A high-precision transmission lead screw defect detection method and system for automotive seats

Abstract

The present invention discloses a method and system for detecting defects of high-precision transmission lead screws of automobile seats, which relates to the technical field of lead screw defects. By monitoring and analyzing the defects in terms of the transmission stability and self-locking property of the lead screw, an operation stability coefficient Zwxs and a self-locking coefficient Zsxs are constructed and used as key input parameters, which are input into a defect prediction model based on the convolutional neural network CNN. The CNN model utilizes the non-linear fitting ability trained by big data. After linear normalization processing, it can output a balance evaluation index Pgzs, which quantifies the balance between the transmission stability and self-locking property of the lead screw, realizes the comprehensive evaluation of the lead screw between the transmission performance and self-locking performance, and makes the prediction result more comprehensive and reliable. In addition, by setting a performance balance threshold Q, this method matches and analyzes the balance evaluation index Pgzs with it, realizing the automatic screening and accurate judgment of the lead screw performance.

Description

Technical Field

[0001] The present invention relates to the technical field of lead screw defects, and specifically to a method and system for detecting defects of a high-precision transmission lead screw for automotive seats. Background Technique

[0002] In recent years, with the increasing requirements of the automotive industry for comfort and safety, the demand for the adjustability and precision control of automotive seat systems has increased significantly. In order to achieve precise adjustment and position control of the seats, the transmission lead screw is an important mechanical transmission component in the automotive seat system, and it is required to have stable transmission performance and good self-locking property. Specifically, the high-precision transmission lead screw used in automotive seats needs to find a balance between maintaining stability and self-locking property to ensure stable operation during seat adjustment and to be able to maintain the set position without external force. Therefore, in practical applications, defect detection of the lead screw has become a key task to ensure the reliability of the seat transmission system, especially for evaluating and analyzing the balance between its self-locking property and stability.

[0003] The deficiencies of the current defect detection methods mainly stem from the limitations of traditional methods in high-precision and dynamic detection. Most of these detection means are based on static analysis, lacking intelligent feature extraction and high-precision real-time monitoring capabilities. And once the defects of the transmission lead screw are not detected in time, especially the situation where the self-locking property and stability are unbalanced, it may lead to problems such as abnormal noise, position drift, or even jamming during seat adjustment. These abnormalities will not only affect the user experience of the seat, but may also seriously affect the safety of passengers, especially in the case of a vehicle impact or sudden stop. Therefore, it is crucial to be able to dynamically detect and comprehensively analyze the balance of the transmission lead screw in terms of self-locking property and stability, and to detect and handle defects in time to ensure the reliability and safety of the seat transmission system. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a method and system for detecting defects of a high-precision transmission lead screw for automotive seats, which solves the problems in the above background technique.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for detecting defects of a high-precision transmission lead screw for automotive seats, including the following steps,

[0006] S1. Rotate the transmission lead screw and capture the video stream in real time from multiple angles to capture the dynamic characteristics of the transmission lead screw. After image processing, a detection data set is generated. Based on the detection data set, it is judged whether the actual rotation axis of the transmission lead screw used in the automotive seat coincides with the standard axis. If not, a sorting instruction is triggered;

[0007] S2. If they coincide, further defect monitoring and analysis of the transmission screw is performed. Based on the detection data set, the transmission stability of the transmission screw when the car seat is adjusted is evaluated to construct the operation stability coefficient Zwxs. At the same time, the self-locking coefficient Zsxs of the transmission screw is constructed by analyzing the appearance conditions of the transmission screw.

[0008] S3. Use convolutional neural network technology to build a defect prediction model, and input the operation stability coefficient Zwxs and the self-locking coefficient Zsxs of the corresponding transmission screw into the defect prediction model. After linear normalization, fit the output balance evaluation index Pgzs;

[0009] S4. Pre-set the performance balance threshold Q and match it with the balance evaluation index Pgzs for analysis to comprehensively judge the balance state between the self-locking property and the transmission stability of the transmission screw. Based on the balance state, comprehensively judge whether the transmission screw installed in the car seat has any defects in performance.

[0010] Preferably, the specific steps of S1 include:

[0011] S11, rotating the transmission screw in advance, and while the transmission screw is rotating, dynamically acquiring a video stream through a multi-view image acquisition device, and extracting a plurality of groups of image frames from the video stream;

[0012] S12. Based on the several groups of image frames obtained in S11, the noise components are removed by using bilateral filtering method, and the pixel values in the several groups of image frames are adjusted to be uniformly distributed by using histogram equalization, and the several groups of image frames after image processing are aggregated into a detection data set, and the detection data set is subjected to feature extraction to obtain the measurement period T, the instantaneous value v(t) of the vibration velocity, the thread angle, the number of intervals J between the threads, and the pitch Jjz at the corresponding interval, and according to the standard transmission screw, the pitch Jjz between the internal threads of the standard transmission screw is obtained. ideal And by applying pressure to the transmission screw, the friction coefficient Mc is obtained;

[0013] S13, pre-divide the transmission screw into two sections evenly and mark them as the front end section and the rear end section respectively, and perform image extraction on several groups of image frames processed by step S12 to obtain the center point coordinates of the front end section and the rear end section of the transmission screw at four angles of 0°, 90°, 180° and 270° respectively.

[0014] Preferably, the specific step S1 also includes:

[0015] S14. Based on the central point coordinates of the front-end cross-section and the rear-end cross-section of the transmission lead screw at four angles of 0°, 90°, 180°, and 270° obtained in S13, take the average value of the four angles on each cross-section to obtain the central point of the corresponding cross-section in three-dimensional space, and based on the central point of the corresponding cross-section in three-dimensional space, construct the actual rotation axis of the transmission lead screw The actual rotation axis of the transmission lead screw Is constituted by the following equation:

[0016]

[0017] In the formula, Represents the three-dimensional coordinates of any point on the actual rotation axis; Represents the starting point coordinates on the actual rotation axis; Represents the direction vector of the actual rotation axis; d represents a scalar;

[0018] In three-dimensional space, we usually use a point (i.e., the starting point) and a direction vector to define a straight line. This straight line (i.e., the rotation axis) can extend to the entire length of the lead screw;

[0019] S141. Analyze the offset trend of the actual rotation axis of the transmission lead screw in three-dimensional space through the vector change between the central points of the front-end cross-section and the rear-end cross-section, calculate the vector from the front end to the rear end, and record it as the direction vector of the actual rotation axis The direction vector of the actual rotation axis Is obtained through the following method: Wherein, I2 represents the central point of the rear-end cross-section, and I1 represents the central point of the front-end cross-section;

[0020] S15. Based on the actual rotation axis of the transmission lead screw obtained in S14 And combined with the standard axis, compare the actual rotation axis With the direction deviation between the standard axis to obtain the included angle θ between the direction vectors of the two axes, which is specifically obtained through the following method:

[0021]

[0022] In the formula, Represents the direction vector of the standard axis; Represents the modulus of the direction vector of the actual rotation axis; Represents the modulus of the direction vector of the standard axis; Represents the direction vector of the actual rotation axis; Is the dot product of the direction vector of the actual rotation axis and the direction vector of the standard axis; cos(θ) represents the actual rotation axis Cosine value of the deviation from the standard axis;

[0023] S16. Preset a safety threshold. By comparing the safety threshold with the cosine value cosθ of the deviation between the actual rotation axis and the standard axis, to determine whether the actual rotation axis of the transmission lead screw used in the vehicle seat coincides with the standard axis. The specific judgment content is as follows:

[0024] If the actual rotation axis and the cosine value cos(θ) of the deviation from the standard axis fall within the safety threshold, it is determined that the actual rotation axis of the transmission lead screw used in the vehicle seat coincides with the standard axis. At this time, the transmission lead screw is further analyzed;

[0025] If the actual rotation axis and the cosine value cos(θ) of the deviation from the standard axis do not fall within the safety threshold, it is determined that the actual rotation axis of the transmission lead screw used in the vehicle seat does not coincide with the standard axis. At this time, a sorting instruction is triggered;

[0026] S17. When the sorting instruction is received, the current transmission lead screw is transferred to the area to be inspected manually.

