A shielding parameter optimization method and system for an electromagnetic interference resistant sensor wire harness

Abstract

The application discloses a shielding parameter optimization method for an electromagnetic interference resistant sensor wire harness, and belongs to the technical field of sensor wire harness design and electromagnetic compatibility simulation. The method determines an electromagnetic interference frequency spectrum range according to a target application scene of the wire harness, constructs a multi-target optimization index system containing shielding effectiveness, cost and flexibility, sets manufacturing feasibility constraint conditions, and outputs a standardized multi-target optimization problem model. A parameterized description is made on a shielding layer structure, a multi-stage electromagnetic simulation framework combining analytical approximate estimation and finite element accurate verification is built, a simulation deviation dynamic calibration mechanism is established, and a shielding performance evaluation capability capable of iterative calling is formed. With the given optimization target and constraint as the optimization criterion, the performance evaluation capability is called to calculate the fitness of population individuals, an improved non-dominated sorting evolutionary algorithm with a dynamic constraint relaxation mechanism is used for iterative optimization, and a Pareto optimal shielding parameter solution set is output, which can be directly used to guide the manufacturing of the electromagnetic interference resistant sensor wire harness.

Description

Technical Field

[0001] This invention relates to the field of sensor harness design and electromagnetic compatibility simulation technology, specifically to a method and system for optimizing the shielding parameters of an electromagnetic interference-resistant sensor harness. Background Technology

[0002] As a critical carrier of signal transmission, the sensor harness's anti-interference performance in environments with strong electromagnetic interference directly affects the sensor's measurement accuracy and system reliability. Currently, the shielding design of sensor harnesses mainly faces the following technical challenges: (i) Shielding design relies on experience and lacks a systematic parameter optimization method. Traditional shielding design mainly relies on engineers' experience to determine parameters such as braiding density, number of shielding layers, and material selection. The design cycle is long and it is difficult to obtain the globally optimal solution. The electromagnetic interference spectrum characteristics of different application scenarios vary significantly, and fixed parameter solutions are difficult to adapt to diverse needs.

[0003] (ii) There is an inherent conflict between shielding performance and multiple objectives such as cost and flexibility. While increasing the braiding density and the number of shielding layers can enhance the shielding effect, it will significantly increase the weight of the harness, manufacturing costs, and reduce flexibility. How to achieve the optimal balance between anti-interference performance, cost, and flexibility is a difficult problem that urgently needs to be solved in engineering practice.

[0004] (III) Existing optimization methods have limitations when applied to shielding parameter design. Although multi-objective evolutionary algorithms, represented by the non-dominated sorting genetic algorithm (NSGA-II), have been applied in many engineering fields, when directly transplanted to shielding parameter optimization, the following specific defects exist: (1) The fixed constraint processing mechanism imposes strong penalties on infeasible solutions in the early stage of evolution, resulting in the premature loss of population diversity and difficulty in escaping local optima. Some researchers have proposed improved schemes for constraint processing methods (such as the ε-constraint method and subsequent variants), but these methods are mostly general constraint processing strategies and have not been specifically adapted to the electromagnetic-mechanical-cost multi-physical coupling characteristics of shielding design; (2) The objective function has not been refined for the electromagnetic-mechanical-cost multi-physical coupling characteristics of shielding design, and the standard algorithm is difficult to converge effectively among shielding efficiency, cost and flexibility; (3) The standard algorithm lacks an effective interaction mechanism with electromagnetic simulation, and the optimization results often deviate from the actual manufacturable parameter range. Some studies employ single-objective genetic algorithms or simple weighted sum methods, but single-objective optimization cannot obtain the Pareto front, and the weight coefficients of the weighted sum method are difficult to determine in advance and can only obtain a single compromise solution.

[0005] (iv) There are differences in accuracy among electromagnetic simulation methods with different fidelity. High-frequency approximate analytical methods are fast but have limited accuracy, while full-wave finite element simulation has high accuracy but high computational cost. There is a lack of effective verification and synergistic utilization mechanism between the two. Summary of the Invention

[0006] To address the aforementioned problems, this invention proposes a method and system for optimizing the shielding parameters of electromagnetic interference-resistant sensor harnesses. The core of this invention lies in the following: constraint violation is handled using dimensionless normalization, directly binding it to physical constraints such as minimum shielding efficiency, maximum outer diameter, and maximum cost in the shielding design; the selection of attenuation parameters is based on the analysis of multi-physical coupling characteristics, exhibiting domain-specific adaptability advantages; simultaneously, a dynamic constraint mechanism is combined with a Kriging surrogate model and a target scaling factor to form a complete shielding parameter optimization framework. Furthermore, addressing the issue that standard non-dominated sorting genetic algorithms cannot directly handle shielding parameter optimization problems with discrete variables in their standard form, this invention employs a partitioned Kriging surrogate model and a hybrid encoding strategy to solve this problem, and achieves continuous improvement in optimization accuracy through a feedback calibration mechanism between low-fidelity and high-fidelity simulations, resulting in a synergistic effect among the three.

[0007] This invention addresses the shortcomings of existing sensor harness shielding designs, such as reliance on experience, difficulty in balancing multiple objectives, and a lack of systematic simulation optimization methods. It provides a simulation optimization method and system for the shielding parameters of electromagnetic interference-resistant sensor harnesses. This invention achieves systematic and parameterized optimization of shielding parameters through a multi-level strategy combining high-frequency approximate analytical methods and finite element simulation. Furthermore, it improves optimization accuracy through a feedback calibration mechanism between low-fidelity and high-fidelity simulations, ultimately achieving optimal overall performance in terms of anti-interference capabilities, cost, and flexibility.

[0008] This invention provides a method for optimizing the shielding parameters of an electromagnetic interference-resistant sensor harness, comprising the following: Step S1: Determine the electromagnetic interference spectrum range based on the target application scenario of the sensor harness, construct a multi-objective optimization index system including shielding effectiveness, cost, and flexibility, set manufacturing feasibility constraints, and output a standardized multi-objective optimization problem model containing optimization objectives and constraints. Step S2: Parametrically describe the shielding layer structure, construct a multi-level electromagnetic simulation framework that includes analytical approximate estimation and finite element accurate verification, establish a dynamic calibration mechanism for simulation deviation, and output a shielding performance evaluation capability that can be iteratively called. Step S3: Using the optimization objective and constraints in the standardized multi-objective optimization problem model output in step S1 as the optimization criteria, the fitness of individuals in the population is calculated by calling the shielding performance evaluation capability output in step S2. An improved non-dominated sorting evolutionary algorithm with dynamic constraint relaxation mechanism is used for iterative optimization, and the Pareto optimal shielding parameter solution set is output to guide the manufacturing of electromagnetic interference sensor harnesses.

[0009] Preferably, in step S1, in the multi-objective optimization index system, the shielding effectiveness index is the weighted sum of the shielding efficiencies of multiple focus frequencies, and the weight coefficient of each frequency is determined according to the proportion of interference spectrum energy in the corresponding frequency band; the cost index includes the material cost and manufacturing cost of the shielding structure; the flexibility index is characterized by bending stiffness, and the optimization direction is to minimize bending stiffness to improve the flexibility of the wire harness.

[0010] Preferably, in step S1, within the standardized multi-objective optimization problem model, the objective function corresponding to the multi-objective optimization index system is expressed in standard minimization form as follows: ; In the formula, To mask the parameters, design a variable vector. To comprehensively assess shielding performance indicators, For cost indicators, As a flexibility indicator, , , This is the target scaling factor used for dimensional normalization.

