Quality inspection method and device for rubber-coated dumbbells

Abstract

The quality inspection method and device for rubber-coated dumbbells provided in the present application relate to the technical field of quality inspection. By configuring static test parameters, monitoring data of the rubber-coated part is collected, and the test pressure value and the monitoring data are associated and mapped; the change rate is screened with the deformation variable, temperature value, and connection status data as the target; repeated loading and unloading pressure cycle tests are performed with the initial test pressure value, test quantity, and initial pressure step length as the initial configuration parameters, and the dynamic test data of the rubber-coated part is collected for step length analysis and then the initial pressure step length is adjusted; the consistency judgment after time alignment of the static and dynamic test data sets is used to determine the pressure-bearing quality inspection result, which solves the technical problems that the existing quality inspection method for rubber-coated dumbbells is not in-depth and intelligent enough, and the inspection result is not accurate enough, resulting in safety hazards in the instrument, and achieves the technical effect of improving the accuracy of quality inspection of rubber-coated dumbbells and thus ensuring safe use.

Description

Technical Field

[0001] The invention relates to the technical field of quality inspection, and in particular to a quality inspection method and device for rubber-coated dumbbells. Background Art

[0002] As a common fitness equipment, the quality of rubber-coated dumbbells is directly related to the user experience and safety. Rubber-coated dumbbells are usually made of a metal dumbbell plate coated with a layer of rubber to improve the anti-slip and safety during use. With the rapid development of the fitness industry, the market demand for rubber-coated dumbbells continues to increase, and consumers have higher and higher requirements for the quality and safety of fitness equipment. Therefore, it is particularly important to conduct quality inspections on rubber-coated dumbbells. At present, the quality inspection of rubber-coated dumbbells mainly includes appearance inspection, size measurement, weight inspection, etc. Although these methods can preliminarily evaluate the quality of the product, there is still a lack of effective testing methods for the special properties of rubber-coated dumbbells, such as compression resistance.

[0003] In summary, the existing quality inspection methods for rubber-coated dumbbells have technical problems such as the quality inspection only stays at the preliminary evaluation, the inspection method is not in-depth and intelligent enough, and the inspection results are not accurate enough, resulting in safety hazards in the equipment. Summary of the invention

[0004] The present application provides a quality inspection method and device for rubber-coated dumbbells, which are used to solve the technical problems that the existing quality inspection methods for rubber-coated dumbbells have quality inspections that only remain at the preliminary evaluation, the inspection method is not in-depth and intelligent enough, and the inspection results are not accurate enough, resulting in safety hazards in the equipment.

[0005] In view of the above problems, the present application provides a quality inspection method and device for rubber-coated dumbbells.

[0006] In a first aspect, the present application provides a quality inspection method for a rubber-coated dumbbell, the method comprising:

[0007] Configure static test parameters, including initial pressure value, pressure adjustment step, and pressure threshold;

[0008] When executing the static test parameters, the high-speed camera, infrared thermal imager and ultrasonic flaw detector are synchronized to collect monitoring data of the rubber-coated part, including deformation, temperature and connection status data, and the test pressure value is associated with the monitoring data to obtain a static test data set;

[0009] Based on the static test data set, the change rate is screened with deformation, temperature value and connection status data as the target, and dynamic test parameters are configured, wherein the dynamic test parameters include initial test pressure value, test quantity, initial pressure step length and dynamic step length adjustment function; repeated loading and unloading pressure cycle test is performed with the initial test pressure value, test quantity and initial pressure step length as initial configuration parameters, and dynamic test data of the rubber-coated part is collected by the high-speed camera, infrared thermal imager and ultrasonic flaw detector; step length analysis is performed according to the dynamic test data through the dynamic step length adjustment function to obtain the dynamic step length; the initial pressure step length is adjusted based on the dynamic step length, and the dynamic test is continued to be performed to obtain dynamic test data until the target pressure threshold is reached to construct a dynamic test data set; timing alignment is performed using the static test data set and the dynamic test data set, and consistency judgment is performed according to the timing alignment data, and the pressure-bearing quality detection result is determined according to the consistency judgment result.

[0010] In a second aspect, the present application provides a quality detection device for a rubber-coated dumbbell, the device comprising:

[0011] A static parameter configuration module is used to configure static test parameters, including an initial pressure value, a pressure adjustment step, and a pressure threshold; a static data acquisition module is used to synchronize a high-speed camera, an infrared thermal imager, and an ultrasonic flaw detector to collect monitoring data of the rubber-coated part when executing the static test parameters, including deformation variables, temperature values, and connection status data, and associate and map the test pressure value with the monitoring data to obtain a static test data set; a dynamic parameter configuration module is used to perform change rate screening based on the static test data set with deformation variables, temperature values, and connection status data as targets, and configure dynamic test parameters, wherein the dynamic test parameters include an initial test pressure value, a test quantity, an initial pressure step, and a dynamic step adjustment function; a dynamic data acquisition module is used to perform change rate screening based on the static test data set with deformation variables, temperature values, and connection status data as targets, and configure dynamic test parameters, wherein the dynamic test parameters include an initial test pressure value, a test quantity, an initial pressure step, and a dynamic step adjustment function; The test quantity and initial pressure step length are used as initial configuration parameters for repeated loading and unloading pressure cycle tests, and the dynamic test data of the rubber-coated part is collected by the high-speed camera, infrared thermal imager, and ultrasonic flaw detector; a step length analysis module is used to perform step length analysis according to the dynamic test data through the dynamic step length adjustment function to obtain the dynamic step length; a step length adjustment module is used to adjust the initial pressure step length based on the dynamic step length, continue to perform dynamic testing, obtain dynamic test data, and construct a dynamic test data set until the target pressure threshold is reached; a test result acquisition module is used to use the static test data set and the dynamic test data set to perform time alignment, and perform consistency judgment based on the time alignment data, and determine the pressure-bearing quality test result based on the consistency judgment result.

