Pattern noise dislocation optimization method based on genetic algorithm

Abstract

The invention relates to the technical field of vehicle tire data analysis, in particular to a pattern noise dislocation optimization method based on a genetic algorithm, and the method comprises the steps: obtaining pattern geometric data through three-dimensional scanning, fitting a grounding mark to generate a nonlinear leading edge shape, generating a time domain signal in combination with a pressure distribution curve, and obtaining a noise spectrum characteristic through Fourier transform; and carrying out dislocation displacement optimization on the pattern RIB unit by utilizing a genetic algorithm, evaluating a noise result through a fitness function, and iteratively searching an optimal scheme. According to the invention, noise prediction accuracy and optimization efficiency are effectively improved, and systematic control of pattern noise is realized.

Description

Technical Field

[0001] This invention relates to the field of tire data analysis technology, and in particular to a method for optimizing tread noise misalignment based on a genetic algorithm. Background Technology

[0002] Existing tire tread noise prediction and optimization technologies are primarily based on a deep understanding of the aerodynamic and structural dynamic mechanisms generated when tread blocks contact the road surface during tire rolling. Noise mainly originates from the rapid compression and release of air within the tread grooves, i.e., the air pumping effect, and the vibrations caused by the tread blocks impacting the ground. Prediction techniques typically employ computational fluid dynamics and acoustic simulation methods to establish a refined digital model of the tire tread pattern, simulating its flow field characteristics and sound field distribution under rolling conditions, thereby quantitatively evaluating the noise levels of different tread designs. Optimization strategies focus on systematically adjusting the tread pitch arrangement, groove geometry, and depth distribution.

[0003] Existing tire tread noise prediction and optimization technologies suffer from the following technical challenges: Firstly, prediction methods fail to adequately consider the nonlinear characteristics of the contact leading edge shape and the non-uniformity of contact pressure distribution in the initial tire tread design phase. This leads to oversimplification of the model, making it unable to accurately simulate real rolling contact excitation and resulting in significant discrepancies between predicted and measured sound pressure levels. Secondly, optimization algorithms are limited to random combinations of pitch arrangements, lacking global search capabilities. This makes the optimization process inefficient and unable to systematically address the impact of tread block misalignment on pump noise. For instance, when engineers apply existing technologies in computer-aided design environments, noise simulations based on the assumption of an approximately straight leading edge and uniform pressure often result in excessive noise levels in actual tire testing. This forces repeated iterations of the design process, extending development time and increasing resource consumption. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a pattern noise misalignment optimization method based on genetic algorithms. This method solves the technical problems of low prediction accuracy and incomplete optimization process caused by existing pattern noise prediction methods not fully considering the shape and pressure distribution of the grounding leading edge and the low efficiency of optimization algorithms.

[0005] To solve the above-mentioned technical problems, the specific contents of the present invention are as follows: This invention provides a pattern noise misalignment optimization method based on a genetic algorithm, comprising: Step 1: Collect the initial design data of the tire tread pattern, obtain the nonlinear leading edge shape by fitting the ground imprint, and use it as a scan line to perform discrete sampling on the tread pattern drawing. Generate a binary data point array with a specified step size, where the intersection of the tread grooves is assigned a value of 0 and the intersection of the tread blocks is assigned a value of 1. Step 2: Using the binary data point array as geometric feature input, obtain the ground pressure distribution curve of the tire, establish the spatial coordinate mapping relationship between the pressure distribution curve and the binary data point array, and multiply the pressure value of the ground pressure distribution curve with the corresponding position value of the binary data point array to generate a time domain signal representing the tread rolling over the ground. Step 3: Use the time-domain signal as the frequency-domain analysis input, perform Fourier transform to convert it into a frequency-domain response, and calculate the sound pressure level in combination with the tire speed parameters to obtain the spectral characteristics of the tread noise. Step 4: The binary data point array is used as a segmentation object and divided into multiple patterned RIB units according to the pattern design ratio, and saved as a structure. Step 5: The segmented pattern RIB units are used as the objects of genetic algorithm processing. Using genetic algorithm, an initial population is randomly generated. Individuals in the population represent the displacement in decimal values. The displacement is assigned to the pattern RIB units. The displacement pattern layout is generated through coordinate transformation, and a spatial displacement optimization scheme is generated. Step 6: For each of the spatial misalignment optimization schemes, reconstruct the misaligned binary data point array, multiply the reconstructed binary data point array with the ground pressure distribution curve to generate a new time-domain signal, perform Fourier transform on the new time-domain signal, and calculate the noise result for each scheme. Step 7: Using the noise results as the basis for fitness calculation, define the fitness function, with the total noise value as the optimization objective, calculate the fitness value of each scheme and sort them. Based on the fitness value sorting results, perform genetic operations, including selection, crossover and mutation, to generate a new generation population. Repeat the process of population generation, noise calculation, fitness evaluation and genetic operations until the termination condition is met. Decode the optimal individual into the actual displacement, output the optimal displacement scheme, and apply it to the initial pattern design.

[0006] Furthermore, in the tread noise error optimization method based on genetic algorithm described in this invention, the initial design data for collecting tire tread patterns includes: Geometric point cloud data of tire tread patterns are obtained using a 3D scanner; Preprocess the geometric point cloud data to remove outliers and noisy data to obtain clean point cloud data; The least squares method is used to fit a surface to the clean point cloud data to generate a continuous leading edge curve. The leading edge curve is discretized into a sequence of equally spaced sampling points, which serve as the scanning path; The pattern drawing is rasterized along the scanning path, and values ​​are assigned according to the spatial relationship between the pattern blocks and the grooves to generate a binary data point array.

[0007] Furthermore, in the tread noise misalignment optimization method based on genetic algorithm described in this invention, obtaining the tire ground pressure distribution curve includes: A pressure sensor array is installed on a tire testing bench to collect the ground pressure distribution of the tire under standard load. The ground pressure distribution data is filtered to eliminate sensor noise and obtain filtered pressure data. The filtered pressure data is interpolated into a continuous pressure distribution surface using the Kriging interpolation algorithm. The pressure distribution surface is resampled into a pressure matrix with the same resolution as the binary data point array; Perform element-wise multiplication of the binary data point array and the pressure matrix to generate a time-domain signal sequence.

[0008] Furthermore, in the pattern noise misalignment optimization method based on genetic algorithm described in this invention, the Fourier transform process includes: The time-domain signal is windowed using the Hanning window function to obtain the windowed time-domain signal. The windowed time-domain signal is processed using the Fast Fourier Transform algorithm and converted into a frequency-domain amplitude spectrum. Based on the ratio of tire circumference to driving speed, the spatial frequency of the frequency domain amplitude spectrum is mapped to the acoustic frequency. The sound pressure level at acoustic frequencies was calculated using a point source radiation model. The sound pressure level data is smoothed to generate a noise spectrum distribution map.

