Intelligent wire harness section design method based on multi-data fusion

Abstract

The invention discloses an intelligent wire harness section design method based on multi-data fusion, and belongs to the field of wire harness section design, and the design method comprises the following specific steps: I, collecting and preprocessing various data of wire harness design; iI, associating information in different data to construct a wire harness design map; iII, analyzing information in the atlas and extracting various characteristic data of the performance of the wire harness; according to the method, local optimum can be effectively avoided, a globally optimal solution can be effectively found, wire harness section design can explore wider design space, optimization pertinence is improved, the two methods can complement each other, the convergence speed is increased, the number of iterations needed by optimization is reduced, comprehensive performance evaluation can be provided, various environment variables can be considered, and the method is suitable for popularization and application. Therefore, the performance of the wire harness in practical application is better reflected, the uncertainty in the design process can be effectively dealt with, more reliable performance estimation is provided for design, and the optimization process of the design scheme is more efficient and accurate.

Description

Technical Field

[0001] The invention relates to the field of wire harness cross-section design, and in particular to an intelligent wire harness cross-section design method based on multi-data fusion. Background Art

[0002] With the increasing complexity of electrical and electronic equipment, the importance of wire harness design, as an important medium for connecting various components, has become increasingly prominent. The design of wire harness cross-section directly affects the performance, reliability and safety of the equipment. Wire harnesses play an indispensable role in modern electronic equipment, and the quality of their design directly affects the performance and reliability of the system. Faced with increasingly complex system requirements, traditional wire harness design methods rely heavily on experience and a single data source, making it difficult to cope with complex working conditions and changing application requirements, and unable to meet the efficiency and accuracy requirements of modern engineering. In order to overcome these challenges, it is particularly important to invent an intelligent wire harness cross-section design method based on multi-data fusion.

[0003] The existing intelligent wire harness cross-section design methods are prone to fall into local optimality and cannot find the global optimal solution, which reduces the targetedness of optimization, reduces the convergence speed, and increases the number of iterations required for optimization; in addition, the existing intelligent wire harness cross-section design methods cannot consider multiple environmental variables, cannot well reflect the performance of wire harnesses in actual applications, cannot cope with uncertainties in the design process, and cannot provide more reliable performance estimates; for this reason, we propose an intelligent wire harness cross-section design method based on multi-data fusion. Summary of the invention

[0004] The purpose of the present invention is to solve the defects in the prior art and to propose an intelligent wire harness cross-section design method based on multi-data fusion.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] An intelligent wire harness cross-section design method based on multi-data fusion, the specific steps of the design method are as follows:

[0007] Ⅰ. Collect and pre-process various data for wiring harness design;

[0008] Ⅱ. Associate information from different data to build a wiring harness design map;

[0009] Ⅲ. Analyze the information in the graph and extract the characteristic data of the wiring harness performance;

[0010] IV. Generate a preliminary wire harness cross-section design and perform iterative updates;

[0011] V. Evaluate the performance of different design solutions in actual use and select them;

[0012] VI. Implement the final wiring harness design and monitor the performance data of the wiring harness;

[0013] Ⅶ. Make subsequent improvements and adjustments based on each group’s monitoring data.

[0014] As a further solution of the present invention, the specific steps of preprocessing the various data of the wiring harness design in step I are as follows:

[0015] Q1.1: Collect various harness design data from various data sources including internal databases, industry standard documents and real-time sensor data. Set a sliding window of size N×N according to the length of different harness design data. Calculate the moving average of the internal data of each position window step by step from the data head.

[0016] Q1.2: Replace the data at the corresponding position in the original data with the calculated moving average, convert the data of different dimensions to the same dimension through the Max-Min normalization method, then detect the missing values ​​in each group of wiring harness design data, and fill the missing data through mean filling or interpolation, and then integrate the data from different sources to establish the wiring harness design set.

