated.

Images