[0009] Furthermore, in the pattern noise misalignment optimization method based on genetic algorithm described in this invention, the pattern segmentation RIB unit includes: Identify the periodic features of patterns in a binary data point array and determine the pitch boundaries; The binary data point array is divided into multiple independent units based on the pitch boundary and the number of circumferential pitches and transverse RIBs of the pattern. Assign a unique identifier to each individual unit and record its location coordinates and geometric dimensions; A hash table data structure is used to store unit information, and an indexing mechanism is established. Verify the mechanical independence of independent elements and confirm the boundary continuity.

[0010] Furthermore, the pattern noise misalignment optimization method based on genetic algorithm described in this invention includes the following spatial misalignment optimization scheme: Set the population size to 100-500 individuals and input the population size parameter into the population initialization process; The initial population was generated using the Latin hypercube sampling method, and the generated initial population was used as the object for displacement assignment. Define the feasible region of the displacement and use the feasible region as a constraint condition for assigning displacement values; The displacement vector is applied to the corresponding RIB element through an affine transformation algorithm, and the coordinates of the transformed RIB element are used as input for geometric rationality verification. Verify the geometric rationality of the pattern after misalignment, and output the verified scheme as the spatial misalignment optimization scheme.

[0011] Furthermore, in the pattern noise misalignment optimization method based on genetic algorithm described in this invention, the calculation of the fitness value includes: Extract the noise spectrum data for each scheme and use the noise spectrum data as input for sound pressure level calculation; Calculate the total sound pressure level in the 1-5000Hz frequency band and use the total sound pressure level value as the input for normalization. The total sound pressure level is normalized, and the normalization result is used as the input for fitness value conversion. The sound pressure level is converted into a fitness value using a negative exponential function, and the converted fitness value is used as the input for the sorting process. The fitness values ​​of all solutions are sorted in descending order using the heap sort algorithm, and the sorting result is used as the input for the genetic operation.

[0012] Furthermore, in the pattern noise misalignment optimization method based on genetic algorithm described in this invention, the genetic operation includes: An elite retention strategy is adopted, and the fitness ranking results are used as the basis for selecting elite individuals; The remaining individuals are selected through a tournament, and the selection results are used as the parent individuals for the crossover operation; The simulated binary crossover algorithm is used, with the genes of the parent individuals as the input for the crossover operation; The polynomial mutation operator is applied, and the offspring individuals after crossover are used as the input for the mutation operation; The crossover and mutation probabilities are dynamically adjusted, and population diversity indicators are used as the basis for parameter adjustment.

[0013] Furthermore, in the pattern noise misalignment optimization method based on genetic algorithm described in this invention, the termination condition judgment includes: Monitor the changing trend of the optimal fitness value and use the fitness value sequence as input for convergence judgment; Calculate the standard deviation of fitness values ​​over 10 consecutive generations, and use the standard deviation as the comparison object for the convergence threshold; Set the maximum number of iterations to 200 generations, and compare the current number of iterations with the maximum number of iterations. Record the average fitness and optimal fitness of each generation of the population, and use the recorded data as the basis for generating the convergence curve; When the convergence condition is met or the maximum number of iterations is reached, the current best individual is used as the output result of the genetic algorithm.

[0014] Furthermore, the pattern noise misalignment optimization method based on genetic algorithm described in this invention, wherein the output optimal misalignment scheme includes: Decode the genetic code of the optimal individual and use the gene sequence as input for the analysis of the fault displacement; The displacement values ​​are converted into actual size units, and the analyzed displacement values ​​are used as input for geometric transformation. Transformation parameters are applied in the CAD system and used as the basis for modifying pattern drawings; The grounding pressure distribution of the optimized pattern was verified through finite element analysis, and the verification results were used as the basis for evaluating the performance of the scheme. Output the final design document, and output the verified solution as the optimization result.

[0015] Beneficial effects of this invention; This invention obtains the nonlinear leading edge shape by fitting the grounding imprint and establishes a spatial coordinate mapping relationship between the grounding pressure distribution curve and the binary data point array to generate a real time-domain signal. This effectively overcomes the prediction bias caused by the simplification of the geometric model and the neglect of pressure distribution in the prior art. At the same time, a genetic algorithm is used to optimize the displacement of the RIB unit of the tread pattern. The initial population is generated by Latin hypercube sampling, the noise results are evaluated by the fitness function, and the crossover and mutation operation is selected to achieve a global and efficient search, which significantly improves the optimization efficiency. Thus, high-precision noise prediction and rapid optimization are achieved in the early stage of tire tread pattern design, solving the technical problems of inaccurate prediction and long time consumption in the prior art. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of discrete sampling of the pattern drawing of the present invention; Figure 2 This is a schematic diagram of time-domain signal generation according to the present invention; Figure 3 This is a schematic diagram of the frequency domain response transformation of the present invention; Figure 4 This is a schematic diagram of the sound pressure level spectrum distribution of the present invention; Figure 5 This is a schematic diagram illustrating the calculation of noise sound pressure level in this invention; Figure 6 This is a schematic diagram of pattern RIB data segmentation according to the present invention; Figure 7 This is a schematic diagram of the patterns acquired by scan lines of different radii according to the present invention; Figure 8 This is a schematic diagram comparing the optimization effects of the present invention. Figure 9 This diagram shows the effect of implementing the optimized solution of the present invention. Detailed Implementation

[0018] To make the technical solution of the present invention clearer, the present invention will be clearly and completely described below with reference to specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The present invention provided by various embodiments will be described in detail below with reference to the accompanying drawings. To better understand the purpose of the present invention, the present invention will be described in further detail below.