[0017] As a further solution of the present invention, the specific steps of constructing the wiring harness design map in step II are as follows:

[0018] Q2.1: Automatically or manually annotate the collected data according to preset rules, identify key entities and their attributes, extract entities from the annotated data using natural language processing technology, and then use named entity recognition algorithms to identify the materials, design parameters, and environmental factors corresponding to each entity;

[0019] Q2.2: The relationships between different entities are extracted through the GCN network, and the extracted entities and relationships are organized into a graph structure, where the nodes in the graph structure represent entities and the edges represent relationships. Attribute values ​​are assigned to each node to complete the construction of the wire harness design graph, and then the constructed wire harness design graph is stored in the graph database, and a query interface is set for subsequent data retrieval and update.

[0020] As a further solution of the present invention, the specific steps of extracting various characteristic data of wiring harness performance in step III are as follows:

[0021] Q3.1: Extract the feature set and labels for training from the harness design graph, where the feature set contains various attributes related to the harness performance and the label is the target variable. Divide each group of extracted data into training set and test set in a ratio of 8:2, initialize the random forest model, and set the number of trees and other hyperparameters.

[0022] Q3.2: Use the training set data to train the random forest model. During the training process, randomly select features and perform multiple sample sampling with replacement to generate multiple sample subsets. Each subset is used to train a decision tree. Then, at each split node of each tree, randomly select a subset of features and split each node according to the selected features and their thresholds.

[0023] Q3.3: Evaluate the quality of different splits, select the split that maximizes purity, repeat sample sampling and feature selection for each child node, and continue splitting until the preset purity threshold is reached;

[0024] Q3.4: Analyze the contribution of each feature to the model prediction and evaluate the importance of the feature. Then evaluate the performance of the model on the test set and calculate the accuracy and mean square error of the prediction. If the model performance does not meet the preset requirements, retrain it. Otherwise, pass the latest design data in the wiring harness design map into the trained random forest model.

[0025] Q3.5: The random forest model evaluates the feature importance of each set of design data received, and based on the results of the feature importance evaluation, selects the feature most relevant to the wiring harness performance. If the feature importance is greater than the preset threshold, the feature is retained, otherwise it is removed.

[0026] As a further solution of the present invention, the specific steps of iterative updating in step IV are as follows:

[0027] Q4.1: Randomly perturb each group of parameters in the preliminary wire harness cross-section design scheme to generate multiple groups of different design schemes, and use each group of design schemes to construct the corresponding design space, and then randomly generate multiple groups of individual initial positions in the design space to represent different design schemes;

[0028] Q4.2: Predefine the fitness function according to the harness design goal, and use the fitness function to evaluate the performance of each group of individuals. Then update the movement speed of each group of individuals, and adjust the position of the individual in the design space at the next moment based on the updated speed of the individual. Then check whether the individual position exceeds the boundary of the design space. If it exceeds the boundary, adjust it back to the boundary, and repeat the speed and position update until the fitness reaches the preset threshold;

[0029] Q4.3: Use the updated positions of each group of individuals as the starting positions, initialize the pheromone concentration of each parameter node in the design scheme, and calculate its corresponding heuristic value. Then calculate the selection probability of the next node based on the pheromone concentration and the heuristic value. Repeat the path selection until the individual constructs a complete design scheme. Evaluate the quality of the path based on the fitness function and update the pheromone concentration. Repeat the path construction and pheromone concentration update until the pheromone concentration converges to the preset range.

[0030] Q4.4: Alternately update the design schemes in steps Q4.2 and Q4.3 until the fitness value reaches the threshold or the maximum number of iterations is reached. After each round of iteration, select the design scheme with the highest fitness as the basis for the next round of optimization. After the iteration, output the optimized design scheme, including the optimal wire harness cross-section parameters and corresponding performance indicators, and provide design documents.