[0011] Preferably, in step S2, the multi-level electromagnetic simulation framework adopts a three-level collaborative architecture: the first level is an analytical approximation model based on plane wave shielding theory, used for rapid shielding efficiency estimation of large-scale population individuals during evolutionary optimization; the second level is a frequency domain finite element simulation model, used for high-precision shielding performance verification of candidate optimization schemes selected through evolution; the third level is a feedback calibration mechanism, which constructs a deviation prediction model based on the deviation data between the frequency domain finite element simulation model and the analytical approximation model, corrects the output results of the analytical approximation model, and the deviation prediction model is dynamically updated with the optimization process.

[0012] Preferably, in step S2, based on the parameterized description of the shielding structure and the multi-level electromagnetic simulation framework, and considering the mixed characteristics of continuous and discrete variables in the shielding design variables, a partitioned Kriging surrogate model is used to accelerate the simulation calculation: the design space is partitioned according to the discrete variable combination of shielding layer type, material type, and number of shielding layers, and a Kriging surrogate model is constructed for each partition; during the evolutionary optimization process, the Kriging surrogate model is called at intervals to evaluate the performance of individuals in the population, and the individual with the highest prediction uncertainty is periodically selected for verification through the analytical approximation model, and the verification results are added to the sample set to update the Kriging surrogate model.

[0013] Preferably, in step S3, the improved non-dominated sorting evolutionary algorithm uses an improved constraint-Pareto dominance relationship to perform hierarchical sorting of individuals. The judgment of the dominance relationship is performed according to three levels of priority: the first priority distinguishes individuals inside and outside the constraint boundary, and individuals that satisfy the constraint boundary are given priority; the second priority performs standard Pareto dominance judgment on individuals inside the constraint boundary; the third priority sorts individuals outside the constraint boundary according to the degree of constraint violation, and individuals with smaller constraint violation are given priority.

[0014] Preferably, the constraint violation degree is a dimensionless scalar, calculated using the following formula: ; In the formula, It is the constraint violation degree, a dimensionless scalar of the output, representing the design variable vector. The corresponding shielding parameter scheme violates the total degree of all manufacturing feasibility constraints. The total number of terms used to create feasibility constraints is the upper limit of the summation operation; The index variable for the constraint term has a range of values. =1, 2, ..., This is used to iterate through each manufacturing feasibility constraint function; For the first A constraint function whose value is less than or equal to 0 indicates that the corresponding constraint is satisfied; For the first The reference deviation of the constraint is used to normalize the constraint violation values ​​of different physical dimensions.

[0015] Preferably, the threshold used to distinguish between the inside and outside of the constraint boundary in the improved constraint-Pareto dominance relationship is the constraint relaxation threshold of dynamic contraction, and the constraint relaxation threshold corresponding to the current generation is calculated by the following formula: ; In the formula, This represents the constraint relaxation threshold corresponding to the current generation. This represents the initial constraint relaxation value; The current generation number; The maximum number of generations; The constraint relaxation threshold is set as the decay control parameter. In the early stages of evolution, the constraint relaxation threshold is relatively large to preserve population diversity and avoid getting trapped in local optima. As the number of generations increases, the constraint relaxation threshold is gradually tightened to ensure that the output Pareto optimal solution satisfies all manufacturing feasibility constraints.

[0016] Preferably, in step S3, the improved non-dominated sorting evolutionary algorithm with dynamic constraint relaxation mechanism uses a hybrid encoding and evolutionary operation rule for the masking parameter design variable vector corresponding to individuals in the population: The coding rules are as follows: real numbers are used for two types of continuous design variables, namely braiding angle and shielding layer thickness; integers are used for four types of discrete design variables, namely, the equivalent number of braids per braid, the number of shielding layers, the shielding layer type, and the material type. The crossover operation rules are as follows: simulated binary crossover is performed on the coding segments corresponding to continuous design variables, and uniform discrete recombination is performed on the coding segments corresponding to discrete design variables. The mutation operation rules are as follows: perform polynomial mutation on the coding segment corresponding to the continuous design variable, and randomly select and replace the coding segment corresponding to the discrete design variable from the preset set of feasible discrete values.

[0017] This invention also provides a shielding parameter optimization system for electromagnetic interference-resistant sensor harnesses, used to implement the above-mentioned method. The system includes: The electromagnetic environment and performance index modeling module is used to determine the electromagnetic interference spectrum range based on the target application scenario of the sensor harness, construct a multi-objective optimization index system including shielding effectiveness, cost, and flexibility, set manufacturing feasibility constraints, and output a standardized multi-objective optimization problem model containing optimization objectives and constraints. The shielding structure parametric modeling and electromagnetic simulation module is used to parametrically describe the shielding layer structure, construct a multi-level electromagnetic simulation framework that includes analytical approximate estimation and finite element accurate verification, establish a dynamic calibration mechanism for simulation deviation, and output a shielding performance evaluation capability that can be iteratively called. The shielding parameter optimization module based on the improved non-dominated sorting evolutionary algorithm is used to calculate the fitness of individuals in the population by calling the shielding performance evaluation capability output by the shielding structure parameterized modeling and electromagnetic simulation module, using the optimization objective and constraints in the standardized multi-objective optimization problem model output by the electromagnetic environment and performance index modeling module as the optimization criteria, and employing the improved non-dominated sorting evolutionary algorithm with dynamic constraint relaxation mechanism for iterative optimization. The module outputs the Pareto optimal shielding parameter solution set to guide the manufacturing of electromagnetic interference sensor harnesses.

[0018] The present invention provides a method and system for optimizing the shielding parameters of an electromagnetic interference-resistant sensor harness, which, compared with the prior art, has the following advantages: 1. Systematic Multi-Objective Optimization Capability: This invention upgrades sensor harness shielding design from experience-driven to parameterized, simulation-driven optimization. Through an improved non-dominated sorting evolutionary algorithm, it simultaneously optimizes three conflicting objectives: shielding efficiency, cost, and flexibility, outputting a complete Pareto optimal solution set for engineering decision-making. This overcomes the shortcomings of traditional designs in balancing multiple conflicting objectives. Compared to the standard non-dominated sorting genetic algorithm's fixed-constraint penalty approach, the dynamic constraint relaxation mechanism of this invention maintains greater population diversity in the early stages of evolution, contributing to a more uniformly distributed Pareto front.

[0019] 2. Dynamic constraint handling mechanism: This invention introduces a dynamic constraint relaxation threshold. In the early stages of evolution, more diverse parameter combinations are allowed to be explored to avoid getting trapped in local optima, while constraints are gradually tightened in the later stages to ensure that the final solution meets all manufacturing feasibility constraints. Compared with existing... - Compared to constraint methods, the constraint violation rate of this invention is higher. The attenuation parameter is directly bound to the physical constraints of the shielding design, such as minimum shielding efficiency, maximum outer diameter, and maximum cost, through normalization processing. The selection is based on the analysis of multi-physics coupling characteristics (when the constraint is steep). When the constraint is gentle It has the advantage of domain-specific adaptation. Attached Figure Description

[0020] Figure 1 This is a flowchart of a method for optimizing the shielding parameters of an electromagnetic interference-resistant sensor harness according to the present invention.

[0021] Figure 2 This is a flowchart of the improved non-dominated sorting evolutionary algorithm in this invention.

[0022] Figure 3 This is a schematic diagram of the overall architecture of the shielding parameter optimization system for an anti-electromagnetic interference sensor harness according to the present invention. Detailed Implementation

[0023] The following detailed implementation of a method and system for optimizing the shielding parameters of an anti-electromagnetic interference sensor harness according to the present invention, with reference to specific embodiments, is provided. The embodiments of the present invention are only used to explain the present invention and are not intended to limit the scope of protection of the present invention.

