Fault early warning method and system for non-uniformly distributed fasteners under complex working conditions

Abstract

The invention relates to the technical field of fault detection, and discloses a non-uniform distributed fastener fault early warning method and system for complex working conditions, and the method comprises the steps: obtaining stress data and image data of to-be-detected equipment; according to the stress data and the image data, identifying areas where the fasteners are distributed in the equipment to obtain a plurality of distribution areas of the fasteners; in each distribution area, combining the influence of the current state of the fastener and the stress on the state of the fastener, predicting the state of the fastener, and when the state of the fastener has a fault, performing fault early warning; according to the method, the stress data and the image data of the equipment under the complex working condition are combined to obtain the accurate position and state recognition result, the state change condition of the fastener is predicted based on the prediction result of the stress change trend of the equipment, fault hidden dangers can be found in advance, early warning can be performed, and the reliability of the fastener is improved. And the timeliness and reliability of equipment fault early warning can be improved.

Description

Technical Field

[0001] This application relates to the technical field of fault detection, and particularly to a method and system for fault early warning of non-uniformly distributed fasteners under complex working conditions. Background Art

[0002] Currently, the fastener fault early warning method relies on real-time image data, with single data and a single judgment method, unable to achieve early prediction of faults for early warning, unable to predict the fault occurrence time and fault type in advance, and unable to ensure the timeliness and effectiveness of fault early warning.

[0003] For example, the Chinese patent application with the publication number CN117094952A discloses a method and device for fault detection of fasteners on a port crane. Among them, the method obtains a denoised port crane fastener image by performing denoising processing on the acquired port crane fastener image, locates the fastener area of the denoised port crane fastener image through the YOLO v5s algorithm to obtain a fastener area image, performs image enhancement processing on it to obtain an enhanced fastener area image, and inputs it into a fault detection model constructed based on the YOLOv5s algorithm for fault detection, thereby outputting a fault detection result.

[0004] The above prior art has the problems raised in this background art: single data and a single judgment method, unable to achieve early prediction of faults for early warning, unable to predict the fault occurrence time and fault type in advance, and unable to ensure the timeliness and effectiveness of fault early warning; to solve at least one of the above problems, this application proposes a method and system for fault early warning of non-uniformly distributed fasteners under complex working conditions. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the main purpose of the present invention is to provide a method and system for fault early warning of non-uniformly distributed fasteners under complex working conditions, which can effectively solve the problems in the background art. The specific technical solutions of the present invention are as follows:

[0006] A method for fault early warning of non-uniformly distributed fasteners under complex working conditions includes:

[0007] Obtain the stress data and image data of the device to be detected;

[0008] Identify the area where the fasteners are distributed in the device according to the stress data and image data to obtain multiple distribution areas of the fasteners;

[0009] In each distribution area, combine the current state of the fastener and the influence of stress on the fastener state to predict the fastener state, and when the fastener state fails, give a fault early warning.

[0010] Specifically, identifying the areas where fasteners are distributed in the device based on the stress data and image data to obtain multiple distribution areas of the fasteners includes:

[0011] Identifying multiple initial areas of the fasteners according to the image data through a preset image recognition model;

[0012] Screening the multiple initial areas according to the stress data to obtain multiple distribution areas of the fasteners.

[0013] Specifically, identifying multiple initial areas of the fasteners according to the image data through a preset image recognition model includes:

[0014] Training the image recognition model using the image data with the positions of the fasteners marked to obtain a pre-trained image recognition model;

[0015] Inputting the current image data into the pre-trained image recognition model to identify multiple initial areas containing the fasteners.

[0016] Specifically, screening the multiple initial areas according to the stress data to obtain multiple distribution areas of the fasteners includes:

[0017] Extracting the stress values of the corresponding areas from the stress data within each initial area to obtain multiple initial area stress values;

[0018] Calculating the initial area stress characteristics based on the initial area stress values;

[0019] Setting a stress characteristic threshold according to the preset stress distribution pattern of the fasteners;

[0020] Eliminating the initial areas where the initial area stress characteristics are less than the stress characteristic threshold to obtain multiple distribution areas of the fasteners.

[0021] Specifically, within each distribution area, combining the current state of the fasteners and the influence of stress on the fastener state to predict the fastener state, and when the fastener state fails, giving a fault warning includes:

[0022] Identifying the current state of each fastener according to the image data within each distribution area;

[0023] Analyzing the influence of stress on the fastener state based on the stress data within a preset time period to predict the state of each fastener and obtain the predicted state of each fastener;

[0024] When the predicted state fails, giving a fault warning.

[0025] Specifically, within each distribution area, the state of the fasteners is identified based on the image data to obtain the current state of each fastener, including:

[0026] Using the image data containing various fastener states to train the fastener state recognition model, and obtaining a pre-trained fastener state recognition model;

[0027] Within each distribution area, according to the pre-trained fastener state recognition model, analyze the image data to obtain the current state of each fastener, where the current state includes normal, damaged, and detached.

[0028] Specifically, within a preset time period, based on the stress data analysis of the influence of stress on the fastener state, predict the state of each fastener to obtain the predicted state of each fastener, including:

[0029] Within a preset time period, calculate the stress gradient value by calculating the gradients of the stress data in space and time;

[0030] According to the stress gradient value, dynamically divide the working area of the fasteners to obtain a multi-level area division result;

[0031] Based on the multi-level area division result, obtain the average stress tensor of each area by performing tensor calculations on the stress in each area;

[0032] According to the average stress tensor of each area and combining the stress influence relationship between different areas, obtain a multi-area stress coupling model;

[0033] According to the multi-area stress coupling model, calculate the comprehensive stress of each area;

[0034] Based on the comprehensive stress of each area and combining the current state of each fastener, predict the change of the fastener state to obtain the predicted state of each fastener.