[0031] As a further solution of the present invention, the specific calculation formula of the fitness function in Q4.2 is as follows:

[0032] F(X i )=w1·f1(X i )+w2·f2(X i )+...+w n ·f n (X i )

[0033] In the formula, F(X i ) represents the fitness of individual i; f n (X i ) represents the performance of individual i on the nth design goal; w n represents the weight of the nth design goal;

[0034] The specific calculation formulas for speed and position update described in Q4.2 are as follows:

[0035] v it =L·v i(t-1) +r1·(p i -X i(t-1) )+r2·(gX i(t-1) )

[0036] X it =X i(t-1) +v it

[0037] In the formula, v it represents the speed of individual i in round t; L represents the acceleration factor; v i(t-1) represents the speed of individual i in round t-1; r1 and r2 represent random numbers in the range [0, 1]; p i represents the best historical position of individual i; g represents the best position in the current design space; X i(t-1) represents the position of individual i in round t-1; X it Represents the position of individual i in round t.

[0038] As a further solution of the present invention, the specific steps of evaluating the performance of different design solutions in actual use in step V are as follows:

[0039] Q5.1: Create a root node based on the initial state of the current design solution, and the root node contains the initial parameters and performance indicators related to the design solution. According to the design state or parameters of the root node, determine the design changes to be made, and randomly select new parameter values ​​within the current parameter range to create new child nodes;

[0040] Q5.2: Starting from the root node, calculate the upper confidence bound of each node, select the child node with the highest upper confidence bound through the UCT selection strategy, and continue to select nodes until the leaf node is selected. Based on the currently selected node, generate a new child node according to the change of design parameters;

[0041] Q5.3: Perform random simulations on newly generated child nodes by running simulation models or empirical formulas, evaluate their performance under different working conditions, transmit simulation results from child nodes back to their parent nodes, update the number of visits and evaluation values ​​of each node, and repeat selection, expansion, simulation, and backtracking until the set number of iterations is reached or the evaluation results converge to the preset threshold;

[0042] Q5.4: Traverse downward from the root node of the tree, check all child nodes of the current node, compare their visit counts, and select the child node with the most visits. Update the current node to the selected child node and continue traversing downward. If the current node is a leaf node, or when the maximum depth is reached, stop traversal and end the selection process. Otherwise, continue the selection process. When the traversal is completed, output the child node with the most visits as the final design solution.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. The intelligent wire harness section design method based on multi-data fusion randomly perturbs each group of parameters in the preliminary wire harness section design scheme to generate multiple groups of different design schemes, and uses each group of design schemes to construct the corresponding design space, and then randomly generates multiple groups of individual initial positions in the design space to represent different design schemes. The fitness function is pre-defined according to the wire harness design goal, and the performance of each group of individuals is evaluated by the fitness function. Then, the moving speed of each group of individuals is updated, and based on the updated speed of the individual, the position of the individual in the design space at the next moment is adjusted. Then, it is checked whether the individual position exceeds the boundary of the design space. If it exceeds the boundary, it is adjusted back to the boundary, and the speed and position updates are repeated until the fitness reaches the preset threshold. The updated position of each group of individuals is used as the starting position, the pheromone concentration of each parameter node in the design scheme is initialized, and the corresponding heuristic is calculated value, and then calculate the selection probability of the next node according to the pheromone concentration and the heuristic value, repeat the path selection until the individual constructs a complete design plan, evaluate the quality of the path according to the fitness function, and update the pheromone concentration, repeat the path construction and pheromone concentration update until the pheromone concentration converges to the preset range, and alternately update the design plans of different methods until the fitness value reaches the threshold or the maximum number of iterations is reached, and after each round of iteration, the design plan with the highest fitness is selected as the basis for the new round of optimization. After the iteration, the optimized design plan is output, including the best wire harness section parameters and corresponding performance indicators, and the design document is provided, which can effectively avoid local optimal solutions and find the global optimal solution, so that the wire harness section design can explore a wider design space and improve the targeted optimization. The two methods can complement each other, improve the convergence speed, and reduce the number of iterations required for optimization.