[0024] Example 1: Implementation of a method for optimizing the shielding parameters of an electromagnetic interference-resistant sensor harness.

[0025] Combined with appendix Figures 1-2 As shown, the present invention provides a method for optimizing the shielding parameters of an electromagnetic interference-resistant sensor harness.

[0026] like Figure 1 As shown, Figure 1 This is a flowchart of a method for optimizing the shielding parameters of an electromagnetic interference-resistant sensor harness according to the present invention.

[0027] Step S1: Modeling the electromagnetic environment and performance indicators.

[0028] Step S1 determines the electromagnetic interference spectrum range based on the target application scenario of the sensor harness, constructs a multi-objective optimization index system including shielding effectiveness, cost, and flexibility, sets manufacturing feasibility constraints, and outputs a standardized multi-objective optimization problem model containing optimization objectives and constraints.

[0029] In step S1, the shielding effectiveness index in the multi-objective optimization index system is a weighted sum of the shielding efficiency of multiple frequencies of interest, and the weight coefficient of each frequency is determined according to the proportion of interference spectrum energy in the corresponding frequency band; the cost index includes the material cost and manufacturing cost of the shielding structure; the flexibility index is characterized by bending stiffness, and the optimization direction is to minimize bending stiffness to improve the flexibility of the wire harness.

[0030] In step S1, within the standardized multi-objective optimization problem model, the objective function corresponding to the multi-objective optimization index system is expressed in standard minimization form as follows: ; In the formula, To mask the parameters, design a variable vector. To comprehensively assess shielding performance indicators, For cost indicators, As a flexibility indicator, , , This is the target scaling factor used for dimensional normalization. The specific process is as follows: Based on the application scenario of the sensor harness (such as engine compartment, near frequency converter, aerospace equipment compartment, etc.), determine the target interference spectrum range and intensity distribution.

[0031] Define shielding effectiveness vector : ; in (Number of frequencies of interest) indicates the frequency range. The shielding efficiency at the location is expressed in dB.

[0032] Comprehensive shielding performance indicators Defined as the shielding efficiency of each frequency of interest Weighted sum: ; in For the first The weighting coefficients for each frequency point are determined by: [the method for determining the weighting coefficients is based on the spectral energy distribution of the interference source in the application scenario]. This distribution can be set through electromagnetic environment simulation under typical operating conditions or engineering experience values, and the weights are calculated according to the energy proportion: ; If spectral energy distribution data cannot be obtained, equal weighting can be used ( The weight distribution can be determined either by the limit curves specified in relevant electromagnetic compatibility standards (such as ISO 11452, MIL-STD-461).

[0033] Cost indicator definition Material cost of shielding structure With manufacturing costs sum: ; in For the design variable vector, the definition is given below, material cost. Based on the volume of the material used With unit price calculate: ; in For the first The density of a type of material, For volume, Cost per unit of quality.

[0034] Manufacturing costs Calculations based on the processing time and unit time cost of each process: ; Total processing time , The weaving time (related to the number of weaving layers, weaving density, and traction speed) ,in Number of shielding layers For cable length, (for traction speed). To squeeze out time, For the detection time; The cost per unit of time is the labor cost. Equipment depreciation costs are allocated based on production volume. Manufacturing costs reflect the impact of process complexity on total costs.

[0035] Flexibility index Defined as bending stiffness, a smaller value indicates better flexibility: ; in The equivalent elastic modulus of the shielding layer. Let the moment of inertia of the cross section be... The characteristic length can be taken as a unit length. In this invention, A uniform definition of bending stiffness is adopted, and the optimization objective is to minimize this index.

[0036] Methods for obtaining flexibility index: Equivalent elastic modulus of shielding layer The orthogonal anisotropy model from composite laminate theory is used for estimation. For the braided shielding layer, its axial bending stiffness... From the weaving angle and coverage Joint decision: ; in This is the Young's modulus of the braided yarn material. This formula applies to the braiding angle. exist The situation within the scope.

[0037] Moment of inertia of cross section Calculated based on the cross-section of the annulus: ; in and These are the outer diameter and inner diameter of the shielding layer, respectively.

[0038] This estimation model can quickly evaluate the flexibility index under different parameter combinations during the optimization process.

[0039] In addition, for weight-sensitive applications such as aerospace, a unit length mass constraint can be added, which is calculated as the sum of the products of the density and cross-sectional area of ​​each layer of material.

[0040] In summary, the multi-objective optimization problem of this invention can be uniformly formulated as standard minimization. form: ; In the formula, To mask the parameters, design a variable vector. To comprehensively assess shielding performance indicators, For cost indicators, As a flexibility indicator, , , This is the target scaling factor used for dimensional normalization, which is used to normalize the dimensions of each target. The default value is 1. Taking a negative sign before the symbol indicates maximizing the shielding efficiency.

[0041] Step S2: Parametric modeling of the shielding structure and multi-stage electromagnetic simulation.

[0042] Step S2 provides a parameterized description of the shielding layer structure, constructs a multi-level electromagnetic simulation framework that includes analytical approximate estimation and finite element precise verification, establishes a dynamic calibration mechanism for simulation deviations, and outputs an iteratively callable shielding performance evaluation capability.

[0043] In step S2, the multi-level electromagnetic simulation framework adopts a three-level collaborative architecture: the first level is an analytical approximation model based on plane wave shielding theory, used for rapid shielding efficiency estimation of large-scale population individuals during evolutionary optimization; the second level is a frequency domain finite element simulation model, used for high-precision shielding performance verification of candidate optimization schemes selected through evolution; the third level is a feedback calibration mechanism, which constructs a deviation prediction model based on the deviation data between the frequency domain finite element simulation model and the analytical approximation model, corrects the output results of the analytical approximation model, and the deviation prediction model is dynamically updated with the optimization process.

[0044] In step S2, based on the parameterized description of the shielding structure and the multi-level electromagnetic simulation framework, and considering the mixed characteristics of continuous and discrete variables in the shielding design variables, a partitioned Kriging surrogate model is used to accelerate the simulation calculation: the design space is partitioned according to the discrete variable combination of shielding layer type, material type, and number of shielding layers, and a Kriging surrogate model is constructed for each partition; during the evolutionary optimization process, the Kriging surrogate model is called at intervals to evaluate the performance of individuals in the population, and the individual with the highest prediction uncertainty is periodically selected for verification through an analytical approximation model, and the verification results are added to the sample set to update the Kriging surrogate model. The specific process is as follows: S2.1, The shielding structure to be optimized is described parametrically, and the design variable vector is: ; The design variables are defined as follows: weaving angle (Value range: 30°~80°, continuous variable); Equivalent number of braided yarns per share (The basic set of values ​​is) (For braiding processes employing dual-spindle or multi-spindle parallel winding structures, the equivalent number of strands can be further expanded); shielding layer thickness. (Continuous variable, typical range 0.05~0.25mm); Number of shielding layers (Values ​​are 1, 2, or 3, discrete variables); Shielding layer type (braided shielding, wrapped shielding, foil plus braided composite shielding), uses discrete encoding; Material type (select from the preset material library: copper, aluminum, tin-plated copper, copper-clad aluminum, etc.), uses discrete encoding.

[0045] Woven coverage In this invention, the variables are not independent design variables, but dependent variables derived from the aforementioned design variables. Their calculation formula is as follows: ; in For the diameter of the braided monofilament, The outer diameter of the cable, including the insulation layer, is a given cable specification parameter. The grounding method, whether single-ended or double-ended, is not an optimization variable but rather a pre-determined input condition by the user based on system specifications, such as automotive CAN bus specifications or aerospace grounding specifications.

