A method for detecting abnormal noise of a horizontal driver of a car seat in a noise interference environment

Abstract

The application discloses a noise interference environment automobile seat horizontal driver abnormal sound detection method, relates to the technical field of fault diagnosis, and comprises the following steps: calculating the structure vibration residual index and the sound residual index of a candidate abnormal segment set, performing abnormal discrimination processing, outputting an initial abnormal sound event set, extracting abnormal sound discrimination features of the initial abnormal sound event set and performing priority sorting, and constructing a verification motion trajectory combination; calculating a recurrence evaluation index of the verification motion trajectory combination, performing abnormal sound confirmation determination on the initial abnormal sound event set according to the recurrence evaluation index, and outputting an abnormal sound state evaluation record. The application realizes essential separation of complex background noise and normal motion acoustic characteristics by constructing counterfactual virtual sequence pairs and verification motion trajectory combinations, improves the reliability and credibility of the evaluation result, and improves the efficiency of fault diagnosis and repair.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, and in particular to a method for detecting abnormal noises from the horizontal drive mechanism of an automobile seat under noise interference conditions. Background Technology

[0002] With the development of electric seat adjustment technology in automobiles, the smoothness of operation and structural reliability of the seat leveling actuator, as a key actuator for adjusting the fore-and-aft position of the seat, have gradually become important indicators for evaluating the overall vehicle comfort. Currently, the automotive industry has increasingly higher requirements for vehicle quietness. Abnormal noises generated during seat adjustment, such as structural friction noise, gap impact noise, and guide rail vibration radiation noise, have become important research objects for vehicle quality control and after-sales fault diagnosis. Existing technologies typically employ detection methods based on acoustic or vibration sensors, collecting sound pressure signals or structural vibration signals during operation and combining them with spectral analysis, energy analysis, or threshold discrimination to identify abnormal responses. With the development of multi-sensor fusion and signal processing technologies, some studies have begun to introduce time-frequency analysis, envelope demodulation, and feature statistics to improve the ability to identify abnormal noises.

[0003] However, existing methods still have certain shortcomings. Traditional detection methods mostly rely on directly observing the amplitude or spectral characteristics of signals to identify anomalies. The detection logic is based on the premise that "the observed signal is the actual response of the structure," and lacks a dynamic reference model for the normal operating state of the structure. When environmental noise or changes in operating conditions cause the observed signal to deviate, misjudgment or missed judgment is likely to occur. Especially under low signal-to-noise ratio conditions, it is difficult to distinguish between abnormal structural responses and external interference signals. Existing methods usually complete anomaly identification based on a single operation process and lack a mechanism for reproducing and verifying suspected anomalies. It is impossible to determine whether the detected anomaly has structural stability and repeatability, thus making it difficult to distinguish between occasional noise disturbances and real structural defects, resulting in insufficient reliability of detection results. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for detecting abnormal noises from a car seat horizontal actuator under noise interference conditions, which solves the problems of poor ability to distinguish between structural abnormalities and external interference signals and insufficient reliability of detection results.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides a method for detecting abnormal noise from a car seat horizontal actuator under noise interference. The method includes: collecting structural response observation data of the car seat horizontal actuator during operation under noise interference; dividing the structural response observation data into motion stages to form aligned slice records; extracting the mechanical excitation features and stage kinematic features of each motion stage from the aligned slice records; calculating the counterfactual structural response of the mechanical excitation features and stage kinematic features; outputting counterfactual virtual sequence pairs; statistically analyzing the frame-level temporal difference between the structural response observation data and the counterfactual virtual sequence pairs; generating structural response residual data; evaluating the noise intensity of the structural response residual data and performing credibility screening to generate a candidate abnormal segment set; calculating the structural vibration residual index and sound residual index of the candidate abnormal segment set, performing anomaly discrimination processing, outputting an initial abnormal noise event set; extracting the abnormal noise discrimination features of the initial abnormal noise event set and prioritizing them; constructing a verification motion trajectory combination; calculating the reproducibility evaluation index of the verification motion trajectory combination; performing abnormal noise confirmation judgment on the initial abnormal noise event set based on the reproducibility evaluation index; and outputting an abnormal noise status evaluation record.

[0008] As a preferred embodiment of the method for detecting abnormal noise from a car seat horizontal drive under noise interference environment as described in this invention, the specific steps for forming aligned slice records are as follows:

[0009] The sampling time positions of the structural response observation data from each acquisition channel are mapped to the same continuous time axis, and equal-interval time position resampling processing is performed to generate a synchronous sampling data sequence.

[0010] By statistically analyzing the changes in driving displacement and motion velocity in the synchronously sampled data sequence, the motion activity index within a continuous time window is calculated. Based on the motion activity index, motion stages are divided, and aligned slice records are output.

[0011] As a preferred embodiment of the method for detecting abnormal noise from a car seat horizontal drive under noise interference environment as described in this invention, the specific steps for extracting the mechanical excitation features and stage kinematic features of each motion stage in the aligned slice record are as follows:

[0012] Read the structural response observation sequence corresponding to each motion stage in the aligned slice record, and calculate the displacement change, instantaneous velocity change and stage acceleration per unit time to generate stage kinematic characteristics;

[0013] By using short-time energy statistics and local peak detection, the high-energy transient impact locations of the structural response observation sequences corresponding to each motion stage in the aligned slice records are identified, and the degree of change in vibration response amplitude, local transient impact response intensity and frequency band energy distribution are statistically analyzed to generate mechanical excitation characteristics.

[0014] As a preferred embodiment of the method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in this invention, the specific steps for outputting counterfactual virtual sequence pairs are as follows:

[0015] The structural excitation correction is calculated based on the synchronous change relationship between mechanical excitation characteristics and stage kinematic characteristics, and the equivalent structural excitation sequence is output.

[0016] Within the same motion phase, the response delay and amplitude ratio between different observation positions in the structural response observation sequence are calculated to form a structural propagation parameter sequence;

[0017] The equivalent structural excitation sequence is subjected to time delay mapping and amplitude mapping according to the structural propagation parameter sequence to generate a counterfactual structural vibration sequence.

[0018] The changes in acoustic radiation generated by the corresponding structural vibration in the counterfactual structural vibration sequence are statistically analyzed to generate a counterfactual structural radiation acoustic sequence, which is then paired with the counterfactual structural vibration sequence according to time position to form a counterfactual virtual sequence pair.