[0035] Specifically, based on the comprehensive stress of each area and combining the current state of each fastener, predict the change of the fastener state to obtain the predicted state of each fastener, including:

[0036] Based on the comprehensive stress of each area, obtain the abnormal stress direction vector of each area through a preset abnormal propagation path prediction model;

[0037] According to the abnormal stress direction vector, determine multiple abnormal areas;

[0038] Within each abnormal area, obtain the stress change trend of each fastener in the abnormal area through a preset stress distribution prediction model;

[0039] Analyze the stress at each moment within a preset time period according to the stress change trend to obtain the instantaneous stress value at each moment;

[0040] Within the preset time period, calculate the cumulative damage effect of the stress on the fasteners according to the stress change trend to obtain the stress cumulative effect;

[0041] Formulate a state mapping rule according to the stress change trend, the instantaneous stress value, and the stress cumulative effect;

[0042] According to the state mapping rule, combined with the current state of each fastener, predict the change situation of the fastener state to obtain the predicted state of each fastener.

[0043] Specifically, the step of calculating the cumulative damage effect of the stress on the fasteners according to the stress change trend within the preset time period to obtain the stress cumulative effect includes:

[0044] Within the preset time period, group the stress change trend according to a preset stress level to obtain multiple stress level groups;

[0045] Calculate the number of cycles of each stress level group;

[0046] Obtain the fatigue life of the fasteners corresponding to each stress level group according to the fastener material;

[0047] By calculating the ratio of the number of cycles to the fatigue life of the fasteners, obtain the damage rate of each stress level group;

[0048] Accumulate the damage rates of each stress level group to obtain the stress cumulative effect.

[0049] A fault warning system for non-uniformly distributed fasteners under complex working conditions, used to implement the method for fault warning of non-uniformly distributed fasteners under complex working conditions, includes:

[0050] A data acquisition module, which acquires the stress data and image data of the device to be detected;

[0051] A region recognition module, which recognizes the regions where the fasteners are distributed in the device according to the stress data and image data to obtain multiple distribution regions of the fasteners;

[0052] A fault warning module, within each distribution region, combined with the current state of the fasteners and the influence of the stress on the fasteners, predicts the state of the fasteners, and when the state of the fasteners fails, issues a fault warning.

[0053] Compared with the prior art, the present application has the following beneficial effects:

[0054] This application combines the stress data and image data of equipment under complex working conditions to identify the position and status of fasteners, can comprehensively and accurately monitor the status of fasteners of equipment under complex working conditions, obtain accurate position and status identification results, and predict the status change of fasteners based on the prediction results of the stress change trend of the equipment, can discover potential faults in advance and give early warnings, helps to improve the timeliness and reliability of equipment fault early warning, reduces equipment damage and production accidents caused by fastener faults, can carry out targeted maintenance and repair, avoid unnecessary over-maintenance, and reduce the maintenance cost of equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flowchart of the method for early warning of non-uniform distributed fastener faults under complex working conditions in Embodiment 1 of the present invention;

[0056] Figure 2 It is a schematic diagram of stress data acquisition in Embodiment 1 of the present invention;

[0057] Figure 3 It is a schematic diagram of abnormal area division of fasteners in Embodiment 1 of the present invention;

[0058] Figure 4 It is a flowchart of the calculation of stress accumulation effect in Embodiment 1 of the present invention;

[0059] Figure 5 It is a schematic diagram of the structure of the system for early warning of non-uniform distributed fastener faults under complex working conditions in Embodiment 2 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] In order to make the above objects, features and advantages of the present invention more obvious and understandable, the following detailed description of the specific embodiments of the present invention will be given in conjunction with the drawings in the specification.

[0061] Many specific details are set forth in the following description in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways different from those described herein, and those skilled in the art may make similar generalizations without departing from the spirit of the present invention, so the present invention is not limited by the specific embodiments disclosed below.

[0062] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure or characteristic that may be included in at least one implementation manner 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 separate or alternative embodiment that excludes other embodiments.

[0063] Embodiment 1

[0064] This embodiment provides a method for early warning of faults of non-uniformly distributed fasteners under complex working conditions, such as Figure 1 The method for early warning of faults of non-uniformly distributed fasteners under complex working conditions includes:

[0065] S101. Obtain the stress data and image data of the device to be detected;

[0066] S102. Identify the areas where the fasteners are distributed in the device according to the stress data and image data, and obtain multiple distribution areas of the fasteners;

[0067] S103. In each distribution area, combine the current state of the fastener and the influence of stress on the state of the fastener to predict the state of the fastener. When the state of the fastener fails, early warning of the fault is carried out.

[0068] During the operation of the device, it is crucial to identify the connection state of the fasteners. At present, the identification of the state of the fasteners is based on the image recognition method. This recognition method is based on a single type of data, and the accuracy of the identified fastener state is relatively low. Moreover, based on the current image data for identification, early warning cannot be realized. This application combines stress data and image data to identify the position and state of the fasteners, improves the accuracy of the identification results, and predicts the change of the state of the fasteners based on the change trend of the stress data and the damage effect of the fasteners, so as to realize early fault warning.

[0069] In this embodiment, such as Figure 2 During the working process of the device, select appropriate stress sensors (such as strain gauges, pressure sensors, etc.) and install them at key positions of the device. The key positions can be the center of the device or each working part of the device to measure the stress change during the working process of the device; select an industrial camera with appropriate resolution and frame rate, install it at a position where the fasteners can be clearly photographed, collect the image data of the device at regular time intervals, and transmit the image data to the data processing center through the network; obtaining stress data and image data at the same time can comprehensively understand the state of the fasteners from both mechanical and visual perspectives and improve the accuracy of fault judgment.