[0046] Regarding the handling of discrete variables: for shielding layer type (type), material type (material), and number of wires per strand... and number of shielding layers Discrete variables are encoded using integers (e.g., 1 represents braided shielding, 2 represents wrapped shielding, and 3 represents composite shielding). Discrete recombination is used in crossover operations, and random replacement from the set of feasible discrete values ​​is used in mutation operations.

[0047] S2.2, Multi-level electromagnetic simulation strategy: This invention employs a multi-level strategy that combines low-fidelity and high-fidelity simulations to achieve a balance between computational efficiency and accuracy.

[0048] Level 1 (Low-fidelity rapid estimation - analytical approximation model): In the main loop of the evolutionary algorithm, for large-scale populations requiring frequent evaluation, an analytical method based on a plane wave shielding model is used for rapid estimation. Shielding efficiency. The expression is: ; The components are defined as follows: Reflection loss Based on the principle of impedance mismatch when a plane wave is incident on an infinitely large flat plate shield, the calculation formula is as follows: ; in, For air wave impedance, The surface impedance of the shielding layer: ; in, For electrical conductivity, Permeability, Angular frequency, For skin depth.

[0049] Absorption loss The formula is: ; in, The shielding layer thickness increases with increasing frequency and thickness.

[0050] Multiple reflection correction term Only when Consider at the time; when hour, It can be ignored.

[0051] This analytical model has reasonable engineering reference value under far-field high-frequency conditions (typically above 30 MHz) and can be used as a rapid estimation tool in the initial screening stage of the optimization process. A single evaluation takes less than 0.1 seconds, enabling generational evaluation of large-scale populations.

[0052] Level 2 (High-fidelity accurate verification - finite element simulation): For the candidate optimal solutions selected during the evolution process, the frequency domain finite element method is used for accurate verification.

[0053] The specific method for establishing the simulation model is as follows: The solution domain is set as a two-dimensional model including the wire harness cross section and the surrounding air layer (used to calculate the transfer impedance per unit length and the cross-sectional field distribution). Boundary conditions: The outer air layer boundary adopts a second-order absorbing boundary condition to simulate the absorption of electromagnetic waves by an infinite space; Excitation and port settings: Based on the characteristic impedance of the signal transmission line in the actual application (such as automotive CAN bus). Radio frequency applications Set the port characteristic impedance A differential port excitation with this characteristic impedance is applied between the conductor and the shielding layer. For preliminary evaluation during the design phase, if the actual characteristic impedance is unknown, a method can be used... As a general reference value; For the candidate optimal solution, the scattering parameters are calculated using a three-dimensional finite element model. The shielding efficiency is from (dB) is calculated; Mesh generation: At least 2-3 cell layers must be maintained along the thickness direction of the shielding layer, and the maximum cell size should not exceed the wavelength corresponding to the highest frequency in the operating band. ( , (at the speed of light) to ensure calculation accuracy.

[0054] Level 3 (Multi-level simulation feedback calibration): A systematic bias exists between low-fidelity analytical models and high-fidelity finite element simulations, and this bias varies with frequency and parameter combinations. To improve the prediction accuracy of low-fidelity models, this invention establishes a feedback calibration mechanism: Select Representative parameter combinations At the same time, an analytical model is adopted. and finite element simulation Calculate its shielding efficiency and obtain the deviation dataset. : ; in , , , Based on this deviation dataset, a deviation prediction model is established. (Using Kriging interpolation or multinomial regression) to correct the output of the analytical model: ; The calibration model is dynamically updated during the optimization process: whenever a new candidate scheme is verified by high-fidelity finite element simulation, its results are added to the deviation dataset, and the deviation prediction model is refitted, so that the prediction accuracy of the low-fidelity model gradually improves with the optimization process.

[0055] The aforementioned multi-level combined strategy allows the computational efficiency advantage of high-frequency approximate analytical methods to be maintained during the optimization process, while the prediction accuracy is continuously improved through feedback calibration of finite element simulation.

[0056] S2.3, Regarding the proxy model acceleration strategy: To further reduce computational overhead during the optimization process, this invention employs a Kriging surrogate model to achieve secondary acceleration based on the low-fidelity analytical model. Addressing the issue of discrete variables within the design variables, this invention adopts a classification sub-model strategy: based on combinations of discrete variables (shielding layer type, material type, number of shielding layers...). For the design space partitioning, Kriging proxy models are established for the main discrete combinatorial partitions. The specific strategy is as follows: (1) The values ​​of discrete variables are set as follows: 3 types of type and 4 types of material (copper, aluminum, tin-plated copper, copper-clad aluminum). Seeds (1st, 2nd, and 3rd layers). There are four basic sets (extended equivalent values ​​are counted separately). In the initial sampling phase before the evolutionary algorithm starts, initial sample points are generated using Latin hypercube sampling and assigned to each partition according to the combination of discrete variables. For partitions with insufficient sample points, neighbor-partition model transfer learning or default material parameters are used as approximations.

[0057] (2) Construct a Kriging proxy model for each major partition based on the training data. The Kriging model employs an anisotropic Matérn-5 / 2 kernel function, and the hyperparameters are optimized through maximum likelihood estimation; for Each focus point is used to construct an independent Kriging model, totaling [number missing]. One model.

[0058] (3) In the main loop of the evolutionary algorithm, the population is evaluated every 9 generations using the Kriging surrogate model. Every 10 generations, the 5 individuals with the highest prediction uncertainty in the current population are selected for analytical model calculation (the analytical model has been calibrated by finite element simulation), and the results are added to the training dataset to update the surrogate model.

[0059] (4) The above two-layer acceleration strategy (analytical model plus Kriging proxy model) can significantly reduce the computational overhead in the optimization process.

[0060] Step S3: Multi-objective optimization based on the improved non-dominated sorting evolutionary algorithm.

[0061] Step S3 uses the optimization objective and constraints in the standardized multi-objective optimization problem model output in step S1 as the optimization criteria, calls the shielding performance evaluation capability output in step S2 to calculate the fitness of individuals in the population, and uses an improved non-dominated sorting evolutionary algorithm with dynamic constraint relaxation mechanism to perform iterative optimization, outputting the Pareto optimal shielding parameter solution set, which is used to guide the manufacturing of electromagnetic interference sensor harnesses.

[0062] In step S3, the improved non-dominated sorting evolutionary algorithm uses an improved constraint-Pareto dominance relationship to perform hierarchical sorting of individuals. The judgment of the dominance relationship is performed according to three levels of priority: the first priority distinguishes individuals inside and outside the constraint boundary, and individuals that satisfy the constraint boundary are given priority; the second priority performs standard Pareto dominance judgment on individuals inside the constraint boundary; the third priority sorts individuals outside the constraint boundary according to the degree of constraint violation, and individuals with smaller constraint violation are given priority.

[0063] In step S3, the improved non-dominated sorting evolutionary algorithm with dynamic constraint relaxation mechanism employs hybrid encoding and evolutionary operation rules for the shielding parameter design variable vectors corresponding to individuals in the population: The encoding rules are as follows: real numbers are used for continuous design variables such as weaving angle and shielding layer thickness, while integers are used for discrete design variables such as the equivalent number of braided filaments per share, the number of shielding layers, the shielding layer type, and the material type; the crossover operation rules are as follows: simulated binary crossover is performed on the encoding segments corresponding to continuous design variables, and uniform discrete recombination is performed on the encoding segments corresponding to discrete design variables; the mutation operation rules are as follows: polynomial mutation is performed on the encoding segments corresponding to continuous design variables, and values ​​are randomly selected from a preset set of feasible discrete values ​​to replace the encoding segments corresponding to discrete design variables. The specific process is as follows: like Figure 2 As shown, Figure 2 The flowchart of the improved non-dominated sorting evolutionary algorithm in this invention shows the complete iterative process from population initialization, fitness evaluation, non-dominated sorting, evolutionary operation to convergence judgment.