[0045] 2. The intelligent wire harness cross-section design method based on multi-data fusion creates a root node according to the initial state of the current design scheme, and the root node contains initial parameters and performance indicators related to the design scheme. According to the design state or parameters of the root node, the design changes are determined, and new parameter values ​​are randomly selected within the current parameter range to create new child nodes. Starting from the root node, the upper confidence bound value of each node is calculated, and the child node with the highest upper confidence bound value is selected through the UCT selection strategy. Node selection is continuously performed until a leaf node is selected. Based on the currently selected node, a new child node is generated according to the change of the design parameters. The newly generated child node is randomly simulated by running a simulation model or an empirical formula to evaluate its performance under different working conditions. The simulation results are transmitted back from the child node to its parent node, and the number of visits to each node is updated. and evaluation value, repeat selection, expansion, simulation and backtracking until the set number of iterations is reached or the evaluation result converges to the preset threshold, traverse downward from the root node of the tree, check all child nodes of the current node, compare their access times, and select the child node with the most access times, update the current node to the selected child node, and continue to traverse downward. If the current node is a leaf node, or the traversal is stopped after the maximum depth is reached, the selection process ends, otherwise, the selection process continues. When the traversal is completed, the child node with the most access times is output as the final design solution, which can provide a comprehensive performance evaluation and can consider a variety of environmental variables to better reflect the performance of the wiring harness in actual applications. It can effectively deal with the uncertainty in the design process, provide a more reliable performance estimate for the design, and make the optimization process of the design solution more efficient and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0047] Figure 1 This is a flowchart of the intelligent wire harness cross-section design method based on multi-data fusion proposed by the present invention. DETAILED DESCRIPTION

[0048] Example 1

[0049] Reference Figure 1 , an intelligent wire harness cross-section design method based on multi-data fusion, the specific steps of the design method are as follows:

[0050] Collect and preprocess various data for wire harness design.

[0051] Specifically, various wire harness design data are collected from various data sources such as internal databases, industry standard documents, and real-time sensor data. A sliding window of size N×N is set according to different lengths of wire harness design data, and the moving average of the internal data of each position window is gradually calculated from the data head. The calculated moving average replaces the data at the corresponding position in the original data, and the data of different dimensions are converted to the same dimension through the Max-Min normalization method. Then, the missing values ​​in each group of wire harness design data are detected, and the missing data are filled by mean filling or interpolation. Finally, the data from different sources are integrated to establish a wire harness design set.

[0052] Correlate information from different data to build a wire harness design picture.

[0053] Specifically, the various collected data are automatically or manually labeled according to preset rules, key entities and their attributes are identified, entities are extracted from the labeled data through natural language processing technology, and then the named entity recognition algorithm is used to identify the materials, design parameters and environmental factors corresponding to each entity. The relationship between different entities is extracted through the GCN network, and the extracted entities and relationships are organized into a graph structure, in which the nodes in the graph structure represent entities and the edges represent relationships. Attribute values ​​are assigned to each node to complete the construction of the wire harness design map, and then the constructed wire harness design map is stored in the graph database, and a query interface is set for subsequent data retrieval and update.

[0054] Analyze the information in the map and extract various characteristic data of wiring harness performance.

[0055] Specifically, feature sets and labels for training are extracted from the harness design atlas, where the feature set contains various attributes related to the harness performance, and the label is the target variable. The extracted data sets are divided into training sets and test sets in a ratio of 8:2. The random forest model is initialized, the number of trees and other hyperparameters are set, and the random forest model is trained using the training set data. At the same time, during the training process, features are randomly selected and sampled with replacement multiple times to generate multiple sample subsets. Each subset is used to train a decision tree. Then, at each split node of each tree, a subset of features is randomly selected, and each node is split according to the selected features and their thresholds. The quality of different splits is evaluated, and the pure The split that maximizes the degree of improvement is performed, sample sampling and feature selection are repeated for each child node, and the split is continued until the preset purity threshold is reached. The contribution of each feature to the model prediction is analyzed, and the importance of the feature is evaluated. Then, the performance of the model is evaluated on the test set, and the accuracy and mean square error of the prediction are calculated. If the model performance does not meet the preset requirements, it is retrained. Otherwise, the latest design data in the wiring harness design atlas is passed into the trained random forest model. The random forest model performs feature importance evaluation on each group of design data received, and based on the results of the feature importance evaluation, selects the feature most relevant to the wiring harness performance. If the feature importance is greater than the preset threshold, the feature is retained, otherwise it is removed.