[0019] As a preferred embodiment of the method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in this invention, the specific steps for generating structural response residual data are as follows:

[0020] By establishing a one-to-one data alignment relationship between structural response observation data and counterfactual virtual sequence pairs according to the motion stage, a frame-level comparative feature group is formed.

[0021] Calculate the residuals of the frame-level comparison feature groups within the same frame, perform residual consistency verification, and output structural response residual data.

[0022] As a preferred embodiment of the method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in this invention, the specific steps for generating a set of candidate abnormal segments are as follows:

[0023] The structural response residual data is divided into multiple residual analysis segments according to the motion stage and corresponding time interval. The residual amplitude fluctuation, low-frequency baseline fluctuation and frequency band energy dispersion of each residual analysis segment are statistically analyzed to generate the noise intensity index corresponding to each residual analysis segment.

[0024] Analyze the continuity of residual changes at adjacent time points for each residual analysis segment and the concentration of residual peaks over time to generate residual stability indices;

[0025] The reliability of each residual analysis segment is ranked based on its noise intensity index and residual stability index, and then merged according to temporal adjacency to form a set of candidate abnormal segments.

[0026] As a preferred embodiment of the method for detecting abnormal noise from a car seat horizontal drive under noise interference environment as described in this invention, the specific steps for outputting the initial set of abnormal noise events are as follows:

[0027] Structural response residual data corresponding to candidate anomaly segments are extracted according to time intervals, and vibration residual sequences and sound residual sequences are separated according to observation channel type;

[0028] The range of residual amplitude variation, concentration of residual peaks, and residual energy distribution characteristics of the vibration residual sequence and sound residual sequence of each candidate abnormal segment are statistically analyzed to generate structural vibration residual index and sound residual index.

[0029] Consistency assessment is performed on the structural vibration residual index and sound residual index of each candidate abnormal segment, and the segments are organized in chronological order to output the initial set of abnormal noise events.

[0030] As a preferred embodiment of the method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in this invention, the specific steps for constructing and verifying the motion trajectory combination are as follows:

[0031] Read the displacement change range, velocity change range, and acceleration change range of each initial abnormal noise event in the initial abnormal noise event set within each time interval to form an abnormal noise trigger interval set for each initial abnormal noise event;

[0032] By extracting the structural vibration residual index, sound residual index and trigger interval duration corresponding to each initial abnormal noise event, the set of abnormal noise trigger intervals is sorted to generate a trigger priority sequence;

[0033] Read the displacement change range, velocity change range, and acceleration change range of each abnormal noise trigger interval in sequence according to the trigger priority sequence, calculate the center value of each range, and generate a combination of verification motion trajectories.

[0034] As a preferred embodiment of the method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in this invention, the specific steps for calculating and verifying the reproducibility evaluation index of the motion trajectory combination are as follows:

[0035] By discretizing the trajectory, the combination of verification motion trajectories is converted into drive control commands. During the execution of the drive control commands, the structural response observation data of the car seat horizontal drive is continuously collected, and the response difference between the drive control command and the corresponding counterfactual virtual sequence pair is calculated to generate a reproducibility matching index.

[0036] Repeatedly execute the same drive control command to form a corresponding set of verification abnormal noise events. Calculate the overlap ratio of the time intervals of the reproducible events in the set of verification abnormal noise events and the residual fluctuation range, and quantify them together with the reproducibility matching degree to generate a reproducibility evaluation index.

[0037] As a preferred embodiment of the method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in this invention, the specific steps for the output abnormal noise status evaluation and recording are as follows:

[0038] Based on the reproduction evaluation indicators, the reproduction reliability of the initial abnormal noise event set is determined, and a confirmation status is assigned to each initial abnormal noise event.

[0039] Summarize the confirmation status, corresponding time interval, reproducibility matching index, and reproducibility stability index of each initial abnormal noise event to generate an abnormal noise status assessment record.

[0040] The beneficial effects of this invention are as follows: by constructing counterfactual virtual sequence pairs, the essential separation of complex background noise and normal motion acoustic characteristics is achieved, improving the accuracy of detection under strong noise background. Simultaneously constructing and verifying motion trajectory combinations transforms one-time, passive observation and detection into an actively controllable and repeatable engineering experiment, realizing the leap from "statistical suspicion" to "engineering confirmation" in detection conclusions, greatly improving the reliability and credibility of evaluation results, as well as the efficiency of fault diagnosis and repair. Attached Figure Description

[0041] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a flowchart of a method for detecting abnormal noise from a car seat horizontal actuator under noise interference conditions.

[0043] Figure 2 The flowchart for generating counterfactual virtual sequence pairs.

[0044] Figure 3 A flowchart for generating a set of candidate abnormal fragments.

[0045] Figure 4 This is a flowchart for confirming and evaluating abnormal noises. Detailed Implementation

[0046] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0047] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0048] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0049] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a method for detecting abnormal noise from a car seat horizontal drive under noise interference environment, including the following steps:

[0050] S1. Collect structural response observation data of the car seat horizontal drive operation process in a noisy environment, divide the motion stage of the structural response observation data, and form an aligned slice record.

[0051] S1.1. Map the sampling time position of the structural response observation data of each acquisition channel to the same continuous time axis, and perform equal-interval time position resampling processing to generate a synchronous sampling data sequence.

[0052] It should be noted that the structural response observation data is used to characterize the structural vibration response generated by the automotive seat horizontal actuator during operation and the acoustic response formed by the radiation of structural vibration to the air medium. The structural response observation data includes structural vibration observation data, acoustic radiation observation data, and drive motion state observation data. Among them, the structural vibration observation data is collected by vibration sensors installed at the transmission housing, guide rail connection position, or fixed support position of the automotive seat horizontal actuator, and is used to record the vibration acceleration change information generated by the structure during operation. The acoustic radiation observation data is collected by acoustic sensors installed at the near field position of the automotive seat installation area, and is used to record the sound pressure change information formed by the propagation of structural vibration to the air. The drive motion state observation data is obtained by reading the displacement detection device or drive feedback signal of the automotive seat horizontal actuator, and is used to record the displacement change information of the actuator during operation and its time correspondence.