[0019] Please see Figures 1 to 9 This invention provides a pattern noise misalignment optimization method based on a genetic algorithm, comprising: Step 1: Collect the initial design data of the tire tread pattern, obtain the nonlinear leading edge shape by fitting the ground imprint, and use it as a scan line to perform discrete sampling on the tread pattern drawing. Generate a binary data point array with a specified step size, where the intersection of the tread grooves is assigned a value of 0 and the intersection of the tread blocks is assigned a value of 1. Step 2: Using the binary data point array as geometric feature input, obtain the ground pressure distribution curve of the tire, establish the spatial coordinate mapping relationship between the pressure distribution curve and the binary data point array, and multiply the pressure value of the ground pressure distribution curve with the corresponding position value of the binary data point array to generate a time domain signal representing the tread rolling over the ground. Step 3: Use the time-domain signal as the frequency-domain analysis input, perform Fourier transform to convert it into a frequency-domain response, and calculate the sound pressure level in combination with the tire speed parameters to obtain the spectral characteristics of the tread noise. Step 4: The binary data point array is used as a segmentation object and divided into multiple patterned RIB units according to the pattern design ratio, and saved as a structure. Step 5: The segmented pattern RIB units are used as the objects of genetic algorithm processing. Using genetic algorithm, an initial population is randomly generated. Individuals in the population represent the displacement in decimal values. The displacement is assigned to the pattern RIB units. The displacement pattern layout is generated through coordinate transformation, and a spatial displacement optimization scheme is generated. Step 6: For each of the spatial misalignment optimization schemes, reconstruct the misaligned binary data point array, multiply the reconstructed binary data point array with the ground pressure distribution curve to generate a new time-domain signal, perform Fourier transform on the new time-domain signal, and calculate the noise result for each scheme. Step 7: Using the noise results as the basis for fitness calculation, define the fitness function, with the total noise value as the optimization objective, calculate the fitness value of each scheme and sort them. Based on the fitness value sorting results, perform genetic operations, including selection, crossover and mutation, to generate a new generation population. Repeat the process of population generation, noise calculation, fitness evaluation and genetic operations until the termination condition is met. Decode the optimal individual into the actual displacement, output the optimal displacement scheme, and apply it to the initial pattern design.

[0020] This invention provides a tread noise misalignment optimization method based on a genetic algorithm, aiming to reduce tread noise by systematically processing tire tread design data and combining acoustic simulation and intelligent optimization algorithms. The core of the method lies in using a genetic algorithm to perform a global search for the misalignment arrangement of tread blocks, thereby optimizing noise performance. The following detailed explanation of each step of the method clarifies the overall process and logical connections of the technical solution.

[0021] In the initial data acquisition phase, geometric point cloud data of the tire tread pattern was acquired using a 3D scanner. This data was preprocessed to remove outliers and noise, resulting in clean point cloud data. The clean point cloud data was then fitted to a surface using the least squares method to generate a continuous leading edge curve. This leading edge curve was discretized into a sequence of equally spaced sampling points, serving as the scanning path. The tread pattern was rasterized along the scanning path, and values ​​were assigned based on the spatial relationship between the tread blocks and grooves, generating a binary data point array. The intersections of the tread and grooves were assigned a value of 0, while the intersections of the tread blocks were assigned a value of 1. This process accurately captures the nonlinear shape of the grounding leading edge, providing a geometric basis for subsequent analysis.

[0022] A binary data point array is used as the geometric feature input to obtain the tire's ground pressure distribution curve. A pressure sensor array is installed on a tire test bench to collect the ground pressure distribution of the tire under standard load. The ground pressure distribution data is filtered to eliminate sensor noise, resulting in filtered pressure data. The filtered pressure data is then interpolated into a continuous pressure distribution surface using the Kriging interpolation algorithm. This pressure distribution surface is resampled into a pressure matrix with the same resolution as the binary data point array. A spatial coordinate mapping relationship between the pressure distribution curve and the binary data point array is established. The pressure values ​​of the ground pressure distribution curve are multiplied by the corresponding values ​​in the binary data point array to generate a time-domain signal sequence representing the tread pattern rolling over the ground. This step incorporates the actual pressure distribution, improving the realism of noise prediction.

[0023] The time-domain signal is used as the input for frequency-domain analysis, and a Fourier transform is performed to convert it into a frequency-domain response. The time-domain signal is then windowed using a Hanning window function to obtain a windowed time-domain signal. A Fast Fourier Transform (FFT) algorithm is used to process the windowed time-domain signal, converting it into a frequency-domain amplitude spectrum. Based on the ratio of tire circumference to driving speed, the spatial frequencies of the frequency-domain amplitude spectrum are mapped to acoustic frequencies. A point source radiation model is used to calculate the sound pressure level at the acoustic frequencies. The sound pressure level data is then smoothed to generate a noise spectrum distribution map. This yields the spectral characteristics of the tire tread noise, providing a quantitative basis for optimization objectives.

[0024] The binary data point array is used as the segmentation object and divided into multiple patterned RIB units according to the pattern design ratio. The periodic characteristics of the pattern in the binary data point array are identified to determine the pitch boundaries. Based on the pitch boundaries and the number of circumferential pitches and lateral RIBs in the pattern, the binary data point array is divided into multiple independent units. A unique identifier is assigned to each independent unit, recording its position coordinates and geometric dimensions. A hash table data structure is used to store the unit information, an indexing mechanism is established, the mechanical independence of the independent units is verified, boundary continuity is confirmed, and the data is saved as a structure. This step provides modular input for the genetic algorithm processing.

[0025] The segmented pattern RIB units are used as the processing objects of a genetic algorithm. An initial population is randomly generated using the genetic algorithm, with each individual representing the displacement in decimal. The population size is set within a certain range, and the initial population is generated using Latin hypercube sampling. The feasible region of the displacement is defined as a constraint for displacement assignment. The displacement is assigned to the pattern RIB units, and the displacement vector is applied to the corresponding RIB units using an affine transformation algorithm. Coordinate transformation is used to generate the displaced pattern layout, verifying the geometric rationality of the displaced pattern and generating a spatial displacement optimization scheme. The introduction of the genetic algorithm enables efficient global search.

[0026] For each spatial misalignment optimization scheme, a new binary data point array is reconstructed. The reconstructed binary data point array is multiplied by the ground pressure distribution curve to generate a new time-domain signal. A Fourier transform is performed on the new time-domain signal to calculate the noise result for each scheme. This step simulates the acoustic performance of the pattern after misalignment, providing data for fitness evaluation.

[0027] Noise results are used as the basis for fitness calculation, and a fitness function is defined with the total noise value as the optimization objective. Noise spectrum data for each scheme is extracted, the total sound pressure level within the frequency band is calculated, the total sound pressure level is normalized, and a negative exponential function is used to convert the sound pressure level into a fitness value. A heap sort algorithm is used to sort the fitness values ​​of all schemes in descending order. Based on the fitness value sorting results, genetic operations, including selection, crossover, and mutation, are performed. An elite retention strategy is adopted, and tournament selection is performed on the remaining individuals. A simulated binary crossover algorithm is used, and a polynomial mutation operator is applied to dynamically adjust the crossover and mutation probabilities. A new generation of population is generated, and the process of population generation, noise calculation, fitness evaluation, and genetic operations is repeated until the termination condition is met. The changing trend of the optimal fitness value is monitored, the standard deviation of the fitness values ​​of consecutive generations is calculated, and a maximum number of iterations is set. When the convergence condition is met or the maximum number of iterations is reached, the optimal individual is decoded into the actual displacement, the optimal displacement scheme is output, and it is applied to the initial pattern design. The final scheme is verified by finite element analysis to verify the grounding pressure distribution and ensure the optimization effect.