[0012] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0013] The quality inspection method for rubber-coated dumbbells provided in the present application configures static test parameters, including an initial pressure value, a pressure adjustment step, and a pressure threshold; when executing the static test parameters, a high-speed camera, an infrared thermal imager, and an ultrasonic flaw detector are synchronized to collect monitoring data of the rubber-coated part, including deformation variables, temperature values, and connection status data, and the test pressure value is associated with the mapping test data to obtain a static test data set; based on the static test data set, the deformation variables, temperature values, and connection status data are used as targets for rate of change screening, and dynamic test parameters are configured, and the dynamic test parameters include an initial test pressure value, a test number, an initial pressure step, and a dynamic step adjustment function; repeated loading and unloading pressure cycle tests are performed with the initial test pressure value, the test number, and the initial pressure step as initial configuration parameters, and the high-speed camera, infrared thermal imager, and ultrasonic flaw detector are used to collect monitoring data of the rubber-coated part, including deformation variables, temperature values, and connection status data, and the test pressure value is associated with the monitoring data to obtain a static test data set; based on the static test data set, the change rate is screened with the deformation variables, temperature values, and connection status data as targets, and dynamic test parameters are configured, and the dynamic test parameters include an initial test pressure value, a test number, an initial pressure step, and a dynamic step adjustment function; repeated loading and unloading pressure cycle tests are performed with the initial test pressure value, the test number, and the initial pressure step as initial configuration parameters, and the high-speed camera, infrared thermal imager, and ultrasonic flaw detector are used to collect monitoring data of the rubber-coated part, including deformation variables, temperature values, and connection status data, and the test pressure value is associated with the monitoring data, and the monitoring data is associated with the mapping test pressure value, temperature value, and connection status ... , an ultrasonic flaw detector collects dynamic test data of the rubber-coated part; according to the dynamic test data, a step length analysis is performed through the dynamic step length adjustment function to obtain a dynamic step length; based on the dynamic step length, the initial pressure step length is adjusted, and the dynamic test is continued to be performed to obtain dynamic test data until a target pressure threshold is reached to construct a dynamic test data set; timing alignment is performed using the static test data set and the dynamic test data set, and consistency judgment is performed based on the timing alignment data, and the pressure-bearing quality inspection result is determined based on the consistency judgment result, which solves the technical problems of the existing quality inspection method for rubber-coated dumbbells, that is, the quality inspection only stays at the preliminary evaluation, the inspection method is not in-depth and intelligent enough, and the inspection result is not accurate enough, resulting in safety hazards in the equipment, and achieves the technical effect of improving the accuracy of quality inspection of rubber-coated dumbbells, thereby ensuring the safety of use and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flow chart of a quality inspection method for rubber-coated dumbbells is provided for this application.

[0015] Figure 2 A structural schematic diagram of a quality inspection device for rubber-coated dumbbells is provided for this application.

[0016] Explanation of the reference numerals: static parameter configuration module 11 , static data acquisition module 12 , dynamic parameter configuration module 13 , dynamic data acquisition module 14 , step length analysis module 15 , step length adjustment module 16 , detection result acquisition module 17 . DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0018] Embodiment 1

[0019] like Figure 1 As shown, the present application provides a quality inspection method for rubber-coated dumbbells, the method comprising:

[0020] Step S100: configuring static test parameters, including initial pressure value, pressure adjustment step, and pressure threshold.

[0021] Specifically, static test parameters are configured to evaluate the pressure-bearing quality of the rubber-coated dumbbell. In order to ensure the accuracy and rationality of the test parameters, the configured static test parameters include initial pressure value, pressure adjustment step, and pressure threshold. Among them, the selection of the initial pressure value should be based on the design specifications, expected usage scenarios, and material properties of the rubber-coated dumbbell to ensure that the selected initial pressure value will not cause immediate damage to the dumbbell, and it must also be able to trigger sufficient deformation or other observable physical phenomena. The determination of the initial pressure value can usually be determined by consulting relevant documents, consulting the manufacturer, or conducting preliminary tests. For example, half of the dumbbell design pressure or a lower safety value can be selected as the initial pressure value. The pressure adjustment step determines the magnitude of the pressure increase during the test. A step that is too large may cause the key change point to be missed, while a step that is too small will increase the test time and cost. Therefore, it is necessary to select a suitable step according to the test purpose and observable physical phenomena. A suitable step can be estimated through preliminary tests or experience. Generally speaking, the step should be small enough to capture subtle changes caused by pressure changes, but should not be too small to avoid too many tests. The pressure threshold is the maximum pressure value allowed to be applied during the test. This value should be determined based on the design specifications, material properties, and safety requirements of the dumbbells to ensure that the pressure threshold does not exceed the dumbbell's tolerance limit to prevent damage or safety accidents. You can consult relevant documents, consult the manufacturer, or refer to industry standards to determine the pressure threshold. Generally speaking, the pressure threshold should be set to a value with a certain safety factor (such as 1.2 or 1.5 times) of the dumbbell design pressure.

[0022] Configuring static test parameters is a key step to ensure the effective pressure-bearing quality inspection of rubber-coated dumbbells. It helps to clarify the test scope, discover potential problems, improve test efficiency, ensure test safety, provide reference for dynamic testing, optimize product design, reduce later repair costs, and improve product reliability and user satisfaction.

[0023] Step S200: When executing the static test parameters, the high-speed camera, infrared thermal imager and ultrasonic flaw detector are synchronized to collect monitoring data of the rubber-coated part, including deformation, temperature and connection status data, and the test pressure value is associated with the monitoring data to obtain a static test data set.

[0024] Optional, ensure the test environment is stable to avoid external factors affecting the test results.