[0056] Example 2

[0057] Reference Figure 1 , an intelligent wire harness cross-section design method based on multi-data fusion, the specific steps of the design method are as follows:

[0058] Generate preliminary wire harness cross-section designs and iteratively update them.

[0059] Specifically, each group of parameters in the preliminary wire harness cross-section design scheme is randomly perturbed to generate multiple groups of different design schemes, and each group of design schemes is used to construct a corresponding design space. Then, multiple groups of initial positions of individuals are randomly generated in the design space to represent different design schemes. The fitness function is predefined according to the wire harness design goal, and the performance of each group of individuals is evaluated by the fitness function. Then, the moving speed of each group of individuals is updated, and based on the updated speed of the individual, the position of the individual in the design space at the next moment is adjusted. Then, it is checked whether the individual position exceeds the boundary of the design space. If it exceeds the boundary, it is adjusted back to the boundary, and the speed and position updates are repeated until the fitness reaches the preset threshold. The updated position of each group of individuals is used as the starting position to initialize The pheromone concentration of each parameter node in the design scheme is calculated, and its corresponding heuristic value is calculated. Then, the selection probability of the next node is calculated based on the pheromone concentration and the heuristic value. The path selection is repeated until the individual constructs a complete design scheme. The quality of the path is evaluated according to the fitness function, and the pheromone concentration is updated. The path construction and pheromone concentration update are repeated until the pheromone concentration converges to the preset range. The design scheme updates in steps Q4.2 and Q4.3 are performed alternately until the fitness value reaches the threshold or the maximum number of iterations is reached. After each round of iteration, the design scheme with the highest fitness is selected as the basis for the new round of optimization. After the iteration, the optimized design scheme is output, including the optimal harness cross-section parameters and corresponding performance indicators, and the design document is provided.

[0060] In this embodiment, the specific calculation formula of the fitness function is as follows:

[0061] F(X i )=w1·f1(X i )+w2·f2(X i )+...+w n ·f n (X i )

[0062] In the formula, F(X i ) represents the fitness of individual i; f n (X i ) represents the performance of individual i on the nth design goal; w n represents the weight of the nth design goal;

[0063] The specific calculation formulas for speed and position update are as follows:

[0064] v it =L·v i(t-1) +r1·(p i -X i(t-1) )+r2·(gX i(t-1) )

[0065] X it =X i(t-1) +v it

[0066] In the formula, v it represents the speed of individual i in round t; L represents the acceleration factor; v i(t-1) represents the speed of individual i in round t-1; r1 and r2 represent random numbers in the range [0, 1]; p i represents the best historical position of individual i; g represents the best position in the current design space; X i(t-1) represents the position of individual i in round t-1; X it Represents the position of individual i in round t.

[0067] Evaluate the performance of different design solutions in actual use and select one.

[0068] Specifically, a root node is created according to the initial state of the current design scheme, and the root node contains initial parameters and performance indicators related to the design scheme. According to the design state or parameters of the root node, the design changes to be made are determined, and new parameter values ​​are randomly selected within the current parameter range to create new child nodes. Starting from the root node, the upper confidence bound value of each node is calculated, and the child node with the highest upper confidence bound value is selected through the UCT selection strategy. Node selection is continued until a leaf node is selected. Based on the currently selected node, a new child node is generated according to the change of the design parameters. The newly generated child nodes are randomly simulated by running simulation models or empirical formulas to evaluate their performance in different working conditions. The performance under the working conditions is transmitted back from the child node to its parent node, the number of visits and evaluation values ​​of each node are updated, and selection, expansion, simulation and backtracking are repeated until the set number of iterations is reached or the evaluation result converges to the preset threshold, and the tree is traversed downward from the root node, all child nodes of the current node are checked, their number of visits is compared, and the child node with the most visits is selected, and the current node is updated to the selected child node, and the traversal continues downward. If the current node is a leaf node, or the traversal is stopped after the maximum depth is reached, the selection process ends. Otherwise, the selection process continues. When the traversal is completed, the child node with the most visits is output as the final design solution.