[0070] Specifically, identify the area of the fasteners according to the image data, and then combine the stress data to verify the area of the fasteners, and eliminate the areas that do not meet the stress conditions of the fasteners to obtain a more accurate distribution area of the fasteners; compared with the traditional method of only identifying the position of the fasteners according to the image data, this application adds stress verification, improves the accuracy of the fastener position identification result, can effectively reduce the misjudgment in image recognition, and improves the accuracy of fastener positioning.

[0071] After determining the distribution area of the fasteners, the status of each fastener is identified using image data and stress data respectively, and combined with the working status of the equipment, the stress change trend in the future time period is predicted to predict the status change of the fasteners; by analyzing the stress change trend and cumulative effect, it is predicted whether faults such as damage or detachment of the fasteners occur. When the status of the identified fastener is a fault, a fault warning mechanism is triggered to notify relevant personnel to take measures in a timely manner; compared with the traditional fastener fault detection method that only judges the status of the fasteners based on the current image, this application can predict the status change of the fasteners in advance through real-time monitoring and status prediction, timely discover the potential faults of the fasteners, avoid the expansion of faults, reduce the equipment downtime and maintenance costs; combining image data and stress data for fault judgment improves the accuracy and reliability of the warning.

[0072] This application combines the stress data and image data of the equipment under complex working conditions to identify the position and status of the fasteners, can comprehensively and accurately monitor the status of the fasteners of the equipment under complex working conditions, obtain accurate position and status recognition results, and predict the status change of the fasteners based on the prediction results of the equipment stress change trend, can discover potential faults in advance and give warnings, which helps to improve the timeliness and reliability of equipment fault warnings, reduce equipment damage and production accidents caused by fastener faults, can carry out targeted maintenance and overhaul, avoid unnecessary over-maintenance, and reduce the maintenance cost of the equipment.

[0073] Further, the method for identifying the distribution area of the fasteners in the equipment according to the stress data and image data to obtain multiple distribution areas of the fasteners includes:

[0074] S201. According to the image data, through a preset image recognition model, multiple initial areas of the fasteners are identified;

[0075] S202. According to the stress data, the multiple initial areas are screened to obtain multiple distribution areas of the fasteners.

[0076] The traditional method for identifying the position of fasteners relies on a single image recognition method. The recognition method is single and only depends on the accuracy of the image recognition model, and cannot guarantee the accuracy of the identified fastener position result. In the embodiment of this application, based on the image recognition result, the stress data is used to further screen the image recognition result, which can effectively remove the misjudged areas in the image recognition, improve the accuracy of the fastener distribution area positioning, and combine the objective physical quantity of stress data for screening, considering multi-dimensional data information, making the obtained distribution area more reliable and reducing misjudgments caused by factors such as image noise and illumination.

[0077] In this embodiment, based on the machine learning algorithm, a large number of labeled fastener image data are used to train the machine learning model, enabling the model to learn the feature patterns of fasteners in the images and obtaining a well-trained image recognition model with high accuracy. The model can master the features that distinguish fasteners from other objects, such as shape, texture, color distribution, etc. When new image data are input, the model searches and locates multiple initial regions containing fasteners in the image according to the learned features; through the machine learning model, the region including the fastener can be quickly located in the image, improving the processing efficiency.

[0078] Specifically, in combination with the stress data of each region, the initial regions are further screened to remove misjudged regions and improve the accuracy of locating the fastener distribution region; the stress data reflect the stress distribution on the object surface. During the operation of the device, the fasteners will bear corresponding stresses, and stress concentration phenomena will occur around the fasteners. By analyzing the stress data, it is judged whether there is obvious stress concentration in the initial region. If so, it is considered that the initial region contains real fasteners and is retained as the distribution region; if not, it is excluded, thereby improving the accuracy of fastener location.

[0079] Further, the recognition of multiple initial regions of fasteners according to the image data through a preset image recognition model includes:

[0080] S301. Use a large number of image data with labeled fastener positions to train the image recognition model to obtain a pre-trained image recognition model;

[0081] S302. Input the current image data into the pre-trained image recognition model to recognize multiple initial regions containing fasteners.

[0082] In this embodiment, a large number of image data containing fasteners and with the fastener positions labeled using professional image annotation tools are used to train the image recognition model. Specifically, the image recognition model is the YOLOv5 model. The labeled image data set is divided into a training set and a test set. Among them, 70% of the data is used as the training set, and 30% is used as the test set. The training set is used for the model to learn features, and the test set is used to evaluate the final performance of the trained model. After training, a pre-trained image recognition model is obtained. Through the training with a large amount of labeled data, the model can learn rich features of fasteners, so as to more accurately identify the positions of fasteners in practical applications. Through the training with a large amount of labeled data, the model can learn rich features of fasteners and more accurately identify the positions of fasteners.

[0083] Specifically, the current image data is input into a pre-trained image recognition model. The image recognition model analyzes the input image according to the learned fastener feature patterns and position information, extracts the features therein, and matches them with the features learned during the training process. According to the matching results, the model marks the areas containing fasteners in the image, and these areas are the initial areas. The pre-trained model can process and recognize the input image in a short time, quickly obtain the initial areas containing fasteners, improving work efficiency; the entire recognition process is automatically completed by the model, reducing manual intervention and lowering labor costs and human errors.

[0084] Further, screening the multiple initial areas according to the stress data to obtain multiple distribution areas of fasteners includes:

[0085] S401. In each initial area, extract the stress value of the corresponding area from the stress data to obtain multiple initial area stress values;

[0086] S402. Calculate the initial area stress characteristics according to the initial area stress values;

[0087] S403. Set a stress characteristic threshold according to the preset fastener stress distribution pattern;

[0088] S404. Eliminate the initial areas whose initial area stress characteristics are less than the stress characteristic threshold to obtain multiple distribution areas of fasteners.