[0025] Prepare the rubber-coated dumbbell sample and place it on the test bench. Start the monitoring equipment such as high-speed camera, infrared thermal imager, ultrasonic flaw detector, etc. synchronously to ensure that they work synchronously with the test pressure. According to the operating instructions of the equipment, set appropriate parameters, such as the frame rate of the camera, the temperature range of the infrared thermal imager, the scanning frequency of the ultrasonic flaw detector, etc. According to the preset static test parameters, gradually increase the pressure to the initial pressure value and keep it stable. At each pressure level, use the monitoring equipment to collect data on the rubber-coated part. In this process, the high-speed camera is used to capture the deformation process of the rubber-coated part and record the deformation data; the infrared thermal imager is used to measure the temperature change of the rubber-coated part and record the temperature value data; the ultrasonic flaw detector is used to detect the internal structure and connection status of the rubber-coated part and record the connection status data. The monitoring data at each pressure level is associated and mapped with the corresponding test pressure value to establish a static test data set containing information such as pressure value, deformation variable, temperature value, connection status data, etc.

[0026] By synchronously collecting monitoring data such as deformation, temperature, and connection status, the performance of the rubber-coated dumbbells under different pressures can be comprehensively evaluated, providing reference and basis for subsequent dynamic tests or other types of tests, and helping to determine more appropriate test parameters and strategies.

[0027] Step S300: Based on the static test data set, the change rate is screened with deformation, temperature value and connection status data as the target, and dynamic test parameters are configured. The dynamic test parameters include initial test pressure value, test quantity, initial pressure step and dynamic step adjustment function.

[0028] For example, the deformation data of the rubber-coated dumbbells at different pressure levels are observed, and the change rate of the deformation is calculated to identify the sensitive areas of the deformation to pressure changes; the temperature value is analyzed with the pressure change, and the temperature change trend and key turning points are determined to understand the thermal performance of the rubber-coated dumbbells during the pressure-bearing process; the connection status data detected by the ultrasonic flaw detector are used to evaluate the stability and connection strength of the internal structure of the rubber-coated dumbbells, and determine the potential risk points that may exist. According to the change rate of the deformation, temperature, and connection status data, the data points that are most sensitive to pressure changes or have obvious change trends are selected. These data points will serve as an important basis for configuring dynamic test parameters to ensure that the dynamic test can cover the key performance areas of the rubber-coated dumbbells. Furthermore, an initial pressure value that can stimulate the key performance changes of the rubber-coated dumbbells is selected as the starting point based on the static test data. According to the test requirements and resource constraints, the number of cycles required for the dynamic test is determined, which can usually be determined based on factors such as the design life and frequency of use of the product. Based on the rate of change in the static test data, a suitable initial pressure step is set. This step should be small enough to capture the details of the performance changes, but should not be too small to avoid too long a test time. A dynamic step adjustment function is designed based on the rate of change trend in static test data. This function can dynamically adjust the pressure step size based on real-time test data to ensure more refined testing in key performance areas, while the step size can be appropriately increased in areas where performance changes are not obvious to save time. Before actually performing dynamic testing, simulation tests or pre-tests can be used to verify whether the configured dynamic test parameters are appropriate. Based on test results and data analysis, necessary adjustments and optimizations are made to the dynamic test parameters to ensure the accuracy and effectiveness of the test.

[0029] Through the above steps, the change rate can be screened based on the static test data set, and reasonable dynamic test parameters can be configured accordingly, thereby ensuring that the dynamic test can comprehensively and accurately evaluate the pressure bearing performance and other key performance of the rubber-coated dumbbell.

[0030] Step S400: Perform repeated loading and unloading pressure cycle tests with the initial test pressure value, test quantity, and initial pressure step length as initial configuration parameters, and collect dynamic test data of the rubber-coated part through the high-speed camera, infrared thermal imager, and ultrasonic flaw detector.

[0031] Further, ensure that the test environment is stable and avoid external interference factors, and install the prepared rubber-coated dumbbell samples on the dynamic test equipment. Check and calibrate high-speed cameras, infrared thermal imagers, ultrasonic flaw detectors and other equipment to ensure that they can work properly and accurately collect data. According to the static test data set analyzed previously, determine the initial test pressure value, test quantity and initial pressure step length, and set these parameters on the dynamic test equipment to ensure that the test can be carried out according to the preset parameters. Start the dynamic test equipment to perform repeated loading and unloading pressure cycle tests. In each pressure cycle, use a high-speed camera to record the deformation process of the rubber-coated part, including the speed, amplitude and recovery of the deformation; use an infrared thermal imager to monitor the temperature change of the rubber-coated part in real time, and record the temperature distribution and temperature change rate; use an ultrasonic flaw detector to detect the internal structure and connection status of the rubber-coated part, especially cracks, defects or connection failures that may occur when the pressure changes. Ensure that high-speed cameras, infrared thermal imagers, ultrasonic flaw detectors and other equipment work synchronously with the dynamic test equipment to collect dynamic test data of the rubber-coated part in real time. Data acquisition should cover the entire pressure cycle process, including loading, pressure holding, unloading and recovery. For each pressure cycle, the complete deformation process, temperature changes and ultrasonic testing results should be recorded and associated with the corresponding pressure value and number of cycles.

[0032] By performing repeated loading and unloading pressure cycle tests and using high-speed cameras, infrared thermal imagers, ultrasonic flaw detectors and other equipment to collect dynamic test data of the rubber-coated parts, the performance and stability of the rubber-coated dumbbells under cyclic pressure can be comprehensively evaluated, providing an important reference for product design and improvement.

[0033] Step S500: performing step length analysis according to the dynamic test data using the dynamic step length adjustment function to obtain a dynamic step length.

[0034] Optionally, analyze the dynamic test data collected by high-speed cameras, infrared thermal imagers, ultrasonic flaw detectors and other equipment, focusing on the performance of the rubber-coated parts during loading, pressure holding, unloading and recovery stages, especially the changing trends of deformation, temperature values ​​and connection status data. Based on the dynamic test data, determine at which pressure levels the rubber-coated dumbbells show obvious performance changes or potential problems. These key performance points will serve as the basis for dynamic step adjustment. According to the previously defined dynamic step adjustment function, combined with the key performance points in the dynamic test data, calculate the dynamic step size that should be used at each pressure level. The dynamic step size adjustment function should be able to dynamically adjust the step size based on real-time test data to allow for more refined testing in key performance areas, while the step size can be appropriately increased to save time in areas where performance changes are not obvious.

