Quality line method data optimization method and device based on cross-correlation and optimization algorithm

Abstract

The invention discloses a cross-correlation and optimization algorithm-based quality line method data optimization method and device, and relates to the technical field of test and analysis, and the method comprises the steps: S1, obtaining frequency response data, carrying out the preprocessing, determining a quality line frequency band, and carrying out the segmentation to obtain sub-frequency bands; s2, taking the sub-band as a candidate solution of an optimization algorithm, taking the cross correlation coefficient of the sub-band as a target value of the optimization algorithm to update the candidate solution, and iterating until an iteration ending condition is met to obtain an optimal sub-band; s3, comparing the cross-correlation numerical value of the optimal sub-frequency band with a preset threshold value, if the cross-correlation numerical value is greater than the preset threshold value, executing S4, otherwise, executing S2 again; and S4, calculating a centroid position according to the optimal sub-frequency band, performing error calculation on the centroid position and an actually measured centroid position, if an error is within a preset error range, the data is valid, otherwise, executing S2 again. According to the method, the frequency band is dynamically selected by using the optimization algorithm based on the cross correlation coefficient between the signals, and the quality line method data meeting the requirements can be quickly and accurately screened out.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile noise, vibration and harshness (NVH) testing and analysis, and in particular to a quality line method data optimization method and device based on cross-correlation and optimization algorithm. Background Art

[0002] In the automotive industry, the control of noise, vibration and harshness (NVH) is crucial to improving the overall performance of the vehicle and the user experience. The knock modal test is one of the important methods used to evaluate and optimize the vibration characteristics of the vehicle structure. By analyzing the vibration response of the vehicle under different conditions, engineers can identify and solve potential NVH problems. However, the knock modal test usually generates a large amount of frequency response data, and the traditional manual screening method has the following shortcomings in practical applications:

[0003] Inefficiency: Traditional manual screening methods rely on operator experience and manual analysis, which is not only time-consuming but also difficult to ensure the consistency of the screening process. At the test site, especially under time constraints, manual screening methods are difficult to quickly complete the screening and evaluation of large amounts of data, affecting test efficiency.

[0004] Difficulty in real-time evaluation of test quality: Due to the complex and time-consuming manual screening process, it is often difficult to evaluate test quality in a timely manner at the test site. If the test quality is poor, problems are often discovered only after data analysis, resulting in the need to repeat the experiment, which increases time costs and wastes resources.

[0005] The consistency and accuracy of data processing are difficult to ensure: Manual screening is prone to human errors when processing large amounts of complex data, making it difficult to ensure the accuracy and consistency of screening, which may affect the final analysis results and decisions. Summary of the invention

[0006] In order to overcome the above-mentioned shortcomings, the present invention proposes a quality line method data optimization method and device based on cross-correlation and optimization algorithm. This method can quickly and accurately screen out data that meets the requirements by dynamically selecting frequency bands and calculating the cross-correlation coefficients between signals, thereby evaluating the test quality in real time. At the test site, even if the test quality is poor, the method can quickly identify and prompt for re-testing, thereby significantly improving the test efficiency and reducing the rework time and resource waste caused by quality problems. Through the automated screening and evaluation process, the accuracy and consistency of data processing are ensured, providing reliable data support for the optimization of automotive NVH performance.

[0007] On the one hand, the quality line method data optimization method based on cross-correlation and optimization algorithm has the following specific steps:

[0008] S1, obtaining the frequency response data of the knocking modal experiment, preprocessing the frequency response data, determining the required quality line frequency band, and dividing the quality line frequency band into sub-frequency bands;

[0009] S2, taking the sub-frequency band as the candidate solution of the optimization algorithm, calculating the mutual correlation coefficient of the output signal of the sub-frequency band in the candidate solution to obtain the mutual correlation value; using the mutual correlation value as the target value of the optimization algorithm to update the candidate solution, iterating until the iteration end condition is met, and obtaining the optimal sub-frequency band in the candidate solution;

[0010] S3, obtaining the mutual correlation value of the optimal sub-frequency band, and comparing it with the preset correlation threshold, if it is greater than the preset correlation threshold, executing S4, otherwise re-executing S2 to re-obtain the optimal sub-frequency band;

[0011] S4, calculating the centroid position according to the obtained optimal sub-frequency band, and performing error calculation between the calculated centroid position and the actually measured centroid position. If the error is within a preset error range, the data of the optimal sub-frequency band is valid, otherwise S2 is executed again.