[0027] Preferably, the specific steps of S1 further include:

[0028] S18. If it is determined that the actual rotation axis of the transmission lead screw used in the vehicle seat coincides with the standard axis, at this time, in combination with the center point of the corresponding cross-section obtained in S14 in the three-dimensional space, analyze the position offset between the actual rotation axis of the transmission lead screw used in the vehicle seat and the standard axis to obtain the eccentricity Pxj. The eccentricity Pxj is obtained by the following method:

[0029]

[0030] In the formula, n represents the number of center points obtained from different cross-section images, i = 1, 2, 3,...., n, I i represents the center point obtained from the i-th cross-section image, represents the standard axis, represents the actual rotation axis The vertical distance between the center point obtained from the i-th cross-section image on the actual rotation axis and the standard axis

[0031] Preferably, the specific steps of S2 include:

[0032] S21. Based on the detection data set, evaluate the transmission stability of the transmission lead screw when the vehicle seat is adjusted in position. After linear normalization, construct the operation stability coefficient Zwxs. The operation stability coefficient Zwxs is obtained through the following formula:

[0033]

[0034] In the formula, Zfz represents the vibration amplitude; Pxj represents the eccentricity; Cx represents the transmission efficiency; α and β are both weight values; C represents the first correction constant. Among them, the specific values of α and β are set by the user according to the situation.

[0035] S211. The vibration amplitude Zfz is obtained through the following formula:

[0036]

[0037] In the formula, T represents the measurement period; v(t) represents the instantaneous value of the vibration velocity.

[0038] S212. The transmission efficiency Cx is obtained through the following formula:

[0039]

[0040] In the formula, Mc represents the friction coefficient; θ represents the thread angle; cos(θ) represents the cosine value of the thread angle; sin(θ) represents the sine value of the thread angle.

[0041] Preferably, the specific steps of S2 further include:

[0042] S22. Based on the detection data set, analyze the appearance conditions of the transmission lead screw. After linear normalization, construct the self-locking coefficient Zsxs of the transmission lead screw. The self-locking coefficient Zsxs is obtained through the following formula:

[0043]

[0044] In the formula, Bccd represents the roughness; θ represents the thread angle; Lpyz represents the pitch deviation factor; a1, a2, and a3 are all weight values. Among them, the specific values of a1, a2, and a3 are set by the user according to the situation; V represents the second correction constant.

[0045] Preferably, the specific steps of S2 further include:

[0046] S221. Use the image Fourier transform technology to perform Fourier transform on several sets of image frames to obtain the amplitude spectrum. The amplitude spectrum contains the intensity information of all frequency components of several sets of image frames. The intensity information includes the frequency coordinates (u, r) and the total number m of high-frequency components. Based on the amplitude spectrum, obtain the roughness Bccd of several sets of image frames, and specifically obtain it in the following manner:

[0047]

[0048] In the formula, m represents the total number of high-frequency components; (u, r) represents the frequency coordinates. The Fourier transform converts several sets of image frames from the spatial domain to the frequency domain, where u and r respectively represent the frequencies in the horizontal and vertical directions in the frequency domain; |G(u, r)| represents the amplitude value at the frequency coordinates (u, r), that is, the signal intensity (or amplitude) at the frequency coordinates (u, r); H represents the range of high-frequency components;

[0049] S222. The pitch deviation factor Lpyz is obtained through the following formula:

[0050]

[0051] In the formula, J represents the number of intervals between threads, j = 1, 2, 3,..., J, Jjz j represents the pitch at the jth interval, Jjz ideal represents the pitch between the internal threads of the standard transmission lead screw.

[0052] Preferably, the specific steps of S3 include:

[0053] S31. Use the convolutional neural network technology to build a preliminary model, train and test the preliminary model with a detection data set, and use the trained preliminary model as an identification model. Respectively obtain the feature information in the identification model, and train and test the obtained feature information on the identification model. Combine whether the sorting instruction is triggered in S1 to use the trained identification model as a defect prediction model. After linear normalization processing, fit and output the balance evaluation index Pgzs. The balance evaluation index Pgzs is obtained in the following manner:

[0054]

[0055] In the formula, F1 and F2 are both weight values, R represents the third correction constant, where the specific values of F1 and F2 are set by the user according to the situation.

[0056] Preferably, the specific steps of S4 include:

[0057] S41. By performing a matching analysis of the balance evaluation index Pgzs with the performance balance threshold Q, comprehensively judge the balance state between the self-locking property and the transmission stability of the transmission lead screw. The specific content is as follows:

[0058] S411. When the balance evaluation index Pgzs falls within the performance balance threshold Q, comprehensively judge that the transmission lead screw is in a balanced state between self-locking property and transmission stability. At this time, transfer the current transmission lead screw to the next defect detection link.

[0059] S412. When the balance evaluation index Pgzs does not fall within the performance balance threshold Q, comprehensively judge that the transmission lead screw is in an unbalanced state between self-locking property and transmission stability. At this time, transfer the current transmission lead screw to the area awaiting manual quality inspection.

[0060] A high-precision defect detection system for automotive seat transmission lead screws, including a dynamic image capture module, an axis detection module, a continuous detection module, a defect comprehensive analysis module, and a defect calibration module.

[0061] The dynamic image capture module is used to rotate the transmission lead screw and capture a real-time video stream from multiple angles to capture the dynamic characteristics of the transmission lead screw. After image processing, a detection data set is generated.

[0062] The axis detection module will, based on the detection data set, judge whether the actual rotation axis of the transmission lead screw used in the automotive seat coincides with the standard axis. If not, trigger a sorting instruction.

[0063] The continuous detection module is used to further perform defect monitoring and analysis on the transmission lead screw when they coincide. Based on the detection data set, evaluate the transmission stability of the transmission lead screw when the automotive seat is adjusted in position to construct an operation stability coefficient Zwxs. At the same time, by analyzing the appearance conditions of the transmission lead screw, construct a self-locking coefficient Zsxs of the transmission lead screw.

[0064] The defect comprehensive analysis module uses convolutional neural network technology to construct a defect prediction model, and inputs the operation stability coefficient Zwxs and the corresponding self-locking coefficient Zsxs of the transmission lead screw into the defect prediction model. After linear normalization processing, fit and output the balance evaluation index Pgzs.