[0053] A unified time axis is established starting from the earliest sampling time, and the sampling time positions of the structural response observation data of all acquisition channels are mapped to the same continuous time axis. Based on the unified time axis, a fixed time interval is set (based on the target sampling interval being less than or equal to the minimum value among the original sampling intervals of each acquisition channel, such as 1 millisecond) as the target sampling interval. Time position interpolation calculation is performed on the structural response observation data of each acquisition channel so that the data of each acquisition channel has corresponding observation values ​​at the same time position. After the unified time position interpolation is completed, the observation values ​​of each acquisition channel at the same time position are organized in chronological order to form a synchronous sampling data sequence with a unified sampling time interval.

[0054] It should also be noted that the specific process of time position interpolation calculation is as follows: for any target sampling time in the structural response observation data of each acquisition channel, the nearest sampling time point before the target sampling time and the nearest sampling time point after the target sampling time in the sampling time marker sequence are retrieved and recorded as the previous sampling time point and the next sampling time point, respectively. The ratio of the difference between the target sampling time and the previous sampling time point and the difference between the next sampling time point and the previous sampling time point is used as the time scaling factor. The time scaling factor is used as the weight to perform a weighted summation on the previous observation value and the next observation value to obtain the observation value at the current sampling time position.

[0055] S1.2. By statistically analyzing the changes in driving displacement and motion velocity in the synchronously sampled data sequence, calculate the motion activity index within a continuous time window, divide the motion stages according to the motion activity index, and output the aligned slice record.

[0056] It should be noted that continuous observations of the driving motion state observation data at a unified time position in the synchronous sampling data sequence are taken as displacement observations. The ratio of the difference in displacement observations at adjacent time positions to the target sampling interval is taken as the change in motion velocity. Within a continuous time window, the fluctuation range of the displacement observation value change and the change in motion velocity are statistically analyzed, and the statistical results are normalized to generate a motion activity index. The length of the continuous time window is not less than a certain multiple of the target sampling interval (e.g., not less than 10 times). The continuous value changes of the motion activity index are read along a unified time axis. When the change in the motion activity index between adjacent time intervals exceeds the average fluctuation range within the continuous time window, or when the motion activity index remains in an interval below the average fluctuation range, the boundary of the motion stage is marked at the corresponding time position. The synchronous sampling data sequence is then divided into time intervals based on the motion stage boundary to form an aligned slice record.

[0057] It should also be noted that, within a continuous time window, the absolute value of the difference between each value of the activity activity index and the average value of the activity activity index is calculated, and the average of all absolute values ​​is taken to obtain the average deviation. The range of values ​​covered by the difference between the average value of the activity activity index and the average deviation, up to the sum of the average value of the activity activity index and the average deviation, is defined as the average fluctuation range.

[0058] S2. Extract the mechanical excitation features and stage kinematic features of each motion stage from the aligned slice record, calculate the counterfactual structural response of the mechanical excitation features and stage kinematic features, and output the counterfactual virtual sequence pairs.

[0059] S2.1 Read the structural response observation sequence corresponding to each motion stage in the aligned slice record, and calculate the displacement change, instantaneous velocity change and stage acceleration per unit time to generate stage kinematic characteristics.

[0060] It should be noted that the aligned slice record contains the structural response observation sequence corresponding to each motion stage and the corresponding unified time position order. The displacement observation value of the driving motion state observation data in the structural response observation sequence of any motion stage is read. The ratio of the difference between the displacement observation values ​​of adjacent time positions to the target sampling interval is taken as the displacement change per unit time. The ratio of the difference between the displacement change per unit time of adjacent time positions to the target sampling interval is taken as the instantaneous motion velocity change. The ratio of the difference between the instantaneous motion velocity change of adjacent time positions to the target sampling interval is taken as the stage acceleration. The displacement change per unit time, the instantaneous motion velocity change, and the stage acceleration are organized in a unified time position order to form the stage kinematic characteristics.

[0061] S2.2. By using short-time energy statistics and local peak detection, identify the high-energy transient impact locations of the structural response observation sequences corresponding to each motion stage in the aligned slice records, and statistically analyze the degree of change in vibration response amplitude, the intensity of local transient impact response, and the frequency band energy distribution to generate mechanical excitation characteristics.

[0062] It should be noted that each motion stage in the aligned slice record corresponds to the structural vibration observation data in the structural response observation sequence. The cumulative value of the square of the vibration observation values ​​within the continuous sliding time interval is used as the short-time energy value. The short-time energy values ​​are sorted in chronological order. When the short-time energy value is in the higher segment of the short-time energy ranking result of the current motion stage (the proportion is determined according to the short-time energy distribution of the current motion stage, such as the top 30% of the ranking result), and the short-time energy value at the corresponding time position is greater than the short-time energy value at the adjacent time position, the corresponding time position is marked as a high-energy transient impact position. Within the time interval corresponding to each high-energy transient impact position, the difference between the maximum and minimum amplitude values ​​of the structural vibration observation data is used as the degree of change in vibration response amplitude. The maximum value among the short-time energy values ​​in the current time interval is read as the local transient impact response intensity. The energy proportion corresponding to different frequency segments is counted as the frequency band energy distribution. The degree of change in vibration response amplitude, local transient impact response intensity, and frequency band energy distribution are organized in a unified time position order to form mechanical excitation characteristics.

[0063] S2.3 Calculate the structural excitation correction based on the synchronous change relationship between mechanical excitation characteristics and stage kinematic characteristics, and output the equivalent structural excitation sequence.

[0064] It should be noted that a one-to-one correspondence between the mechanical excitation characteristics and the kinematic characteristics of each motion stage is established sequentially along a unified time axis. The difference in values ​​of the mechanical excitation characteristics between adjacent time positions is taken as the change in mechanical excitation characteristics, and the difference in values ​​of the kinematic characteristics between corresponding adjacent time positions is taken as the change in kinematics. The changes in mechanical excitation characteristics and kinematics at each time position are organized sequentially according to the unified time axis to form a sequence of mechanical excitation characteristic changes and a sequence of kinematics changes. The ratio of the changes in mechanical excitation characteristics to the changes in kinematics is calculated at time positions where the change in kinematics is not zero, thus obtaining a sequence of change ratios.

[0065] The sequence of change ratios is sorted according to the size of the values. The difference between the change ratios corresponding to adjacent sorting positions is calculated to form an adjacent ratio difference sequence. The average value of the adjacent ratio difference sequence is calculated. In the sorted sequence of change ratios, the continuous range of change ratios where the adjacent ratio difference does not exceed the average value is identified as the main clustering interval of the change ratios. The ratio relationship corresponding to the main clustering interval is used as the proportional relationship of the change in mechanical excitation feature relative to the change in kinematics.