[0012] In the preferred S1, the preprocessing of the frequency response data includes denoising and normalization; the denoising includes filter technology or wavelet denoising technology.

[0013] Preferably, in S2, the optimization algorithm includes a genetic algorithm or a particle swarm optimization algorithm.

[0014] Preferably, in S2, the mutual correlation coefficient calculation of the output signals of the sub-frequency bands in the candidate solution to obtain the mutual correlation value is specifically: the mutual correlation coefficients between all the output signals of the sub-frequency bands are calculated, and the average value of all the mutual correlation values ​​obtained is the mutual correlation value of the sub-frequency band.

[0015] Preferably, in S2, the iteration end condition is the maximum number of iterations or fitness convergence.

[0016] Preferably, S3 further includes a marking and storage step; the marking and storage step occurs after the comparison with the preset correlation threshold, and the comparison result of the mutual correlation value of the optimal sub-frequency band and the preset correlation threshold is marked, classified and data stored, as follows;

[0017] If the mutual correlation value of the optimal sub-frequency band is greater than the preset correlation threshold value, the data is marked as "preferred data"; if the mutual correlation value of the optimal sub-frequency band is less than the preset correlation lower threshold value, the data is marked as "rejected data"; the remaining data is marked as "data to be reviewed"; the preset correlation lower threshold value is less than the preset correlation lower threshold value;

[0018] The labeled data is stored in the database for subsequent retrieval and analysis.

[0019] Preferably, in S4, the centroid position is calculated according to the obtained optimal sub-band that meets the preset correlation threshold, specifically as follows:

[0020] Perform transfer function analysis on the optimal sub-frequency band to obtain the amplitude information of each output point;

[0021] Obtain the geometric position of each output point from the frequency response data of the impact modal experiment;

[0022] The geometric position and amplitude of each output point are weighted averaged to calculate the centroid position.

[0023] Preferably, the error between the calculated center of mass position and the actual measured center of mass position is obtained by the Euclidean distance between the calculated center of mass position and the actual measured center of mass position.

[0024] On the other hand, the quality line method data optimization device based on cross-correlation and optimization algorithm includes the following:

[0025] A data preprocessing module is used to obtain the frequency response data of the knocking modal experiment, preprocess the frequency response data, determine the required quality line frequency band, and divide the quality line frequency band into sub-frequency bands;

[0026] The optimization algorithm selection module is used to use the sub-frequency band as a candidate solution of the optimization algorithm, calculate the mutual correlation coefficient of the output signal of the sub-frequency band in the candidate solution to obtain the mutual correlation value; update the candidate solution using the mutual correlation value as the target value of the optimization algorithm, iterate until the iteration end condition is met, and obtain the optimal sub-frequency band in the candidate solution;

[0027] A threshold screening module is used to obtain the mutual correlation value of the optimal sub-frequency band and compare it with a preset correlation threshold. If it is greater than the preset correlation threshold, the centroid position error calculation module is executed, otherwise the optimization algorithm selection module is re-executed to re-obtain the optimal sub-frequency band;

[0028] The centroid position error calculation module is used to calculate the centroid position according to the obtained optimal sub-frequency band, and calculate the error between the calculated centroid position and the actual measured centroid position. If the error is within the preset error range, the data of the optimal sub-frequency band is valid, otherwise the optimization algorithm selection module is re-executed.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] (1) The present invention can quickly and accurately screen out data that meets the requirements by dynamically selecting frequency bands and calculating the correlation coefficients between signals, thereby evaluating the test quality in real time. At the test site, even if the test quality is poor, the method can quickly identify and prompt for retesting, thereby significantly improving the test efficiency and reducing the rework time and resource waste caused by quality problems.