[0065] The defect calibration module is used to preset the performance balance threshold Q and perform a matching analysis with the balance evaluation index Pgzs to comprehensively judge the balance state between the self-locking property and the transmission stability of the transmission lead screw. Based on the balance state, comprehensively judge whether there are defects in the performance of the transmission lead screw set in the automotive seat.

[0066] The present invention provides a high-precision defect detection method and system for automotive seat transmission lead screws, having the following beneficial effects:

[0067] (1) Through the capture of multi-angle real-time video streams and image processing technology, this method can capture the dynamic characteristics of the lead screw during the rotation operation and generate a set of detection data. This real-time and multi-angle detection method ensures the integrity and accuracy of the data, enabling a more accurate judgment on whether the actual rotation axis of the lead screw coincides with the standard axis. If a deviation is detected, the sorting instruction can be triggered in a timely manner to separate the defective lead screws, improving the detection efficiency. Further, through the defect monitoring and analysis of the transmission stability and self-locking properties of the lead screw, the operation stability coefficient Zwxs and the self-locking coefficient Zsxs are constructed and used as key input parameters, which are input into the defect prediction model based on the convolutional neural network CNN. Using the non-linear fitting ability trained by big data, the CNN model can output the balance evaluation index Pgzs after linear normalization processing. This index quantifies the balance between the transmission stability and self-locking properties of the lead screw, realizing a comprehensive evaluation of the lead screw between the transmission performance and self-locking performance, making the prediction results more comprehensive and reliable. In addition, by setting the performance balance threshold Q, this method conducts a matching analysis between the balance evaluation index Pgzs and it, realizing the automatic screening and accurate judgment of the lead screw performance. When the balance evaluation index Pgzs meets the threshold, it indicates that the transmission lead screw meets the established balance requirements in terms of performance and can be safely used for the high-precision adjustment of automotive seats; otherwise, it can be determined that there are potential defects in the self-locking or transmission stability of the lead screw, and the risky lead screw products can be removed in a timely manner, thereby improving the safety and adjustment accuracy of automotive seats, further reducing production costs, and ensuring the stability and consistency of product quality.

[0068] (2) The central point data collected at different angles through the front-end cross-section and the rear-end cross-section are used to calculate the actual rotation axis of the transmission lead screw. This detection system realizes the quantification of the direction deviation by calculating the cosine value of the angle between the actual rotation axis and the direction vector of the standard axis. Using the cosine value of the angle as the deviation determination basis can ensure the measurement accuracy of the deviation. By setting a safety threshold, the system can automatically compare the cosine value of the deviation between the actual rotation axis and the standard axis. If the deviation value falls within the safety threshold range, it is determined that the actual rotation axis meets the standard; otherwise, the sorting instruction will be automatically triggered. This automated judgment mechanism reduces the error of manual judgment and improves the objectivity and standardization of judgment. The automated judgment process effectively reduces the dependence on manual intervention, enhances the reliability and repeatability of the detection process, and improves the efficiency. When the deviation value exceeds the safety threshold, the system immediately triggers the sorting instruction and sends the current transmission lead screw to the manual quality inspection area for further inspection. This mechanism ensures that only the lead screws with deviations will enter the manual inspection stage, further reducing the quality inspection workload and ensuring that all lead screws with large deviations can be identified and isolated in a timely manner. In short, through precise detection and screening, this method ensures the accuracy of the actual rotation axis of the transmission lead screw, further avoiding the risk of wear and failure of the transmission system caused by eccentricity or misalignment. The accuracy of detection and screening directly improves the service life of the transmission lead screw and the overall reliability of the system. By preventing potential installation deviations and structural abnormalities, the reliability of the automotive seat transmission system is improved, the failure rate is reduced, and the stability of the overall seat is guaranteed.

[0069] (3) The system uses convolutional neural network technology to build a preliminary model, and the preliminary model is trained and tested through the detection data set to obtain an identification model. On the basis of the identification model, feature information is further obtained and extracted, and the model is deeply trained to build the final defect prediction model. This model can automatically identify the defect state of the transmission lead screw and fit the balance evaluation index Pgzs based on this. By using deep learning technology to build the defect prediction model, the system has an adaptive ability during the defect detection process, can efficiently capture potential defects of the transmission lead screw, and significantly improves the accuracy and efficiency of defect detection. Through the extraction of feature information and model training, the balance evaluation index Pgzs output by the system can quantify the balance state between the self-locking property and the transmission stability of the transmission lead screw. This index integrates multiple characteristic factors of self-locking property and stability. After linear normalization processing, it can comprehensively quantify and evaluate the balance state of the lead screw. Description of the Drawings

[0070] Figure 1 Schematic flow chart of a method for detecting defects of a high-precision transmission lead screw of an automotive seat according to the present invention;

[0071] Figure 2This is a block diagram of a high-precision transmission lead screw defect detection system for an automotive seat according to the present invention. Specific embodiments

[0072] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work shall fall within the protection scope of the present invention.

[0073] Embodiment 1

[0074] Please refer to Figure 1 , the present invention provides a method for detecting defects in a high-precision transmission lead screw of an automotive seat, including the following steps:

[0075] S1. Rotate the transmission lead screw and capture the video stream in real time from multiple angles to capture the dynamic characteristics of the transmission lead screw. After image processing, a detection data set is generated. Based on the detection data set, it is judged whether the actual rotation axis of the transmission lead screw used in the automotive seat coincides with the standard axis. If not, a sorting instruction is triggered.

[0076] S2. If they coincide, further monitor and analyze the defects of the transmission lead screw. Based on the detection data set, evaluate the transmission stability of the transmission lead screw when the automotive seat is adjusted in position to construct an operation stability coefficient Zwxs. At the same time, by analyzing the appearance conditions of the transmission lead screw, construct a self-locking coefficient Zsxs of the transmission lead screw.

[0077] S3. Use the convolutional neural network technology to construct a defect prediction model, and input the operation stability coefficient Zwxs and the self-locking coefficient Zsxs of the corresponding transmission lead screw into the defect prediction model. After linear normalization processing, the balance evaluation index Pgzs is fitted and output.

[0078] S4. Preset a performance balance threshold Q and perform matching analysis with the balance evaluation index Pgzs to comprehensively judge the balance state between the self-locking property and the transmission stability of the transmission lead screw. Based on the balance state, comprehensively judge whether there are defects in the performance of the transmission lead screw set in the automotive seat.