[0028] The process of acquiring initial design data for tire tread patterns involves the precise acquisition and processing of geometric data. After capturing the geometric point cloud data of the tire tread pattern using a 3D scanner, the point cloud data is preprocessed to eliminate outliers and noise interference, resulting in clean point cloud data. Least squares method is used for surface fitting to generate a continuous leading-edge curve, ensuring the accuracy of the geometric shape. The leading-edge curve is discretized into a sequence of equally spaced sampling points as the scanning path. The tread pattern drawing is rasterized along this path, and values ​​are assigned based on the spatial relationship between the tread blocks and grooves, generating a binary data point array. These steps are sequentially linked; preprocessing ensures data quality, surface fitting restores nonlinear characteristics, discretization achieves digital sampling, and the final binary array provides structured input for subsequent analysis.

[0029] The steps for obtaining the tire ground pressure distribution curve are based on experiments and data processing. A pressure sensor array is installed on a tire test bench to collect the ground pressure distribution under standard load. The ground pressure distribution data is filtered to eliminate sensor noise, resulting in filtered pressure data. Kriging interpolation is used to interpolate the filtered data into a continuous pressure distribution surface, improving spatial continuity. The pressure distribution surface is resampled into a pressure matrix with the same resolution as the binary data point array. Element-by-element multiplication of the binary data point array and the pressure matrix is ​​performed to generate a time-domain signal sequence. Filtering ensures data reliability, interpolation enhances surface smoothness, resampling achieves coordinate alignment, and multiplication integrates geometric and pressure information to generate an excitation signal.

[0030] The Fourier transform process converts a time-domain signal into a frequency-domain response. The time-domain signal is windowed using a Hanning window function to reduce spectral leakage, resulting in a windowed time-domain signal. The windowed signal is then processed using a Fast Fourier Transform (FFT) algorithm, converting it into a frequency-domain amplitude spectrum. Based on the ratio of tire circumference to driving speed, spatial frequencies are mapped to acoustic frequencies. A point source radiation model is used to calculate the sound pressure level at the acoustic frequencies. The sound pressure level data is then smoothed to generate a noise spectrum distribution map. Windowing optimizes spectral accuracy, Fourier transform enables frequency domain conversion, frequency mapping correlates with physical parameters, sound pressure calculation quantifies noise levels, and smoothing improves spectral readability.

[0031] The process of segmenting patterned RIB units enables modular processing of pattern data. It involves identifying the periodic characteristics of the pattern in the binary data point array to determine the pitch boundaries; dividing the array into multiple independent units based on the pitch boundaries, the number of circumferential pitches, and the number of transverse RIBs; assigning a unique identifier to each independent unit and recording its position coordinates and geometric dimensions; using a hash table data structure to store unit information and establishing an indexing mechanism; and verifying the mechanical independence of the independent units to confirm boundary continuity. Periodic identification ensures segmentation accuracy, unit division supports modular operations, identifiers facilitate management, hash tables optimize storage efficiency, and the verification step guarantees unit independence.

[0032] The steps for generating spatial displacement optimization schemes utilize a genetic algorithm for displacement optimization. The population size is set to a certain range of individuals, and this population size parameter is input into the initialization process. An initial population is generated using the Latin hypercube sampling method, serving as the object for displacement assignment. The feasible region of the displacement is defined as a constraint condition for displacement assignment. The displacement vector is applied to the corresponding RIB element through an affine transformation algorithm, and the transformed coordinates serve as input for geometric rationality verification. The geometric rationality of the pattern after displacement is verified, and a satisfactory scheme is output. Population settings balance search efficiency, sampling generates diverse initial solutions, feasible region constraints ensure physical feasibility, affine transformation implements displacement application, and verification steps select a reasonable layout.

[0033] The fitness calculation step evaluates the merits of different schemes based on noise results. Noise spectrum data for each scheme is extracted and used as input for sound pressure level (SPL) calculation. The total SPL within a specific frequency band is calculated, and this value serves as input for normalization. The total SPL is normalized, and the result is used as input for fitness value conversion. A negative exponential function is used to convert the SPL into fitness values, and these converted values ​​are used as input for ranking. A heap sort algorithm is used to sort the fitness values ​​of all schemes in descending order, and the ranking result serves as input for the genetic operation. Spectrum extraction provides the raw data, sound pressure level calculation quantifies noise, normalization enables standard comparison, function transformation maps fitness, and ranking supports selection operations.

[0034] Genetic operations are performed to achieve population evolution and optimization. An elite retention strategy is employed, using fitness ranking results as the basis for selecting elite individuals. A tournament selection process is used for the remaining individuals, with the results serving as the parents for crossover. A simulated binary crossover algorithm is used, with the genes of the parents as input for the crossover operation. A polynomial mutation operator is applied, with the offspring individuals after crossover as input for the mutation operation. Crossover and mutation probabilities are dynamically adjusted, using population diversity indicators as the basis for parameter adjustments. Elite retention preserves superior genes, tournament selection promotes competitive selection, the crossover algorithm generates new combinations, mutation introduces diversity, and the search process is dynamically adjusted and optimized.

[0035] The termination condition judgment step monitors the convergence of the optimization process. It monitors the trend of the best fitness value, using the fitness value sequence as input for convergence judgment; calculates the standard deviation of fitness values ​​over multiple generations, using this standard deviation as a comparison object for the convergence threshold; sets a maximum number of iterations, comparing the current iteration number with the maximum; records the average fitness and best fitness of each generation, using this data as the basis for generating the convergence curve; and outputs the current best individual when the convergence condition is met or the maximum number of iterations is reached. Trend monitoring identifies stable states, standard deviation quantifies the degree of convergence, iteration limits prevent infinite loops, recording supports analysis, and termination conditions ensure efficiency.

[0036] The optimal displacement scheme is output through a series of steps, applying the optimization results to the actual design. The genetic code of the optimal individual is decoded, and the gene sequence is used as input for displacement analysis. The displacement values ​​are converted to actual unit dimensions, and the analyzed values ​​are used as input for geometric transformation. Transformation parameters are applied in the CAD system, serving as the basis for modifying the pattern drawings. Finite element analysis is used to verify the ground pressure distribution of the optimized pattern, and the verification results serve as the basis for scheme performance evaluation. The final design document is output, with the verified scheme considered the optimization result. Decoding enables displacement extraction, unit conversion ensures design compatibility, CAD integration supports drawing updates, finite element analysis verifies performance, and the output completes the optimization process.