[0089] In this embodiment, the initial areas are screened according to the stress distribution of the identified initial areas to improve the accuracy of fastener position recognition. According to the stress information of the position obtained by the stress sensor and combining the distance between the fastener position and the stress sensor, the stress value of each initial area is obtained. The stress data is a two-dimensional matrix, and each element represents the stress value at the corresponding position in the image. According to the coordinate range of the initial area, all the stress values in the corresponding area are extracted from the stress data matrix.

[0090] Specifically, by calculating the stress characteristics of the stress values, information such as the stress concentration degree and change trend in this area can be comprehensively reflected, and it can be more accurately judged whether there is a stress concentration phenomenon caused by fasteners in this area. Specific stress characteristic indicators include average stress, maximum stress, stress standard deviation, etc. Calculate the stress characteristics of each initial area to more comprehensively describe the stress state of the initial area; obtain the stress distribution pattern of fasteners under different working conditions through experimental measurement, numerical simulation or empirical summary, etc., and find the key parameters of the stress characteristics in this pattern, such as the range of average stress, the upper limit of maximum stress, etc. According to the analysis results, set corresponding thresholds for each stress characteristic indicator. By setting the characteristic thresholds, areas that, although within the initial area, do not conform to the stress distribution pattern of fasteners in terms of stress characteristics can be excluded, reducing misjudgment.

[0091] Specifically, screen the initial areas according to the set stress characteristic thresholds, compare the stress characteristics of each initial area with the corresponding thresholds. When the stress characteristics of the initial area are less than the preset thresholds, it indicates that the stress concentration degree in this area is relatively low and does not conform to the stress distribution pattern that fasteners should have when stressed. The corresponding areas are not the real fastener distribution areas. Eliminate these areas to obtain a more accurate fastener distribution area, thereby improving the accuracy of fastener distribution area positioning.

[0092] Furthermore, within each distribution area, combine the current state of the fasteners and the influence of stress on the fastener state to predict the fastener state. When a fault occurs in the fastener state, a fault warning is issued, including:

[0093] S501. Within each distribution area, identify the fastener state according to the image data to obtain the current state of each fastener;

[0094] S502. Within a preset time period, analyze the influence of stress on the fastener state according to the stress data, predict the state of each fastener, and obtain the predicted state of each fastener;

[0095] S503. When a fault occurs in the predicted state, issue a fault warning.

[0096] Traditional methods for identifying fastener states analyze the real-time state of fasteners based on the current image and issue an alarm when a fault occurs in the real-time state, unable to give an early warning of the fault situation; this application identifies the current state of fasteners and predicts the state change of fasteners in the future time period to achieve early warning, issue a warning before the fault occurs, can timely detect and handle the potential faults of fasteners, and helps to ensure the safe and stable operation of the equipment.

[0097] In this embodiment, within each distribution area, the state of the fasteners in the image data is identified. The image data contains the appearance information of the fasteners. Different states will present different visual features in the image. For example, a loose fastener will show position deviation, increased gap, etc., and a damaged fastener will have obvious features such as fracture and deformation. Based on the image recognition method, the current state of the fasteners is identified through the image feature information, and the current state of each fastener is obtained.

[0098] Specifically, as the working time of the equipment increases and the stress effect accumulates, the state of the fasteners will change accordingly. For example, when a fastener is loose, the stress distribution it bears will change; when a fastener is about to be damaged, the stress will show abnormal fluctuations or accumulations. According to the working process of the equipment, the stress change trend of the equipment is predicted, and based on the stress change trend, the state change of the fasteners is predicted. By analyzing the stress data sequence within a preset time period and using the time series analysis method, a prediction model between the stress data and the fastener state is established, so as to predict the state of the fasteners in the future for a period of time, and the predicted state of each fastener is obtained.

[0099] Specifically, when the state of the fasteners predicted by the stress data is a fault state, relevant personnel are reminded to take measures in time, triggering a fault warning mechanism. The warning mechanism can be realized in various ways, such as SMS notification, email reminder, etc., to ensure that relevant personnel can learn the fault information in time. By combining the image data and the stress data, the current state of the fasteners can be intuitively identified, and the future state can be predicted through the stress data, improving the comprehensiveness and accuracy of fault detection. Through the prediction function of the stress data, a warning can be issued before the fault occurs, providing sufficient time for the maintenance and repair of the equipment, realizing the transformation from passive maintenance to active prevention.

[0100] Further, the identifying the state of the fasteners according to the image data in each distribution area to obtain the current state of each fastener includes:

[0101] S601. Using a large amount of image data containing various fastener states to train a fastener state recognition model, and obtaining a pre-trained fastener state recognition model;

[0102] S602. In each distribution area, analyzing the image data according to the pre-trained fastener state recognition model to obtain the current state of each fastener, where the current state includes normal, damaged, and detached.

[0103] In this embodiment, based on the deep learning model, the status of the fasteners in each distribution area is identified to obtain the current status of the fasteners. Through the deep learning model, the current status of each fastener can be quickly and accurately identified in the image. In industrial production, timely detection of damage or falling off of fasteners can avoid equipment failures and safety accidents, and ensure the safety and stability of the production process.

[0104] Specifically, a large number of images containing fasteners in various annotated states (normal, damaged, and fallen off) are used to train the fastener state recognition model. These images should be diverse, including different lighting conditions, shooting angles, background environments, etc., to improve the generalization ability of the model. The annotated image data set is divided into a training set and a test set. 70% of the image data is used as the training set and 30% of the image data is used as the test set to train the fastener state recognition model. Specifically, the fastener state recognition model is a CNN model. CNN can automatically extract meaningful features from images. Through training with a large amount of annotated image data, the model can learn the characteristic patterns of fasteners in different states (normal, damaged, and fallen off). When processing new images, the model can accurately identify the state of the fastener based on the learned features.

