A quality inspection method and system
By establishing a library of product quality requirements and configuration test solutions, combining conventional and destructive testing, and using the correlation analysis network to generate comprehensive quality inspection results, the problem of limited sampling ratio in the quality inspection of high-damping shock-absorbing rubber gloves is solved, and the entire batch is achieved Accurate evaluation of sub-product quality.
Patent Information
- Application Number
- CN202411119178.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-08-15
AI Technical Summary
The quality inspection of high-damping shock-absorbing rubber gloves in the prior art is difficult to accurately evaluate the quality status of the entire batch of products due to the limited sampling ratio of destructive detection.
By establishing the product quality requirements of the target product, configuring the test solution library, combining routine inspection and destructive inspection, using the correlation analysis network to perform data correlation, generating comprehensive quality inspection results, inferring the performance indicators of the gloves that have not been destructively tested, and achieving completeness Accurate evaluation of batch product quality.
On the basis of limited damage detection, a comprehensive and accurate assessment of the quality of the entire batch of products was achieved, the limitations of insufficient destructive sampling samples were overcome, and the accuracy and comprehensiveness of the detection were improved.
Smart Images

Figure CN118656757B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of quality inspection, and in particular to a quality inspection method and system. Background Art
[0002] High-damping shock-absorbing rubber gloves are personal protective equipment widely used in the fields of vibration tool operation, heavy machinery assembly, etc. Their shock-absorbing performance, appearance quality and durability are directly related to the occupational safety and health of users. In order to ensure the quality of glove products, it is necessary to conduct factory quality inspection on mass-produced gloves. In the prior art, high-damping shock-absorbing rubber gloves are mainly quality inspected through comprehensive routine inspection and sampling destructive inspection. After all gloves are routinely inspected, a certain proportion of samples are taken for destructive inspection, and then the quality level of the entire batch of products is inferred and evaluated through proportion calculation. However, since destructive inspection will cause permanent damage to the samples, the sampling ratio is usually low, which limits the accuracy of the quality inspection results and makes it difficult to fully reflect the true quality status of the entire batch of products. Summary of the invention
[0003] The present application provides a quality inspection method and system, aiming to solve the technical problem in the prior art that it is difficult to accurately evaluate the quality of an entire batch of products due to the limited sampling ratio of destructive inspection.
[0004] In view of the above problems, the present application provides a quality detection method and system.
[0005] The first aspect disclosed in the present application provides a quality inspection method, which includes: establishing product quality requirements for a target product, wherein the product quality requirements include shock absorption performance requirements, appearance requirements, and aging durability requirements; configuring a test solution library through big data, matching test solutions based on product quality requirements, and establishing a test solution set, wherein the test solution set includes a destruction solution set and a conventional inspection solution set, and the destruction solution set is provided with a random sampling ratio; performing a full product inspection of the target product with the conventional inspection solution set, establishing a product full inspection data set, clustering quality inspections of the target product based on the product full inspection data set, and generating a full inspection quality inspection result with the quality inspection clustering result; performing random sampling of the target product based on the random sampling ratio, and establishing a mapping identifier between the random sampling result and the quality inspection clustering result, performing a destruction solution set inspection of the random sampling result, and establishing a random inspection quality inspection result; inputting the test solution set into an association analysis network, performing an association analysis between the conventional inspection solution set and the destruction inspection solution set, performing a full product satisfaction prediction of the destruction inspection based on the association analysis result, the random inspection quality inspection result, and the mapping identifier, and generating a comprehensive quality inspection result according to the satisfaction prediction result, the full inspection quality inspection result, and the product quality requirements.
[0006] Another aspect disclosed in the present application provides a quality inspection system, which includes: a demand establishment module, which is used to establish product quality requirements for a target product, wherein the product quality requirements include shock absorption performance requirements, appearance requirements, and aging durability requirements; a test scheme matching module, which is used to configure a test scheme library through big data, match test schemes based on product quality requirements, and establish a test scheme set, wherein the test scheme set includes a destruction scheme set and a conventional inspection scheme set, and the destruction scheme set is provided with a random sampling ratio; a conventional inspection module, which is used to perform a full product inspection of the target product with the conventional inspection scheme set, establish a full product inspection data set, and perform quality control of the target product based on the full product inspection data set. Quantity detection clustering, using quality detection clustering results to generate full inspection quality detection results; destruction detection module, used to perform random sampling of target products based on random sampling ratios, and establish mapping identifiers between random sampling results and quality detection clustering results, perform destruction scheme set detection of random sampling results, and establish random inspection quality detection results; detection result generation module, used to input the test scheme set into the association analysis network, perform association analysis between conventional detection scheme set and destruction detection scheme set, perform full product satisfaction prediction of destruction detection based on association analysis results, random inspection quality detection results and mapping identifiers, and generate comprehensive quality detection results based on satisfaction prediction results, full inspection quality detection results and product quality requirements.
[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:
[0008] Due to the adoption of establishing product quality requirements for target products, the evaluation dimensions and indicator system for quality testing of high-damping shock-absorbing rubber gloves are established, providing a basis for matching and quality evaluation of the test solution set; configuring the test solution library through big data, matching test solutions based on product quality requirements, and establishing a test solution set, wherein the test solution set includes a destruction solution set and a conventional detection solution set, and the destruction solution set is set with a random sampling ratio to provide support for subsequent quality testing; performing routine full inspections on all glove products, obtaining a complete product full inspection data set, and performing quality clustering analysis to reveal the inherent distribution law of product quality and generate full inspection quality test results; performing destructive sampling testing, and establishing a mapping relationship between the sampling samples and the conventional detection quality clustering results, integrating conventional detection information, mining the quality performance law of destructive detection samples, and establishing sampling quality test results; inputting the test solution set into the relevant A correlation analysis network is used to perform correlation analysis between conventional test schemes and destructive test schemes, and a full product satisfaction prediction of destructive test is performed based on the correlation analysis results, random inspection quality test results and mapping identifiers. A comprehensive quality test result is generated according to the satisfaction prediction results, full inspection quality test results and product quality requirements. A correlation analysis network is used to establish a correlation model between conventional test schemes and destructive test schemes, and the performance indicators of gloves that have not been destructively tested are predicted and inferred. The test data and analysis results of each link are then integrated to form a technical solution for comprehensive quality evaluation results. This solves the technical problem in the prior art that it is difficult to accurately evaluate the quality of a whole batch of products due to the limited sampling ratio of destructive testing. By analyzing the correlation of the test scheme set and generating a comprehensive quality test result by predicting the satisfaction of the whole product, the technical effect of accurately evaluating the quality of a whole batch of products based on limited destructive testing sampling is achieved.