[0066] At each time point, the signs of the changes in mechanical excitation characteristics and kinematic changes are read. When the signs of the changes in mechanical excitation characteristics and kinematic changes are the same, the product of the original value of the mechanical excitation characteristic and the proportional relationship of the change is used as the enhancement correction amount. When the signs of the changes in mechanical excitation characteristics and kinematic changes are opposite, the product of the original value of the mechanical excitation characteristic and the proportional relationship of the change is determined as the suppression correction base value, and the direction is adjusted according to the correction direction opposite to the direction of change of the original value of the mechanical excitation characteristic to obtain the suppression correction amount. The range of values ​​covered by the maximum and minimum values ​​of the mechanical excitation characteristics in the current motion stage is determined as the overall value interval. The values ​​at each time point are compared... The corresponding enhancement or suppression correction is superimposed with the original value of the corresponding mechanical excitation feature to obtain candidate correction values ​​at each time position. When the candidate correction value is within the overall value range, it is used as the corrected mechanical excitation feature value. When the candidate correction value exceeds the overall value range, it is adjusted to the corresponding interval boundary value. The difference between the corrected mechanical excitation feature value and the original value of the corresponding mechanical excitation feature is determined as the structural excitation correction amount at each time position. The original value of the mechanical excitation feature and the structural excitation correction amount are superimposed according to a unified time axis order to generate an equivalent structural excitation sequence.

[0067] S2.4 Within the same motion phase, calculate the response delay and amplitude ratio between different observation positions in the structural response observation sequence to form a structural propagation parameter sequence.

[0068] It should be noted that, within the same motion phase, any observation position in the structural response observation sequence is selected as the reference observation position. Cross-correlation matching calculation is performed on the structural vibration observation data at any non-reference observation position and the structural vibration observation data at the reference observation position, and the time offset when the cross-correlation value reaches its maximum is determined as the response delay. After aligning the structural vibration observation data at the non-reference observation position according to the response delay, the amplitude ratio of the structural vibration observation data at the non-reference observation position and the structural vibration observation data at the reference observation position at the corresponding time position is calculated. The response delay and amplitude ratio are organized in a unified time position order to form a structural propagation parameter sequence.

[0069] The expression for calculating the response delay is:

[0070] ;

[0071] in, Indicates response delay. Represents a set of time offsets. Indicates the time offset as The cross-correlation values ​​at time, Indicates the time offset. Indicates the reference observation location in the time position index. Structural vibration observation values ​​were obtained at the location. Indicates the observation location index in time location The structural vibration observation values ​​were obtained at the location.

[0072] It should also be noted that the time offset represents the relative displacement of the structural vibration observation data at the non-reference observation position on the time axis relative to the structural vibration observation data at the reference observation position. It is used to characterize the time delay that occurs during the propagation of the structural vibration response from the reference observation position to the non-reference observation position.

[0073] S2.5. Perform time delay mapping and amplitude mapping on the equivalent structural excitation sequence according to the structural propagation parameter sequence to generate the counterfactual structural vibration sequence.

[0074] It should be noted that, using the reference observation position as the time base, the continuous values ​​of the equivalent structural excitation sequence on a unified time axis are read. Based on the response delay values ​​corresponding to each observation position in the structural propagation parameter sequence, the equivalent structural excitation sequence is time-shifted, shifting the effective time of the equivalent structural excitation on the time axis towards the propagation arrival time of the corresponding observation position, thus forming a time-delay mapping excitation sequence for each observation position. After completing the time-delay mapping, the amplitude ratio values ​​corresponding to each observation position are read, and amplitude scaling is performed on the values ​​of the time-delay mapping excitation sequence at each observation position, making the excitation values ​​at each time position proportional to the corresponding amplitude ratio, thus forming an amplitude mapping excitation sequence for each observation position. The amplitude mapping excitation sequences for each observation position are organized in a unified time axis order as the structural vibration response value sequence that should be generated at each observation position under no abnormal contact conditions, and arranged in time-position order to form a counterfactual structural vibration sequence.

[0075] S2.6. Statistically analyze the changes in acoustic radiation generated by the corresponding structural vibration in the counterfactual structural vibration sequence, generate a counterfactual structural radiation acoustic sequence, and pair it with the counterfactual structural vibration sequence according to the time position to form a counterfactual virtual sequence pair.

[0076] It should be noted that during the motion phase, the difference in structural vibration observation data between adjacent time positions is read along a unified time axis and used as the change in acoustic radiation coupled vibration. Simultaneously, the difference in acoustic radiation observation data between corresponding adjacent time positions is read along the unified time axis and used as the change in acoustic radiation. This forms a sequence of acoustic radiation coupled vibration changes and a sequence of acoustic radiation changes across all time positions in the current motion phase. The ratio of the acoustic radiation change to the acoustic radiation coupled vibration change at each time position is calculated to form a sequence of influence degree ratios. The median of the influence degree ratio sequences is taken as the influence degree value. This is based on the counterfactual structural vibration sequence. The counterfactual structural vibration changes at adjacent time positions are calculated along a unified time axis. The product of the counterfactual structural vibration changes and the influence degree values ​​is taken as the counterfactual acoustic radiation changes. All counterfactual acoustic radiation changes are integrated to form a counterfactual acoustic radiation change sequence. Then, a time-position cumulative operation is performed along the unified time axis to obtain the counterfactual structural radiation sound sequence. The counterfactual structural vibration sequence and the counterfactual structural radiation sound sequence are paired one by one according to the corresponding time positions of the unified time axis, so that each time position corresponds to a set of counterfactual structural vibration values ​​and counterfactual structural radiation sound values, forming a counterfactual virtual sequence pair.

[0077] S3. Frame-level temporal difference between statistical structural response observation data and counterfactual virtual sequence pairs is used to generate structural response residual data. The noise intensity of the structural response residual data is evaluated and credibility screening is performed to generate a set of candidate anomalous segments.

[0078] S3.1 Establish a one-to-one data alignment relationship between the structural response observation data and the counterfactual virtual sequence pairs according to the motion stage, forming a frame-level comparison feature group.