[0037] This invention systematically solves key technical problems in pattern noise prediction and optimization by integrating high-precision geometric modeling, realistic pressure coupling, and intelligent optimization algorithms. In the prediction stage, a method of fitting ground imprints is used to obtain the nonlinear leading edge shape, replacing the simplified approximation of straight lines or arcs in existing technologies, accurately restoring the actual geometric contour of tire ground contact. Simultaneously, by acquiring the ground pressure distribution curve, a spatial coordinate mapping relationship with the binary data point array is established. The pressure value is multiplied point-by-point with the geometric data to generate a time-domain signal, thereby introducing the influence of pressure distribution non-uniformity on noise excitation and improving the physical realism of the prediction model. In the optimization stage, a genetic algorithm is applied to replace the existing random combination strategy. An initial population is generated through Latin hypercube sampling, using displacement as the optimization variable. Affine transformation is combined to adjust the displacement of the pattern RIB units, and the fitness function is used to evaluate the merits of the scheme based on the total noise value. Global efficient search is achieved through selection, crossover, and mutation operations. During the iteration process, noise calculation and genetic operations are repeatedly executed until the convergence condition is met, avoiding local optima and significantly shortening the optimization cycle. The present invention uses a data flow coherent processing approach, from geometric feature acquisition, pressure integration, frequency domain analysis to genetic optimization, to form a closed-loop feedback mechanism, ensuring a synergistic improvement in prediction accuracy and optimization efficiency.

[0038] This invention addresses the technical problems of existing tire tread noise prediction methods, which fail to adequately consider the nonlinear characteristics of the ground contact front shape and the uneven distribution of ground pressure, as well as the low efficiency of optimization algorithms leading to prediction errors and time-consuming optimization processes. It provides a specific implementation of a tread noise misalignment optimization method based on a genetic algorithm. The following description, presented in paragraph form, incorporates application scenarios from the background art and supports the interpretation of the claims through subordinate technical solutions.

[0039] In the initial stage of tire tread design, geometric point cloud data of the tire tread is acquired using a 3D scanner. This data is preprocessed to remove outliers and noise, resulting in clean point cloud data. The clean point cloud data is then fitted to a surface using the least squares method to generate a continuous leading edge curve. This leading edge curve is discretized into a sequence of equally spaced sampling points, serving as the scanning path. The tread pattern is then rasterized along this scanning path, and values ​​are assigned based on the spatial relationship between the tread blocks and grooves, generating a binary data point array. The intersections of the tread grooves are assigned a value of 0, while the intersections of the tread blocks are assigned a value of 1. This process accurately recreates the nonlinear shape of the ground contact leading edge, overcoming the simplification bias of straight line or circular arc approximations in existing technologies.

[0040] When acquiring the tire's ground pressure distribution curve, a pressure sensor array is installed on a tire test bench to collect the ground pressure distribution of the tire under standard load. The ground pressure distribution data is filtered to eliminate sensor noise, resulting in filtered pressure data. The filtered pressure data is then interpolated into a continuous pressure distribution surface using the Kriging interpolation algorithm. This surface is resampled into a pressure matrix with the same resolution as the binary data point array. A spatial coordinate mapping relationship is established between the pressure distribution curve and the binary data point array. The pressure values ​​on the ground pressure distribution curve are multiplied by the corresponding values ​​in the binary data point array to generate a time-domain signal sequence representing the tread pattern rolling over the ground. This step introduces the non-uniformity of the actual pressure distribution, improving the realism of the noise-excited simulation.

[0041] The time-domain signal is used as input for frequency-domain analysis, and a Fourier transform is performed to convert it into a frequency-domain response. The time-domain signal is then windowed using a Hanning window function to reduce spectral leakage, resulting in a windowed time-domain signal. A Fast Fourier Transform (FFT) algorithm is used to process the windowed time-domain signal, converting it into a frequency-domain amplitude spectrum. Based on the ratio of tire circumference to driving speed, the spatial frequencies of the frequency-domain amplitude spectrum are mapped to acoustic frequencies. A point source radiation model is used to calculate the sound pressure level at the acoustic frequencies. The sound pressure level data is smoothed to generate a noise spectrum distribution map, thereby obtaining the spectral characteristics of the tire pattern noise and providing a quantitative basis for optimization objectives.

[0042] The binary data point array is used as the segmentation object, divided into multiple pattern RIB units according to the pattern design ratio. The periodic characteristics of the pattern in the binary data point array are identified to determine the pitch boundaries. Based on the pitch boundaries and the number of circumferential pitches and lateral RIBs in the pattern, the binary data point array is divided into multiple independent units. A unique identifier is assigned to each independent unit, recording its position coordinates and geometric dimensions. A hash table data structure is used to store the unit information, establishing an indexing mechanism. The mechanical independence of the independent units is verified, boundary continuity is confirmed, and the data is saved as a structure. This step achieves modular processing of the pattern data, providing structured input for the genetic algorithm.

[0043] When optimizing the segmented patterned RIB units using a genetic algorithm, a population size of a certain number of individuals is set, and this population size parameter is input into the population initialization process. The initial population is generated using Latin hypercube sampling, and the displacement of each individual is represented by a decimal value. A feasible region for the displacement is defined and used as a constraint for assigning displacement values. The displacement vector is applied to the corresponding RIB unit using an affine transformation algorithm, and the coordinates of the transformed RIB unit are used as input for geometric rationality verification. The geometric rationality of the pattern after displacement is verified, and the verified scheme is output as the spatial displacement optimization scheme. The global search capability of the genetic algorithm replaces the existing random combination, improving optimization efficiency.

[0044] For each spatial misalignment optimization scheme, a new binary data point array is reconstructed. The reconstructed binary data point array is multiplied by the ground pressure distribution curve to generate a new time-domain signal. A Fourier transform is performed on the new time-domain signal to calculate the noise result for each scheme. This step simulates the acoustic performance of the pattern after misalignment, achieving closed-loop evaluation.

[0045] Noise levels are used as the basis for fitness calculation, and a fitness function is defined with the total noise value as the optimization objective. Noise spectrum data for each scheme is extracted, and the total sound pressure level within a specific frequency band is calculated. The total sound pressure level is normalized, and the normalization result is used as input for fitness value transformation. A negative exponential function is used to convert the sound pressure level into a fitness value, and the transformed fitness value is used as input for sorting. A heap sort algorithm is used to sort the fitness values ​​of all schemes in descending order, and the sorting result is used as input for the genetic operation.

[0046] When performing genetic operations, an elite preservation strategy is adopted, using fitness ranking results as the basis for selecting elite individuals. A tournament selection process is used for the remaining individuals, and the selection results are used as the parent individuals for the crossover operation. A simulated binary crossover algorithm is used, with the parent individual genes as input for the crossover operation. A polynomial mutation operator is applied, with the offspring individuals after crossover as input for the mutation operation. The crossover and mutation probabilities are dynamically adjusted, using population diversity indicators as the basis for parameter adjustment. A new generation of population is generated, and the processes of population generation, noise calculation, fitness evaluation, and genetic operations are repeated.