[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] Figure 1 A flow chart of a quality inspection method is provided for an embodiment of the present application;
[0011] Figure 2 A structural schematic diagram of a quality detection system is provided for an embodiment of the present application.
[0012] Explanation of the accompanying drawings: requirement establishment module 11, test scheme matching module 12, conventional detection module 13, destruction detection module 14, detection result generation module 15. DETAILED DESCRIPTION
[0013] The overall idea of the technical solution provided by this application is as follows:
[0014] The embodiments of the present application provide a quality detection method and system.
[0015] First, starting from the established quality requirements of high-damping shock-absorbing rubber gloves, the test solution library is configured through big data to automatically configure the test solution set. In the test execution stage, routine tests are first carried out on all high-damping shock-absorbing rubber gloves products to establish a complete product full inspection data set, and quality cluster analysis is performed to explore the quality distribution law inherent in the data and generate full inspection quality test results. At the same time, destructive spot checks are carried out according to the preset random sampling ratio, and a mapping identifier is established between the random sampling results and the quality inspection cluster results to obtain the spot inspection quality test results. In the data processing stage, a deep learning network is used to perform correlation analysis on the conventional test solution set and the destructive test solution set, and the correlation analysis results, the spot inspection quality test results and the mapping identifier are integrated to intelligently predict and infer the destructive indicators of the unsampled samples. Subsequently, the test data and analysis results of each link are integrated to generate accurate comprehensive quality test results for the entire batch of products.
[0016] The present application combines a small amount of destructive sampling with conventional testing, uses a correlation analysis network to perform correlation analysis on the two, and then combines the sampling results and mapping identifiers to predict the destructive testing indicators of the entire batch of products, thereby generating a comprehensive quality inspection result that takes into account both conventional testing and destructive testing, and predicting the destructive testing of the entire batch of products with limited destructive testing, thereby comprehensively and accurately evaluating the quality of the entire batch of high-damping shock-absorbing rubber gloves.
[0017] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification. Example 1
[0018] like Figure 1 As shown, an embodiment of the present application provides a quality detection method, the method comprising:
[0019] S1: Establish product quality requirements for the target product, wherein the product quality requirements include shock absorption performance requirements, appearance requirements, and aging durability requirements;
[0020] Specifically, the target product is high-damping shock-absorbing rubber gloves. According to the product characteristics and application requirements of high-damping shock-absorbing rubber gloves, product quality requirements are established. Product quality requirements refer to the indicators and standards for measuring and evaluating the product quality of high-damping shock-absorbing rubber gloves, which are the basis and basis for carrying out testing work.
[0021] The product quality requirements for high-damping shock-absorbing rubber gloves include three aspects: shock-absorbing performance requirements, appearance requirements, and aging durability requirements. Among them, the shock-absorbing performance requirements refer to the standards that the performance indicators of the gloves, such as the shock-absorbing effect and shock-absorbing durability, need to meet, such as the target value or allowable range of parameters such as the maximum impact absorption force, the impact energy attenuation ratio, and the performance decline after the number of consecutive impacts; the appearance requirements refer to the appearance quality requirements of the gloves, such as whether the surface is smooth and flat, whether the color is uniform, and whether there are obvious defects; the aging durability requirements refer to the stability requirements of the performance of the gloves during long-term use or storage, such as anti-oxidation, anti-corrosion, anti-aging capabilities, as well as the retention rate of mechanical strength and shock-absorbing performance, etc. By establishing complete product quality requirements for target products, a basis is provided for subsequent comprehensive quality inspections.
[0022] S2: configuring a test solution library through big data, matching test solutions based on the product quality requirements, and establishing a test solution set, wherein the test solution set includes a destruction solution set and a conventional detection solution set, and the destruction solution set is set with a random sampling ratio;
[0023] Specifically, using big data technology, based on massive test data and empirical knowledge, a test solution library covering various test scenarios and methods is configured as a big data configuration test solution library. The big data configuration test solution library contains multiple test solutions, each of which specifies the corresponding test items, test standards, test methods, sample quantity and other elements for one or several specific quality requirements.
[0024] After establishing product quality requirements, based on the established product quality requirements, information retrieval, semantic analysis, pattern matching and other technologies are used to match and filter out test solutions suitable for the target product from the big data configuration test solution library to form a test solution set. The test solution set is a series of test solutions tailored to the quality requirements of the target product, covering all dimensions and levels required for a comprehensive assessment of product quality requirements.
[0025] The test solution set consists of two parts: the destruction solution set and the routine test solution set. The destruction solution set refers to a series of solutions that require destructive testing of samples, such as tensile testing, tear testing, impact testing, aging testing, etc. These tests require the samples to be destroyed or their original shape and properties to be changed to obtain the ultimate performance indicators of the gloves. Due to the irreversibility and high cost of destructive tests, a certain random sampling ratio is set in the destruction solution set, that is, a certain number or proportion of samples are randomly selected from the entire batch of products to perform destructive tests. The routine test solution set refers to various test solutions that will not cause damage or permanent changes to the samples, such as appearance inspection, dimensional measurement, hardness testing, fatigue testing, etc. These tests can be fully tested on the entire batch of products to obtain the conventional quality indicators of the products.
[0026] By configuring the test plan library with big data and matching it to form a test plan set, we can fully utilize the existing mature plans and experience to quickly develop a targeted, comprehensive and systematic testing strategy, and by classifying the destruction plan set and the conventional testing plan set, we can achieve a balance between comprehensiveness and economy, thus laying the foundation for the plan to achieve comprehensive and accurate quality testing.