[0079] It should be noted that, within any motion phase, the corresponding observation values ​​of structural vibration observation data and acoustic radiation observation data at the current time position are read along a unified time axis at each time position. At the same time, the corresponding values ​​of the counterfactual structural vibration sequence and the counterfactual structural radiated sound sequence at the current time position are read from the counterfactual virtual sequence pair. The structural vibration observation value, counterfactual structural vibration value, acoustic radiation observation value, and counterfactual structural radiated sound value at the same time position are grouped according to the time position index, so that each time position forms a set of frame-level comparison features. The frame-level comparison features of all motion phases at each time position are integrated to form a frame-level comparison feature group.

[0080] S3.2 Calculate the residual of the frame-level comparison feature group within the same frame, perform residual consistency verification, and output structural response residual data.

[0081] It should be noted that each time position in the frame-level comparison feature group includes structural vibration observation values, counterfactual structural vibration values, acoustic radiation observation values, and counterfactual structural radiated acoustic values. At the same time position index, the difference between the structural vibration observation value and the counterfactual structural vibration value is taken as the frame-level structural vibration residual value, and the difference between the acoustic radiation observation value and the counterfactual structural radiated acoustic value is taken as the frame-level acoustic radiation residual value. The frame-level structural vibration residual values ​​and the frame-level acoustic radiation residual values ​​are arranged in a unified time axis order to form a frame-level structural vibration residual sequence and a frame-level acoustic radiation residual sequence. Within the corresponding time interval of the same motion stage, The residual differences between the frame-level structural vibration residual sequence and the frame-level acoustic radiation residual sequence at adjacent time positions are calculated separately. The proportion of time positions with the same sign of the residual difference between the two sequences is counted out of the total number of comparisons, which is used as the directional consistency ratio. At the same time, the time positions with the highest residual amplitude of the two sequences (such as the top 30%) are extracted to form their respective peak time sets. The proportion of the number of time positions that overlap in the two peak time sets is counted out of the number of peaks, which is used as the peak concentration degree. All frame-level structural vibration residual sequences and frame-level acoustic radiation residual sequences that pass the residual consistency check are summarized according to the motion stage and time order to form structural response residual data.

[0082] It should also be noted that, within the corresponding time interval of the same motion phase, the frame-level structural vibration residual sequence and the frame-level acoustic radiation residual sequence are subjected to random time position rearrangement, and the directional consistency ratio and peak concentration are recalculated after each random rearrangement. This process is repeated multiple times to obtain the random consistency value distribution. Then, the directional consistency ratio and peak concentration are calculated for the frame-level structural vibration residual sequence and the frame-level acoustic radiation residual sequence that have not undergone random time position rearrangement. When the calculated directional consistency ratio and peak concentration of the frame-level structural vibration residual sequence and the frame-level acoustic radiation residual sequence that have not undergone random time position rearrangement are both higher than the concentration range of the corresponding random consistency value distribution, the frame-level structural vibration residual sequence and the frame-level acoustic radiation residual sequence corresponding to the current motion phase are determined to have passed the residual consistency check.

[0083] It should also be noted that all values ​​in the random consistency value set are sorted according to the size of the values; the absolute value of the difference between each value in the sorted result and the average value of the random consistency value set is calculated to form a sequence of absolute difference values, and the sequence of absolute difference values ​​is sorted according to the size of the absolute difference values; the set of values ​​with the smallest absolute difference value is selected as the concentrated value interval of the random consistency value distribution.

[0084] S3.3. Divide the structural response residual data into multiple residual analysis segments according to the motion stage and corresponding time interval, and statistically analyze the residual amplitude fluctuation, low-frequency baseline fluctuation and frequency band energy dispersion of each residual analysis segment to generate the noise intensity index corresponding to each residual analysis segment.

[0085] It should be noted that the structural response residual data includes frame-level structural vibration residual sequences and frame-level acoustic radiation residual sequences arranged in a unified time axis order, and the aligned slice records give the time interval boundaries corresponding to each motion stage; within any motion stage, the frame-level structural vibration residual sequences and frame-level acoustic radiation residual sequences are segmented along the unified time axis according to continuous time intervals, so that each continuous time interval forms a residual analysis segment; all residual values ​​of the frame-level structural vibration residual sequences and frame-level acoustic radiation residual sequences in each residual analysis segment are read, and the difference between the maximum residual amplitude and the minimum residual amplitude is taken as the residual amplitude fluctuation degree; The continuous trend of the residual value sequence within the time interval is taken as the low-frequency baseline sequence, and the difference between the maximum and minimum values ​​of the low-frequency baseline sequence is taken as the low-frequency baseline fluctuation degree. The energy proportion of the residual value sequence in the current residual analysis segment in different frequency segments is statistically analyzed, and the dispersion of the energy proportion of each frequency segment relative to the average energy proportion (the ratio of standard deviation to mean) is taken as the frequency band energy dispersion degree. The residual amplitude fluctuation degree, the low-frequency baseline fluctuation degree, and the frequency band energy dispersion degree are normalized respectively, and the square root of the sum of the squares of the three after normalization is taken as the noise intensity index corresponding to the residual analysis segment.

[0086] S3.4 Analyze the continuity of residual changes at adjacent time positions for each residual analysis segment and the concentration of residual peaks over time to generate residual stability indices.

[0087] It should be noted that within each residual analysis segment, a sequence of absolute values ​​of the differences between residual values ​​at adjacent time positions in the frame-level structural vibration residual sequence is calculated along a unified time axis. The average of this absolute value sequence is used as the mean amplitude of the structural vibration residual variation. Simultaneously, a sequence of absolute values ​​of the differences between residual values ​​at adjacent time positions in the frame-level acoustic radiation residual sequence is calculated, and the average of this absolute value sequence is used as the mean amplitude of the acoustic radiation residual variation. The mean amplitudes of the structural vibration residual variation and the mean amplitudes of the acoustic radiation residual variation are then ratioed to the average absolute value of the residual values ​​within the corresponding residual analysis segment. The process yields the values ​​for the continuity of residual changes. Within the same residual analysis segment, the time positions of the top-ranked (e.g., the top 20%) residual amplitudes in the frame-level structural vibration residual sequence and the frame-level acoustic radiation residual sequence are extracted to form a peak time set. The ratio of the coverage span of the peak time set within the time interval of the residual analysis segment to the number of elements in the peak time set is used as the peak time set value. The values ​​for the continuity of residual changes and the peak time set values ​​are normalized, and the square root of the sum of the squares of the two normalized values ​​is used as the residual stability index of the current residual analysis segment.