[0069] Implement final harness design and monitor harness performance data.

[0070] Make subsequent improvements and adjustments based on each group’s monitoring data.

Claims

1. An intelligent wire harness cross-section design method based on multi-data fusion, characterized in that: The specific steps of this design method are as follows: Ⅰ. Collect and pre-process various data for wiring harness design; Ⅱ. Associate information from different data to build a wiring harness design map; Ⅲ. Analyze the information in the graph and extract the characteristic data of the wiring harness performance; IV. Generate a preliminary wire harness cross-section design and perform iterative updates; V. Evaluate the performance of different design solutions in actual use and select them; VI. Implement the final wiring harness design and monitor the performance data of the wiring harness; Ⅶ. Make subsequent improvements and adjustments based on each group’s monitoring data.

2. The intelligent wire harness cross-section design method based on multi-data fusion according to claim 1 is characterized in that: The specific steps of preprocessing various data of wiring harness design in step I are as follows: Q1.1: Collect various harness design data from various data sources including internal databases, industry standard documents and real-time sensor data. Set a sliding window of size N×N according to the length of different harness design data. Calculate the moving average of the internal data of each position window step by step from the data head. Q1.2: Replace the data at the corresponding position in the original data with the calculated moving average, convert the data of different dimensions to the same dimension through the Max-Min normalization method, then detect the missing values ​​in each group of wiring harness design data, and fill the missing data through mean filling or interpolation, and then integrate the data from different sources to establish the wiring harness design set.

3. The intelligent wire harness cross-section design method based on multi-data fusion according to claim 2 is characterized in that: The specific steps for constructing the wiring harness design map described in step II are as follows: Q2.1: Automatically or manually annotate the collected data according to preset rules, identify key entities and their attributes, extract entities from the annotated data using natural language processing technology, and then use named entity recognition algorithms to identify the materials, design parameters, and environmental factors corresponding to each entity; Q2.2: The relationships between different entities are extracted through the GCN network, and the extracted entities and relationships are organized into a graph structure, where the nodes in the graph structure represent entities and the edges represent relationships. Attribute values ​​are assigned to each node to complete the construction of the wire harness design graph, and then the constructed wire harness design graph is stored in the graph database, and a query interface is set for subsequent data retrieval and update.

4. The intelligent wire harness cross-section design method based on multi-data fusion according to claim 3 is characterized in that: The specific steps of extracting various characteristic data of wiring harness performance in step III are as follows: Q3.1: Extract the feature set and labels for training from the harness design graph, where the feature set contains various attributes related to the harness performance and the label is the target variable. Divide each group of extracted data into training set and test set in a ratio of 8:2, initialize the random forest model, and set the number of trees and other hyperparameters. Q3.2: Use the training set data to train the random forest model. During the training process, randomly select features and perform multiple sample sampling with replacement to generate multiple sample subsets. Each subset is used to train a decision tree. Then, at each split node of each tree, randomly select a subset of features and split each node according to the selected features and their thresholds. Q3.3: Evaluate the quality of different splits, select the split that maximizes purity, repeat sample sampling and feature selection for each child node, and continue splitting until the preset purity threshold is reached; Q3.4: Analyze the contribution of each feature to the model prediction and evaluate the importance of the feature. Then evaluate the performance of the model on the test set and calculate the accuracy and mean square error of the prediction. If the model performance does not meet the preset requirements, retrain it. Otherwise, pass the latest design data in the wiring harness design map into the trained random forest model. Q3.5: The random forest model evaluates the feature importance of each set of design data received, and based on the results of the feature importance evaluation, selects the feature most relevant to the wiring harness performance. If the feature importance is greater than the preset threshold, the feature is retained, otherwise it is removed.