[0027] S3: Perform a full product inspection of the target product using the conventional inspection solution set to establish a product full inspection data set, perform quality inspection clustering of the target product based on the product full inspection data set, and generate a full inspection quality inspection result based on the quality inspection clustering result;
[0028] Specifically, after establishing the test scheme set, the conventional test scheme set is first used to conduct a comprehensive test on the entire batch of target products, that is, a full product inspection. The full product inspection covers all the test items and indicators specified in the conventional test scheme set. Each product is tested and data collected one by one to obtain a product full inspection data set that reflects the product quality status. After obtaining the product full inspection data set, the product full inspection data set is clustered for quality inspection. For example, clustering algorithms such as k-means, hierarchical clustering, and DBSCAN are used to divide the quality characteristics of all glove products into several clusters to obtain quality inspection clustering results. The glove products within each cluster have a high similarity in the test indicators, while there are obvious quality differences between different clusters.
[0029] Through quality inspection clustering, the quality level and category of glove products in routine inspection are identified and classified, revealing the overall distribution and internal rules of product quality. Based on the quality inspection clustering results, the quality inspection conclusion of the full inspection of glove products is further generated, including the proportion and distribution of samples of different quality levels such as qualified products, defective products, and unqualified products, etc., as the full inspection quality inspection result, reflecting the preliminary evaluation of the quality of the entire batch of products at the level of routine inspection.
[0030] S4: Perform random sampling of the target product based on the random sampling ratio, establish a mapping identifier between the random sampling result and the quality inspection clustering result, perform a destruction scheme set inspection of the random sampling result, and establish a random inspection quality inspection result;
[0031] Specifically, after completing the routine full inspection and quality clustering, a certain number or proportion of samples are randomly selected from the entire batch of products as random inspection samples for destructive testing according to the random sampling ratio set in the destruction plan set, as the random sampling result. Since the random inspection samples are randomly selected from the entire batch of products, they originally belong to different quality clusters divided in the quality inspection clustering results. At the same time of random sampling, by recording and identifying the cluster corresponding to each random inspection sample, a mapping identifier is established between the random inspection sample and the corresponding quality inspection clustering result, preparing for subsequent association analysis.
[0032] Then, the random samples randomly selected from the random sampling results are subjected to various destructive tests specified in the destruction scheme set, including tensile, tearing, impact, aging and other tests, to obtain various index data reflecting the intrinsic material performance and reliability level of the product, and obtain destructive test data. Based on the destructive test data, the sampling quality test results of the sampled samples can be obtained, including statistical indicators such as the qualified rate, average level, and discrete degree of the sampled samples in the destructive test items.
[0033] Through random sampling and mapping identification, based on limited destructive detection, a data and logical foundation is laid for the overall correlation analysis of quality inspection results.
[0034] S5: Input the test plan set into the association analysis network, perform association analysis between the conventional inspection plan set and the destructive inspection plan set, perform full product satisfaction prediction for destructive inspection based on the association analysis results, random inspection quality inspection results and mapping identifiers, and generate comprehensive quality inspection results based on the satisfaction prediction results, full inspection quality inspection results and product quality requirements.
[0035] Specifically, due to the relatively limited number of random sampling samples, the random sampling quality inspection results are not sufficient to fully judge the performance of the entire batch of products in terms of destructive indicators. However, through the mapping of random sampling results and quality inspection clustering results, the destructive inspection results of limited sampling can be linked to the conventional inspection results of all samples, thereby making a prediction and inference on the destructive indicator level of the entire batch of products.
[0036] The association analysis network is used to mine the association features and mapping relationships between conventional detection schemes and destructive detection schemes in terms of detection objects, detection indicators, detection conditions, etc., and based on this, realize the migration inference from conventional detection schemes to destructive detection schemes. First, each detection scheme in the set of conventional detection schemes and the set of destructive detection schemes is abstractly represented as a structured feature vector composed of elements such as detection objects, detection indicators, and detection conditions. Through technologies such as feature engineering and vectorization, the detection scheme is converted into a numerical form that is convenient for machine learning processing. Secondly, based on the knowledge and expert experience in the field of quality inspection, a group of rule templates reflecting the association relationship between conventional detection schemes and destructive detection schemes are pre-defined, such as "products with qualified appearance usually also meet the standards for strength indicators", etc., to characterize the correlation and causality of the two types of detection schemes in result judgment. Subsequently, based on the feature representation of the detection scheme and the association rule template, the basic architecture of the association analysis network is designed. For example, a multi-layer feedforward neural network, convolutional neural network, graph neural network and other models are used to establish a mapping mechanism between the detection scheme features and the association rules by setting the input layer, hidden layer, output layer and their connection weights. Afterwards, the existing product testing scheme data is used to train the association analysis network model through supervised learning or semi-supervised learning. Optimization algorithms such as back propagation and gradient descent are used to continuously adjust the parameters and structure of the network so that it can fit and learn the association characteristics and laws between the testing schemes to the greatest extent, thereby obtaining the association analysis network.
[0037] The established test scheme set, including the conventional detection scheme set and the destructive detection scheme set, is input into the association analysis network, and the association analysis network is used to perform association analysis on the input test scheme set. Through forward propagation and feature interaction in the association analysis network, the association strength and association rules between the schemes in the conventional detection scheme set and the schemes in the destructive detection scheme set are mined. As the association analysis result, the correlation and causality of the two types of detection schemes in the result judgment are characterized, and the intrinsic connection between the conventional detection indicators and the destructive detection indicators is revealed. Next, based on the association analysis results obtained by the association analysis, combined with the obtained sampling quality inspection results, and the mapping identifiers between the sampling samples and the quality inspection clustering results, the satisfaction of the whole batch of products on the destructive inspection indicators is predicted and inferred. Specifically, using the quality clustering cluster where the sampling samples are located, according to the migration mapping relationship from the conventional detection scheme to the destructive detection scheme mined by the association analysis network, the destructive indicator performance of other products that have not been destructively inspected in the same clustering cluster is inferred, and then the satisfaction prediction probability distribution of the whole batch of products on the destructive indicators is obtained, and the satisfaction prediction results are obtained. Afterwards, the prediction results will be integrated with the full inspection quality test results of routine testing, and compared with the established product quality requirements, the final judgment will be made on the overall quality level of the target product from the two dimensions of routine testing and destructive testing to generate comprehensive quality test results.