[0064] This invention uses an improved non-dominated sorting evolutionary algorithm as the core optimization engine to perform multi-objective optimization of the masking parameters. Compared with the standard non-dominated sorting genetic algorithm, the improvements of this invention are reflected in: (1) the introduction of a dynamic constraint relaxation mechanism to overcome the defect of premature loss of population diversity caused by fixed constraints in the standard algorithm; (2) the design of a special three-objective function structure and objective scaling factor for the multi-physical coupling characteristics of the masking parameters; and (3) the realization of efficient optimization under limited computational budget by combining a multi-level simulation strategy.

[0065] (1) Population coding and initialization.

[0066] Each individual corresponds to a set of masking parameter vectors Continuous variables Real number encoding, discrete variables ,type,material Integer encoding is used. Initial population. The population is uniformly generated in the parameter space using Latin hypercube sampling, and the population size is... .

[0067] (2) Fitness assessment.

[0068] For each individual in the population The electromagnetic simulation module (analytical approximation model, Kriging proxy model, or finite element simulation) is called to calculate its shielding efficiency vector, and then the three objective function values ​​are calculated. : ; Minimize (Equivalent to maximizing shielding efficiency), minimizing cost, and minimizing bending stiffness (equivalent to optimizing flexibility). Default. .

[0069] (3) Improved non-dominated sorting.

[0070] This invention employs an improved non-dominated sorting strategy to stratify individuals in the population. This strategy is specifically designed for manufacturing feasibility constraints (such as minimum shielding efficiency requirements, maximum outer diameter limits, maximum cost budgets, and maximum mass per unit length) in shielding parameter optimization.

[0071] Definition (Improved Constraint - Pareto Dominance): For any two individuals and ,say Improved constraint – Pareto dominance (recorded as) ), judged step by step according to the following priority: First priority (distinguishing between inside and outside the constraint boundary): If ,but That is, those that satisfy the boundary constraints take precedence over those that do not. Second priority (all within the constraint boundaries): If and and (Standard Pareto dominance), then ; Third priority (all exceeding constraint boundaries): If ,but That is, the one with the smaller degree of constraint violation takes priority; if and ,but .

[0072] The above three-layered judgment logic constitutes a complete determination of the dominance relationship between individuals, and there is no situation that cannot be judged.

[0073] in This refers to the constraint violation rate. To eliminate the problem that constraints cannot be directly added due to their different dimensions, the constraint functions are normalized to be dimensionless. Defined as the sum of all normalization constraint violations: ; in (where the number of constraints is the first) A constraint function, For the first Reference deviation of each constraint (for dimensionless representation): For the minimum shielding efficiency constraint Reference deviation measurement For the maximum outer diameter constraint Reference deviation measurement For the maximum cost constraint Reference deviation measurement Yuan; for the maximum unit length mass constraint Reference deviation measurement .

[0074] in: The minimum shielding efficiency constraint function, whose value is less than or equal to 0 indicates that the constraint is satisfied; : Design variable vector, including braiding angle, number of filaments per strand, shielding layer thickness, number of shielding layers, shielding layer type, and material type; : The minimum shielding efficiency threshold required for the application scenario, in dB; Frequency index, with a value range of: ; In parameter combinations Lower, frequency The predicted shielding efficiency at the location is expressed in dB. : No. One focus point; The lowest shielding efficiency among all frequencies of interest; : Maximum outer diameter constraint function, a value less than or equal to 0 indicates that the constraint is satisfied; In parameter combinations The total outer diameter of the lower wire harness, in mm; : The maximum allowable outer diameter of the wire harness in the application scenario, in mm; The maximum cost constraint function, whose value is less than or equal to 0, indicates that the constraint is satisfied; In parameter combinations The total cost (the sum of material costs and manufacturing costs); : The maximum cost budget allowed by the application scenario; : Maximum unit length mass constraint function, whose value is less than or equal to 0 indicates that the constraint is satisfied; In parameter combinations The mass per unit length of the wire harness is given in units of... ; : The maximum mass per unit length allowed in the application scenario, in units of .

[0075] After the above normalization process, constraint violation degrees of different dimensions can be compared and summed in a unified manner. It becomes a dimensionless scalar.

[0076] Each constraint function Specific forms include, but are not limited to: Minimum shielding efficiency constraint: ,in Minimum shielding efficiency required for the application scenario; Maximum outer diameter constraint: Maximum cost constraint: Maximum mass per unit length constraint (aerospace scenario): .

[0077] For the first The dynamic constraint relaxation threshold of the generation: ; in, Initial constraint relaxation value (typical value) (corresponding to an initial allowance of 50% normalization constraint violation), Gen is the maximum number of generations. , These are the attenuation control parameters. The selection of [the constraint] is related to the steepness of the constraint characteristics: when the objective function changes drastically near the constraint boundary, it is recommended [to use a specific constraint]. When the constraint boundaries are gentle, it is recommended to Typical value This dynamic constraint mechanism allows for the exploration of more diverse parameter combinations in the early stages of evolution, while gradually tightening the constraints in the later stages to ensure that the final solution meets all manufacturing feasibility constraints.

[0078] (4) Crowded distance calculation.

[0079] Within the same non-dominated layer, individuals are sorted according to their objective function values. The crowding distance between boundary individuals (the first and last individuals after sorting) is set as... This ensures that they are given priority for retention. Intermediate individuals. crowded distance for: ; in The number of objective functions. and The first The maximum and minimum values ​​of the objective function in the current layer. For the first On each objective function, and with the individual The objective function value of the next adjacent individual (after sorting by the objective function value). For the first On each objective function, and with the individual The objective function value of the adjacent previous individual (after sorting by the objective function value).

[0080] (5) Evolutionary operation.

[0081] Selection operator: Binary tournament selection is adopted, and individual selection is based on non-dominated level and crowding distance (those with lower non-dominated level are given priority; when the levels are the same, those with larger crowding distance are given priority).

[0082] Crossover operator: Continuous variables use simulated binary crossover, crossover probability Distribution index :

[0083] in For random variables, the distribution index is used. control; Index for variable dimensions. For the first parent individual in the first The variable value in the dimension. For the second parent individual in the first The variable value in the dimension. For the first offspring individual in the first The variable value in the dimension. For the second offspring individual in the first The variable value in the dimension.

[0084] Discrete variables are rearranged using discrete crossover (uniform crossover).

[0085] Mutation operator: Continuous variables undergo polynomial mutation, with mutation probability... ( (for design variable dimensions), distribution index 20: ; in For distribution index Controlled random disturbances. For the first The maximum value that a dimension variable can take. For the first The minimum value that a dimension variable can take. For the individual after mutation, on the 1st The variable value in the dimension.

[0086] Discrete variables are replaced randomly from the set of feasible discrete values.

[0087] (6) Elite retention and convergence judgment.

[0088] Each generation merges the parent and offspring populations (size) ), and select the optimal one through improved non-dominated sorting and crowding distance. Each individual enters the next generation. Evolution stops when any of the following termination conditions are met: reaching the maximum number of generations. ;continuous In the middle of the generation, the relative change of the hypervolume index at the Pareto front was less than the threshold. (That is, the rate of change of the supervolume is less than 0.1%).

[0089] The reference point used for hypervolume calculation adopts a fixed reference point: the statistical maximum value of each target dimension in the initial population is used as the benchmark, multiplied by a coefficient of 1.1 and kept unchanged to ensure that the hypervolume values ​​of each generation are comparable.