[0035] Through the above steps, the step length analysis can be performed through the dynamic step length adjustment function according to the dynamic test data to obtain the appropriate dynamic step length, so as to more accurately evaluate the performance and stability of the rubber-coated dumbbell under cyclic pressure.

[0036] Step S600: adjusting the initial pressure step length based on the dynamic step length, continuing to perform dynamic testing, acquiring dynamic testing data, and constructing a dynamic testing data set until a target pressure threshold is reached.

[0037] Next, according to the dynamic step size previously calculated by the dynamic step size adjustment function, the initially set pressure step size is adjusted to ensure that the new pressure step size can better reflect the performance changes of the rubber-coated dumbbells at different pressure levels, especially in the key performance areas. Continue to perform dynamic testing using the adjusted pressure step size. In each pressure cycle, high-speed cameras, infrared thermal imagers, ultrasonic flaw detectors and other equipment are used to collect dynamic test data of the rubber-coated parts to ensure the accuracy and completeness of data collection and cover the entire pressure cycle process. During the test, regularly evaluate the completed test scope and progress, check whether the key performance areas have been covered, and evaluate whether the target pressure threshold is approached or reached. If it is found during the test that the performance changes at certain pressure levels are not obvious or more refined testing is required, the dynamic step size or other test parameters can be further adjusted according to the actual situation. Such adjustments should be based on the results of test data and performance analysis to improve the efficiency and accuracy of the test.

[0038] Repeat the process of dynamic testing, evaluating test progress, and adjusting test parameters until the target pressure threshold is reached. During the entire process, dynamic test data is continuously accumulated and updated to provide an important basis for subsequent data analysis and product improvement. When the target pressure threshold is reached, the dynamic test is stopped, and all dynamic test data collected during the entire test process are sorted, classified, and stored to build a complete dynamic test data set. This data set should include key indicators such as deformation, temperature, and connection status data under each pressure cycle, as well as corresponding information such as pressure values ​​and number of cycles.

[0039] By adjusting the initial pressure step size based on the dynamic step size, continuing to perform dynamic tests and obtaining more dynamic test data until the target pressure threshold is reached, it helps to build a complete dynamic test data set, providing strong support for product performance analysis and improvement.

[0040] Step S700: performing time alignment using the static test data set and the dynamic test data set, performing consistency judgment based on the time alignment data, and determining the pressure-bearing quality test result based on the consistency judgment result.

[0041] Specifically, the data in the static test data set and the dynamic test data set are aligned in time so that they can be compared and analyzed. Specifically, the static test data (such as performance parameters under initial conditions) and the dynamic test data (such as performance changes under different pressure cycles) are feature spliced. Using deformation convolution or other appropriate algorithms, biases are calculated based on these features, and these biases are used to adjust the feature distribution of dynamic test data to achieve temporal alignment. Since static test data may only contain a single level of semantic information, pyramid pooling operations can be considered for multi-scale semantic alignment to improve the accuracy of alignment.

[0042] Furthermore, based on the time-aligned data, the consistency between the static and dynamic test results is evaluated to determine the stability and reliability of the pressure-bearing quality. Specifically, statistical methods (such as the Kappa statistic) are used to calculate the degree of consistency between the static and dynamic test results. The closer the Kappa value is to 1, the better the consistency; the closer the Kappa value is to -1, the worse the consistency. Other consistency assessment methods can also be considered, such as basic feature comparison, treatment effect modifiers-covariates, reference measure comparisons, etc. to comprehensively evaluate the consistency of the results. For data points with poor consistency, in-depth analysis should be conducted to determine whether there are potential performance issues or test errors.

[0043] Finally, the pressure-bearing quality is comprehensively evaluated and judged based on the consistency judgment results. If the consistency judgment results show good consistency between the static and dynamic test results (such as a high Kappa value), the pressure-bearing quality can be considered stable and reliable, and the test results are valid. If the consistency is poor, it is necessary to further analyze the reasons and consider retesting or adjusting the test parameters. The qualified standards and judgment basis for pressure-bearing quality testing can be formulated based on specific industry standards and specifications combined with the consistency judgment results.

[0044] By using static test data sets and dynamic test data sets for timing alignment and making consistency judgments based on the timing alignment data, the pressure quality inspection results can be determined, which helps to improve the accuracy and reliability of the test and provide strong support for the performance evaluation and improvement of pressure equipment.

[0045] Furthermore, the static test data set and the dynamic test data set are used to perform timing alignment, and consistency judgment is performed according to the timing alignment data. Step S700 of the present application further includes:

[0046] Step S710: performing deformation data alignment on the static test data set and the dynamic test data set through a dynamic time planning algorithm to determine a data alignment relationship.

[0047] Step S720: Based on the data alignment relationship, calculate the Euclidean distance between the static test data and the dynamic test data.

[0048] Step S730: construct a distance matrix based on the calculated Euclidean distances between the static test data and the dynamic test data, obtain the minimum cumulative distance from the starting point to each point, and construct a cumulative distance matrix.

[0049] Step S740: trace back from the end point of the cumulative distance matrix to the starting point, search for the minimum cumulative distance path, obtain the minimum distance, and perform consistency judgment based on the minimum distance.

[0050] Furthermore, the searching for the minimum cumulative distance path to obtain the minimum distance, step S740 of the present application further includes:

[0051] By formula Get the minimum distance, where , is the i-th shape variable in the static test data, is the jth deformation variable in the dynamic test data, is the Euclidean distance, Indicates arrival The minimum cumulative distance of points, i and j are all positive integers.

[0052] Exemplarily, in order to evaluate the effectiveness of the similarity and consistency between two test data, data alignment is performed through a dynamic time planning algorithm. Specifically, first define the static test data set as S and the dynamic test data set as D, both of which are sets of time series. Use the DTW algorithm (i.e., dynamic time planning algorithm) to compare the two time series, and stretch or compress the time series in a nonlinear way to achieve the best match between them. DTW calculates the cumulative distance between all possible alignment points in the two sequences, finds a path that minimizes the cumulative distance, and thus achieves the goal of finding the best alignment path on the time axis for two test data sets of different lengths.