[0047] The algorithm monitors the trend of the optimal fitness value and uses the fitness value sequence as input for convergence judgment. It calculates the standard deviation of fitness values ​​over multiple generations and uses this standard deviation as a comparison object for the convergence threshold. A maximum number of iterations is set, and the current iteration number is compared with the maximum number of iterations. The average fitness and optimal fitness of each generation are recorded, and this recorded data is used as the basis for generating the convergence curve. When the convergence condition is met or the maximum number of iterations is reached, the current best individual is output as the result of the genetic algorithm.

[0048] When outputting the optimal displacement scheme, the genetic code of the optimal individual is decoded, and the gene sequence is used as input for displacement analysis. The displacement values ​​are converted into actual size units, and the analyzed displacement values ​​are used as input for geometric transformation. Transformation parameters are applied in the CAD system, and these parameters serve as the basis for modifying the pattern drawing. The ground pressure distribution of the optimized pattern is verified through finite element analysis, and the verification results are used as the basis for scheme performance evaluation. The final design file is output, completing the pattern noise optimization process.

[0049] Through the above-described embodiments, this invention systematically integrates geometric feature acquisition, pressure distribution coupling, frequency domain analysis, and genetic algorithm optimization, effectively improving prediction accuracy and optimization efficiency, and solving the technical pain points in the background art.

[0050] In the initial stage of tire tread design, this invention acquires geometric point cloud data of the tire tread pattern using a 3D scanner. The geometric point cloud data is preprocessed to remove outliers and noise, resulting in clean point cloud data. The clean point cloud data is then fitted to a surface using the least squares method to generate a continuous leading edge curve. This leading edge curve is discretized into a sequence of equally spaced sampling points, serving as the scanning path. The tread pattern is then rasterized along the scanning path, and values ​​are assigned based on the spatial relationship between the tread blocks and grooves, generating a binary data point array. The intersections of the tread grooves are assigned a value of 0, while the intersections of the tread blocks are assigned a value of 1. This process accurately reconstructs the nonlinear shape of the ground contact leading edge, providing a geometric basis for subsequent analysis.

[0051] When acquiring the tire's ground pressure distribution curve, a pressure sensor array is installed on a tire test bench to collect the ground pressure distribution of the tire under standard load. The ground pressure distribution data is filtered to eliminate sensor noise, resulting in filtered pressure data. The filtered pressure data is then interpolated into a continuous pressure distribution surface using the Kriging interpolation algorithm. This surface is resampled into a pressure matrix with the same resolution as the binary data point array. A spatial coordinate mapping relationship is established between the pressure distribution curve and the binary data point array. The pressure values ​​on the ground pressure distribution curve are multiplied by the corresponding values ​​in the binary data point array to generate a time-domain signal sequence representing the tread pattern rolling over the ground. This step introduces the non-uniformity of the actual pressure distribution, improving the realism of the noise-excited simulation.

[0052] The time-domain signal is used as input for frequency-domain analysis, and a Fourier transform is performed to convert it into a frequency-domain response. The time-domain signal is then windowed using a Hanning window function to reduce spectral leakage, resulting in a windowed time-domain signal. A Fast Fourier Transform (FFT) algorithm is used to process the windowed time-domain signal, converting it into a frequency-domain amplitude spectrum. Based on the ratio of tire circumference to driving speed, the spatial frequencies of the frequency-domain amplitude spectrum are mapped to acoustic frequencies. A point source radiation model is used to calculate the sound pressure level at the acoustic frequencies. The sound pressure level data is smoothed to generate a noise spectrum distribution map, thereby obtaining the spectral characteristics of the tire pattern noise and providing a quantitative basis for optimization objectives.

[0053] The binary data point array is used as the segmentation object, divided into multiple pattern RIB units according to the pattern design ratio. The periodic characteristics of the pattern in the binary data point array are identified to determine the pitch boundaries. Based on the pitch boundaries and the number of circumferential pitches and lateral RIBs in the pattern, the binary data point array is divided into multiple independent units. A unique identifier is assigned to each independent unit, recording its position coordinates and geometric dimensions. A hash table data structure is used to store the unit information, establishing an indexing mechanism. The mechanical independence of the independent units is verified, boundary continuity is confirmed, and the data is saved as a structure. This step achieves modular processing of the pattern data, providing structured input for the genetic algorithm.

[0054] When optimizing the segmented patterned RIB units using a genetic algorithm, a population size of a certain number of individuals is set, and this population size parameter is input into the population initialization process. The initial population is generated using Latin hypercube sampling, and the displacement of each individual is represented by a decimal value. A feasible region for the displacement is defined and used as a constraint for assigning displacement values. The displacement vector is applied to the corresponding RIB unit using an affine transformation algorithm, and the coordinates of the transformed RIB unit are used as input for geometric rationality verification. The geometric rationality of the pattern after displacement is verified, and the verified scheme is output as the spatial displacement optimization scheme. The global search capability of the genetic algorithm replaces the existing random combination, improving optimization efficiency.

[0055] For each spatial misalignment optimization scheme, a new binary data point array is reconstructed. The reconstructed binary data point array is multiplied by the ground pressure distribution curve to generate a new time-domain signal. A Fourier transform is performed on the new time-domain signal to calculate the noise result for each scheme. This step simulates the acoustic performance of the pattern after misalignment, achieving closed-loop evaluation.

[0056] Noise levels are used as the basis for fitness calculation, and a fitness function is defined with the total noise value as the optimization objective. Noise spectrum data for each scheme is extracted, and the total sound pressure level within a specific frequency band is calculated. The total sound pressure level is normalized, and the normalization result is used as input for fitness value transformation. A negative exponential function is used to convert the sound pressure level into a fitness value, and the transformed fitness value is used as input for sorting. A heap sort algorithm is used to sort the fitness values ​​of all schemes in descending order, and the sorting result is used as input for the genetic operation.

[0057] When performing genetic operations, an elite preservation strategy is adopted, using fitness ranking results as the basis for selecting elite individuals. A tournament selection process is used for the remaining individuals, and the selection results are used as the parent individuals for the crossover operation. A simulated binary crossover algorithm is used, with the parent individual genes as input for the crossover operation. A polynomial mutation operator is applied, with the offspring individuals after crossover as input for the mutation operation. The crossover and mutation probabilities are dynamically adjusted, using population diversity indicators as the basis for parameter adjustment. A new generation of population is generated, and the processes of population generation, noise calculation, fitness evaluation, and genetic operations are repeated.

[0058] The algorithm monitors the trend of the optimal fitness value and uses the fitness value sequence as input for convergence judgment. It calculates the standard deviation of fitness values ​​over multiple generations and uses this standard deviation as a comparison object for the convergence threshold. A maximum number of iterations is set, and the current iteration number is compared with the maximum number of iterations. The average fitness and optimal fitness of each generation are recorded, and this recorded data is used as the basis for generating the convergence curve. When the convergence condition is met or the maximum number of iterations is reached, the current best individual is output as the result of the genetic algorithm.