[0090] (7) Output of Pareto optimal solution set.

[0091] After evolution, all individuals in the first non-dominated layer are output as the Pareto optimal solution set. The population size is... In the last generation, the typical number of non-dominated solutions is 15-40 (depending on the distribution characteristics of the objective space and the setting of constraint relaxation during the optimization process). If the number of non-dominated solutions exceeds 30, a uniform sampling strategy (such as k-means clustering or grid uniform sampling) can be used to select 15-20 representative solutions for engineers to choose according to their actual preferences.

[0092] Example 2: Implementation of a shielding parameter optimization system for an anti-electromagnetic interference sensor harness.

[0093] Combined with appendix Figure 3 As shown, Figure 3 This is a schematic diagram of the overall architecture of a shielding parameter optimization system for an electromagnetic interference-resistant sensor harness according to the present invention. The diagram illustrates the complete process from electromagnetic environment modeling to shielding parameter optimization, as well as the data flow between each module. The present invention also provides a shielding parameter optimization system for an electromagnetic interference-resistant sensor harness, used to implement the above-mentioned method. The system includes: The electromagnetic environment and performance index modeling module is used to determine the electromagnetic interference spectrum range based on the target application scenario of the sensor harness, construct a multi-objective optimization index system including shielding effectiveness, cost, and flexibility, set manufacturing feasibility constraints, and output a standardized multi-objective optimization problem model containing optimization objectives and constraints. The shielding structure parametric modeling and electromagnetic simulation module is used to parametrically describe the shielding layer structure, construct a multi-level electromagnetic simulation framework that includes analytical approximate estimation and finite element accurate verification, establish a dynamic calibration mechanism for simulation deviation, and output a shielding performance evaluation capability that can be iteratively called. The shielding parameter optimization module based on the improved non-dominated sorting evolutionary algorithm is used to calculate the fitness of individuals in the population by calling the shielding performance evaluation capability output by the shielding structure parameterized modeling and electromagnetic simulation module, using the optimization objective and constraints in the standardized multi-objective optimization problem model output by the electromagnetic environment and performance index modeling module as the optimization criteria, and employing the improved non-dominated sorting evolutionary algorithm with dynamic constraint relaxation mechanism for iterative optimization. The module outputs the Pareto optimal shielding parameter solution set to guide the manufacturing of electromagnetic interference sensor harnesses.

[0094] The shielding parameter optimization system for electromagnetic interference-resistant sensor harnesses proposed in this invention consists of the following core modules: Electromagnetic Environment and Performance Index Modeling Module: Based on the application scenarios of sensor harnesses, establish a target electromagnetic interference spectrum model and a shielding performance index system.

[0095] Parametric Modeling and Electromagnetic Simulation Module for Shielding Structures: This module provides a parametric description of the shielding layer structure, employs a high-frequency approximate analytical method for rapid estimation, and utilizes the finite element method for high-precision verification, forming a multi-level simulation strategy.

[0096] The shielded parameter optimization module based on the improved non-dominated sorting evolutionary algorithm: The improved non-dominated sorting evolutionary algorithm is used to search for the Pareto optimal shielded parameter solution set. By introducing a dynamic constraint relaxation mechanism and a surrogate model acceleration strategy, the shortcomings of the standard algorithm in this type of problem, such as difficulty in convergence and excessive computational cost, are solved.

[0097] This embodiment is a software-based implementation system corresponding to the shielding parameter optimization method described in Embodiment 1, applied to the shielding parameter optimization design of pressure sensor harnesses. The system adopts a hierarchical collaborative architecture, consisting of three top-level core modules: an electromagnetic environment and performance index modeling module, a shielding structure parameter modeling and electromagnetic simulation module, and a shielding parameter optimization module based on an improved non-dominated sorting evolutionary algorithm. Data is sequentially transferred and linked in a closed loop between modules, realizing fully automated calculation from application scenario input to optimal shielding parameter output.

[0098] 1. Electromagnetic environment and performance index modeling module.

[0099] This module serves as the system's input and problem definition layer, used for standardized modeling of the optimization problem. The module receives the target application scenario of the sensor harness from the user, determines the spectral range and intensity distribution of the corresponding electromagnetic interference, constructs a multi-objective optimization index system including shielding effectiveness, cost, and flexibility, and sets manufacturing feasibility constraints such as minimum shielding efficiency, maximum cost, and maximum outer diameter. Finally, it outputs a standardized multi-objective optimization problem model containing the optimization objectives and constraints, serving as a unified criterion for subsequent shielding parameter optimization. In this embodiment, the module inputs the target scenario as a car engine compartment, determines the frequency band of interest as 30MHz~1GHz, sets the weight coefficients for each frequency point of interest according to the proportion of interference spectrum energy under typical operating conditions, with the 100MHz~300MHz strong interference band having the highest weight; constructs three types of optimization indicators: weighted comprehensive shielding effectiveness, material and manufacturing cost, and bending stiffness; sets constraints such as a minimum shielding efficiency of 55dB and a maximum cost of 200 yuan / meter for the 100MHz frequency point, and outputs a standardized multi-objective optimization problem model.

[0100] 2. Parametric modeling and electromagnetic simulation module for shielded structures.

[0101] This module serves as the core computational support layer of the system, providing iteratively invoked masked performance evaluation capabilities for the optimization process. Internally, it comprises five collaborative sub-modules, collectively forming a complete evaluation system of "parameter definition - multi-level simulation - deviation calibration - surrogate acceleration." The specific details of each sub-module are as follows: (1) Shielding structure parameterized modeling submodule.

[0102] This method is used to parametrically describe the shielding layer structure, defining a vector of shielding parameter design variables containing both continuous and discrete variables, deriving derived dependent variables such as braiding coverage, and outputting a parametric model in a unified format. In this embodiment, two types of continuous variables, namely braiding angle and shielding layer thickness, are parametrically defined, along with two types of discrete variables, namely the equivalent number of braids per strand and the number of shielding layers. The braiding coverage is derived from the design variables, and the coverage constraint range is set to 75%~95%.

[0103] (2) Analytical approximate simulation submodule.

[0104] This is a low-fidelity rapid evaluation unit that, based on plane wave shielding theory, quickly obtains an estimated shielding efficiency value through the superposition of reflection loss, absorption loss, and multiple reflection correction terms. In this embodiment, this submodule is used for rapid shielding efficiency estimation of individuals in large-scale populations during the main evolutionary cycle. A single evaluation takes less than 0.1 seconds, supporting rapid generation-by-generation evaluation of populations of hundreds in scale, ensuring computational efficiency for optimization iterations.

[0105] (3) Finite element simulation verification submodule To achieve high-fidelity and accurate verification of the unit, the frequency domain finite element method is used to verify the shielding performance with high precision. In this embodiment, the submodule constructs a two-dimensional simulation model including the wire harness cross-section and the surrounding air layer. A second-order absorbing boundary is set around the air layer to simulate an infinite space; the characteristic impedance of the differential port is set according to the automotive CAN bus standard; the thickness of the shielding layer ensures 2-3 layers of mesh elements, and the maximum element size does not exceed 1 / 10 of the wavelength corresponding to the highest frequency in the operating band, ensuring calculation accuracy. For the candidate optimal scheme selected through evolution, the submodule further uses a three-dimensional finite element model to calculate the scattering parameter S. 21 By SE=-20lg|S 21 We obtained high-precision shielding efficiency results and completed the final accurate verification of the candidate schemes.

[0106] (4) Multi-level simulation feedback calibration submodule.