[0053] Quantify the difference between the static test data and the aligned dynamic test data at each time point. For each pair of data points aligned by DTW , calculate the Euclidean distance between them. The Euclidean distance formula is . Further, create a distance matrix in which each element is the Euclidean distance between a point in the static test data set and a point in the dynamic test data set. Using the DTW algorithm, starting from the upper left corner (starting point) of the distance matrix, the cumulative distance to the lower right corner (end point) is calculated step by step and filled into the cumulative distance matrix. The calculation of the cumulative distance takes into account all possible paths and retains the minimum cumulative distance so far. Then, trace back from the end point of the cumulative distance matrix to the starting point, and select the adjacent grid point with the smallest cumulative distance in each step. The final path is the minimum cumulative distance path, and its corresponding cumulative distance is the minimum distance. According to the size of the minimum distance, the consistency between the static test data set and the dynamic test data set can be judged. A smaller minimum distance indicates a higher consistency, while a larger minimum distance may indicate a larger difference between the two. In practical applications, it is usually necessary to set one or more consistency thresholds. If the minimum distance is less than or equal to a certain threshold, the static test data set and the dynamic test data set are considered to be consistent; otherwise, they are considered to be inconsistent.

[0054] Specifically, the formula Get the minimum distance, where , is the i-th shape variable in the static test data, is the jth deformation variable in the dynamic test data, is the Euclidean distance, Indicates arrival The minimum cumulative distance of points, i and j are all positive integers.

[0055] By performing time alignment on the static test data set and the dynamic test data set, and performing consistency judgment based on the time alignment data, the accuracy and correspondence of data analysis can be guaranteed, thereby ensuring the test quality.

[0056] Furthermore, based on the static test data set, the change rate is screened with the deformation, temperature value, and connection status data as the target, and the dynamic test parameters are configured. Step S300 of the present application also includes:

[0057] Step S310: According to the static test data set, obtain test pressure data items, deformation data items, temperature data items, and connection status data items.

[0058] Step S320: respectively fitting the change relationship of the deformation data item, the temperature data item, and the connection status data item to determine the relationship type, which includes a linear relationship and a nonlinear relationship.

[0059] Step S330: perform interval segmentation according to the linear relationship and nonlinear relationship, determine the linear deformation interval, nonlinear acceleration interval, and critical value interval, determine the test pressure value based on the association mapping, partition the test pressure data items, and configure the dynamic test pressure partition value.

[0060] Step S340: determining the initial pressure step length of each partition according to the distribution discrete relationship of the pressure values ​​in the dynamic test pressure partition values.

[0061] Specifically, the test pressure data item, deformation variable data item, temperature data item, and connection status data item are obtained from the static test data set. The deformation variable data item is fitted with the test pressure to determine whether it presents a linear relationship or a nonlinear relationship. The fitting method can choose to use regression analysis methods, such as linear regression, polynomial regression, etc. Similarly, the temperature data item is fitted with the test pressure to determine the type of relationship. If the connection status data is quantitative (such as connection strength, etc.), similar fitting can be performed; if it is non-quantitative (such as whether the connection is broken), logical judgment may be required based on the actual test results.

[0062] Further, according to the linear relationship and the nonlinear relationship, the interval is segmented. After segmentation, a linear deformation interval, a nonlinear acceleration interval, and a critical value interval are obtained. Among them, the linear deformation interval refers to determining one or more linear deformation intervals within the interval where the deformation variable and the test pressure are linearly related; the nonlinear acceleration interval refers to determining a nonlinear acceleration interval within the interval where the rate of change of the deformation variable or the temperature value increases; the critical value interval refers to determining a critical value interval according to the threshold of the connection state data or the deformation variable and the temperature value, and these intervals may represent potential problems or failure points. Then, based on the above intervals, the test pressure data items are partitioned, and each interval corresponds to a dynamic test pressure partition value. By associative mapping, it is ensured that each pressure partition value corresponds to a specific deformation variable, temperature or connection state change. Finally, according to the distribution discrete relationship of the pressure value in the dynamic test pressure partition value, the initial pressure step of each partition is determined. In the linear deformation interval, since the change is relatively stable, a larger initial pressure step can be set to speed up the test speed; in the nonlinear acceleration interval or when approaching the critical value interval, a smaller initial pressure step should be set to more accurately capture the performance change.

[0063] Through the above steps, dynamic test parameters can be configured according to the static test data set to ensure that the dynamic test can comprehensively and accurately evaluate the pressure-bearing quality.

[0064] Furthermore, the step S300 of configuring the dynamic test parameters in the present application further includes:

[0065] Step S350:

[0066] , in, is the current pressure step, is the deformation rate , is the temperature change rate , is the connection status detection data, indicating the looseness of the dumbbell under the current pressure, and N is the new step size of dynamic adjustment.

[0067] Step S360: When the deformation rate, temperature change rate or connection status detection data exceeds a set high threshold, the step size is reduced.

[0068] Step S370: When the deformation rate, temperature change rate and connection status detection data are all lower than the set low threshold, increase the step size.

[0069] Step S380: In other cases, keep the current step length unchanged.

[0070] In this embodiment, the pressure step size can be configured by Conduct, among which, is the current pressure step, is the deformation rate , is the temperature change rate , is the connection state detection data, indicating the degree of looseness of the dumbbell under the current pressure, and N is the new step length for dynamic adjustment. When the deformation rate, temperature change rate or connection state detection data exceeds the set high threshold, the step length is reduced to form a new reduced step length; when the deformation rate, temperature change rate and connection state detection data are all lower than the set low threshold, the step length is increased to form a new increased step length; in other cases, the current step length is kept unchanged.

[0071] Configuring specific adjustment strategies helps to more accurately control pressure and avoid potential performance issues during testing, ensuring the accuracy and safety of the testing process.