[0079] In this embodiment, in step S1 of the method, by capturing the video stream in real time from multiple angles, the dynamic characteristics of the lead screw during rotation can be captured from multiple aspects, and a detection data set is generated. Further, through the analysis of the coincidence degree with the standard axis, it is determined whether the lead screw deviates from the preset standard axis. If an axis deviation is found, the defective lead screw can be marked and sent for manual inspection through the sorting instruction, further reducing the risk of defects such as installation eccentricity and assembly error, and effectively improving the centering accuracy of the lead screw. In step S2, the method deeply analyzes the transmission stability and appearance conditions of the lead screw through the detection data set, constructs the operation stability coefficient Zwxs and the self-locking coefficient Zsxs respectively, and forms a more comprehensive performance evaluation system. The method further breaks through the limitation of the single nature of the traditional detection method, can consider two key parameters of transmission stability and self-locking at the same time, ensure that the lead screw can maintain stable transmission and effectively self-lock when the power is cut off, and further improve the overall transmission efficiency and safety. In steps S3 and S4, the method uses the convolutional neural network technology to construct a defect prediction model, takes the operation stability coefficient Zwxs and the self-locking coefficient Zsxs as input features, and after linear normalization processing, outputs the balance evaluation index Pgzs. By matching and analyzing this index with the preset performance balance threshold Q, the system can comprehensively judge the balance state between the self-locking and transmission stability of the transmission lead screw, and further avoid the risk of possible product defects flowing into the use link. In short, through multi-angle video acquisition, characteristic coefficient analysis and intelligent defect prediction, the method further realizes the balanced detection and evaluation of the multi-dimensional performance of the transmission lead screw, meets the requirements of the high-precision automotive seat for the stability and self-locking of the transmission lead screw, and provides a reliable guarantee for the safety and comfort of the automotive seat.

[0080] Embodiment 2

[0081] Please refer to Figure 1 , specifically: The specific steps of S1 include:

[0082] S11. First, perform a rotation operation on the transmission lead screw, and while the transmission lead screw is performing the rotation operation, dynamically obtain a video stream through a multi-view image acquisition device, and extract several groups of image frames from the video stream;

[0083] S12. Based on the several groups of image frames obtained in S11, the noise components are removed by using bilateral filtering method, and the pixel values in the several groups of image frames are adjusted to be uniformly distributed by histogram equalization to improve the contrast of the image, so that the brightness distribution is more uniform, which is suitable for images with uneven illumination. The several groups of image frames after image processing are aggregated into a detection data set, and the detection data set is subjected to feature extraction to obtain the measurement period T, the instantaneous value v(t) of the vibration speed, the thread angle, the number of intervals J between the threads and the pitch Jjz at the corresponding interval, and according to the standard transmission screw, the pitch Jjz between the internal threads of the standard transmission screw is obtained. ideal And by applying pressure to the transmission screw, the friction coefficient Mc is obtained;

[0084] S13. Divide the transmission screw into two sections in advance and mark them as the front end section and the rear end section respectively, and perform image extraction on several groups of image frames processed by step S12 to obtain the center point coordinates of the front end section and the rear end section of the transmission screw at four angles of 0°, 90°, 180° and 270° respectively; wherein the center point coordinates refer to the coordinates of the geometric center of each cross section of the transmission screw at different angles, which is a point in three-dimensional space. Specifically, the center point is calculated based on the cross-sectional image of the screw by image processing technology.

[0085] In dynamic acquisition, a rotary encoder will be connected to the lead screw to monitor its real-time rotation speed and angle. Whenever the lead screw rotates to a specific angle, the encoder sends a trigger signal to the high-speed camera to ensure that each frame of the image is captured at the same rotation position. The external trigger controls the start and end time of the shooting, which is suitable for capturing the dynamic performance of the lead screw during acceleration or deceleration.

[0086] In this embodiment, in S11, a multi-view image acquisition device is used to dynamically obtain a video stream and extract several groups of image frames, so that the state of the transmission lead screw during actual rotation operation is continuously captured. This dynamic acquisition method can ensure the acquisition of stable geometric information in the working state of the lead screw. Especially when rotating at high frequency, key details can still be extracted, providing a reliable data source for subsequent analysis and evaluation. In S12, image noise is removed by the bilateral filtering method and histogram equalization adjustment is performed, improving the contrast and clarity of the image. This processing makes the brightness of the image more uniform. Even in the case of unstable lighting conditions or the presence of local shadows, the structural information of the lead screw can be accurately extracted. Therefore, even in a complex lighting environment, the quality of the data is effectively guaranteed, contributing to more accurate detection results. In S13, the lead screw is divided into front and rear sections, and the geometric center point coordinates of the front and rear ends at four angles are obtained respectively. Using this multi-angle and two-section center point data, the actual rotation axis of the lead screw can be accurately described. These accurate center point coordinates provide basic data for judging defects such as coaxiality, misalignment, and eccentricity of the lead screw. In particular, by analyzing the center point positions of the front and rear sections, it can be evaluated whether the lead screw has bending, eccentricity, or other structural defects. Through the image processing and coordinate extraction in steps S12 and S13, multiple frames of images are converted into a structured detection data set, which directly reflects the spatial geometric characteristics of the lead screw and can be used for further numerical analysis, modeling, and prediction. The efficiency of this process lays the foundation for the subsequent defect detection and prediction model implementation, reducing the data processing time and improving the real-time performance of the detection system. The image preprocessing technology included in this detection method, combined with the acquisition of double-section multi-angle center point coordinates, ensures the reliability of the detection system in a complex environment. This method can not only be used for static detection but also meet the precise measurement requirements in a dynamic rotation environment, effectively improving the applicability of the transmission lead screw detection system.

[0087] Embodiment 3

[0088] Please refer to Figure 1 , specifically: The specific steps of S1 also include:

[0089] S14. Based on the center point coordinates of the front cross-section and the rear cross-section of the transmission lead screw obtained in S13 at four angles of 0°, 90°, 180°, and 270°, the average value of the four angles is taken on each cross-section to obtain the center point of the corresponding cross-section in three-dimensional space, and based on the center point of the corresponding cross-section in three-dimensional space, the actual rotation axis of the transmission lead screw is constructed The actual rotation axis of the transmission lead screw Is constituted by the following equation:

[0090]

[0091] In the formula, represents the three-dimensional coordinates of any point on the actual rotation axis; represents the starting point coordinates on the actual rotation axis; represents the direction vector of the actual rotation axis; d represents a scalar, which is a parameter used to determine different positions on the actual rotation axis to adjust the distance moved in the direction;

[0092] In three-dimensional space, we usually use a point (i.e., the starting point) and a direction vector to define a straight line. This straight line (i.e., the rotation axis) can extend to the entire length of the lead screw. The starting point coordinates on the actual rotation axis This is a fixed point selected on a straight line. Any point on this straight line (the center points of the front end section and the rear end section) can be selected as the starting point, as long as this point is indeed on the straight line, because no matter which center point is selected, the actual rotation axis obtained is the same, only the starting point is different and the direction is the same;

[0093] Example: x0 + d * d x represents the coordinate component along the x direction, y0 + d * d y represents the coordinate component along the y direction, z0 + d * d z represents the coordinate component along the z direction; therefore, is the coordinate position of any point on the actual rotation axis in three-dimensional space, and its direction is controlled by the direction vector and extends in space along the direction;

[0094] S141. Analyze the offset trend of the actual rotation axis of the drive lead screw in three-dimensional space through the vector change between the center points of the front end section and the rear end section, calculate the vector from the front end to the rear end, and record it as the direction vector of the actual rotation axis The direction vector of the actual rotation axis is obtained through the following method: where I2 represents the center point of the rear end section, and I1 represents the center point of the front end section; the trend of the two center points (the center points I of the front end section and the rear end section) is used to determine the direction vector;