[0059] When outputting the optimal displacement scheme, the genetic code of the optimal individual is decoded, and the gene sequence is used as input for displacement analysis. The displacement values ​​are converted into actual size units, and the analyzed displacement values ​​are used as input for geometric transformation. Transformation parameters are applied in the CAD system, and these parameters serve as the basis for modifying the pattern drawing. The ground pressure distribution of the optimized pattern is verified through finite element analysis, and the verification results are used as the basis for scheme performance evaluation. The final design file is output, completing the pattern noise optimization process.

[0060] When applying this invention in the early stages of tire tread design, engineers use a 3D scanner to obtain the geometric point cloud data of the tread pattern of a certain model of car tire. After preprocessing the point cloud data to remove outliers, the least squares method is used to fit the nonlinear leading edge curve of the ground contact imprint.

[0061] The leading edge curve is discretized into a sequence of sampling points as the scanning path. The pattern drawing is rasterized along the path to generate a binary data point array, where the groove positions are assigned a value of 0 and the block positions are assigned a value of 1. Simultaneously, the ground pressure distribution under standard load is collected on a tire test bench, and a continuous pressure surface is generated by Kriging interpolation. This surface is then resampled into a pressure matrix with the same resolution as the binary array.

[0062] After establishing the spatial coordinate mapping relationship, the pressure matrix is ​​multiplied point by point with the binary array to generate a time-domain signal sequence representing the tread rolling excitation. This time-domain signal is then processed by Hanning windowing and converted into a frequency domain amplitude spectrum through fast Fourier transform. Based on the parameters of tire circumference and driving speed of 80 km / h, it is mapped to acoustic frequencies, and the noise spectrum of the 1-5000 Hz frequency band is calculated using a point source model.

[0063] This implementation process solves the prediction bias problem caused by geometric simplification in the background technology by coupling the real pressure distribution with the nonlinear leading edge shape.

[0064] During the optimization phase, the binary data point array is divided into multiple RIB units according to the pattern periodicity characteristics, and a hash table structure is used to store the unit coordinates and size information.

[0065] The genetic algorithm population was set to 300 individuals. Initial displacement combinations were generated using Latin hypercube sampling, with the displacement range constrained within a feasible region of ±5 mm. The displacement vectors were applied to the corresponding RIB elements through affine transformation to verify the groove connectivity and mechanical rationality of the pattern after displacement.

[0066] For each shift scheme, a new binary array is constructed and the total noise sound pressure level is calculated. The sound pressure level is converted into a fitness value using a negative exponential function, and then sorted in descending order using heap sort. During iteration, an elite retention strategy and tournament selection are employed, combined with simulated binary crossover and polynomial mutation, dynamically adjusting the crossover probability to 0.8 and the mutation probability to 0.05. Optimization terminates when the fitness standard deviation is less than 0.01 for 10 consecutive generations or after 200 iterations. Finally, the optimal individual is decoded to obtain the shift scheme.

[0067] This embodiment shortens the optimization cycle from several weeks in traditional methods to within 72 hours by replacing random combination with systematic genetic search.

[0068] For optimizing tire tread noise in commercial vehicles, this invention, when dealing with complex lateral groove structures, divides the tread pattern into 12 circumferential pitches and 4 lateral RIB units by identifying the pitch boundaries in the binary array. Each unit is assigned a unique identifier, and affine transformation is used to achieve coordinated optimization of lateral misalignment and circumferential offset. Pressure distribution data is collected through a 256-point sensor array, interpolated to generate a 500×500 resolution pressure matrix, and the time-domain signal generated by multiplication with the binary array reaches 2500 data points.

[0069] In the frequency domain analysis phase, a 4096-point FFT was used to calculate the sound pressure level. Combined with a vehicle speed of 60 km / h, the spatial frequency was mapped to a 1 / 3 octave band spectrum of 63-4000 Hz. The genetic algorithm's fitness function focused on the main noise band of 300-2000 Hz, using negative exponential transformation to strengthen the weighting coefficients for low-frequency noise. The optimized pattern scheme was used in the CAD system to adjust the RIB unit positions. Finite element analysis verified that this improved the uniformity of grounding pressure distribution and reduced the measured total sound pressure level by 3.2 dB.

[0070] This embodiment verifies the method's adaptability to heavy-duty tires by refining the pressure mapping and frequency band weighting strategy.

[0071] In special cases dealing with pattern periodic variable pitch designs, this invention dynamically adjusts the RIB cell segmentation ratio through a hash table indexing mechanism. When three different pitch lengths are detected in the binary array, a differentiated feasible domain for misalignment displacement is automatically allocated, with large pitch cells constrained to ±8 mm and small pitch cells limited to ±3 mm.

[0072] The genetic algorithm population initialization uses Latin hypercube sampling to ensure the spatial uniformity of displacement combinations. During the crossover and mutation phase, a population diversity monitoring index is introduced; when the Hamming distance falls below a threshold, the mutation probability is automatically increased to 0.1. In the noise assessment stage, the time-domain signal undergoes multiple windowing processing, and a three-dimensional sound pressure level spectrum is generated based on different tire rolling speeds (40 / 80 / 120 km / h). The final optimized scheme is verified through finite element analysis, showing that the variable pitch tread pattern reduces the peak ground pressure, and the staggered layout does not affect drainage performance. This embodiment demonstrates the method's scalability to complex tread structures, maintaining optimization stability through dynamic parameter adjustment.