[0038] By introducing the association analysis network, the intelligent association mining and mapping migration of conventional inspection schemes and destructive inspection schemes are realized, and this is used as a bridge to extend the sampling quality inspection results of destructive inspection to the entire batch of products. The results of the two types of inspection schemes are then integrated, which not only ensures the comprehensiveness of the inspection conclusions, but also overcomes the limitation of insufficient destructive sampling samples, and obtains comprehensive and accurate comprehensive quality inspection results at a lower destructive inspection cost.
[0039] Furthermore, the embodiment of the present application also includes:
[0040] Establishing periodic compression force parameters of a universal testing machine based on the set of failure scenarios, and controlling the universal testing machine to perform compression testing on random sampling results using the periodic compression force parameters;
[0041] During the dynamic loading process, the stress data and strain data of the random sampling results are recorded, and a cyclic curve is established through the stress data and strain data. The hysteresis loss angle is calculated using the cyclic curve and periodic compression force parameters, and the damping performance evaluation result is generated using the hysteresis loss angle. The damping performance evaluation result is added to the sampling quality inspection result.
[0042] In a feasible implementation, by conducting special destructive testing and evaluation on the damping and vibration reduction performance of the high damping and shock absorbing rubber gloves, the quality characteristics of the product can be more comprehensively reflected.
[0043] First, in the established failure scheme set, a compression test scheme is specially designed and formulated for the damping and vibration reduction performance of high-damping shock-absorbing rubber gloves, and the cyclic compression load conditions of the actual shock-absorbing working conditions of high-damping shock-absorbing rubber gloves are specified, including parameters such as compression strain amplitude, loading frequency, and number of cycles. Secondly, the cyclic compression load conditions in the failure scheme set are converted into control instructions and loading curves of the universal testing machine, and the cyclic compression force parameters of the universal testing machine are established. Among them, the universal testing machine is a general material mechanical property testing equipment, which applies various forms of stress or strain to the sample according to the preset cyclic compression force parameters. The sample samples in the random sampling results are placed between the pressure plates of the universal testing machine, and the movement of the pressure plates is controlled to make the sample samples subject to dynamic compression deformation specified by the cyclic compression force parameters. During the entire compression test process, the universal testing machine strictly applies loads to the sample samples according to the cyclic compression force parameters, simulating the cyclic compression stress that the gloves are subjected to during actual use, so as to evaluate their damping performance under dynamic conditions.
[0044] In the process of periodic compression of the sampled samples in the random sampling results, the high-precision force sensor and displacement sensor provided by the universal testing machine are used to collect the force response data and deformation response data of the sampled samples in real time to obtain stress data and strain data. Among them, the force sensor continuously records the magnitude of the compressive stress of the sampled samples and its changing process, and obtains the stress-time data sequence of the sampled samples as stress data; the displacement sensor continuously records the thickness change and its changing process of the sampled samples under the compressive load, and obtains the strain-time data sequence of the sample as strain data. Subsequently, the stress data and strain data in the time domain are segmented according to the loading cycle, and the stress-strain data subset in each loading cycle is extracted. The stress-strain data subset pairs of each cycle are plotted on the same coordinate graph to obtain a cyclic curve reflecting the dynamic viscoelastic characteristics of the material, which intuitively shows the hysteresis behavior of the high-damping shock-absorbing rubber gloves under periodic compression. Then, based on the obtained cyclic curve and the set periodic compression force loading parameters, the hysteresis loss angle is introduced as a quantitative indicator to evaluate the damping and shock absorption performance of rubber gloves. It describes the hysteresis degree of the strain signal relative to the stress signal in the dynamic deformation response of the rubber gloves. Its value reflects the damping dissipation capacity of the rubber gloves under dynamic loading. By calculating the average value and statistical distribution of the hysteresis loss angle of each sample in multiple loading cycles, the quantitative evaluation results reflecting the overall damping performance level of the random sampling samples can be obtained, and the damping performance evaluation results can be obtained. Among them, the larger the value of the hysteresis loss angle, the stronger the damping energy dissipation effect of the high damping and shock absorption rubber gloves, and the better the shock absorption performance. Subsequently, the damping performance evaluation results obtained are added to the sampling quality inspection results to provide a basis for the comprehensive evaluation of the quality of high damping and shock absorption rubber gloves.
[0045] Furthermore, the embodiment of the present application also includes:
[0046] Configuring vibration test frequencies based on a destructive test scheme set, wherein the vibration test frequencies include a first vibration test frequency, a second vibration test frequency, and a third vibration test frequency, and the first vibration test frequency is higher than the second vibration test frequency, and the second vibration test frequency is higher than the third vibration test frequency;
[0047] Performing vibration testing on the random sampling results at the vibration test frequency, and performing vibration monitoring on the random sampling results to establish a response vibration data set;
[0048] establishing a vibration reduction characteristic spectrum based on the response vibration data set and the mapped vibration test frequency, and generating a vibration transfer evaluation result through the vibration reduction characteristic spectrum;
[0049] The vibration transmission assessment result is added to the sampling quality inspection result.
[0050] In a feasible implementation, first, in the destructive test scheme set, for the shock absorption performance evaluation of high damping shock absorbing rubber gloves, a multi-frequency vibration test scheme is designed, the vibration test frequency is configured, and three different levels of vibration excitation frequencies are specified, namely the first vibration test frequency, the second vibration test frequency and the third vibration test frequency. The three vibration test frequencies are in a decreasing relationship, that is, the first vibration test frequency is the highest, the second vibration test frequency is the second, and the third vibration test frequency is the lowest. Then, the determined vibration test frequencies are applied to the randomly selected sampling samples in turn, and the sampling samples are subjected to a vibration table loading test using professional vibration test equipment to simulate the dynamic response process of the sampling samples under vibration excitation of different frequencies. Specifically, the sampling samples are fixed on the vibration table platform, and the vibration table is controlled to generate sinusoidal vibration excitation of corresponding frequency and amplitude according to the preset vibration test frequency. During the entire vibration test process, the vibration response of the sampling samples is monitored in real time using high-precision acceleration sensors, velocity sensors, displacement sensors, etc., and the acceleration, velocity and displacement time history signals of the samples under different frequency excitations are collected, and the corresponding frequency parameters are recorded to obtain a response vibration data set that fully describes the dynamic vibration behavior of the samples. This response vibration dataset comprehensively records the vibration response characteristics of each sample under three levels of vibration test frequency excitation, including multi-dimensional information such as the vibration time domain waveform, frequency domain spectrum, amplitude statistical parameters, etc., reflecting the suppression effect and transfer characteristics of high-damping shock-absorbing rubber gloves on vibrations of different frequencies.