[0095] S15. Based on the actual rotation axis of the drive lead screw obtained in S14 and combined with the standard axis, compare the direction deviation between the actual rotation axis and the standard axis to obtain the included angle θ between the direction vectors of the two axes, which is specifically obtained through the following method:

[0096]

[0097] In the formula, Expressed as the direction vector of the standard axis; Expressed as the magnitude (length) of the direction vector of the actual rotation axis; The magnitude (length) of the direction vector expressed as the standard axis; Expressed as the direction vector of the actual rotation axis; It is the dot product of the direction vector of the actual rotation axis and the direction vector of the standard axis, indicating the directional projection relationship between the two vectors. The larger the dot product result, the closer the two vectors are to the same direction. cos(θ) represents the actual rotation axis. The cosine of the deviation from the standard axis;

[0098] S16, pre-set the safety threshold, by comparing the safety threshold with the actual rotation axis The cosine value cosθ of the deviation from the standard axis is compared to determine whether the actual rotation axis of the transmission screw used in the car seat coincides with the standard axis. The specific determination is as follows:

[0099] If the actual rotation axis If the cosine value cos(θ) of the deviation from the standard axis falls within the safety threshold, it is determined that the actual rotation axis of the transmission screw used in the car seat coincides with the standard axis, and the transmission screw is further analyzed.

[0100] If the actual rotation axis If the cosine value cos(θ) of the deviation from the standard axis does not fall within the safety threshold, it is determined that the actual rotation axis of the transmission screw used in the car seat does not coincide with the standard axis, and the sorting instruction will be triggered;

[0101] The above safety threshold is set as follows: according to the corresponding actual rotation axis obtained by several groups of transmission screws during defect detection The cosine value cos(θ) of the deviation from the standard axis, the mean value of the cosine value cos(θ) and the standard deviation of the cosine value cos(θ) are obtained, and the lower limit of the safety threshold is set by the mean value of the cosine value cos(θ) and the standard deviation of the cosine value cos(θ): the lower limit is: the mean value of the cosine value cos(θ) - K* the standard deviation of the cosine value cos(θ); the upper limit is: the maximum value of the cosine value cos(θ), that is, the value 1; wherein K is a constant, usually taking a value of 1-3, corresponding to different confidence levels, and the specific value is set by the user (according to the actual situation); according to the setting method of the safety threshold here, the balance threshold Q in S4 is obtained in the same way;

[0102] Explanation: A cos(θ) value close to 1 indicates that the direction of the actual axis of rotation is very close to the standard axis, almost parallel, with a small directional deviation; a cos(θ) value much less than 1 indicates a large deviation between the direction of the actual axis of rotation and the standard axis, possibly indicating eccentricity or installation error. This value helps us determine whether the direction of the actual axis of rotation is consistent with the standard axis in the design.

[0103] S17. After receiving the sorting instruction, the current drive lead screw is transferred to the area awaiting manual quality inspection.

[0104] In this embodiment, the detection system calculates the central position of the cross-section by obtaining the center point data of the front and rear cross-sections of the lead screw at multiple angles, and constructs the actual axis of rotation of the drive lead screw. Compared with traditional single-point or simple manual visual inspection methods, it can improve the accuracy of axis detection as much as possible, especially in scenarios with a small detection space and high installation requirements, it can accurately evaluate the centering and accuracy of the lead screw. By constructing the actual axis of rotation and comparing it with the standard axis, the system can accurately calculate the deviation angle between the actual axis and the standard axis. The cosine value of this directional deviation can reflect whether there is installation eccentricity or directional deviation of the lead screw, providing a highly sensitive installation status judgment standard to ensure the stable operation of the equipment. The system presets a safety threshold. When the cosine value of the deviation between the actual axis and the standard axis exceeds the threshold, it indicates that the lead screw has serious eccentricity or direction error and does not meet the installation standard. At this time, the system will automatically trigger a sorting instruction to send the drive lead screw to the area for manual quality inspection for further detection and analysis. Through such an intelligent sorting mechanism, unqualified lead screws can be effectively filtered out, reducing the risk of equipment failure and improving the stability of the drive system. The automated detection and sorting functions of this system reduce the dependence on manual detection.

[0105] Through high-precision axis deviation detection, it can ensure the smooth operation of the drive lead screw during actual installation. Since the lead screw is a crucial drive component in parts such as car seats, ensuring its axis centering can directly affect the drive effect and the service life of the product. The application of this system improves the stability of the overall product quality and ensures high-standard assembly quality. By archiving the detection data, including the cosine value of the deviation, the position of the center point of each cross-section, and the axis direction, etc., a complete detection data record library can be established. These data can be used for subsequent quality traceability and analysis, providing important reference bases for subsequent product improvement, fault prediction, etc. This system can effectively detect the axis centering of the drive lead screw, judge whether the direction meets the standard through the cosine value of the deviation, and automatically sort unqualified products. The entire process reduces human error, improves the detection accuracy and efficiency, guarantees the installation quality of the lead screw and the stability of the system, and improves the automation and reliability of production as much as possible.

[0106] Embodiment 4

[0107] Please refer to Figure 1 , specifically: The specific steps of S1 also include:

[0108] S18. If it is determined that the actual rotation axis of the transmission lead screw used in the vehicle seat coincides with the standard axis, at this time, the center point of the corresponding cross-section obtained in S14 in the three-dimensional space will be combined to analyze the position offset between the actual rotation axis and the standard axis of the transmission lead screw used in the vehicle seat, so as to obtain the eccentricity Pxj. The eccentricity Pxj is obtained through the following method:

[0109]

[0110] In the formula, n represents the number of center points obtained from different cross-section images, i = 1, 2, 3,...., n, I i represents the center point obtained from the i-th cross-section image, represents the standard axis, represents the actual rotation axis The vertical distance between the center point obtained from the i-th cross-section image on and the standard axis

[0111] The specific steps of S2 include:

[0112] S21. Based on the detection data set, evaluate the transmission stability of the transmission lead screw of the vehicle seat during position adjustment. After linear normalization processing, a working stability coefficient Zwxs is constructed. The working stability coefficient Zwxs is obtained through the following formula:

[0113]

[0114] In the formula, Zfz represents the vibration amplitude, which reflects the average vibration ability; Pxj represents the eccentricity, Cx represents the transmission efficiency, α and β are both weight values, C represents the first correction constant. Among them, the specific values of α and β are set by the user according to the situation;

[0115] S211. The vibration amplitude Zfz is obtained through the following formula:

[0116]

[0117] In the formula, T represents the measurement period, v(t) represents the instantaneous value of the vibration velocity, dt is the tiny increment of the time variable t. In the integral symbol, it means that the entire measurement period T is divided into countless extremely small time segments, and the square values of the instantaneous vibration velocity v(t) within each time segment are accumulated (integrated);

[0118] S212. The transmission efficiency Cx is obtained through the following formula:

[0119]

[0120] In the formula, Mc represents the friction coefficient, θ represents the thread angle, the value of the transmission efficiency Cx ranges from 0 to 1, and the closer it is to 1, the higher the efficiency; cos(θ) represents the cosine value of the thread angle, and sin(θ) represents the sine value of the thread angle.