[0107] This submodule is used to achieve accuracy coordination between low-fidelity and high-fidelity models. It selects representative parameter combinations covering the corners and center of the design space, calculates the shielding efficiency through both analytical approximation simulation and finite element simulation verification submodules, and constructs a deviation dataset. Kriging interpolation is used to fit the deviation prediction model, correcting the output of the analytical approximation model. During the optimization process, whenever a new candidate scheme passes finite element simulation verification, the result is automatically added to the deviation dataset, and the deviation prediction model is refitted, ensuring that the prediction accuracy of the low-fidelity model continuously improves with the optimization process.

[0108] (5) Partition Kriging Agent Acceleration Submodule.

[0109] To further reduce the computational overhead of the optimization process, this submodule partitions the design space according to discrete variable combinations of shielding layer type, material type, and number of shielding layers. For each major partition, a Kriging surrogate model based on the anisotropic Matérn-5 / 2 kernel function is constructed. During the evolutionary optimization process, the surrogate model is called at intervals to complete the overall population evaluation. At fixed intervals, several individuals with the highest current prediction uncertainty are selected, and precise calculations are performed using a calibrated analytical model. The results are then added to the training sample set to update the surrogate model, achieving a secondary improvement in computational efficiency.

[0110] 3. A masked parameter optimization module based on an improved non-dominated sorting evolutionary algorithm.

[0111] This module is the core optimization engine of the system. It uses the standardized multi-objective optimization problem model output by the electromagnetic environment and performance index modeling module as the optimization criterion, and utilizes the shielding performance evaluation capabilities provided by the shielding structure parameterized modeling and electromagnetic simulation module. It employs an improved non-dominated sorting evolutionary algorithm with a dynamic constraint relaxation mechanism to complete iterative optimization. The module contains five functional units: (1) Population initialization unit.

[0112] Latin hypercube sampling is used to uniformly generate the initial population in the parameter space. A hybrid encoding strategy is adopted for the shielding parameter design variable vector: continuous variables such as weaving angle and shielding layer thickness are encoded with real numbers, while discrete variables such as the number of filaments per strand and the number of shielding layers are encoded with integers. In this embodiment, the initial population size is set to 100.

[0113] (2) Fitness assessment unit.

[0114] The evaluation interface of the parametric modeling and electromagnetic simulation module for shielding structures is called to calculate the shielding effectiveness, cost, and flexibility of each individual in the population, and obtain the corresponding three-objective fitness function values.

[0115] (3) Improved constraint-Pareto sorting unit.

[0116] An improved constraint-Pareto dominance relationship with a three-level priority is used to perform non-dominated stratification of individuals in the population. At the same time, a dynamically shrinking constraint relaxation threshold is introduced: the constraint threshold is larger in the early stage of evolution, allowing for the exploration of more diverse parameter combinations to maintain population diversity and avoid getting trapped in local optima; the constraint threshold is gradually tightened as the number of generations increases, ensuring that the final output scheme meets all manufacturing feasibility constraints.

[0117] (4) Evolutionary operation unit.

[0118] Individual selection is accomplished using a binary tournament method; in the crossover operation, continuous variables undergo simulated binary crossover, while discrete variables undergo uniform discrete recombination; in the mutation operation, continuous variables undergo polynomial mutation, while discrete variables are randomly replaced from a preset set of feasible values ​​to generate the offspring population.

[0119] (5) Convergence judgment and result output unit.

[0120] Each generation merges the parent and offspring populations, and elites are retained through non-dominated sorting and crowding distance calculation, selecting the best individuals to enter the next generation. When the maximum number of generations is reached or the Pareto front hypervolume index continuously and stably meets the convergence threshold, evolution stops, and all individuals in the first non-dominated layer are output as the Pareto optimal shielding parameter solution set, which is then selected by engineering designers according to actual preferences to directly guide the manufacturing of electromagnetic interference sensor harnesses.

[0121] When this system is running, the electromagnetic environment and performance index modeling module first completes the definition of the optimization problem, then the shielding structure parameter modeling and electromagnetic simulation module builds a multi-level evaluation capability, and finally the shielding parameter optimization module completes the iterative solution. The three steps are progressive and the data is passed layer by layer to form a complete closed loop of shielding parameter simulation optimization.

[0122] Example 3: Simulation optimization of shielding parameters for automotive engine compartment pressure sensor wiring harness.

[0123] Step 1: Electromagnetic environment modeling.

[0124] A certain model of automobile has broadband electromagnetic interference in its engine compartment from devices such as the ignition system, alternator, and electric motor. The 30MHz band is of particular concern. 1 GHz band. Based on simulation data of a typical engine nacelle electromagnetic environment, the energy distribution of the interference spectrum was obtained, and the weighting coefficients for each frequency of interest were determined as shown in the table below: Table 1

[0125] in The frequency band with the most severe interference is assigned the highest weight of 0.30.

[0126] Step 2: Parameterization of the shielding structure and establishment of the simulation model.

[0127] The pressure sensor wiring harness needs to be highly flexible to accommodate the wiring requirements of the confined space inside the engine compartment. Cable outer diameter... mm, diameter of braided monofilament The pre-selected shielding type is braided shielding (tinned copper wire), and the grounding method is determined to be double-ended grounding according to the vehicle manufacturer's specifications.

[0128] The shielding parameters to be optimized include: braiding angle. Equivalent number of braided yarns per share (Using single-spindle or multi-spindle braiding process), number of shielding layers Shielding layer thickness .

[0129] Woven coverage The dependent variable is . To balance shielding efficiency and flexibility, this embodiment sets the coverage constraint range to . .

[0130] Minimum shielding efficiency constraint (100 MHz frequency requirement) Maximum cost constraint Yuan / meter, Yuan.

[0131] Step 3: Initial calibration of multi-level simulation.

[0132] Before optimization, 10 representative parameter combinations (covering the corners and center of the design space) were selected, and their shielding efficiency was calculated using both analytical approximation models and the two-dimensional frequency domain finite element method. The frequency range was solved using finite element simulation. The characteristic impedance of the port is set according to the automotive CAN bus standard. Based on the calculation results of the two methods, an initial deviation dataset and a deviation prediction model (using Kriging interpolation) are established.

[0133] Step 4: Multi-objective evolutionary optimization.

[0134] An improved non-dominated sorting evolutionary algorithm is used for optimization: population size Maximum number of generations Initial constraint relaxation value Attenuation control parameters Crossover probability Probability of mutation ( In this embodiment, to design the variable dimensions, Cross-distribution index Variation distribution index Target scaling factor .

[0135] Fitness evaluation employs a multi-stage combined strategy: the main loop uses a calibrated analytical approximation model for rapid estimation (single evaluation). (seconds); every 10 generations, the 5 individuals with the highest prediction uncertainty in the current population are selected, and finite element simulation is used for accurate verification and to update the bias prediction model; every 9 generations, a Kriging surrogate model is used for population evaluation to reduce the number of analytical model calls. After 500 generations of evolution, a Pareto optimal solution set containing multiple non-dominated solutions is obtained.

[0136] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0137] The above embodiments are merely illustrative examples and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for optimizing the shielding parameters of an electromagnetic interference-resistant sensor harness, characterized in that, Including the following: Step S1: Determine the electromagnetic interference spectrum range based on the target application scenario of the sensor harness, construct a multi-objective optimization index system including shielding effectiveness, cost, and flexibility, set manufacturing feasibility constraints, and output a standardized multi-objective optimization problem model containing optimization objectives and constraints. Step S2: Parametrically describe the shielding layer structure, construct a multi-level electromagnetic simulation framework that includes analytical approximate estimation and finite element accurate verification, establish a dynamic calibration mechanism for simulation deviation, and output a shielding performance evaluation capability that can be iteratively called. Step S3: Using the optimization objective and constraints in the standardized multi-objective optimization problem model output in step S1 as the optimization criteria, the fitness of individuals in the population is calculated by calling the shielding performance evaluation capability output in step S2. An improved non-dominated sorting evolutionary algorithm with dynamic constraint relaxation mechanism is used for iterative optimization, and the Pareto optimal shielding parameter solution set is output to guide the manufacturing of electromagnetic interference sensor harnesses.