[0051] Then, the acceleration, velocity and displacement time history signals of each sample sample in the response vibration data set are analyzed in the time-frequency domain, and the time domain signals are converted into frequency domain spectrograms by Fourier transform and other methods to obtain the vibration response amplitude distribution of the sample sample at different frequency components. Subsequently, the frequency domain spectrogram is mapped one by one with the corresponding vibration test frequency to form a shock absorption characteristic spectrum with frequency as the horizontal coordinate and vibration response amplitude as the vertical coordinate, showing the shock absorption effect and vibration transfer characteristics of the sample sample under different frequency excitations. Afterwards, according to the morphological characteristics of the shock absorption characteristic spectrum, combined with the vibration transfer theory and the dynamic model of the shock absorbing material, the vibration transfer characteristics and shock absorption efficiency of the sample sample of the high damping shock absorbing rubber gloves are quantitatively evaluated, and the vibration transfer evaluation results are obtained, including vibration transmission rate, vibration attenuation coefficient, vibration isolation efficiency, etc., which respectively describe the vibration suppression and isolation capabilities of the sample sample of the high damping shock absorbing rubber gloves from the perspectives of amplitude ratio, energy ratio and frequency ratio. After that, the vibration transfer assessment results will be incorporated into the random quality inspection results, realizing the organic integration of the shock absorbing performance of high damping shock absorbing rubber gloves and other quality characteristics, enriching the perspective of quality inspection.
[0052] Furthermore, the embodiment of the present application also includes:
[0053] Perform usage data interaction on the target product to establish a location key value;
[0054] Adaptively clustering the regions according to the position key values, and configuring clustering regions;
[0055] Acquire regional images of the clustering area by an image acquisition device to establish an image set mapped with the area;
[0056] Mapping verification is performed on the image set, and a product full inspection data set is established based on the mapping verification result.
[0057] In a feasible implementation, in the actual use scenario of the gloves, due to the differences in factors such as hand movements and holding postures, there is an obvious distribution pattern of the pressure, friction, deformation and other effects on different parts of the gloves. Usually, the palm, finger pulp, finger joints and other parts are the main stress and wear areas, while the back of the hand, finger joints and other parts are relatively less affected by external forces. By exchanging data with the high-damping shock-absorbing rubber gloves that have been put into use, the position information data of the gloves in the actual use process, that is, the frequency of use of each part, the force distribution, etc., are obtained, and the position key value reflecting the use part of the gloves is established accordingly. Then, according to the geometric structure and partition requirements of the gloves, several area division schemes to be clustered are preset on the surface of the gloves, for example, the surface of the gloves is divided into several partitions such as the palm area, the metacarpophalangeal joint area, the back of the finger area, and the tiger's mouth area. Then, the position key value parameters in each area are input into the cluster analysis model, and by calculating the data similarity between different areas, the areas with similar position key value features are dynamically classified into the same cluster cluster, and the final cluster partition scheme is generated to obtain the configuration cluster area. Among them, cluster analysis models include K-means, DBSCAN, etc. Through adaptive cluster analysis, the glove surface is divided into several cluster areas with usage characteristics. The position key values within each cluster area are relatively close, reflecting the consistency of the force distribution of the area during use; while the position key values between different cluster areas are quite different, reflecting the differences in the intensity of use and degree of wear of different parts. Through the function-oriented regional division method, more accurate and targeted data samples can be matched for subsequent quality inspection and predictive analysis.
[0058] In the process of executing the conventional detection scheme, a high-resolution image acquisition device is used to scan and image the clustered areas on the surface of the glove one by one, and high-definition image data of each area is obtained. Through image metadata annotation, a one-to-one mapping relationship between the image and the area is established to form a regional mapping image set covering all clustered areas. Each image is mapped to a specific clustered area, and the quality status of each key part is intuitively displayed. After that, each image in the regional mapping image set is preprocessed, including image enhancement, noise removal, background segmentation, etc., to improve the image quality and feature recognizability. Then, a series of feature parameters of each image, such as texture features, morphological features, color features, etc., are extracted using an image processing algorithm to establish a standardized image feature vector. After obtaining the image feature data, the correlation between each image feature vector and the position key value of its mapping area is analyzed to evaluate the representativeness of the image to the quality characteristics of its mapping area. At the same time, the image feature distribution of different areas is cross-validated to identify potential mapping errors or abnormal data. For images that fail to pass the verification, re-collection or mapping relationship correction is performed to ensure a high degree of consistency between the regional division and the image data. After mapping verification, high-confidence and highly representative regional mapping image data are obtained as the mapping verification result. Afterwards, the verified image data in the mapping verification results are summarized according to regional attributes to form a complete product full inspection data set.
[0059] Furthermore, the embodiment of the present application also includes:
[0060] Image recognition convolution kernels by clustering region configuration and region mapping;
[0061] When an image in the image set is called, the corresponding image recognition convolution kernel is synchronously started according to the mapping relationship to perform image traversal recognition of the called image;
[0062] The mapping verification is completed based on the image traversal recognition results.