5. The intelligent wire harness cross-section design method based on multi-data fusion according to claim 1 is characterized in that: The specific steps of iterative updating described in step IV are as follows: Q4.1: Randomly perturb each group of parameters in the preliminary wire harness cross-section design scheme to generate multiple groups of different design schemes, and use each group of design schemes to construct the corresponding design space, and then randomly generate multiple groups of individual initial positions in the design space to represent different design schemes; Q4.2: Predefine the fitness function according to the harness design goal, and use the fitness function to evaluate the performance of each group of individuals. Then update the movement speed of each group of individuals, and adjust the position of the individual in the design space at the next moment based on the updated speed of the individual. Then check whether the individual position exceeds the boundary of the design space. If it exceeds the boundary, adjust it back to the boundary, and repeat the speed and position update until the fitness reaches the preset threshold; Q4.3: Use the updated positions of each group of individuals as the starting positions, initialize the pheromone concentration of each parameter node in the design scheme, and calculate its corresponding heuristic value. Then calculate the selection probability of the next node based on the pheromone concentration and the heuristic value. Repeat the path selection until the individual constructs a complete design scheme. Evaluate the quality of the path based on the fitness function and update the pheromone concentration. Repeat the path construction and pheromone concentration update until the pheromone concentration converges to the preset range. Q4.4: Alternately update the design schemes in steps Q4.2 and Q4.3 until the fitness value reaches the threshold or the maximum number of iterations is reached. After each round of iteration, select the design scheme with the highest fitness as the basis for the next round of optimization. After the iteration, output the optimized design scheme, including the optimal wire harness cross-section parameters and corresponding performance indicators, and provide design documents.

6. The intelligent wire harness cross-section design method based on multi-data fusion according to claim 5 is characterized in that: The specific calculation formula of the fitness function described in Q4.2 is as follows: F(X i )=w1·f1(X i )+w2·f2(X i )+...+w n ·f n (X i ) In the formula, F(X i ) represents the fitness of individual i; f n (X i ) represents the performance of individual i on the nth design goal; w n represents the weight of the nth design goal; The specific calculation formulas for speed and position update described in Q4.2 are as follows: v it =L·v i(t-1) +r1·(p i -X i(t-1) )+r2·(g-X i(t-1) ) X it =X i(t-1) +v it In the formula, v it represents the speed of individual i in round t; L represents the acceleration factor; v i(t-1) represents the speed of individual i in round t-1; r1 and r2 represent random numbers in the range of [0, 1]; p i represents the best historical position of individual i; g represents the best position in the current design space; X i(t-1) represents the position of individual i in round t-1; X it Represents the position of individual i in round t.

7. The intelligent wire harness cross-section design method based on multi-data fusion according to claim 5 is characterized in that: The specific steps for evaluating the performance of different design solutions in actual use described in step V are as follows: Q5.1: Create a root node based on the initial state of the current design solution, and the root node contains the initial parameters and performance indicators related to the design solution. According to the design state or parameters of the root node, determine the design changes to be made, and randomly select new parameter values ​​within the current parameter range to create new child nodes; Q5.2: Starting from the root node, calculate the upper confidence bound of each node, select the child node with the highest upper confidence bound through the UCT selection strategy, and continue to select nodes until the leaf node is selected. Based on the currently selected node, generate a new child node according to the change of design parameters; Q5.3: Perform random simulations on newly generated child nodes by running simulation models or empirical formulas, evaluate their performance under different working conditions, transmit simulation results from child nodes back to their parent nodes, update the number of visits and evaluation values ​​of each node, and repeat selection, expansion, simulation, and backtracking until the set number of iterations is reached or the evaluation results converge to the preset threshold; Q5.4: Traverse downward from the root node of the tree, check all child nodes of the current node, compare their visit counts, and select the child node with the most visits. Update the current node to the selected child node and continue traversing downward. If the current node is a leaf node, or when the maximum depth is reached, stop traversal and end the selection process. Otherwise, continue the selection process. When the traversal is completed, output the child node with the most visits as the final design solution.

Images