[0105] Specifically, the pre-trained fastener state recognition model analyzes the input image data in each distribution area according to the learned characteristic patterns of fasteners in different states, extracts the features, and matches them with the features learned during the training process. Based on the matching results, the model outputs the current state of each fastener (normal, damaged, or detached). The pre-trained model can process and analyze the input images in a short time, quickly obtain the current state of each fastener, and improve work efficiency; the entire recognition process is automatically completed by the model, reducing manual intervention, labor costs, and human errors.

[0106] Furthermore, within the preset time period, the influence of stress on the state of the fastener is analyzed according to the stress data, and the state of each fastener is predicted to obtain the predicted state of each fastener, including:

[0107] S701, obtaining a stress gradient value by calculating the gradient of stress data in space and time within a preset time period;

[0108] S702, dynamically dividing the working area of ​​the fastener according to the stress gradient value to obtain a multi-level area division result;

[0109] S703, based on the multi-level region division result, performing tensor calculation on the stress in each region to obtain an average stress tensor of each region;

[0110] S704, obtaining a multi-region stress coupling model according to the average stress tensor of each region and the stress influence relationship between different regions;

[0111] S705, calculating the comprehensive stress of each region according to the multi-region stress coupling model;

[0112] S706: Based on the comprehensive stress of each region and in combination with the current state of each fastener, predict the change of the fastener state to obtain the predicted state of each fastener.

[0113] In this embodiment, the stress data during the operation of the equipment is analyzed, and the fasteners under complex working conditions are analyzed by region. This can reduce the difficulty of analysis, clarify the status of fasteners in different regions, and analyze the stress influence between different regions in combination with the stress coupling relationship between different regions. This can more accurately capture the status changes of the fasteners, and when the fastener status is abnormal, an early warning can be issued in time.

[0114] Specifically, within a preset time period, the finite difference method is used to analyze the stress data and calculate the gradient change of stress. By calculating the gradient of stress in space and time, the changing trend of stress at different locations and at different times can be comprehensively considered, so as to analyze the stage of drastic stress change; according to the calculated stress gradient value, a gradient threshold is set, and the specific gradient threshold can be set according to the actual calculation accuracy requirements. When the gradient value in a certain area is greater than the gradient threshold and the stress continues to rise, it is divided into a high-risk area; if the gradient value is less than the gradient threshold and the stress tends to be stable, it is divided into a low-risk area. The high-risk area indicates that the stress changes drastically and there is a high risk of failure; the stress in the low-risk area is relatively stable, and different levels of risk areas are obtained through dynamic division. Different monitoring and analysis strategies can be adopted for different areas to make the analysis more targeted.

[0115] Specifically, after the fasteners under complex working conditions are divided into regions, the stress in each divided region is tensor calculated to calculate the average stress tensor of each region. The calculation formula is as follows:

[0116] ;

[0117] In the formula, is the average stress tensor, N is the number of data points in the region, is the stress tensor component of the kth data point. In three-dimensional space, i and j take values ​​of 1, 2, and 3, representing the stress tensors in various directions. The stress tensor contains 9 components in three-dimensional space. By calculating the average stress tensor in each region, the influence of local stress fluctuations in the region can be eliminated, and the representative stress state of the region can be obtained.

[0118] Furthermore, analyze the stress influence relationship between different regions, combine the average stress tensors of each region, calculate the comprehensive stress of each region to obtain a more accurate comprehensive stress value. According to the mutual influence relationship of the stress between different regions of the fastener, establish a coupling coefficient matrix to describe the stress influence degree between different regions, combine the average stress tensors of each region, and calculate the comprehensive stress of each region. The calculation formula is as follows:

[0119] ;

[0120] In the formula, is the comprehensive stress tensor of region m considering coupling, is the average stress tensor of region m, is the average stress tensor of region n, is the stress influence degree between region m and region n. Based on the multi-region stress coupling model, by superimposing the self-stress of each region and the coupling stress of other regions on it, a more realistic comprehensive stress value of each region can be obtained. Taking the comprehensive stress of each region as an influencing factor, combined with the current state of the fastener, predict the future state of the fastener. Considering stress and current state comprehensively, accurately predict the state change of the fastener; by combining the current state and stress data of the fastener for prediction, it avoids the limitation of judging based on a single factor and improves the accuracy of state prediction; it can predict potential failure risks in advance before the actual change of the fastener state, providing a basis for preventive maintenance of the equipment.

[0121] Furthermore, based on the comprehensive stress of each region, combined with the current state of each fastener, predict the state change of the fastener to obtain the predicted state of each fastener, including:

[0122] S801. Based on the comprehensive stress of each region, through a preset abnormal propagation path prediction model, obtain the abnormal stress direction vector of each region;

[0123] S802. Determine multiple abnormal regions according to the abnormal stress direction vector;

[0124] S803. In each abnormal region, through a preset stress distribution prediction model, obtain the stress change trend of each fastener in the abnormal region;

[0125] S804. According to the stress change trend, analyze the stress at each moment in a preset time period to obtain the instantaneous stress value at each moment;

[0126] S805. Within a preset time period, calculate the cumulative damage effect of stress on the fastener according to the stress change trend to obtain the stress cumulative effect;

[0127] S806. Develop a state mapping rule based on the stress change trend, instantaneous stress value, and stress accumulation effect;

[0128] S807. Predict the change in the fastener state based on the state mapping rule and the current state of each fastener to obtain the predicted state of each fastener.