[0063] In a feasible implementation, first, according to the shape, size, topological structure and other attributes of the clustering area, the basic architecture of the convolution kernel is determined, including parameters such as the number of convolution layers, the number of pooling layers, the filter size, and the step size, to obtain a set of image recognition convolution kernels that are mapped one-to-one with the clustering areas. Among them, clustering areas with larger areas and more regular shapes can use deeper convolution structures to extract more abstract and semantic features; while areas with smaller areas and irregular shapes are suitable for shallower convolution structures, focusing on extracting local and detailed features. When the image set is called, the image recognition convolution kernel corresponding to the current image mapping area is dynamically loaded, and the convolution kernel is used to perform sliding window feature extraction on the image in the entire image domain to obtain high-dimensional feature representations of each local area of the image for subsequent mapping relationship verification. Specifically, first obtain the cluster region ID mapped by the image according to the metadata tag of the calling image, then retrieve the image recognition convolution kernel matching the current region ID and start loading; after the image recognition convolution kernel is loaded, immediately perform feature extraction based on convolution operation on the calling image, using a sliding window method, slide the convolution kernel row by row and column by column on the image with a certain step size, and perform convolution transformation on the local area of the image to obtain a set of high-dimensional feature maps, each of which corresponds to a local area of the calling image, represents the projection of the area in the feature space defined by the convolution kernel, and depicts the complex feature information of the area such as texture, structure, and semantics. By repeating the above convolution extraction process on the entire image, a local feature map covering the entire image domain is obtained as the image traversal recognition result.
[0064] Subsequently, for the call image, the cluster region ID recorded in its metadata tag is first extracted as the prior mapping region of the call image, and then the generated local feature map is analyzed to examine the correlation between the feature representation of each local region of the image and the prior mapping region. Taking feature similarity as an example, the cosine similarity between the feature vector of each local region of the image and the representative feature vector of the prior mapping region is calculated to obtain a set of similarity scores. The higher the score, the closer the quality mode of the local region is to the prior mapping region. After obtaining the correlation index of all local regions, the distribution law of the index is comprehensively analyzed to give the matching degree of the whole image and the prior mapping region. If the correlation index of most local regions is high and the distribution is relatively balanced, it can be considered that the call image matches the mapping region indicated by its label, and the mapping relationship between the two is accurate and reliable; on the contrary, if the correlation index of most local regions is low, or the distribution of the index shows obvious differentiation and abnormality, it indicates that the image and the region annotated by the label are semantically inconsistent, and there is mapping error or label noise. Based on the above analysis, the verification results between each called image and its prior mapping area can be obtained, and the images can be divided into two categories: mapping match and mapping mismatch. The metadata tags in the regional mapping image set are confirmed or corrected accordingly, thereby completing the mapping verification.
[0065] Furthermore, the embodiment of the present application also includes:
[0066] Conduct random inspection and verification on the random inspection quality test results and establish random inspection quality rating;
[0067] Matching an additional sampling factor based on the sampling quality rating, and matching an additional random number;
[0068] Determine additional sampling with the sampling coefficient and additional random number, and perform additional sampling verification;
[0069] The prediction results are updated based on the additional sampling verification results, and the quality inspection is completed based on the updated prediction results.
[0070] In a preferred embodiment, first, a part of the samples are randomly selected from the original sample samples according to a certain proportion (such as 10%) to form a review sampling set, and the same detection method and judgment criteria as the original detection are used to perform independent quality detection on the review sampling samples to obtain a new set of test results. By comparing the consistency of the review test results with the original test results, calculating the matching rate, correlation and other indicators between the two, the accuracy of the original sampling quality data can be quantitatively evaluated. Based on the review evaluation results, with reference to the preset sampling quality grade assessment standards (such as a matching rate higher than 90% is grade A, 80-90% is grade B, 70-80% is grade C, and less than 70% is grade D), the overall credibility of the original sampling quality test is graded and judged to obtain the sampling quality rating. The higher the quality grade, the higher the reliability of the original sampling result, and vice versa. Subsequently, for different quality grades (such as A, B, C, D), the corresponding additional sampling coefficient is pre-set, indicating the sampling ratio that needs to be increased. The lower the sampling quality rating, the lower the credibility, the larger the additional sampling coefficient, and the more additional samples are needed for verification. At the same time, in order to improve the randomness and coverage of sampling, before each additional sampling, an additional random number is matched according to the sampling quality rating. Different sampling quality ratings correspond to different random number value ranges. The lower the sampling quality rating, the larger the fluctuation space of the random number is set to expand the diversity of the sample.
[0071] Next, the additional sampling coefficient and additional random number obtained by matching are used to determine the specific sample size for this round of additional sampling verification, and the required number of samples are randomly selected from the original batch of products. Quality inspection is carried out according to the same inspection standard as the original sampling inspection. The additional sampling samples are inspected one by one to obtain the inspection results of the additional samples. The inspection results of all additional samples are comprehensively analyzed, and the qualified product rate is statistically calculated to obtain the overall quality level of this round of additional sampling, and compared with the original sampling results to verify the consistency of both parties and obtain the additional sampling verification results. If the difference between the two sampling results is large, the reason needs to be analyzed, and if necessary, the additional sampling volume needs to be expanded for further verification. After that, the sampling quality inspection results are updated according to the additional sampling verification results, and the mapping of the association analysis results and the random sampling results with the quality inspection clustering results is recombined to identify the destruction of the full product satisfaction prediction of the inspection, and a new satisfaction prediction result is obtained.
[0072] Furthermore, the embodiment of the present application also includes:
[0073] Establishing a risk warning for the product based on the comprehensive quality test results, and extracting key abnormal features in the comprehensive quality test results;
[0074] The abnormal process is located by using the key abnormal features, and the risk warning and the abnormal process are reported as abnormal.
[0075] In a feasible implementation, first, statistical analysis is performed on the comprehensive quality inspection results to calculate the qualified rate, abnormal rate, deviation between the predicted qualified rate and the actual value of each quality indicator, and issue risk warnings for indicators that exceed the limit or deviate greatly. At the same time, the comprehensive quality inspection results are analyzed to find out the main feature combinations that cause quality problems or abnormalities, such as key process deviations, key parameter exceeding the limit, and frequent key defect modes, as key abnormal features of product quality.
[0076] Subsequently, based on the identified key abnormal characteristics, through retrospective analysis of the product production process, the abnormal process links that cause the abnormal characteristics are quickly found and located, and the risk warnings and abnormal process information are summarized into an abnormal report, which is sent to relevant functional departments and personnel, requiring timely response measures.