[0121] The above-mentioned friction coefficient Mc is obtained through the following formula: F f represents the frictional force, that is, the parallel resistance acting on the contact surface; F n represents the normal force, that is, the pressure applied perpendicular to the contact surface.

[0122] In this embodiment, after analyzing the coincidence of the actual rotation axis of the transmission lead screw and the standard axis, the eccentricity of each section relative to the standard axis is further calculated. The eccentricity Pxj is obtained by the perpendicular distance from each center point in the multi-section data to the standard axis, which reflects the position offset of the actual rotation axis of the lead screw relative to the standard axis. The eccentricity Pxj quantifies the eccentricity of the transmission lead screw, can identify it before the occurrence of a tiny eccentricity phenomenon, helps to improve the alignment accuracy of the installation of the transmission lead screw, and ensures the smoothness of the vehicle seat adjustment system. By comprehensively considering the three factors of vibration amplitude, eccentricity, and transmission efficiency, this solution generates a job stability coefficient Zwxs through linear normalization processing. This coefficient quantifies the stability of the transmission lead screw during the vehicle seat adjustment process, ensures that the job stability is more comprehensive and reliable. The calculation of the vibration amplitude can more truly reflect the vibration characteristics and transmission smoothness of the transmission lead screw during the actual working process, helps to identify vibration abnormalities in advance, prevent potential failures, and thus extend the service life of the lead screw and the seat adjustment system. The calculation of the transmission efficiency comprehensively considers two key factors of the friction coefficient and the thread angle, can truly reflect the loss situation during the transmission process. The closer the transmission efficiency is to 1, the smaller the transmission loss and the higher the efficiency.

[0123] Embodiment 5

[0124] Please refer to Figure 1 , specifically: The specific steps of S2 also include:

[0125] S22. Based on the detection data set, analyze the appearance conditions of the transmission lead screw, and after linear normalization processing, construct the self-locking coefficient Zsxs of the transmission lead screw. The self-locking coefficient Zsxs is obtained through the following formula:

[0126]

[0127] Wherein, Bccd represents roughness, θ represents thread angle, Lpyz represents pitch deviation factor, a1, a2 and a3 are all weight values, wherein the specific values of a1, a2 and a3 are set by the user according to the situation, and V represents the second correction constant.

[0128] The specific steps of S2 also include:

[0129] S221. Using the image Fourier transform technology, perform Fourier transform on several groups of image frames to obtain amplitude spectrum, wherein the amplitude spectrum contains the intensity information of all frequency components of several groups of image frames, and the intensity information includes the frequency coordinates (u, r) and the total number m of high-frequency components. The roughness is usually reflected in the amplitude of the high-frequency components. Based on the amplitude spectrum, the roughness Bccd of several groups of image frames is obtained, which is specifically obtained in the following manner:

[0130]

[0131] Wherein, m represents the total number of high-frequency components, which represents the number of frequency components in the high-frequency range in the amplitude spectrum after Fourier transform; (u, r) represents the frequency coordinate, and Fourier transform converts several groups of image frames from the spatial domain to the frequency domain, where u and r represent the horizontal and vertical frequencies in the frequency domain, respectively; |G(u, r)| represents the amplitude value at the frequency coordinate (u, r), that is, the signal strength (or amplitude) at the frequency coordinate (u, r); H represents the range of high-frequency components, which is the frequency range specified in the frequency domain, including the higher frequency part. Generally, the high-frequency components are distributed in the four corners of the spectrum diagram, representing the details of the image that change dramatically.

[0132] The surface roughness of the transmission screw is indirectly represented by the roughness Bccd of several groups of image frames.

[0133] S222, pitch deviation factor Lpyz is obtained by the following formula:

[0134]

[0135] Where J represents the number of intervals between threads, j = 1, 2, 3, ..., J, Jjz j Expressed as the pitch at the jth interval, Jjz ideal Expressed as the pitch between the internal threads of a standard transmission screw.

[0136] In this embodiment, through the analysis of the appearance conditions of the transmission lead screw, considering multiple factors such as roughness, thread angle, and pitch deviation factor, the system calculates the self-locking coefficient Zsxs. The construction of this coefficient integrates surface roughness and thread geometric features, making the evaluation of self-locking performance more comprehensive and accurate. The quantitative evaluation of the self-locking coefficient Zsxs improves the judgment accuracy of the self-locking performance of the lead screw, ensuring that the transmission lead screw can be stably locked during actual operation, further avoiding slippage or loosening phenomena, and thus improving the safety and reliability of the car seat during use. Through multi-level appearance detection, including the analysis of multiple factors such as roughness, thread angle, and pitch deviation factor, the system can evaluate the quality of the transmission lead screw from multiple dimensions. Compared with single-parameter detection, this multi-factor evaluation method is more comprehensive and can capture subtle changes in the appearance and geometric parameters of the lead screw. The multi-level analysis improves the coverage and accuracy of the detection of the transmission lead screw, helps to identify possible defects at an early stage, ensures that the product meets the quality standards, and reduces the risk of failures after installation.

[0137] Embodiment 6

[0138] Please refer to Figure 1 , specifically: The specific steps of S3 include:

[0139] S31. Use convolutional neural network technology to build a preliminary model, train and test the preliminary model with a detection data set, and use the trained preliminary model as an identification model. Respectively obtain the feature information in the identification model, and use the obtained feature information to train and test the identification model. Combine whether the sorting instruction is triggered in S1 to use the trained identification model as a defect prediction model. After linear normalization processing, fit and output the balance evaluation index Pgzs. The balance evaluation index Pgzs is obtained in the following way:

[0140]

[0141] In the formula, both F1 and F2 are weight values, and R represents the third correction constant. Among them, the specific values of F1 and F2 are set by the user according to the situation.

[0142] The specific steps of S4 include:

[0143] S41. Through the matching analysis of the balance evaluation index Pgzs and the performance balance threshold Q, comprehensively judge the balance state between the self-locking property and the transmission stability of the transmission lead screw. The specific content is as follows:

[0144] S411. If the balance evaluation index Pgzs falls within the performance balance threshold Q, comprehensively judge that the transmission lead screw is in a balanced state between self-locking property and transmission stability. At this time, transfer the current transmission lead screw to the next defect detection link;

[0145] S412. When the balance evaluation index Pgzs does not fall within the performance balance threshold Q, it is comprehensively judged that the transmission lead screw is in an unbalanced state between self-locking and transmission stability. At this time, the current transmission lead screw is sent to the area to be manually inspected.