[0031] (2) The present invention ensures the accuracy and consistency of data processing through an automated screening and evaluation process, providing reliable data support for the optimization of automobile NVH performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The present invention is further described in detail below in conjunction with the accompanying drawings;

[0033] Figure 1 It is a flow chart of a quality line method data optimization method based on cross-correlation and optimization algorithm according to an embodiment of the present invention;

[0034] Figure 2 It is a flow chart of the implementation of the quality line method data optimization method based on cross-correlation and optimization algorithm in an embodiment of the present invention;

[0035] Figure 3 A frequency response function curve of a quality line method data optimization method based on cross-correlation and optimization algorithm according to an embodiment of the present invention;

[0036] Figure 4 It is a structural block diagram of a quality line method data optimization device based on cross-correlation and optimization algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The present invention is further described below through specific implementation modes.

[0038] See also Figure 1 and Figure 2 As shown, the specific steps of the quality line method data optimization method based on cross-correlation and optimization algorithm are as follows:

[0039] S1, obtain the frequency response data of the knocking modal experiment, pre-process the frequency response data, determine the required quality line frequency band, and divide the quality line frequency band into sub-frequency bands.

[0040] S11, data import.

[0041] Import frequency response data from the knock modal experiment. The so-called frequency response data means that in the knock modal experiment, the experimental object is knocked at different knock points, and the frequency response data from each knock point to multiple output points is recorded.

[0042] S12, data preprocessing.

[0043] Preprocess the imported frequency response data. The preprocessing operation includes denoising and normalization. The details are as follows:

[0044] Denoising: Use filters or wavelet denoising techniques to remove high-frequency noise and other interfering signals from the data.

[0045] Normalization processing: Normalize the data to ensure that the signals at each output point are at the same level to avoid affecting subsequent correlation calculations due to amplitude differences.

[0046] S13, frequency band selection and frequency band segmentation.

[0047] See also Figure 3 As shown, according to the obtained data, the rigid body modal frequency and the first-order elastic modal frequency of the tested piece are determined through preliminary modal analysis, and the frequency range required for analysis, namely the mass line frequency band, is determined and stored in a matrix form.

[0048] Perform frequency band segmentation. Divide the frequency band to be analyzed into several sub-bands, for example, each sub-band covers 5Hz. Each sub-band will be used as a candidate band for subsequent optimization and calculation.

[0049] Specifically, different sub-band segmentation methods can be selected according to the accuracy requirements for data analysis and the design of optimization algorithms, as follows:

[0050] Average interval division: To simplify the analysis and calculation process, the frequency band to be analyzed can be evenly divided into several sub-bands of equal width. For example, if the frequency band range is 30Hz (such as from 10Hz to 40Hz), it can be divided into 6 5Hz wide sub-bands: 10-15Hz, 15-20Hz and 20-25Hz, etc.

[0051] Non-even interval segmentation: In some cases, if the importance of certain frequency bands is known in advance (for example, some frequency bands may contain more modal information), non-even segmentation can be performed based on this information. For example, smaller intervals can be used for refinement within the important frequency bands, while wider intervals can be used within the less important frequency bands.

[0052] Overlapping area segmentation: At the same time, in order to ensure smooth transitions between frequency bands and avoid missing key information at the sub-band boundaries, overlapping frequency bands can be designed; for example, 10-15Hz, 13-18Hz, and 15-20H, etc. This overlapping design can form transition areas between frequency bands, thereby better capturing possible modal features.

[0053] S2, taking the sub-frequency band as the candidate solution of the optimization algorithm, calculating the mutual correlation coefficient of the output signal of the sub-frequency band in the candidate solution to obtain the mutual correlation value; using the mutual correlation value as the target value of the optimization algorithm to update the candidate solution, iterating until the iteration end condition is met, and obtaining the optimal sub-frequency band in the candidate solution.

[0054] Select an optimization algorithm (such as a genetic algorithm or a particle swarm optimization algorithm) to optimize the selection of sub-bands. The algorithm uses a strategy of selecting and calculating at the same time, selecting different sub-bands for calculation in each iteration and evaluating their correlation. The goal is to find the optimal frequency band that can maximize the cross-correlation coefficient.