[0088] S3.5. Based on the noise intensity index and residual stability index of each residual analysis segment, the credibility is sorted and merged according to the temporal adjacency relationship to form a set of candidate abnormal segments.

[0089] It should be noted that interval normalization is performed on the residual stability index of each residual analysis segment, and inverse normalization is performed on the noise intensity index of each residual analysis segment to generate normalized noise suppression value and normalized residual stability value. The product of the normalized residual stability value and the normalized noise suppression value is used as the confidence index. The residual analysis segments are sorted by confidence index value from high to low. The difference in confidence index between adjacent residual analysis segments after sorting is calculated, and the sorting position corresponding to the peak value of the difference is identified in the difference sequence as the truncation position. The residual analysis segments before the truncation position are identified as high-confidence residual analysis segments. The high-confidence residual analysis segments with adjacent time intervals or time intervals not exceeding the target sampling interval are subjected to interval union, and the corresponding frame-level structural vibration residual sequence and frame-level acoustic radiation residual sequence are simultaneously merged to form a candidate anomaly segment set.

[0090] S4. Calculate the structural vibration residual index and sound residual index of the candidate abnormal segment set, perform abnormal discrimination processing, output the initial abnormal sound event set, extract the abnormal sound discrimination features of the initial abnormal sound event set and sort them by priority, and construct the verification motion trajectory combination.

[0091] S4.1 Extract the structural response residual data corresponding to the candidate abnormal segments according to the time interval, and separate the vibration residual sequence and the sound residual sequence according to the observation channel type.

[0092] It should be noted that the residual value sequence arranged in a uniform time axis is read from the structural response residual data, and continuous residual values ​​within the corresponding time interval are extracted according to the time interval boundary of the candidate anomaly segment to form the structural response residual subsequence corresponding to the current candidate anomaly segment. The structural response residual data also retains the observation channel identification information corresponding to each residual value, where the observation channel type includes structural vibration observation channel and acoustic radiation observation channel. The structural response residual subsequence is traversed along the time sequence, and the residual values ​​are classified according to the observation channel identification. The residual values ​​belonging to the structural vibration observation channel are organized in time sequence to form a vibration residual sequence, and the residual values ​​belonging to the acoustic radiation observation channel are organized in time sequence to form a sound residual sequence.

[0093] S4.2. Statistically analyze the residual amplitude variation range, residual peak concentration degree and residual energy distribution characteristics of the vibration residual sequence and sound residual sequence of each candidate abnormal segment, and generate structural vibration residual index and sound residual index.

[0094] It should be noted that, for any candidate anomaly segment, the vibration residual sequence and sound residual sequence within the corresponding time interval are read, and all residual values ​​are extracted along a unified time axis. The difference between the maximum and minimum residual values ​​in the vibration residual sequence and the sound residual sequence is calculated as the range of residual amplitude variation corresponding to the candidate anomaly segment. The distribution density of the residual peak positions in each residual sequence within the time interval of the candidate anomaly segment is identified and statistically analyzed, and the ratio of the distribution width of the residual peak positions to the segment time span is used as a measure of the concentration of residual peaks. Frequency decomposition processing is performed on each residual sequence, and the proportion of residual energy in different frequency segments to the total residual energy is statistically analyzed to form residual energy distribution characteristics. The residual amplitude variation range, residual peak concentration, and residual energy distribution characteristics are organized in a unified time interval order to form structural vibration residual index and sound residual index.

[0095] S4.3. Perform a consistency assessment on the structural vibration residual index and sound residual index for each candidate abnormal segment, and organize them according to the time sequence of the candidate abnormal segments to output the initial abnormal sound event set.

[0096] It should be noted that, on a unified time axis, the differences between the structural vibration residual index and the sound residual index at each time position are calculated to form a residual difference sequence; the average value of the residual difference sequence is calculated, and the absolute value of the difference between the residual difference at each time position and the average value is calculated to form a residual difference absolute offset sequence; the residual difference absolute offset sequence is sorted according to the numerical size, and the difference between each adjacent value in the sorting result is used as the interval, and the average value of all intervals is used as the continuity judgment criterion. Continuous value segments in the sorting result where adjacent intervals do not exceed the continuity judgment criterion are identified, and the residual difference range covered by this continuous value segment is taken as the consistent change range; within the time interval corresponding to the candidate abnormal segment, the proportion of time positions where the residual difference falls into the consistent change range is statistically analyzed, and the candidate abnormal segments with a high proportion (e.g., 60%) are identified as candidate abnormal segments where the vibration response and sound response change are consistent; all candidate abnormal segments are organized in chronological order to form an initial abnormal event set.

[0097] S4.4 Read the displacement change range, velocity change range and acceleration change range of each initial abnormal noise event in the initial abnormal noise event set within each time interval, and form the abnormal noise trigger interval set of each initial abnormal noise event.

[0098] It should be noted that the time interval boundary of each initial abnormal noise event in the initial abnormal noise event set is read, and the displacement change range, velocity change range and acceleration change range in the corresponding time interval are extracted based on the motion state observation data in that time interval. These are then organized according to the time interval of the initial abnormal noise event to form an abnormal noise trigger interval set corresponding to each initial abnormal noise event.

[0099] S4.5 By extracting the structural vibration residual index, sound residual index and trigger interval duration corresponding to each initial abnormal noise event, the set of abnormal noise trigger intervals is sorted to generate a trigger priority sequence.

[0100] It should be noted that the structural vibration residual index, sound residual index, and trigger interval duration for each initial abnormal noise event are read within the corresponding time interval, and the three types of values ​​are matched one-to-one under a unified time interval identifier. The sets of structural vibration residual index values, sound residual index values, and trigger interval duration values ​​for all initial abnormal noise events are sorted from smallest to largest, and the relative position of each initial abnormal noise event in the three sorting results is used as the sorting rank. The three sorting ranks corresponding to the same initial abnormal noise event are jointly compared, and the overall priority of the three sorting ranks (the higher the ranking, the higher the priority) is used as the triggering priority of the initial abnormal noise event. All initial abnormal noise events are sorted from high to low triggering priority to generate a triggering priority sequence.

[0101] S4.6. Read the displacement change range, velocity change range and acceleration change range of each abnormal noise trigger interval in sequence according to the trigger priority sequence, calculate the center value of each range, and generate the verification motion trajectory combination.