[0072] Furthermore, the ultrasonic flaw detector collects monitoring data of the rubber-encapsulated part, and step S200 of the present application also includes:

[0073] Step S210: collecting the reflected waveform of the rubber-coated part by the ultrasonic flaw detector, and extracting the amplitude, arrival time, and waveform shape from the reflected waveform.

[0074] Step S220: using the amplitude, arrival time, and waveform shape as processing data, performing identification processing through a classifier, and determining loose points and loose classification information, wherein the loose classification information is used to characterize the degree of looseness.

[0075] Step S230: generating the connection status data according to the loose point and the loose classification information.

[0076] Optionally, an ultrasonic flaw detector is used to transmit ultrasonic waves to the rubber-coated part, and the ultrasonic wave signal reflected from the rubber-coated part is received to form a reflected waveform. Specifically, the ultrasonic flaw detector transmits ultrasonic pulses through the transmitter, and the ultrasonic wave propagates in the rubber-coated part. When it encounters different media or structures, it will be reflected. The reflected ultrasonic wave is received by the receiver of the flaw detector and converted into an electrical signal. After these electrical signals are processed, a visual reflected waveform can be formed. Then, key information such as amplitude, arrival time, and waveform shape are extracted from the reflected waveform as processing data. Among them, the amplitude can be obtained by measuring the peak value or effective value of the reflected waveform; the arrival time refers to the recorded time difference from the ultrasonic wave emission to the reception of the reflected wave; the waveform shape can be described by analyzing the overall shape of the reflected waveform, the number of peaks, the frequency distribution and other characteristics.

[0077] Next, the extracted processed data is input into the classifier, and the loose points and loose classification information are identified by the classifier. Specifically, a suitable classifier is selected or trained, such as a support vector machine (SVM), random forest, neural network, etc., and the extracted processed data is used as the input of the classifier. The classifier identifies whether there are loose points based on the input data and gives loose classification information. The loose classification information can be a numerical value (such as 1-5 levels) or a descriptive label (such as slightly loose, moderately loose, severely loose). Based on the identified loose points and loose classification information, the connection status data of the lagging part is generated. The location, quantity and loose classification information of the identified loose points are integrated into the connection status data. The connection status data can be a structure or a database record that contains all the necessary information about the connection status of the lagging part. The connection status data can be used for subsequent analysis, reporting or alarm triggering.

[0078] Furthermore, before the identification process is performed by the classifier, step S220 of the present application further includes:

[0079] Step S221: Collect ultrasonic flaw detection monitoring data including normal and different looseness degrees to construct a sample data set.

[0080] Step S222: normalize the sample data set, and encode the loose classification labels of the sample data set using one-hot encoding.

[0081] Step S223: extract the processed sample data set to construct a training sample set and a test sample set.

[0082] Step S224: construct a gradient boosting tree model, initialize model parameters, train and test the gradient boosting tree model through the training sample set and the test sample set, respectively, to obtain the classifier, which is a gradient boosting tree model after training convergence, including:

[0083] Step S2244: Divide the training sample set into K subsets, use one of the subsets as a test set and the remaining subsets as training sets, verify the initialization model parameters, judge the verification performance results, set the search direction and search step length based on the initialization model parameters to determine multiple parameter combinations, repeat the verification for each group of parameter combinations, obtain K verification results, and select the best parameter combination.

[0084] Step S2245: Divide each subset into M subsets, configure the best parameter combination as the current model parameters, use the M subsets to verify the current model parameters, and determine the optimal parameter combination.

[0085] Exemplarily, ultrasonic flaw detection monitoring data containing normal and different degrees of looseness are collected to construct a sample data set. That is, ultrasonic flaw detectors are used to collect ultrasonic data of the lagging parts with different degrees of looseness, and these data are matched with known looseness labels (normal, slightly loose, moderately loose, severely loose, etc.), and a sample data set containing features (such as amplitude, arrival time, waveform shape, etc.) and labels (loose classification) is constructed. The sample data set is normalized, and the loose classification labels of the sample data set are encoded using one-hot encoding. Among them, the normalization refers to scaling all feature data to the same scale so that the model can better process them. Common normalization methods include minimum-maximum normalization and Z-score normalization. The one-hot encoding means that for classification labels, if they are in text form (such as "normal", "slightly loose", etc.), they need to be converted into numerical form using one-hot encoding. One-hot encoding creates a new binary column for each category and sets 1 (if the sample belongs to the category) or 0 (if it does not belong) on ​​the category for each sample. Furthermore, the processed sample data set is extracted to construct a training sample set and a test sample set. The training set is used to train the model, and the test set is used to evaluate the performance of the model.

[0086] Then, a gradient boosting tree model is constructed. Gradient boosting trees are an ensemble learning method based on decision trees. It constructs multiple decision trees and combines their results to produce the final prediction. Before starting training, some initial parameters need to be set for the gradient boosting tree model to initialize, such as learning rate, number of trees, tree depth, subsampling rate, etc. These parameters will affect the training process and final performance of the model. The training sample set is divided into K subsets (or "folds"), and one of the subsets is selected as the validation set each time, and the remaining K-1 subsets are used as the training set. This process is repeated K times, and a different subset is selected as the validation set each time. In each round of K-fold cross-validation, the model is trained using the current parameter combination, and its performance (such as accuracy, recall, F1 score, etc.) is evaluated on the validation set. Then, the model parameters are adjusted based on the evaluation results. This step is usually completed through a search strategy (such as grid search, random search, or Bayesian optimization). The strategy starts with the initial parameters, sets the search direction and step size, determines multiple parameter combinations, and repeatedly verifies the performance of each set of parameters. After completing K-fold cross-validation, you will get K validation results (one for each validation set). Compare these results and select the parameter combination that optimizes the model performance. That is, select the parameter combination that gives the highest performance indicator on the validation set (such as average accuracy, average F1 score, etc.) as the best parameter combination.

[0087] The above steps can improve model performance, prevent overfitting and underfitting, enhance model robustness, select optimal parameters for the model, and provide reliable model evaluation.