[0055] The optimization algorithm used in this embodiment is the particle swarm optimization algorithm (PSO, Particle Swarm Optimization), and the specific steps are as follows:

[0056] S21, initialization.

[0057] In the frequency range between the rigid body mode and the first-order elastic mode (ie, the mass line frequency band to be analyzed), a group of particles are initialized, each particle representing a candidate frequency band (including the starting frequency and the frequency band width).

[0058] S22, fitness calculation.

[0059] Each particle calculates the average value of the mutual correlation coefficient of the signals in its frequency band as the fitness value. The higher the fitness value, the stronger the correlation of the signals in the frequency band.

[0060] S23, speed and position update.

[0061] The particle adjusts its speed and position according to its historical best position (p_best) and the global best position (g_best). The speed update formula takes into account the particle's current position, individual experience, and group experience, so that it gradually moves toward the optimal sub-band.

[0062] Each particle updates its speed and position according to its current speed, its own historical optimal position (local optimality), and the global optimal position in the group (global optimality).

[0063] S24, iterative update.

[0064] Dynamically adjust the acceleration coefficients (c1 and c2) to ensure that global exploration is increased in the early stage and fine optimization of the best frequency band is strengthened in the later stage. Specific inertia and acceleration adjustment: Set a larger inertia weight (w) in the early stage to explore the frequency band widely, and gradually reduce it with iterations to increase convergence accuracy.

[0065] The fitness calculation and speed position update steps are repeated until the stopping condition (such as the maximum number of iterations or fitness convergence) is reached.

[0066] The optimization algorithm of this embodiment can adjust the frequency band selection strategy according to the result of the cross-correlation calculation, and gradually approach the optimal frequency band. The frequency band finally selected should be able to provide the highest signal correlation. The selection of the optimal frequency band is based on maximizing the cross-correlation coefficient. The average cross-correlation coefficient of all output signals in the frequency band is the highest, indicating that the similarity and consistency between the signals are the strongest. This strategy of selecting and calculating while dynamically adjusting the position and width of the frequency band ensures that the optimal frequency band is gradually approached and the signal correlation is maximized.

[0067] The optimization algorithm of this embodiment may also use other algorithms besides the genetic algorithm and the particle swarm optimization algorithm to optimize the selection of sub-frequency bands, which is not limited in this embodiment.

[0068] The cross-correlation calculation is as follows:

[0069] Usually in a knock modal test, multiple knock points are selected on the test object to comprehensively evaluate the vibration characteristics of the entire structure. Each of the multiple knock points will generate a set of independent output signals. The cross-correlation between the output signals of each knock point is calculated separately to ensure the independence and reliability of the data at each knock point.

[0070] For each candidate sub-band, the mutual correlation coefficients between all output signals in the band are calculated. The mutual correlation coefficient is used to measure the similarity and time synchronization between two output signals, and then used to evaluate the signal quality of the sub-band. The highest average value of the mutual correlation coefficients of all output signals in the band indicates the strongest similarity and consistency between the signals.

[0071] S3, obtaining the mutual correlation value of the optimal sub-frequency band, and comparing it with the preset correlation threshold, if it is greater than the preset correlation threshold, executing S4, otherwise re-executing S2 to re-obtain the optimal sub-frequency band.

[0072] The preset correlation threshold is determined based on historical data or experimental optimization, as follows:

[0073] Based on historical data: Analyze the data from previous similar experiments, calculate the mean and variance of the cross-correlation coefficient between signals, and determine a threshold that can effectively distinguish high-quality data from low-quality data.

[0074] Experimental optimization: Perform multiple measurements in the initial experiment, and set an initial threshold that can ensure data quality and experimental stability by observing the distribution of cross-correlation coefficients in different frequency bands.

[0075] In addition, based on the actual measurement and analysis results, the threshold is continuously adjusted and optimized to ensure that it can accurately filter out data that meets the requirements.

[0076] Furthermore, use smart data tags to filter signal data:

[0077] Automatically label each set of data based on the correlation coefficient results calculated in the steps. The labels include:

[0078] "Preferred data": Data that meets the preset relevance threshold is marked as "preferred data" and is suitable for subsequent analysis.