[0102] It should be noted that, according to the trigger priority sequence, the set of abnormal noise trigger intervals corresponding to each initial abnormal noise event is read sequentially, and the displacement change range, velocity change range, and acceleration change range corresponding to each abnormal noise trigger interval are extracted; within each abnormal noise trigger interval, the midpoint between the upper and lower bounds of the displacement change range is taken as the target displacement value, the midpoint between the upper and lower bounds of the velocity change range is taken as the target velocity value, and the midpoint between the upper and lower bounds of the acceleration change range is taken as the target acceleration value. The target displacement value, target velocity value, and target acceleration value corresponding to each abnormal noise trigger interval are combined in a unified time order; while maintaining the continuity of the time order between adjacent abnormal noise trigger intervals, the combined values ​​are sequentially spliced ​​to form a combination of verification motion trajectories that can cover the central state path of all abnormal noise trigger intervals.

[0103] S5. Calculate and verify the reproducibility evaluation index of the motion trajectory combination, and perform abnormal noise confirmation judgment on the initial abnormal noise event set according to the reproducibility evaluation index, and output the abnormal noise status evaluation record.

[0104] S5.1. The combination of verification motion trajectories is converted into drive control commands through trajectory discretization. During the execution of drive control commands, structural response observation data of the car seat horizontal drive is continuously collected, and the response difference between the drive control command and the corresponding counterfactual virtual sequence pair is calculated to generate a reproducibility matching index.

[0105] It should be noted that, in chronological order, the ratio of the difference in target displacement values ​​to the corresponding time interval in the verification motion trajectory combination is used as the displacement change rate value, the ratio of the difference in target velocity values ​​to the corresponding time interval is used as the velocity change rate value, and the ratio of the difference in target acceleration values ​​to the corresponding time interval is used as the acceleration change rate value. The displacement change rate value, velocity change rate value, and acceleration change rate value corresponding to each time position are organized in chronological order and converted into drive execution quantity change commands to form a drive control command sequence. During the execution of the drive control command sequence, structural response observation data of the vehicle seat horizontal drive at a unified time position are continuously collected and organized in chronological order to form a verification response time sequence. At each time position, the verification response time sequence is compared with the counterfactual virtual sequence at the corresponding time position to perform differential calculation, resulting in a response difference sequence. The amplitude distribution characteristics and change persistence characteristics of the response difference sequence in the continuous time interval are statistically analyzed to generate a reproducibility matching index.

[0106] S5.2 Repeatedly execute the same drive control command to form a corresponding set of verification abnormal noise events. Calculate the overlap ratio of the time intervals of the reproducible events in the set of verification abnormal noise events and the residual fluctuation range, and quantify them together with the reproducibility matching degree to generate a reproducibility evaluation index.

[0107] It should be noted that the same drive control command is repeatedly executed, and structural response observation data is collected during each execution to form a corresponding verification response time sequence. Simultaneously, the corresponding response difference sequence is calculated according to a uniform time position order in each execution. Within the response difference sequence corresponding to each execution, the time interval where the response difference continuously increases and remains stable is identified as the verification abnormal noise event time interval. All verification abnormal noise event time intervals formed by each execution constitute the verification abnormal noise event set. The overlapping portions of the time intervals in different executions are calculated on a uniform time axis. The proportion of the total coverage area of ​​the corresponding time interval is used to characterize the overlap ratio of the time intervals of the reproduced event. At the same time, the continuous value changes of the response difference sequence are read within the corresponding time interval of each execution, and the range of change of the difference between the maximum and minimum values ​​of the response difference within the time interval is statistically analyzed between different executions to characterize the residual fluctuation range. The overlap ratio of the time intervals of the reproduced event and the residual fluctuation range are normalized and expressed according to a unified dimension, and combined with the reproduction matching degree index for joint quantification to generate a reproduction evaluation index to characterize the stability of abnormal response reproduction under repeated execution of drive control instructions.

[0108] S5.3. Based on the reproduction evaluation index, perform a reproduction reliability determination on the initial abnormal noise event set and assign a confirmation status to each initial abnormal noise event.

[0109] It should be noted that the sequence of recurrence evaluation index values ​​within the corresponding time interval of each initial abnormal noise event is read, and the difference between the maximum and minimum values ​​of the recurrence evaluation index within the corresponding time interval is taken as the value distribution range. At the same time, the absolute value of the difference between the values ​​of the recurrence evaluation index at adjacent time positions within the time interval is calculated, and the average of all absolute values ​​is taken as the continuous change amplitude representation. The value distribution range and the continuous change amplitude representation are compared with the corresponding statistical results of the recurrence evaluation index within the overall time axis. When the value distribution range of the recurrence evaluation index within the time interval corresponding to an initial abnormal noise event is smaller than the overall distribution range, and the continuous change amplitude is lower than the overall continuous change amplitude, the initial abnormal noise event is determined to have recurrence stability characteristics and is assigned a confirmed state. Initial abnormal noise events that do not meet the corresponding conditions are assigned an unconfirmed state.

[0110] S5.4 Summarize the confirmation status, corresponding time interval, reproducibility matching index and reproducibility stability index of each initial abnormal noise event, and generate an abnormal noise status assessment record.

[0111] It should be noted that the time intervals corresponding to each initial abnormal noise event in the initial abnormal noise event set are read, and the confirmation status, reproduction matching index, and reproduction stability index corresponding to the current time interval are extracted one by one. According to the unified time axis order, the start position of the time interval, the end position of the time interval, the confirmation status value, the reproduction matching index value, and the reproduction stability index value corresponding to each initial abnormal noise event are correlated, and the structured arrangement is performed in chronological order, so that each initial abnormal noise event forms an abnormal noise status evaluation record with time positioning information and reproduction judgment information.

[0112] In summary, this invention achieves the essential separation of complex background noise from the acoustic features of normal motion by constructing counterfactual virtual sequence pairs, thereby improving the accuracy of detection in strong noise environments. Simultaneously, it constructs and verifies motion trajectory combinations, transforming one-off, passive observation and detection into an actively controllable and repeatable engineering experiment. This enables a leap from "statistically suspected" to "engineering confirmed" detection conclusions, greatly improving the reliability and credibility of evaluation results, as well as the efficiency of fault diagnosis and repair.