[0077] By quickly tracing the source of quality problems, shortening the response chain of quality control, and realizing risk warning-driven abnormal management, the front-end quality monitoring and the back-end process improvement closed-loop connection can be connected to improve the quality control of high-damping shock-absorbing rubber gloves.
[0078] In summary, the quality inspection method provided in the embodiment of the present application has the following technical effects:
[0079] Establish product quality requirements for target products, clarify the evaluation dimensions and indicators of quality inspection of high-damping shock-absorbing rubber gloves, and provide a basis and standard for inspection scheme matching and quality evaluation. Configure the test scheme library through big data, match the test scheme based on product quality requirements, and establish a test scheme set, where the test scheme set includes a destruction scheme set and a conventional inspection scheme set, and the destruction scheme set is set with a random sampling ratio, automatically build a test scheme covering various quality requirements, optimize the inspection strategy, and strike a balance between comprehensiveness and economy. Perform full inspection of target products with the conventional inspection scheme set, establish a product full inspection data set, cluster the quality inspection of target products based on the product full inspection data set, generate full inspection quality inspection results based on the quality inspection clustering results, fully grasp the conventional quality feature data of the entire batch of products, and explore the quality distribution law, providing a basis for subsequent quality prediction. Perform random sampling of target products based on the random sampling ratio, and establish a mapping identifier between the random sampling results and the quality inspection clustering results, perform the destruction scheme set detection of the random sampling results, establish the sampling quality inspection results, obtain the destructive indicator detection data, and establish a mapping relationship with the conventional inspection data to provide support for the full product satisfaction prediction of the destruction detection. The test plan set is input into the association analysis network, and the association analysis of the conventional inspection plan set and the destructive inspection plan set is performed. Based on the association analysis results, the sampling quality inspection results and the mapping identifiers, the full product satisfaction prediction of destructive inspection is carried out. According to the satisfaction prediction results, the full inspection quality inspection results and the product quality requirements, the comprehensive quality inspection results are generated. The internal connection between the conventional inspection data and the destructive sampling data is deeply explored, and the limitation of the destructive inspection sampling ratio is broken through, so as to improve the accuracy and comprehensiveness of the quality assessment of the whole batch. Example 2
[0080] Based on the same inventive concept as a quality detection method in the aforementioned embodiment, Figure 2 As shown, an embodiment of the present application provides a quality detection system, which includes:
[0081] The demand establishment module 11 is used to establish product quality requirements of the target product, wherein the product quality requirements include shock absorption performance requirements, appearance requirements, and aging durability requirements;
[0082] A test scheme matching module 12 is used to configure a test scheme library through big data, match test schemes based on the product quality requirements, and establish a test scheme set, wherein the test scheme set includes a destruction scheme set and a conventional detection scheme set, and the destruction scheme set is set with a random sampling ratio;
[0083] A conventional inspection module 13 is used to perform a full inspection of a target product with the conventional inspection scheme set, establish a full inspection data set, perform quality inspection clustering of the target product based on the full inspection data set, and generate a full inspection quality inspection result with the quality inspection clustering result;
[0084] A damage detection module 14 is used to perform random sampling of target products based on the random sampling ratio, establish a mapping identifier between the random sampling results and the quality detection clustering results, perform damage scheme set detection on the random sampling results, and establish a random inspection quality detection result;
[0085] The detection result generation module 15 is used to input the test plan set into the association analysis network, perform association analysis between the conventional detection plan set and the destructive detection plan set, perform full product satisfaction prediction for destructive detection based on the association analysis results, random inspection quality inspection results and mapping identifiers, and generate comprehensive quality inspection results based on the satisfaction prediction results, full inspection quality inspection results and product quality requirements.
[0086] Furthermore, the damage detection module 14 includes the following execution steps:
[0087] Establishing periodic compression force parameters of a universal testing machine based on the set of failure scenarios, and controlling the universal testing machine to perform compression testing on random sampling results using the periodic compression force parameters;
[0088] During the dynamic loading process, the stress data and strain data of the random sampling results are recorded, and a cyclic curve is established through the stress data and strain data. The hysteresis loss angle is calculated using the cyclic curve and periodic compression force parameters, and the damping performance evaluation result is generated using the hysteresis loss angle. The damping performance evaluation result is added to the sampling quality inspection result.
[0089] Furthermore, the damage detection module 14 further includes the following execution steps:
[0090] Configuring vibration test frequencies based on a destructive test scheme set, wherein the vibration test frequencies include a first vibration test frequency, a second vibration test frequency, and a third vibration test frequency, and the first vibration test frequency is higher than the second vibration test frequency, and the second vibration test frequency is higher than the third vibration test frequency;
[0091] Performing vibration testing on the random sampling results at the vibration test frequency, and performing vibration monitoring on the random sampling results to establish a response vibration data set;
[0092] establishing a vibration reduction characteristic spectrum based on the response vibration data set and the mapped vibration test frequency, and generating a vibration transfer evaluation result through the vibration reduction characteristic spectrum;
[0093] The vibration transmission assessment result is added to the sampling quality inspection result.
[0094] Furthermore, the conventional detection module 13 includes the following execution steps:
[0095] Perform usage data interaction on the target product to establish a location key value;
[0096] Adaptively clustering the regions according to the position key values, and configuring clustering regions;
[0097] Acquire regional images of the clustering area by an image acquisition device to establish an image set mapped with the area;
[0098] Mapping verification is performed on the image set, and a product full inspection data set is established based on the mapping verification result.
[0099] Furthermore, the conventional detection module 13 also includes the following execution steps:
[0100] Image recognition convolution kernels by clustering region configuration and region mapping;
[0101] When an image in the image set is called, the corresponding image recognition convolution kernel is synchronously started according to the mapping relationship to perform image traversal recognition of the called image;
[0102] The mapping verification is completed based on the image traversal recognition results.
[0103] Furthermore, the embodiment of the present application also includes a prediction result update module, which includes the following execution steps:
[0104] Conduct random inspection and verification on the random inspection quality test results and establish random inspection quality rating;
[0105] Matching an additional sampling factor based on the sampling quality rating, and matching an additional random number;
[0106] Determine additional sampling with the sampling coefficient and additional random number, and perform additional sampling verification;
[0107] The prediction results are updated based on the additional sampling verification results, and the quality inspection is completed based on the updated prediction results.