[0088] Furthermore, the initialization model parameters, step S224 of the present application also includes:

[0089] Step S2241: Obtain the range of various parameters, including the number of trees, learning rate, and maximum depth.

[0090] Step S2242: perform minimum unit numerical segmentation based on the parameter ranges, configure a search grid, and configure numerical position coding according to the row and column relationship of the grid.

[0091] Step S2243: Perform position encoding acquisition through a random number function, extract random parameter values, and obtain the initialization model parameters.

[0092] Specifically, when initializing the model parameters, first determine the range of each parameter, including the number of trees, learning rate, and maximum depth. Among them, the learning rate controls the contribution of each tree to the final result and is usually set to a smaller range; the maximum depth of the tree can be set according to the complexity of the data and the number of trees. For each parameter, determine a minimum unit value split, that is, for continuous parameters (such as learning rate), a step size will be selected to generate a series of discrete values; for discrete parameters (such as the number of trees and maximum depth), a suitable interval can be selected to generate a series of possible values. Next, based on the above split values, configure the search grid. The search grid is a list or array containing all possible parameter combinations. Then, according to the row and column relationship of the grid, a numerical position code is configured for each parameter combination. These codes can track or record the position of the parameter combination during the search process. Finally, a random position code is selected using a random number function, that is, a random parameter combination corresponding to the search grid is obtained, and then the corresponding parameter value is extracted to initialize the model parameters. The above random initialization process can avoid over-reliance on specific parameter selections, and also help explore different areas in the parameter space to find better model configurations.

[0093] Through the technical solutions of the above-mentioned embodiments, the quality inspection method for rubber-coated dumbbells provided in the present application solves the technical problems of the existing quality inspection methods for rubber-coated dumbbells, that is, the quality inspection only stays at the preliminary evaluation, the inspection method is not in-depth and intelligent enough, and the inspection results are not accurate enough, resulting in safety hazards in the equipment. The technical effect of improving the accuracy of quality inspection of rubber-coated dumbbells and thus ensuring the safety of use and user experience is achieved.

[0094] Embodiment 2

[0095] Based on the same inventive concept as the quality inspection method for rubber-coated dumbbells in the aforementioned embodiment, Figure 2 As shown, the present application provides a quality detection device for rubber-coated dumbbells, the device comprising:

[0096] The static parameter configuration module 11 is used to configure static test parameters, including an initial pressure value, a pressure adjustment step, and a pressure threshold.

[0097] The static data acquisition module 12 is used to synchronize the high-speed camera, infrared thermal imager, and ultrasonic flaw detector to collect monitoring data of the rubber-coated part when executing the static test parameters, including deformation, temperature, and connection status data, and associate and map the test pressure value with the monitoring data to obtain a static test data set.

[0098] The dynamic parameter configuration module 13 is used to perform change rate screening based on the static test data set, with deformation, temperature value, and connection status data as the target, and configure dynamic test parameters. The dynamic test parameters include initial test pressure value, test quantity, initial pressure step and dynamic step adjustment function.

[0099] The dynamic data acquisition module 14 is used to perform repeated loading and unloading pressure cycle tests with the initial test pressure value, test quantity, and initial pressure step length as initial configuration parameters, and collect dynamic test data of the rubber-coated part through the high-speed camera, infrared thermal imager, and ultrasonic flaw detector.

[0100] The step length analysis module 15 is used to perform step length analysis according to the dynamic test data through the dynamic step length adjustment function to obtain a dynamic step length.

[0101] The step length adjustment module 16 is used to adjust the initial pressure step length based on the dynamic step length, continue to perform dynamic testing, and obtain dynamic testing data until a target pressure threshold is reached to construct a dynamic testing data set.

[0102] The test result acquisition module 17 is used to perform time alignment using the static test data set and the dynamic test data set, perform consistency judgment based on the time alignment data, and determine the pressure-bearing quality test result based on the consistency judgment result.

[0103] Furthermore, the detection result obtaining module 17 is further configured to perform the following steps:

[0104] The static test data set and the dynamic test data set are aligned with each other in shape variables by using a dynamic time planning algorithm to determine a data alignment relationship.

[0105] Based on the data alignment relationship, the Euclidean distance between the static test data and the dynamic test data is calculated.

[0106] According to the calculated Euclidean distance between the static test data and the dynamic test data, a distance matrix is ​​constructed, the minimum cumulative distance from the starting point to each point is obtained, and the cumulative distance matrix is ​​constructed.

[0107] The minimum cumulative distance path is searched from the end point of the cumulative distance matrix back to the starting point to obtain the minimum distance, and consistency judgment is performed based on the minimum distance.

[0108] Furthermore, the detection result obtaining module 17 is further configured to perform the following steps:

[0109] By formula Get the minimum distance, where , The i-th shape variable in the static test data, is the jth deformation variable in the dynamic test data, is the Euclidean distance, Indicates arrival The minimum cumulative distance of points, i and j are all positive integers.

[0110] Furthermore, the dynamic parameter configuration module 13 is further configured to perform the following steps:

[0111] According to the static test data set, a test pressure data item, a deformation variable data item, a temperature data item, and a connection status data item are obtained.

[0112] The change relationship fitting is performed on the deformation data item, the temperature data item, and the connection state data item respectively to determine the relationship type, and the relationship type includes a linear relationship and a nonlinear relationship.

[0113] According to the linear relationship and nonlinear relationship, interval segmentation is performed to determine the linear deformation interval, nonlinear acceleration interval, and critical value interval, the test pressure value is determined based on the association mapping, the test pressure data item is partitioned, and the dynamic test pressure partition value is configured.

[0114] The initial pressure step length of each partition is determined according to the distribution discrete relationship of the pressure values ​​in the dynamic test pressure partition values.

[0115] Furthermore, the dynamic parameter configuration module 13 is further configured to perform the following steps:

[0116] , in, is the current pressure step, is the deformation rate , is the temperature change rate , is the connection status detection data, indicating the looseness of the dumbbell under the current pressure, and N is the new step size of dynamic adjustment.