[0129] In this embodiment, as Figure 3 , determine the abnormal area according to the stress direction vector of each area. In the figure, according to the abnormal stress direction vector, the dotted area is determined as the abnormal area. Analyze the state of the fasteners in the abnormal area to achieve accurate area analysis. Conduct personalized analysis on the abnormal stress conditions of the fasteners in the abnormal area, which can improve the accuracy of fastener state prediction, reduce misjudgment and missed judgment, and at the same time avoid analyzing all fasteners, improving the efficiency of abnormal state detection.

[0130] Specifically, train the abnormal propagation path prediction model with a large amount of historical data and stress change data to obtain a preset abnormal propagation path prediction model, and learn the mode of stress anomaly propagation between regions under different comprehensive stress conditions. In this embodiment, the abnormal propagation path prediction model is a Bayesian network model based on machine learning. Input the comprehensive stress data of each area into the preset abnormal propagation path prediction model, and the model outputs the predicted direction of abnormal stress occurrence to obtain the abnormal stress direction vector of each area; the abnormal stress direction vector indicates the direction of stress anomaly propagation. According to the predicted abnormal stress direction vector, determine the abnormal area with stress anomaly from the fasteners in complex working conditions. When the abnormal stress direction vector of a certain area exceeds a certain threshold, it indicates that this area is greatly affected by abnormal stress, and it is determined as the abnormal area; by quickly locating the area greatly affected by abnormal stress, in-depth analysis can be carried out on the abnormal area, improving the efficiency and accuracy of fault prediction.

[0131] Specifically, a large amount of historical stress data of abnormal regions is used to train the stress distribution prediction model, and a preset stress distribution prediction model is obtained. In this embodiment, the stress distribution prediction model is an LSTM model. This model learns the stress distribution characteristics within the region, the positions of the fasteners, and the mechanical relationships between them. By inputting the characteristics of the abnormal region and the relevant information of the fasteners within the region, the stress distribution prediction model can predict the stress change trend of each fastener in the future for a period of time, including whether the stress increases, decreases, or remains stable, etc. The stress change trend reflects the change law of stress over time. According to the predicted stress change trend, the stress values corresponding to each moment over time are calculated to obtain the corresponding instantaneous stress values. Through the instantaneous stress values, the impact of instantaneous stress changes on the fasteners can be evaluated, and the stress state of the fasteners within the preset time period can be understood more accurately, which helps to detect in a timely manner the situation where the fasteners are damaged due to stress peaks.

[0132] Specifically, within the preset time period, according to the predicted stress change trend, the total damage degree suffered by the fasteners during this period is calculated. The fastener material will generate fatigue damage under the action of alternating stress. The damage caused to the fasteners by the accumulation of stress over time is calculated, so as to evaluate the change situation of the fastener state. By considering the cumulative effect of stress, the damage degree of the fasteners within the preset time period can be evaluated more comprehensively, avoiding only focusing on instantaneous stress and ignoring the influence of long-term damage, and detecting potential failures such as fatigue damage existing in the fasteners in advance, providing a basis for preventive maintenance.

[0133] Specifically, after obtaining the stress change trend, the instantaneous stress value, and the stress cumulative effect, comprehensively considering the three influencing factors of the stress change trend, the instantaneous stress value, and the stress cumulative effect, the state of the fasteners is predicted, and corresponding state mapping rules are set. According to the stress change trend, the corresponding stress change rate is calculated, and the calculation formula is as follows:

[0134] ;

[0135] In the formula, is the stress change rate per unit time, is the stress value at the current moment, is the stress value at the previous moment, is the preset unit time interval. According to the state of the fasteners under actual working conditions, a fault threshold is set to obtain the stress change rate threshold, the instantaneous stress threshold, and the stress cumulative effect threshold. According to the stress change trend, the instantaneous stress, and the stress cumulative effect, combined with the current state of each fastener, state mapping rules are formulated, and the state mapping rules are as follows:

[0136] The current state is normal, and the stress change rate of the predicted stress change trend is less than the stress change rate threshold, the instantaneous stress value is less than the instantaneous stress threshold, and the stress accumulation effect is less than the stress accumulation effect threshold, then the predicted state is still normal;

[0137] The current state is normal, but the stress change rate of the predicted stress change trend is greater than the stress change rate threshold, or the predicted instantaneous stress value is greater than the instantaneous stress threshold, or the stress accumulation effect is greater than the stress accumulation effect threshold, then the predicted state is damaged;

[0138] The current state is damaged, and the predicted instantaneous stress value is greater than the instantaneous stress threshold, or the stress accumulation effect is greater than the stress accumulation effect threshold, then the predicted state is detachment. Different thresholds are set to determine the state of the fastener; the stress change trend reflects the dynamic change of stress over time, the instantaneous stress value reflects the magnitude of stress at the current moment, and the stress accumulation effect takes into account the damage accumulation under long-term stress; combined with the current state of the fastener, the future state of the fastener can be predicted more accurately. When the predicted state shows damage or detachment, an early warning is issued in a timely manner, and the staff can replace the corresponding fastener in a timely manner.

[0139] Furthermore, as Figure 4 stated, within the preset time period, the cumulative damage effect of stress on the fastener is calculated according to the stress change trend to obtain the stress accumulation effect, including:

[0140] S901. Within the preset time period, group the stress change trend according to the preset stress level to obtain multiple stress level groups;

[0141] S902. Calculate the number of cycles for each stress level group;

[0142] S903. According to the fastener material, obtain the fatigue life of the fastener corresponding to each stress level group;

[0143] S904. By calculating the ratio of the number of cycles to the fatigue life of the fastener, obtain the damage rate for each stress level group;

[0144] S905. Accumulate the damage rates of each stress level group to obtain the stress accumulation effect.