[0079] “Removed data”: Data that does not meet the correlation threshold is marked as “removed data” and is not suitable for subsequent analysis.

[0080] “Data to be reviewed”: Data whose relevance is close to the threshold is marked as “data to be reviewed” and requires further inspection.

[0081] In this embodiment, a preset correlation lower threshold is set, and the preset correlation threshold and the preset correlation lower threshold constitute an approximate interval; data whose correlation is greater than the preset correlation threshold is data that meets the preset correlation threshold; data whose correlation is less than the preset correlation lower threshold (i.e., data that does not reach the correlation threshold) is eliminated data; data whose correlation is within the approximate interval (i.e., data whose correlation is close to the threshold) is data to be reviewed.

[0082] For scenarios with high accuracy requirements, the approximate interval can be set smaller; if a certain error is allowed, the approximate interval can be appropriately relaxed. Under different experimental conditions, the approximate interval can be adjusted dynamically. If too much or too little "data to be reviewed" is found in the experiment, the approximate interval can be automatically adjusted to optimize the efficiency and accuracy of data screening.

[0083] Further, use data storage and management:

[0084] The tagged data is stored in the database for subsequent retrieval and analysis. The database supports fast retrieval and efficient tag management to improve data processing efficiency.

[0085] S4, calculating the centroid position according to the obtained optimal sub-frequency band, and performing error calculation between the calculated centroid position and the actually measured centroid position. If the error is within a preset error range, the data of the optimal sub-frequency band is valid, otherwise S2 is executed again.

[0086] S41, actual centroid position measurement.

[0087] The actual center of mass position depends on the measurement of suspension methods, such as using a force sensor to record the force on the suspension point, and combining the position data in the coordinate system to perform a moment balance calculation to ultimately obtain the center of gravity position.

[0088] S42, calculate and obtain the center of mass position.

[0089] Based on the selected preferred data, high-precision center of mass position data is obtained.

[0090] The calculated center of mass position depends on the selected optimal data, which provides the vibration response information of the experimental object at each output point. The center of mass position can be calculated by weighted average of the vibration response and the output point position. The details are as follows:

[0091] S421, transfer function analysis based on the preferred data.

[0092] The selected preferred data are used, which usually include the transfer function (frequency response function) from the striking point to each output point, reflecting the vibration response of the experimental object at different positions.

[0093] These transfer functions are interpreted to obtain amplitude and phase information at each output point to understand the vibration intensity and response characteristics at each location.

[0094] S422, vibration response analysis.

[0095] Calculate the vibration amplitude of the experimental object at each output point. These amplitudes represent the response strength of each output point when struck, and are key data for calculating the center of mass position.

[0096] At each output point the corresponding amplitude value is recorded and combined with the geometric position of that point.

[0097] S423, calculate the centroid position using the weighted average method.

[0098] The position of each output point and its vibration response (amplitude) are weighted averaged. The center of mass position of the experimental object is calculated by the following formula:

[0099]

[0100] The output point position represents the geometric position of the known output point on the experimental object in the frequency response data of the tapping modal experiment; the amplitude represents the vibration response intensity of each output point calculated in the preferred data.

[0101] The weighted average method can use the intensity of the vibration response to determine the actual center of mass position of the experimental object, because the center of mass position usually corresponds to the concentrated area of ​​the vibration response.

[0102] S43, comparison of centroid positions.

[0103] The calculated centroid position is compared with the actual measured centroid position. If the error between the two is within a preset adjustable threshold (e.g., 5%), the data is considered valid; if the error exceeds the threshold, the frequency band is readjusted and a new calculation is performed.

[0104] Specifically, the error is quantified by calculating the Euclidean distance between the actual center of mass position and the calculated center of mass position, so as to determine whether the accuracy of the center of mass position calculation meets the requirements. This method can intuitively reflect the size and direction of the center of mass position deviation. The error calculation formula is expressed as:

[0105]

[0106] Among them, (X 计算 ,Y 计算 ,Z 计算 ) is the coordinate for calculating the centroid position; (X 实际 ,Y 实际 ,Z 实际 ) is the coordinate of the actual measured centroid position.