2. The method for optimizing the shielding parameters of an anti-electromagnetic interference sensor harness according to claim 1, characterized in that, In step S1, the shielding effectiveness index in the multi-objective optimization index system is a weighted sum of the shielding efficiency of multiple frequencies of interest, and the weight coefficient of each frequency is determined according to the proportion of interference spectrum energy in the corresponding frequency band; the cost index includes the material cost and manufacturing cost of the shielding structure; the flexibility index is characterized by bending stiffness, and the optimization direction is to minimize bending stiffness to improve the flexibility of the wire harness.

3. The method for optimizing the shielding parameters of an anti-electromagnetic interference sensor harness according to claim 2, characterized in that, In step S1, within the standardized multi-objective optimization problem model, the objective function corresponding to the multi-objective optimization index system is expressed in standard minimization form as follows: ； In the formula, To mask the parameters, design a variable vector. To comprehensively assess shielding performance indicators, For cost indicators, As a flexibility indicator, , , This is the target scaling factor used for dimensional normalization.

4. The method for optimizing the shielding parameters of an anti-electromagnetic interference sensor harness according to claim 1, characterized in that, In step S2, the multi-level electromagnetic simulation framework adopts a three-level collaborative architecture: the first level is an analytical approximation model based on plane wave shielding theory, which is used to quickly estimate the shielding efficiency of large-scale population individuals during the evolutionary optimization process; The second level is the frequency domain finite element simulation model, which is used to verify the shielding performance of the candidate optimization schemes selected through evolution; The third level is a feedback calibration mechanism, which constructs a deviation prediction model based on the deviation data between the frequency domain finite element simulation model and the analytical approximation model, corrects the output results of the analytical approximation model, and dynamically updates the deviation prediction model as the optimization process progresses.

5. The method for optimizing the shielding parameters of an anti-electromagnetic interference sensor harness according to claim 1, characterized in that, In step S2, based on the parameterized description of the shielding structure and the multi-level electromagnetic simulation framework, and considering the mixed characteristics of continuous and discrete variables in the shielding design variables, a partitioned Kriging surrogate model is adopted to accelerate the simulation calculation: the design space is partitioned according to the discrete variable combination of shielding layer type, material type, and number of shielding layers, and a Kriging surrogate model is constructed for each partition; during the evolutionary optimization process, the Kriging surrogate model is called at intervals to evaluate the performance of individuals in the population, and the individuals with the highest prediction uncertainty are periodically selected for verification through the analytical approximation model, and the verification results are added to the sample set to update the Kriging surrogate model.

6. The method for optimizing the shielding parameters of an anti-electromagnetic interference sensor harness according to claim 1, characterized in that, In step S3, the improved non-dominated sorting evolutionary algorithm uses an improved constraint-Pareto dominance relationship to perform hierarchical sorting of individuals. The judgment of the dominance relationship is performed according to three levels of priority: the first priority distinguishes individuals inside and outside the constraint boundary, and individuals that satisfy the constraint boundary are given priority; the second priority performs standard Pareto dominance judgment on individuals inside the constraint boundary; the third priority sorts individuals outside the constraint boundary according to the degree of constraint violation, and individuals with smaller constraint violation are given priority.

7. The method for optimizing the shielding parameters of an anti-electromagnetic interference sensor harness according to claim 6, characterized in that, The constraint violation degree is a dimensionless scalar, calculated using the following formula: ； In the formula, It is the constraint violation degree, a dimensionless scalar of the output, representing the design variable vector. The corresponding shielding parameter scheme violates the total degree of all manufacturing feasibility constraints. The total number of terms used to create feasibility constraints is the upper limit of the summation operation; The index variable for the constraint term has a range of values. =1, 2, ..., This is used to iterate through each manufacturing feasibility constraint function; For the first A constraint function whose value is less than or equal to 0 indicates that the corresponding constraint is satisfied; For the first The reference deviation of the constraint is used to normalize the constraint violation values ​​of different physical dimensions.

8. The method for optimizing the shielding parameters of an anti-electromagnetic interference sensor harness according to claim 6, characterized in that, In the improved constraint-Pareto dominance relationship, the threshold used to distinguish between the inside and outside of the constraint boundary is the constraint relaxation threshold for dynamic contraction. The constraint relaxation threshold corresponding to the current generation is calculated by the following formula: ； In the formula, This represents the constraint relaxation threshold corresponding to the current generation. This represents the initial constraint relaxation value; The current generation number; The maximum number of generations; The constraint relaxation threshold is set as the decay control parameter. In the early stages of evolution, the constraint relaxation threshold is relatively large to preserve population diversity and avoid getting trapped in local optima. As the number of generations increases, the constraint relaxation threshold is gradually tightened to ensure that the output Pareto optimal solution satisfies all manufacturing feasibility constraints.

9. The method for optimizing the shielding parameters of an anti-electromagnetic interference sensor harness according to claim 1, characterized in that, In step S3, the improved non-dominated sorting evolutionary algorithm with dynamic constraint relaxation mechanism uses hybrid encoding and evolutionary operation rules for the masking parameter design variable vectors corresponding to individuals in the population: The coding rules are as follows: real numbers are used for two types of continuous design variables, namely braiding angle and shielding layer thickness; integers are used for four types of discrete design variables, namely, the equivalent number of braids per braid, the number of shielding layers, the shielding layer type, and the material type. The crossover operation rules are as follows: simulated binary crossover is performed on the coding segments corresponding to continuous design variables, and uniform discrete recombination is performed on the coding segments corresponding to discrete design variables. The mutation operation rules are as follows: perform polynomial mutation on the coding segment corresponding to the continuous design variable, and randomly select and replace the coding segment corresponding to the discrete design variable from the preset set of feasible discrete values.

10. A shielding parameter optimization system for an anti-electromagnetic interference sensor harness, used to implement the method described in any one of claims 1-9, characterized in that, The system includes: The electromagnetic environment and performance index modeling module is used to determine the electromagnetic interference spectrum range based on the target application scenario of the sensor harness, construct a multi-objective optimization index system including shielding effectiveness, cost, and flexibility, set manufacturing feasibility constraints, and output a standardized multi-objective optimization problem model containing optimization objectives and constraints. The shielding structure parametric modeling and electromagnetic simulation module is used to parametrically describe the shielding layer structure, construct a multi-level electromagnetic simulation framework that includes analytical approximate estimation and finite element accurate verification, establish a dynamic calibration mechanism for simulation deviation, and output a shielding performance evaluation capability that can be iteratively called. The shielding parameter optimization module based on the improved non-dominated sorting evolutionary algorithm is used to calculate the fitness of individuals in the population by calling the shielding performance evaluation capability output by the shielding structure parameterized modeling and electromagnetic simulation module, using the optimization objective and constraints in the standardized multi-objective optimization problem model output by the electromagnetic environment and performance index modeling module as the optimization criteria, and employing the improved non-dominated sorting evolutionary algorithm with dynamic constraint relaxation mechanism for iterative optimization. The module outputs the Pareto optimal shielding parameter solution set to guide the manufacturing of electromagnetic interference sensor harnesses.

Images