[0145] In this embodiment, the cumulative damage effect of the fastener is calculated according to the stress change trend. By calculating the damage situation using the Miner's rule, the fatigue damage degree caused by the stress on the fastener within the preset time period can be accurately evaluated. Specifically, the fatigue damage characteristics of the fastener are different under different stress levels. The stress change trend is grouped according to the preset stress levels, and the damage situations under different stress levels are calculated respectively to obtain the influence of different stress intervals on the fastener. According to the design requirements, material characteristics, and actual working condition experience of the fastener, a series of preset stress levels are determined. Within the preset time period, the stress values in the stress change trend are read one by one, and each stress value is classified into the corresponding preset stress level interval to obtain multiple stress level groups.

[0146] Specifically, at each stress level, the cumulative effect is calculated. Within each stress level group, the stress will cycle. The fatigue damage caused by the stress to the fastener is closely related to the number of stress cycles. The number of complete stress cycles within each stress level group is counted to obtain the stress cycle number. Specifically, the rain flow counting method is used to calculate the stress cycle number. The rain flow counting method finds the peak and valley points of the stress by comparing the magnitude relationship of adjacent stress values, and determines the cycle number according to the number of peak and valley points. The cycle number is one of the key parameters for calculating fatigue damage. Accurately calculating the cycle number of each stress level group can more precisely evaluate the damage effect of this stress level on the fastener.

[0147] Specifically, different materials of the fastener have different fatigue lives under different stress levels. According to the fastener material, the calculation method of the fatigue life of the fastener under different stress actions can be determined. Specifically, the stress-life curve (S-N curve) of the corresponding fastener material can be obtained by referring to material manuals, experimental data, or relevant research literature. According to the stress-life curve, the fatigue life of the fastener can be calculated. The calculation formula is as follows:

[0148] ;

[0149] In the formula, N is the fatigue life, S is the stress level, and z and C are constants related to the fastener material, which need to be determined according to the specific fastener material. The fatigue life of the corresponding fastener is calculated through the formula. The fatigue performances of different materials vary greatly. Determining the fatigue life according to the fastener material can more accurately reflect the fatigue damage situation of the fastener in the actual situation.

[0150] Specifically, according to Miner's rule, the cumulative fatigue damage of the fastener under different stress levels can be calculated. The damage rate of each stress level group is defined as the ratio of the number of cycles of that stress level group to the corresponding fatigue life, indicating the degree of damage caused by that stress level group to the fastener within a preset time period. The damage rate provides a quantitative index for the damage of each stress level group, facilitating the comparison of the damage magnitudes of different stress levels to the fastener. After obtaining the damage rate of each stress level group, the total fatigue damage of the fastener can be calculated. According to Miner's rule, the total fatigue damage of the material is the linear accumulation of the damages at each stress level. By summing up the damage rates of each stress level group, the cumulative damage effect of the stress on the fastener within the preset time period can be obtained, reflecting the overall fatigue damage degree of the fastener during this time period. The specific calculation formula is as follows:

[0151] ;

[0152] In the formula, D is the stress cumulative damage effect, Q is the number of stress level groups, is the number of cycles of the p-th stress level group, is the fatigue life corresponding to the p-th stress level group. By calculating the stress cumulative effect, the damage effects of different stress levels on the fastener within the preset time period are comprehensively considered, enabling a comprehensive assessment of the fatigue damage degree of the fastener, and then determining whether the fastener is approaching fatigue failure, so as to conduct early fault warning and maintenance decision-making.

[0153] Embodiment 2

[0154] In this embodiment, as Figure 5 , a fault warning system for non-uniformly distributed fasteners under complex working conditions is provided, which is used to implement the above-mentioned fault warning method for non-uniformly distributed fasteners under complex working conditions, including:

[0155] A data acquisition module that acquires stress data and image data of the device to be detected;

[0156] A region recognition module that recognizes the regions where the fasteners are distributed in the device according to the stress data and image data, and obtains multiple distribution regions of the fasteners;

[0157] A fault warning module that predicts the state of the fastener by combining the current state of the fastener and the influence of stress on the state of the fastener in each distribution region, and issues a fault warning when the state of the fastener fails.

[0158] In this embodiment, the data acquisition module includes a stress sensor, an image acquisition device, a data transmission unit, and a data preprocessing sub-module, which can acquire the stress data and image data of the equipment under complex working conditions in real time and accurately, and preliminarily process and transmit the data, providing a reliable data basis for subsequent area recognition and fault warning. Through the collaborative work of multiple sensors and flexible data transmission methods, it is ensured that the required data can be effectively collected under different complex working conditions.

[0159] Specifically, the area recognition module includes an image recognition sub-module, a stress analysis sub-module, and an area screening and decision-making unit, which accurately identify the distribution area of the fasteners by combining the stress data and image data. The fastener area is initially determined through image recognition, and then these areas are screened and verified using the stress data to remove misjudged areas, and finally a reliable fastener distribution area is obtained, providing accurate position information for subsequent fault warning; the fault warning module includes a status recognition sub-module, a stress prediction sub-module, a status prediction sub-module, and a warning trigger unit, which accurately identify the status of the fasteners in each distribution area and predict their future status based on the stress data. When it is predicted that the fastener status fails, a fault warning is issued in a timely manner to remind relevant personnel to take corresponding measures to avoid equipment failures and safety accidents and ensure the normal operation of the equipment and the continuity of production.

[0160] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for early warning of non-uniformly distributed fastener failures in complex working conditions, characterized in that: include: Acquire stress data and image data of the equipment to be tested; Identify the distribution area of ​​fasteners in the device according to the stress data and the image data to obtain multiple distribution areas of the fasteners; In each distribution area, the fastener state is predicted by combining the current state of the fastener and the influence of stress on the fastener state, and when the fastener state fails, a fault warning is issued.