Claims

1. A pattern noise misalignment optimization method based on genetic algorithm, characterized in that, include: Step 1: Collect the initial design data of the tire tread pattern, obtain the nonlinear leading edge shape by fitting the ground imprint, and use it as a scan line to perform discrete sampling on the tread pattern drawing. Generate a binary data point array with a specified step size, where the intersection of the tread grooves is assigned a value of 0 and the intersection of the tread blocks is assigned a value of 1. Step 2: Using the binary data point array as geometric feature input, obtain the ground pressure distribution curve of the tire, establish the spatial coordinate mapping relationship between the pressure distribution curve and the binary data point array, and multiply the pressure value of the ground pressure distribution curve with the corresponding position value of the binary data point array to generate a time domain signal representing the tread rolling over the ground. Step 3: Use the time-domain signal as the frequency-domain analysis input, perform Fourier transform to convert it into a frequency-domain response, and calculate the sound pressure level in combination with the tire speed parameters to obtain the spectral characteristics of the tread noise. Step 4: The binary data point array is used as a segmentation object and divided into multiple patterned RIB units according to the pattern design ratio, and saved as a structure. Step 5: The segmented pattern RIB units are used as the objects of genetic algorithm processing. Using genetic algorithm, an initial population is randomly generated. Individuals in the population represent the displacement in decimal values. The displacement is assigned to the pattern RIB units. The displacement pattern layout is generated through coordinate transformation, and a spatial displacement optimization scheme is generated. Step 6: For each of the spatial misalignment optimization schemes, reconstruct the misaligned binary data point array, multiply the reconstructed binary data point array with the ground pressure distribution curve to generate a new time-domain signal, perform Fourier transform on the new time-domain signal, and calculate the noise result for each scheme. Step 7: Using the noise results as the basis for fitness calculation, define the fitness function, with the total noise value as the optimization objective, calculate the fitness value of each scheme and sort them. Based on the fitness value sorting results, perform genetic operations, including selection, crossover and mutation, to generate a new generation population. Repeat the process of population generation, noise calculation, fitness evaluation and genetic operations until the termination condition is met. Decode the optimal individual into the actual displacement, output the optimal displacement scheme, and apply it to the initial pattern design.

2. The pattern noise misalignment optimization method based on genetic algorithm according to claim 1, characterized in that, The initial design data for collecting tire tread patterns includes: Geometric point cloud data of tire tread patterns are obtained using a 3D scanner; Preprocess the geometric point cloud data to remove outliers and noisy data to obtain clean point cloud data; The least squares method is used to fit a surface to the clean point cloud data to generate a continuous leading edge curve. The leading edge curve is discretized into a sequence of equally spaced sampling points, which serve as the scanning path; The pattern drawing is rasterized along the scanning path, and values ​​are assigned according to the spatial relationship between the pattern blocks and the grooves to generate a binary data point array.

3. The pattern noise misalignment optimization method based on genetic algorithm according to claim 2, characterized in that, The method for obtaining the tire's ground pressure distribution curve includes: A pressure sensor array is installed on a tire testing bench to collect the ground pressure distribution of the tire under standard load. The ground pressure distribution data is filtered to eliminate sensor noise and obtain filtered pressure data. The filtered pressure data is interpolated into a continuous pressure distribution surface using the Kriging interpolation algorithm. The pressure distribution surface is resampled into a pressure matrix with the same resolution as the binary data point array; Perform element-wise multiplication of the binary data point array and the pressure matrix to generate a time-domain signal sequence.

4. The pattern noise misalignment optimization method based on genetic algorithm according to claim 3, characterized in that, The Fourier transform process includes: The time-domain signal is windowed using the Hanning window function to obtain the windowed time-domain signal. The windowed time-domain signal is processed using the Fast Fourier Transform algorithm and converted into a frequency-domain amplitude spectrum. Based on the ratio of tire circumference to driving speed, the spatial frequency of the frequency domain amplitude spectrum is mapped to the acoustic frequency. The sound pressure level at acoustic frequencies was calculated using a point source radiation model. The sound pressure level data is smoothed to generate a noise spectrum distribution map.

5. The pattern noise misalignment optimization method based on genetic algorithm according to claim 4, characterized in that, The segmented pattern RIB unit includes: Identify the periodic features of patterns in a binary data point array and determine the pitch boundaries; The binary data point array is divided into multiple independent units based on the pitch boundary and the number of circumferential pitches and transverse RIBs of the pattern. Assign a unique identifier to each individual unit and record its location coordinates and geometric dimensions; A hash table data structure is used to store unit information, and an indexing mechanism is established. Verify the mechanical independence of independent elements and confirm the boundary continuity.

6. The pattern noise misalignment optimization method based on genetic algorithm according to claim 5, characterized in that, The generated spatial misalignment optimization scheme includes: Set the population size to 100-500 individuals and input the population size parameter into the population initialization process; The initial population was generated using the Latin hypercube sampling method, and the generated initial population was used as the object for displacement assignment. Define the feasible region of the displacement and use the feasible region as a constraint condition for assigning displacement values; The displacement vector is applied to the corresponding RIB element through an affine transformation algorithm, and the coordinates of the transformed RIB element are used as input for geometric rationality verification. Verify the geometric rationality of the pattern after misalignment, and output the verified scheme as the spatial misalignment optimization scheme.

7. The pattern noise misalignment optimization method based on genetic algorithm according to claim 6, characterized in that, The calculation of fitness values ​​includes: Extract the noise spectrum data for each scheme and use the noise spectrum data as input for sound pressure level calculation; Calculate the total sound pressure level in the 1-5000Hz frequency band and use the total sound pressure level value as the input for normalization. The total sound pressure level is normalized, and the normalization result is used as the input for fitness value conversion. The sound pressure level is converted into a fitness value using a negative exponential function, and the converted fitness value is used as the input for the sorting process. The fitness values ​​of all solutions are sorted in descending order using the heap sort algorithm, and the sorting result is used as the input for the genetic operation.

8. The pattern noise misalignment optimization method based on genetic algorithm according to claim 7, characterized in that, The genetic operations include: An elite retention strategy is adopted, and the fitness ranking results are used as the basis for selecting elite individuals; The remaining individuals are selected through a tournament, and the selection results are used as the parent individuals for the crossover operation; The simulated binary crossover algorithm is used, with the genes of the parent individuals as the input for the crossover operation; The polynomial mutation operator is applied, and the offspring individuals after crossover are used as the input for the mutation operation; The crossover and mutation probabilities are dynamically adjusted, and population diversity indicators are used as the basis for parameter adjustment.

9. The pattern noise misalignment optimization method based on genetic algorithm according to claim 8, characterized in that, The termination condition determination includes: Monitor the changing trend of the optimal fitness value and use the fitness value sequence as input for convergence judgment; Calculate the standard deviation of fitness values ​​over 10 consecutive generations, and use the standard deviation as the comparison object for the convergence threshold; Set the maximum number of iterations to 200 generations, and compare the current number of iterations with the maximum number of iterations. Record the average fitness and optimal fitness of each generation of the population, and use the recorded data as the basis for generating the convergence curve; When the convergence condition is met or the maximum number of iterations is reached, the current best individual is used as the output result of the genetic algorithm.

10. The pattern noise misalignment optimization method based on genetic algorithm according to claim 9, characterized in that, The optimal error correction scheme for output includes: Decode the genetic code of the optimal individual and use the gene sequence as input for the analysis of the fault displacement; The displacement values ​​are converted into actual size units, and the analyzed displacement values ​​are used as input for geometric transformation. Transformation parameters are applied in the CAD system and used as the basis for modifying pattern drawings; The grounding pressure distribution of the optimized pattern was verified through finite element analysis, and the verification results were used as the basis for performance evaluation of the scheme. Output the final design document, and output the verified solution as the optimization result.

Images