[0146] In this embodiment, a preliminary recognition model is constructed through a convolutional neural network, and the model is trained and tested in combination with the detection data set, so that the model has the ability to automatically identify defects. Through the extraction and training of feature information, the recognition model is further developed into a defect prediction model, and finally the balance evaluation index Pgzs is fitted. This CNN-based prediction model not only improves the accuracy of defect detection, but also realizes automatic defect prediction. The defect prediction model automatically identifies potential defects in the transmission lead screw, significantly improving the detection efficiency, reducing the need for manual inspection, and ensuring efficient and high-precision quality control. The system quantifies the balance state between the self-locking and transmission stability of the transmission lead screw by calculating the balance evaluation index Pgzs. The system sets a performance balance threshold Q, and by matching the balance evaluation index Pgzs with this threshold, it judges whether the self-locking and transmission stability of the transmission lead screw reach the expected balance state, further realizing the compatibility verification between self-locking and transmission stability, ensuring the safety of the transmission lead screw during operation, reducing the risk caused by imbalance, and effectively extending the service life of the transmission system. When the balance evaluation index Pgzs does not fall within the performance balance threshold Q, the system will automatically send the transmission lead screw to the area to be manually inspected again to further confirm its defect state. This intelligent sorting mechanism effectively filters out the lead screws that do not meet the standards and reduces the burden of manual quality inspection. In short, through the feature extraction and prediction model of the convolutional neural network, combined with the matching analysis of the balance evaluation index and the performance balance threshold, this method further realizes the intelligent comprehensive judgment of the self-locking and transmission stability of the transmission lead screw. Through the automatic defect prediction and sorting mechanism, the system significantly improves the detection efficiency, reduces the need for manual operation, ensures the high quality and stability of the transmission lead screw, and provides a strong guarantee for the safe operation of the transmission system.

[0147] Embodiment 7

[0148] Please refer to Figure 2 , specifically: a high-precision transmission lead screw defect detection system for an automotive seat, including a dynamic image capture module, an axis detection module, a continuous detection module, a defect comprehensive analysis module, and a defect calibration module;

[0149] The dynamic image capture module is used to rotate the transmission lead screw and capture a video stream in real time from multiple angles to capture the dynamic characteristics of the transmission lead screw. After image processing, a detection data set is generated;

[0150] Based on the detection data set, the axis detection module determines whether the actual rotation axis of the transmission lead screw used in the vehicle seat coincides with the standard axis. If not, a sorting instruction is triggered;

[0151] The continuous detection module is used to further analyze the defect of the transmission lead screw when they coincide. Based on the detection data set, it evaluates the transmission stability of the transmission lead screw when the vehicle seat is adjusted in position, so as to construct the operation stability coefficient Zwxs. At the same time, by analyzing the appearance condition of the transmission lead screw, the self-locking coefficient Zsxs of the transmission lead screw is constructed;

[0152] The defect comprehensive analysis module uses the convolutional neural network technology to construct a defect prediction model, and inputs the operation stability coefficient Zwxs and the self-locking coefficient Zsxs of the corresponding transmission lead screw into the defect prediction model. After linear normalization processing, the balance evaluation index Pgzs is fitted and output;

[0153] The defect calibration module is used to preset the performance balance threshold Q, and match and analyze it with the balance evaluation index Pgzs to comprehensively judge the balance state between the self-locking property and the transmission stability of the transmission lead screw. Based on the balance state, it comprehensively judges whether there are defects in the performance of the transmission lead screw set in the vehicle seat.

[0154] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for detecting defects of a high-precision transmission lead screw of an automotive seat, characterized in that: Including the following steps, S1. Rotate the transmission lead screw and capture the video stream in real time from multiple angles to capture the dynamic characteristics of the transmission lead screw. After image processing, a detection data set is generated. Based on the detection data set, it is judged whether the actual rotation axis of the transmission lead screw used in the car seat coincides with the standard axis. If not, a sorting instruction is triggered; S2. If they coincide, further defect monitoring and analysis are carried out on the transmission lead screw. Based on the detection data set, the transmission stability of the transmission lead screw when the car seat is adjusted in position is evaluated to construct the operation stability coefficient Zwxs. At the same time, by analyzing the appearance conditions of the transmission lead screw, the self-locking coefficient Zsxs of the transmission lead screw is constructed; S3. Use the convolutional neural network technology to construct a defect prediction model, and input the operation stability coefficient Zwxs and the self-locking coefficient Zsxs of the corresponding transmission lead screw into the defect prediction model. After linear normalization processing, the balance evaluation index Pgzs is fitted and output; S4. Preset the performance balance threshold Q in advance and perform matching analysis with the balance evaluation index Pgzs to comprehensively judge the balance state between the self-locking property and the transmission stability of the transmission lead screw. Based on the balance state, it is comprehensively judged whether there are defects in the performance of the transmission lead screw set in the car seat; The specific steps of S2 include: S21. Based on the detection data set, evaluate the transmission stability of the transmission lead screw when the car seat is adjusted in position. After linear normalization processing, the operation stability coefficient Zwxs is constructed. The operation stability coefficient Zwxs is obtained through the following formula: In the formula, is expressed as the vibration amplitude; is expressed as the eccentricity, is expressed as the transmission efficiency, and are both weight values, is expressed as the first correction constant, where and The specific values are set by the user according to the situation; S211. The vibration amplitude is obtained by the following formula: where T represents the measurement period, represents the instantaneous value of the vibration velocity; S212. The transmission efficiency is obtained by the following formula: In the formula, is expressed as the coefficient of friction, is expressed as the thread angle; is expressed as the cosine value of the thread angle, is expressed as the sine value of the thread angle; The specific steps of S2 also include: S22. Based on the detection data set, analyze the appearance conditions of the transmission lead screw and, after linear normalization processing, construct the self-locking coefficient Zsxs of the transmission lead screw. The self-locking coefficient Zsxs is obtained through the following formula: In the formula, is expressed as the roughness, is expressed as the thread angle, is expressed as the pitch deviation factor, , and are all weight values. Among them, , and The specific values are set by the user according to the situation, and V is expressed as the second correction constant.

2. The defect detection method for the high-precision transmission lead screw of an automotive seat according to claim 1, characterized in that: The specific steps of S1 include: S11. First, rotate the transmission lead screw. While the transmission lead screw is rotating, dynamically obtain the video stream through a multi-view image acquisition device and extract several groups of image frames from the video stream; S12, based on the several groups of image frames obtained in S11, the noise component is removed by using bilateral filtering method, and the pixel values in the several groups of image frames are adjusted to be uniformly distributed by histogram equalization, and the several groups of image frames after image processing are aggregated into a detection data set, and the detection data set is subjected to feature extraction to obtain the instantaneous value of the measurement period T and the vibration speed. , thread angle 、Number of intervals between threads and the pitch at the corresponding interval , and according to the standard transmission screw, obtain the pitch between the internal threads of the standard transmission screw and obtain the friction coefficient by applying pressure to the transmission screw ; S13. Divide the transmission lead screw into two equal segments in advance, and mark them as the front-end section and the rear-end section respectively. Then, perform image extraction on several groups of image frames processed in step S12 to obtain the center point coordinates of the front-end section and the rear-end section of the transmission lead screw at , , 18 and the center point coordinates at four angles.