2. The method for early warning of non-uniformly distributed fastener failures in complex working conditions according to claim 1 is characterized in that: The method of identifying the distribution area of ​​fasteners in the device according to the stress data and the image data to obtain multiple distribution areas of fasteners includes: According to the image data, a plurality of initial regions of the fastener are identified by using a preset image recognition model; The multiple initial regions are screened according to the stress data to obtain multiple distribution regions of the fastener.

3. The method for early warning of non-uniformly distributed fastener failures in complex working conditions according to claim 2 is characterized in that: The method of identifying a plurality of initial regions of the fastener based on the image data by using a preset image recognition model includes: Using the image data with the fastener positions marked, the image recognition model is trained to obtain a pre-trained image recognition model; The current image data is input into a pre-trained image recognition model to identify multiple initial regions containing fasteners.

4. The method for early warning of non-uniformly distributed fastener failures in complex working conditions according to claim 2 is characterized in that: The method of screening the multiple initial regions according to the stress data to obtain multiple distribution regions of the fasteners includes: In each initial region, the stress value of the corresponding region is extracted from the stress data to obtain multiple initial region stress values; According to the initial regional stress value, the initial regional stress characteristic is calculated; According to the preset fastener stress distribution mode, the stress characteristic threshold is set; Initial regions whose stress characteristics of the initial regions are less than the stress characteristic threshold are eliminated to obtain a plurality of distribution regions of the fastener.

5. The method for early warning of non-uniformly distributed fastener failures in complex working conditions according to claim 1, characterized in that: In each distribution area, the fastener state is predicted by combining the current state of the fastener and the influence of stress on the fastener state, and when the fastener state fails, a fault warning is performed, including: In each distribution area, the fastener status is identified based on the image data to obtain the current status of each fastener; Within a preset time period, the influence of stress on the state of fasteners is analyzed according to stress data, and the state of each fastener is predicted to obtain the predicted state of each fastener; When a fault occurs in the predicted state, a fault warning is issued.

6. The method for early warning of non-uniformly distributed fastener failures in complex working conditions according to claim 5 is characterized in that: In each distribution area, the fastener status is identified according to the image data to obtain the current status of each fastener, including: Using image data containing various fastener states, a fastener state recognition model is trained to obtain a pre-trained fastener state recognition model; In each distribution area, the image data is analyzed according to the pre-trained fastener state recognition model to obtain the current state of each fastener, wherein the current state includes normal, damaged and fallen off.

7. The method for early warning of non-uniformly distributed fastener failures in complex working conditions according to claim 5, characterized in that: The method of analyzing the influence of stress on the state of fasteners according to stress data within a preset time period, predicting the state of each fastener, and obtaining the predicted state of each fastener includes: In a preset time period, the stress gradient value is obtained by calculating the gradient of stress data in space and time; According to the stress gradient value, the working area of ​​the fastener is dynamically divided to obtain a multi-level area division result; Based on the multi-level region division result, the stress in each region is calculated as a tensor to obtain an average stress tensor of each region; According to the average stress tensor of each region and the stress influence relationship between different regions, a multi-region stress coupling model is obtained; According to the multi-region stress coupling model, the comprehensive stress of each region is calculated; Based on the comprehensive stress in each area and combined with the current state of each fastener, the fastener state change is predicted to obtain the predicted state of each fastener.

8. The method for early warning of non-uniformly distributed fastener failures in complex working conditions according to claim 7, characterized in that: The method predicts the change of the fastener state based on the comprehensive stress of each region and the current state of each fastener to obtain the predicted state of each fastener, including: Based on the comprehensive stress of each area, the abnormal stress direction vector of each area is obtained through the preset abnormal propagation path prediction model; Determining a plurality of abnormal regions according to the abnormal stress direction vector; In each abnormal area, the stress change trend of each fastener in the abnormal area is obtained through the preset stress distribution prediction model; According to the stress variation trend, the stress at each moment of the preset time period is analyzed to obtain the instantaneous stress value at each moment; In a preset time period, the cumulative damage effect of stress on the fastener is calculated according to the stress change trend to obtain the stress accumulation effect; Formulate state mapping rules according to the stress change trend, instantaneous stress value and stress accumulation effect; According to the state mapping rule, combined with the current state of each fastener, the fastener state change is predicted to obtain the predicted state of each fastener.

9. The method for early warning of non-uniformly distributed fastener failures in complex working conditions according to claim 8, characterized in that: The method of calculating the cumulative damage effect of stress on the fastener according to the stress change trend within the preset time period to obtain the stress accumulation effect includes: Within a preset time period, the stress change trends are grouped according to preset stress levels to obtain multiple stress level groups; Calculate the number of cycles for each stress level group; According to the fastener material, the fastener fatigue life corresponding to each stress level group is obtained; The damage rate of each stress level group is obtained by calculating the ratio of the number of cycles to the fatigue life of the fastener; The damage rate of each stress level group is accumulated to obtain the stress accumulation effect.

10. A non-uniformly distributed fastener fault warning system for complex working conditions, characterized in that: The method for realizing the early warning method for non-uniformly distributed fastener failures in complex working conditions as claimed in claims 1 to 9 comprises: A data acquisition module, which acquires stress data and image data of the device to be tested; A region identification module, which identifies the regions where fasteners are distributed in the equipment according to the stress data and the image data, and obtains a plurality of distribution regions of the fasteners; The fault warning module predicts the state of the fastener in each distribution area by combining the current state of the fastener and the influence of stress on the state of the fastener, and issues a fault warning when the fastener state fails.

Images