[0107] S44, Recalculation and evaluation.

[0108] If the data are invalid, the cross-correlation coefficients are recalculated based on the newly adjusted frequency bands and calculation strategy, and the centroid position comparison is performed again until the error is within an acceptable range.

[0109] Furthermore, batch processing and parallel computing are used in the implementation of the quality line method data optimization method based on cross-correlation and optimization algorithm in this embodiment:

[0110] Batch data processing. Supports batch processing of multiple hit point data. For each hit point data set, preprocessing, frequency band selection and correlation calculation can be automatically performed.

[0111] Parallel computing optimization. Utilize parallel computing technology (such as multithreading or distributed computing) to process multiple data sets simultaneously, significantly improving data processing efficiency. Especially when processing large-scale data sets, screening results can be obtained quickly.

[0112] Through the above specific implementation steps, the method of this embodiment can quickly and accurately screen out data that meets the requirements by dynamically selecting frequency bands and calculating the mutual correlation coefficients between signals, thereby evaluating the test quality in real time; at the test site, even if the test quality is poor, the method can quickly identify and prompt for retesting, thereby significantly improving the test efficiency and reducing the rework time and resource waste caused by quality problems; through the automated screening and evaluation process, the accuracy and consistency of data processing are ensured, providing reliable data support for the optimization of automotive NVH performance.

[0113] See also Figure 4 As shown, the present invention also discloses a quality line method data optimization device based on cross-correlation and optimization algorithm, comprising:

[0114] The data preprocessing module 401 is used to obtain the frequency response data of the knocking modal experiment, preprocess the frequency response data, determine the required quality line frequency band, and divide the quality line frequency band into sub-frequency bands;

[0115] The optimization algorithm selection module 402 is used to use the sub-band as a candidate solution of the optimization algorithm, calculate the mutual correlation coefficient of the output signal of the sub-band in the candidate solution to obtain a mutual correlation value; update the candidate solution using the mutual correlation value as the target value of the optimization algorithm, iterate until the iteration end condition is met, and obtain the optimal sub-band in the candidate solution;

[0116] Threshold screening module 403, used to obtain the mutual correlation value of the optimal sub-frequency band and compare it with the preset correlation threshold. If it is greater than the preset correlation threshold, the centroid position error calculation module is executed, otherwise the optimization algorithm selection module is re-executed to re-obtain the optimal sub-frequency band;

[0117] The centroid position error calculation module 404 is used to calculate the centroid position according to the obtained optimal sub-frequency band, and perform error calculation between the calculated centroid position and the actual measured centroid position. If the error is within the preset error range, the data of the optimal sub-frequency band is valid, otherwise the optimization algorithm selection module is re-executed.

[0118] The specific implementation of the quality line method data optimization device based on cross-correlation and optimization algorithm is the same as the quality line method data optimization method based on cross-correlation and optimization algorithm, and will not be repeated in this embodiment.

[0119] The above is only a specific implementation of the present invention, but the design concept of the present invention is not limited thereto. Any non-substantial changes to the present invention using this concept shall be deemed as an infringement of the protection scope of the present invention.

Claims

1. A quality line method data optimization method based on cross-correlation and optimization algorithm, characterized in that: The specific steps are as follows: S1, obtaining the frequency response data of the knocking modal experiment, preprocessing the frequency response data, determining the required quality line frequency band, and dividing the quality line frequency band into sub-frequency bands; S2, taking the sub-frequency band as a candidate solution of the optimization algorithm, calculating the mutual correlation coefficient of the output signal of the sub-frequency band in the candidate solution to obtain the mutual correlation value; The mutual correlation value is used as the target value of the optimization algorithm to update the candidate solution, and the iteration is performed until the iteration end condition is met to obtain the optimal sub-band in the candidate solution; S3, obtaining the mutual correlation value of the optimal sub-frequency band, and comparing it with the preset correlation threshold, if it is greater than the preset correlation threshold, executing S4, otherwise re-executing S2 to re-obtain the optimal sub-frequency band; S4, calculating the centroid position according to the obtained optimal sub-frequency band, and performing error calculation between the calculated centroid position and the actually measured centroid position. If the error is within a preset error range, the data of the optimal sub-frequency band is valid, otherwise S2 is executed again.