[0108] Furthermore, the embodiment of the present application also includes an abnormal warning module, which includes the following execution steps:
[0109] Establishing a risk warning for the product based on the comprehensive quality test results, and extracting key abnormal features in the comprehensive quality test results;
[0110] The abnormal process is located by using the key abnormal features, and the risk warning and the abnormal process are reported as abnormal.
[0111] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.
[0112] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to multiple elements that can be selected individually or in full. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.
Claims
1. A method for detecting high damping shock absorbing rubber gloves, characterized in that: The method comprises: Establish product quality requirements for target products, wherein the product quality requirements include shock absorption performance requirements, appearance requirements, and aging durability requirements; The test scheme library is configured through big data, and the test scheme is matched based on the product quality requirements to establish a test scheme set, wherein the test scheme set includes a destruction scheme set and a conventional detection scheme set, and the destruction scheme set is set with a random sampling ratio; Performing a full product inspection of the target product with the conventional inspection scheme set, establishing a product full inspection data set, performing quality inspection clustering of the target product based on the product full inspection data set, and generating a full inspection quality inspection result with the quality inspection clustering result; Based on the random sampling ratio, random sampling of the target product is performed, and a mapping identifier between the random sampling result and the quality inspection clustering result is established, and a destruction scheme set test of the random sampling result is performed to establish a random inspection quality inspection result; The test plan set is input into the association analysis network, and the association analysis of the conventional detection plan set and the destructive detection plan set is performed. Based on the association analysis results, the random inspection quality inspection results and the mapping identifier, the full product satisfaction prediction of the destructive inspection is performed, and the comprehensive quality inspection results are generated according to the satisfaction prediction results, the full inspection quality inspection results and the product quality requirements.
2. The method according to claim 1, characterized in that The execution of the random sampling results of the destruction scheme set detection and the establishment of the sampling quality detection results include: Establishing periodic compression force parameters of a universal testing machine based on the set of failure scenarios, and controlling the universal testing machine to perform compression testing on random sampling results using the periodic compression force parameters; During the dynamic loading process, the stress data and strain data of the random sampling results are recorded, and a cyclic curve is established through the stress data and strain data. The hysteresis loss angle is calculated using the cyclic curve and periodic compression force parameters, and the damping performance evaluation result is generated using the hysteresis loss angle. The damping performance evaluation result is added to the sampling quality inspection result.
3. The method according to claim 2, characterized in that The performing of the destruction scheme set detection of the random sampling results and establishing the sampling quality detection results also includes: Configuring vibration test frequencies based on a destructive test scheme set, wherein the vibration test frequencies include a first vibration test frequency, a second vibration test frequency, and a third vibration test frequency, and the first vibration test frequency is higher than the second vibration test frequency, and the second vibration test frequency is higher than the third vibration test frequency; Performing vibration testing on the random sampling results at the vibration test frequency, and performing vibration monitoring on the random sampling results to establish a response vibration data set; establishing a vibration reduction characteristic spectrum based on the response vibration data set and the mapped vibration test frequency, and generating a vibration transfer evaluation result through the vibration reduction characteristic spectrum; The vibration transmission assessment result is added to the sampling quality inspection result.
4. The method according to claim 1, characterized in that The method of performing a full product inspection of the target product using the conventional inspection scheme set to establish a full product inspection data set includes: Perform usage data interaction on the target product to establish a location key value; Adaptively clustering the regions according to the position key values, and configuring clustering regions; Acquire regional images of the clustering area by an image acquisition device to establish an image set mapped with the area; Mapping verification is performed on the image set, and a product full inspection data set is established based on the mapping verification result.
5. The method according to claim 4, characterized in that The mapping verification is performed on the image set, and a product full inspection data set is established based on the mapping verification result, including: Image recognition convolution kernels by clustering region configuration and region mapping; When an image in the image set is called, the corresponding image recognition convolution kernel is synchronously started according to the mapping relationship to perform image traversal recognition of the called image; The mapping verification is completed based on the image traversal recognition results.
6. The method according to claim 1, characterized in that The method further comprises: Conduct random inspection and verification on the random inspection quality test results and establish random inspection quality rating; Matching an additional sampling factor based on the sampling quality rating, and matching an additional random number; Determine additional sampling with the sampling coefficient and additional random number, and perform additional sampling verification; The prediction results are updated based on the additional sampling verification results, and the quality inspection is completed based on the updated prediction results.
7. The method according to claim 1, characterized in that The method further comprises: Establishing a risk warning for the product based on the comprehensive quality test results, and extracting key abnormal features in the comprehensive quality test results; The abnormal process is located by using the key abnormal features, and the risk warning and the abnormal process are reported as abnormal.
8. A detection system for high damping shock absorbing rubber gloves, characterized in that: For implementing the method according to any one of claims 1 to 7, the system comprises: A demand establishment module, the demand establishment module is used to establish product quality requirements of the target product, wherein the product quality requirements include shock absorption performance requirements, appearance requirements, and aging durability requirements; A test scheme matching module, the test scheme matching module is used to configure a test scheme library through big data, match test schemes based on the product quality requirements, and establish a test scheme set, wherein the test scheme set includes a destruction scheme set and a conventional detection scheme set, and the destruction scheme set is set with a random sampling ratio; A conventional inspection module, wherein the conventional inspection module is used to perform a full product inspection of a target product with the conventional inspection scheme set, establish a full product inspection data set, perform quality inspection clustering of the target product based on the full product inspection data set, and generate a full inspection quality inspection result with the quality inspection clustering result; A damage detection module, the damage detection module is used to perform random sampling of target products based on the random sampling ratio, establish a mapping identifier between the random sampling results and the quality detection clustering results, perform a damage scheme set detection of the random sampling results, and establish a random inspection quality detection result; The detection result generation module is used to input the test plan set into the association analysis network, perform association analysis between the conventional detection plan set and the destructive detection plan set, perform full product satisfaction prediction for destructive detection based on the association analysis results, random inspection quality inspection results and mapping identifiers, and generate comprehensive quality inspection results based on the satisfaction prediction results, full inspection quality inspection results and product quality requirements.