[0113] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting abnormal noise from a car seat horizontal actuator under noise interference conditions, characterized in that, include: Structural response observation data of a car seat horizontal drive during operation were collected in a noisy environment. The motion stages of the structural response observation data were divided to form an aligned slice record. Extract the mechanical excitation features and stage kinematic features for each motion stage from the aligned slice record, calculate the counterfactual structural responses of the mechanical excitation features and stage kinematic features, and output counterfactual virtual sequence pairs. The specific steps are as follows: The structural excitation correction is calculated based on the synchronous change relationship between mechanical excitation characteristics and stage kinematic characteristics, and the equivalent structural excitation sequence is output. Within the same motion phase, the response delay and amplitude ratio between different observation positions in the structural response observation sequence are calculated to form a structural propagation parameter sequence; The equivalent structural excitation sequence is subjected to time delay mapping and amplitude mapping according to the structural propagation parameter sequence to generate a counterfactual structural vibration sequence. The changes in acoustic radiation generated by the corresponding structural vibration in the counterfactual structural vibration sequence are statistically analyzed to generate a counterfactual structural radiation acoustic sequence, which is then paired with the counterfactual structural vibration sequence according to the time position to form a counterfactual virtual sequence pair. The frame-level temporal difference between the statistical structural response observation data and the counterfactual virtual sequence pairs is used to generate structural response residual data. The noise intensity of the structural response residual data is evaluated and confidence screening is performed to generate a set of candidate anomalous segments. Calculate the structural vibration residual index and sound residual index of the candidate abnormal segment set, perform abnormal discrimination processing, output the initial abnormal sound event set, extract the abnormal sound discrimination features of the initial abnormal sound event set and sort them by priority, and construct the verification motion trajectory combination; Calculate and verify the reproducibility evaluation index of the motion trajectory combination, and perform abnormal noise confirmation judgment on the initial abnormal noise event set according to the reproducibility evaluation index, and output the abnormal noise status evaluation record.

2. The method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in claim 1, characterized in that, The specific steps for forming the aligned slice record are as follows: The sampling time positions of the structural response observation data from each acquisition channel are mapped to the same continuous time axis, and equal-interval time position resampling processing is performed to generate a synchronous sampling data sequence. By statistically analyzing the changes in driving displacement and motion velocity in the synchronously sampled data sequence, the motion activity index within a continuous time window is calculated. Based on the motion activity index, motion stages are divided, and aligned slice records are output.

3. The method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in claim 2, characterized in that, The specific steps for extracting the mechanical excitation features and stage kinematic features of each motion stage from the aligned slice record are as follows: Read the structural response observation sequence corresponding to each motion stage in the aligned slice record, and calculate the displacement change, instantaneous velocity change and stage acceleration per unit time to generate stage kinematic characteristics; By using short-time energy statistics and local peak detection, the high-energy transient impact locations of the structural response observation sequences corresponding to each motion stage in the aligned slice records are identified, and the degree of change in vibration response amplitude, local transient impact response intensity and frequency band energy distribution are statistically analyzed to generate mechanical excitation characteristics.

4. The method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in claim 3, characterized in that, The specific steps for generating the structural response residual data are as follows: By establishing a one-to-one data alignment relationship between structural response observation data and counterfactual virtual sequence pairs according to the motion stage, a frame-level comparative feature group is formed. Calculate the residuals of the frame-level comparison feature groups within the same frame, perform residual consistency verification, and output structural response residual data.

5. The method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in claim 4, characterized in that, The specific steps for generating the candidate abnormal fragment set are as follows: The structural response residual data is divided into multiple residual analysis segments according to the motion stage and corresponding time interval. The residual amplitude fluctuation, low-frequency baseline fluctuation and frequency band energy dispersion of each residual analysis segment are statistically analyzed to generate the noise intensity index corresponding to each residual analysis segment. Analyze the continuity of residual changes at adjacent time points for each residual analysis segment and the concentration of residual peaks over time to generate residual stability indices; The reliability of each residual analysis segment is ranked based on its noise intensity index and residual stability index, and then merged according to temporal adjacency to form a set of candidate abnormal segments.

6. The method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in claim 1, characterized in that, The specific steps for outputting the initial set of abnormal noise events are as follows: Structural response residual data corresponding to candidate anomaly segments are extracted according to time intervals, and vibration residual sequences and sound residual sequences are separated according to observation channel type; The range of residual amplitude variation, concentration of residual peaks, and residual energy distribution characteristics of the vibration residual sequence and sound residual sequence of each candidate abnormal segment are statistically analyzed to generate structural vibration residual index and sound residual index. Consistency assessment is performed on the structural vibration residual index and sound residual index of each candidate abnormal segment, and the segments are organized in chronological order to output the initial set of abnormal noise events.

7. The method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in claim 6, characterized in that, The specific steps for constructing and verifying the motion trajectory combination are as follows: Read the displacement change range, velocity change range, and acceleration change range of each initial abnormal noise event in the initial abnormal noise event set within each time interval to form an abnormal noise trigger interval set for each initial abnormal noise event; By extracting the structural vibration residual index, sound residual index and trigger interval duration corresponding to each initial abnormal noise event, the set of abnormal noise trigger intervals is sorted to generate a trigger priority sequence; Read the displacement change range, velocity change range, and acceleration change range of each abnormal noise trigger interval in sequence according to the trigger priority sequence, calculate the center value of each range, and generate a combination of verification motion trajectories.

8. The method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in claim 1, characterized in that, The specific steps for calculating and verifying the reproducibility evaluation index of the motion trajectory combination are as follows: By discretizing the trajectory, the combination of verification motion trajectories is converted into drive control commands. During the execution of the drive control commands, the structural response observation data of the car seat horizontal drive is continuously collected, and the response difference between the drive control command and the corresponding counterfactual virtual sequence pair is calculated to generate a reproducibility matching index. Repeatedly execute the same drive control command to form a corresponding set of verification abnormal noise events. Calculate the overlap ratio of the time intervals of the reproducible events in the set of verification abnormal noise events and the residual fluctuation range, and quantify them together with the reproducibility matching degree to generate a reproducibility evaluation index.

9. The method for detecting abnormal noise from a car seat horizontal actuator under noise interference environment as described in claim 1, characterized in that, The specific steps for evaluating and recording the output abnormal noise status are as follows: Based on the reproduction evaluation indicators, the reproduction reliability of the initial abnormal noise event set is determined, and a confirmation status is assigned to each initial abnormal noise event. Summarize the confirmation status, corresponding time interval, reproducibility matching index, and reproducibility stability index of each initial abnormal noise event to generate an abnormal noise status assessment record.

Images