2. The quality line method data optimization method based on cross-correlation and optimization algorithm according to claim 1 is characterized in that: In S1, the preprocessing of the frequency response data includes denoising and normalization; the denoising includes filter technology or wavelet denoising technology.

3. The quality line method data optimization method based on cross-correlation and optimization algorithm according to claim 1 is characterized in that: In S2, the optimization algorithm includes a genetic algorithm or a particle swarm optimization algorithm.

4. The quality line method data optimization method based on cross-correlation and optimization algorithm according to claim 1 is characterized in that: In S2, the mutual correlation coefficient calculation of the output signals of the sub-frequency bands in the candidate solution to obtain the mutual correlation value is specifically: the mutual correlation coefficients between all output signals of the sub-frequency bands are calculated, and the average value of all the mutual correlation values ​​obtained is the mutual correlation value of the sub-frequency band.

5. The quality line method data optimization method based on cross-correlation and optimization algorithm according to claim 1 is characterized in that: In S2, the iteration end condition is the maximum number of iterations or fitness convergence.

6. The quality line method data optimization method based on cross-correlation and optimization algorithm according to claim 1 is characterized in that: S3 further includes a marking and storage step; the marking and storage step occurs after the comparison with the preset correlation threshold, and the comparison result between the mutual correlation value of the optimal sub-frequency band and the preset correlation threshold is marked, classified and data stored, as follows; If the mutual correlation value of the optimal sub-frequency band is greater than the preset correlation threshold value, the data is marked as "preferred data"; if the mutual correlation value of the optimal sub-frequency band is less than the preset correlation lower threshold value, the data is marked as "rejected data"; the remaining data is marked as "pending data"; the preset correlation lower threshold value is less than the preset correlation lower threshold value; The labeled data is stored in the database for subsequent retrieval and analysis.

7. The quality line method data optimization method based on cross-correlation and optimization algorithm according to claim 1 is characterized in that: In S4, the centroid position is calculated according to the obtained optimal sub-band that meets the preset correlation threshold, as follows: Perform transfer function analysis on the optimal sub-frequency band to obtain the amplitude information of each output point; Obtain the geometric position of each output point from the frequency response data of the impact modal experiment; The geometric position and amplitude of each output point are weighted averaged to calculate the centroid position.

8. The quality line method data optimization method based on cross-correlation and optimization algorithm according to claim 1 is characterized in that: The error between the calculated centroid position and the actual measured centroid position is obtained by the Euclidean distance between the calculated centroid position and the actual measured centroid position.

9. A quality line method data optimization device based on cross-correlation and optimization algorithm, comprising the following: A data preprocessing module is used to obtain the frequency response data of the knocking modal experiment, preprocess the frequency response data, determine the required quality line frequency band, and divide the quality line frequency band into sub-frequency bands; The optimization algorithm selection module is used to use the sub-frequency band as a candidate solution of the optimization algorithm, and calculate the mutual correlation coefficient of the output signal of the sub-frequency band in the candidate solution to obtain the mutual correlation value; The mutual correlation value is used as the target value of the optimization algorithm to update the candidate solution, and the iteration is performed until the iteration end condition is met to obtain the optimal sub-band in the candidate solution; A threshold screening module is used to obtain the mutual correlation value of the optimal sub-frequency band and compare it with a preset correlation threshold. If it is greater than the preset correlation threshold, the centroid position error calculation module is executed, otherwise the optimization algorithm selection module is re-executed to re-obtain the optimal sub-frequency band; The centroid position error calculation module is used to calculate the centroid position according to the obtained optimal sub-frequency band, and calculate the error between the calculated centroid position and the actual measured centroid position. If the error is within the preset error range, the data of the optimal sub-frequency band is valid, otherwise the optimization algorithm selection module is re-executed.

Images