3. The method for detecting defects of the high-precision transmission screw rod of an automotive seat according to claim 2, wherein: The specific steps of S1 also include: S14. Based on the center point coordinates of the front cross-section and the rear cross-section of the transmission lead screw obtained in S13 at , , 18 and the center point coordinates at four angles, take the average value of the four angles on each cross-section to obtain the center point of the corresponding cross-section in three-dimensional space, and based on the center point of the corresponding cross-section in three-dimensional space, construct the actual rotation axis of the transmission lead screw , the actual rotation axis of the transmission lead screw is formed by the following equation: In the formula, represents the three-dimensional coordinates of an arbitrary point on the actual axis of rotation; represents the starting point coordinates on the actual axis of rotation; represents the direction vector of the actual axis of rotation; represents a scalar; S141. Analyze the offset trend of the actual rotation axis of the transmission lead screw in three-dimensional space through the vector change between the center points of the front cross-section and the rear cross-section, calculate the vector from the front end to the rear end, and denote it as the direction vector of the actual rotation axis , the direction vector of the actual rotation axis is obtained by the following method: wherein, represents the center point of the rear cross-section, represents the center point of the front cross-section; S15. Based on the actual rotation axis of the transmission lead screw obtained in S14 , and in combination with the standard axis, compare the actual rotation axis with the direction deviation between the actual rotation axis and the standard axis to obtain the angle between the direction vectors of the two axes , which is obtained specifically through the following method: In the formula, represents the direction vector of the standard axis; represents the modulus of the direction vector of the actual rotation axis; represents the modulus of the direction vector of the standard axis; represents the direction vector of the actual rotation axis; is the dot product of the direction vector of the actual rotation axis and the direction vector of the standard axis; represents the actual rotation axis and the cosine value of the deviation between the standard axes; S16. Preset a safety threshold, and by comparing the safety threshold with the cosine value of the deviation between the actual rotation axis and the standard axis to determine whether the actual rotation axis of the transmission lead screw used in the vehicle seat coincides with the standard axis. The specific judgment content is as follows: If the cosine value of the deviation between the actual rotation axis and the standard axis falls within the safety threshold, it is determined that the actual rotation axis of the transmission lead screw used in the vehicle seat coincides with the standard axis. At this time, the transmission lead screw is further analyzed; If the cosine value of the deviation between the actual rotation axis and the standard axis does not fall within the safety threshold, it is determined that the actual rotation axis of the transmission lead screw used in the vehicle seat does not coincide with the standard axis, and at this time, a sorting instruction will be triggered; S17. After receiving the sorting instruction, transfer the current transmission lead screw to the area to be manually inspected.

4. The method for detecting defects of the high-precision transmission screw rod of an automotive seat according to claim 3, characterized in that: The specific steps of S1 also include: S18. If it is judged that the actual rotation axis of the transmission lead screw used in the car seat coincides with the standard axis, at this time, in combination with the center point of the corresponding cross-section obtained in S14 in the three-dimensional space, analyze the position offset between the actual rotation axis and the standard axis of the transmission lead screw used in the car seat to obtain the eccentricity Pxj. The eccentricity Pxj is obtained in the following way: Wherein, n represents the number of center points obtained from different cross-sectional images, and i = 1, 2, 3,...., n, represents the center point obtained from the i-th cross-sectional image, represents the standard axis, represents the actual rotation axis The vertical distance between the center point obtained from the i-th cross-sectional image on and the standard axis therebetween.

5. The method for detecting defects of the high-precision transmission lead screw of an automotive seat according to claim 1, characterized in that: The specific steps of S2 also include: S221. Using the image Fourier transform technology, perform Fourier transform on several groups of image frames to obtain the amplitude spectrum. Among them, the amplitude spectrum contains the intensity information of all frequency components of several groups of image frames, and the intensity information includes the frequency coordinates and the total number of high-frequency components , and based on the amplitude spectrum, obtain the roughness of several groups of image frames , which is obtained specifically in the following manner: wherein, represents the total number of high-frequency components; represents the frequency coordinate. The Fourier transform converts several sets of image frames from the spatial domain to the frequency domain, where and r respectively represent the frequencies in the horizontal and vertical directions in the frequency domain; represents the frequency coordinate at the amplitude value; represents the range of high-frequency components; S222. The pitch deviation factor is obtained by the following formula: In the formula, represents the number of intervals between threads, j = 1, 2, 3, ..., , represents the pitch at the j-th interval, represents the pitch between the internal threads of the standard transmission lead screw.

6. The defect detection method for the high-precision transmission lead screw of an automotive seat according to claim 1, wherein: The specific steps of S3 include: S31. Use the convolutional neural network technology to build a preliminary model, train and test the preliminary model with the detection data set, and use the trained preliminary model as the recognition model. Respectively obtain the feature information in the recognition model, and use the obtained feature information to train and test the recognition model. Combine whether the sorting instruction is triggered in S1 to use the trained recognition model as the defect prediction model. After linear normalization processing, fit and output the balance evaluation index Pgzs. The balance evaluation index Pgzs is obtained in the following way: In the formula, and are both weight values, is expressed as the third correction constant, where and The specific values are set by the user according to the situation.

7. The defect detection method for the high-precision transmission lead screw of an automotive seat according to claim 1, characterized in that: The specific steps of S4 include: S41. Through the matching analysis of the balance evaluation index Pgzs and the performance balance threshold Q, comprehensively judge the balance state between the self-locking property and the transmission stability of the transmission lead screw. The specific content is as follows: S411. When the balance evaluation index Pgzs falls within the performance balance threshold Q, comprehensively judge that the transmission lead screw is in a balanced state between the self-locking property and the transmission stability. At this time, transfer the current transmission lead screw to the next defect detection link; S412. When the balance evaluation index Pgzs does not fall within the performance balance threshold Q, comprehensively judge that the transmission lead screw is in an unbalanced state between the self-locking property and the transmission stability. At this time, transfer the current transmission lead screw to the area to be manually inspected for quality.

8. An automotive seat high-precision transmission lead screw defect detection system for implementing the automotive seat high-precision transmission lead screw defect detection method described in any one of the above claims 1 to 7, characterized in that: It includes a dynamic image capture module, an axis detection module, a continuous detection module, a defect comprehensive analysis module and a defect calibration module; The dynamic image capture module is used to rotate the transmission lead screw and capture the video stream in real time from multiple angles to capture the dynamic characteristics of the transmission lead screw. After image processing, a detection data set is generated; The axis detection module will judge whether the actual rotation axis of the transmission lead screw used in the car seat coincides with the standard axis based on the detection data set. If they do not coincide, trigger the sorting instruction; The continuous detection module is used to further analyze the defect monitoring of the transmission lead screw when they coincide. Based on the detection data set, evaluate the transmission stability of the transmission lead screw when the car seat is adjusted in position to construct the operation stability coefficient Zwxs. At the same time, by analyzing the appearance condition of the transmission lead screw, construct the self-locking coefficient Zsxs of the transmission lead screw; The defect comprehensive analysis module uses the convolutional neural network technology to build a defect prediction model, and inputs the operation stability coefficient Zwxs and the corresponding self-locking coefficient Zsxs of the transmission lead screw into the defect prediction model. After linear normalization processing, fit and output the balance evaluation index Pgzs; The defect calibration module is used to preset the performance balance threshold Q and perform matching analysis with the balance evaluation index Pgzs to comprehensively judge the balance state between the self-locking property and the transmission stability of the transmission lead screw. Based on the balance state, comprehensively judge whether there are defects in the performance of the transmission lead screw set in the car seat.

Images