[0117] When the deformation rate, temperature change rate or connection status detection data exceeds the set high threshold, the step size is reduced.

[0118] When the deformation rate, the temperature change rate and the connection state detection data are all lower than the set low threshold, the step size is increased.

[0119] In other cases, keep the current step size unchanged.

[0120] Furthermore, the static data acquisition module 12 is further configured to perform the following steps:

[0121] The ultrasonic flaw detector is used to collect the reflected waveform of the rubber-encapsulated part, and the amplitude, arrival time and waveform shape are extracted from the reflected waveform.

[0122] The amplitude, arrival time, and waveform shape are used as processing data, and are identified and processed by a classifier to determine the loose point and loose classification information, and the loose classification information is used to characterize the degree of looseness.

[0123] The connection status data is generated according to the loose point and the loose classification information.

[0124] Furthermore, the static data acquisition module 12 is further configured to perform the following steps:

[0125] Ultrasonic flaw detection monitoring data including normal and different looseness degrees are collected to construct a sample data set.

[0126] The sample data set is normalized, and the loose classification labels of the sample data set are encoded using one-hot encoding.

[0127] The processed sample data set is extracted to construct training sample sets and test sample sets.

[0128] Construct a gradient boosting tree model, initialize model parameters, train and test the gradient boosting tree model through the training sample set and the test sample set, respectively, to obtain the classifier, which is a gradient boosting tree model after training convergence, including:

[0129] The training sample set is divided into K subsets, one of which is used as a test set and the remaining subsets are used as training sets. The initialization model parameters are verified, and the verification performance results are judged. The search direction and search step size are set with the initialization model parameters as the starting point to determine multiple parameter combinations. The verification is repeated for each group of parameter combinations to obtain K verification results, and the best parameter combination is selected.

[0130] Each subset is divided into M subsets, the optimal parameter combination is configured as the current model parameters, the current model parameters are verified using the M subsets, and the optimal parameter combination is determined.

[0131] Furthermore, the static data acquisition module 12 is further configured to perform the following steps:

[0132] Get the range of various parameters, including the number of trees, learning rate, and maximum depth.

[0133] Based on the parameter ranges, the minimum unit value segmentation is performed, a search grid is configured, and the value position coding is configured according to the row and column relationship of the grid.

[0134] The position encoding is acquired through a random number function, and the random parameter value is extracted to obtain the initialization model parameter.

[0135] Through the above detailed description of the quality inspection method for rubber-coated dumbbells in this specification, those skilled in the art can clearly know the quality inspection device for rubber-coated dumbbells in this embodiment. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0136] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A quality inspection method for rubber-coated dumbbells, characterized in that: The quality inspection method for the rubber-coated dumbbells comprises: Configure static test parameters, including initial pressure value, pressure adjustment step, and pressure threshold; When executing the static test parameters, the high-speed camera, infrared thermal imager and ultrasonic flaw detector are synchronized to collect monitoring data of the rubber-coated part, including deformation, temperature and connection status data, and the test pressure value is associated with the monitoring data to obtain a static test data set; Based on the static test data set, the change rate is screened with deformation, temperature and connection status data as the target, and dynamic test parameters are configured, wherein the dynamic test parameters include initial test pressure value, test quantity, initial pressure step and dynamic step adjustment function; Repeated loading and unloading pressure cycle tests are performed with the initial test pressure value, test quantity, and initial pressure step length as initial configuration parameters, and dynamic test data of the rubber-coated part is collected by the high-speed camera, infrared thermal imager, and ultrasonic flaw detector; Performing step length analysis according to the dynamic test data through the dynamic step length adjustment function to obtain a dynamic step length; Adjusting the initial pressure step length based on the dynamic step length, continuing to perform dynamic testing, acquiring dynamic testing data, and constructing a dynamic testing data set until a target pressure threshold is reached; The static test data set and the dynamic test data set are used to perform time alignment, and consistency judgment is performed based on the time alignment data, and the pressure-bearing quality detection result is determined based on the consistency judgment result.

2. The quality inspection method for rubber-coated dumbbells according to claim 1, characterized in that: Performing timing alignment using the static test data set and the dynamic test data set, and performing consistency judgment according to the timing alignment data, including: Performing shape variable data alignment on the static test data set and the dynamic test data set by a dynamic time planning algorithm to determine a data alignment relationship; Based on the data alignment relationship, calculating the Euclidean distance between the static test data and the dynamic test data; According to the calculated Euclidean distance between the static test data and the dynamic test data, a distance matrix is ​​constructed, the minimum cumulative distance from the starting point to each point is obtained, and a cumulative distance matrix is ​​constructed; The minimum cumulative distance path is searched from the end point of the cumulative distance matrix back to the starting point to obtain the minimum distance, and consistency judgment is performed based on the minimum distance.

3. The quality inspection method for rubber-coated dumbbells according to claim 2, characterized in that: The searching for the minimum cumulative distance path to obtain the minimum distance includes: By formula Get the minimum distance, where , is the i-th shape variable in the static test data, is the jth deformation variable in the dynamic test data, is the Euclidean distance, Indicates arrival The minimum cumulative distance of points, i and j are all positive integers.

4. The quality inspection method for rubber-coated dumbbells according to claim 1, characterized in that: Based on the static test data set, the change rate is screened with the deformation, temperature value, and connection status data as the target, and the dynamic test parameters are configured, including: According to the static test data set, obtaining test pressure data items, deformation data items, temperature data items, and connection status data items; Performing change relationship fitting on the deformation data item, the temperature data item, and the connection state data item respectively to determine the relationship type, wherein the relationship type includes a linear relationship and a nonlinear relationship; According to the linear relationship and the nonlinear relationship, interval segmentation is performed to determine the linear deformation interval, the nonlinear acceleration interval, and the critical value interval, the test pressure value is determined based on the association mapping, the test pressure data item is partitioned, and the dynamic test pressure partition value is configured; The initial pressure step length of each partition is determined according to the distribution discrete relationship of the pressure values ​​in the dynamic test pressure partition values.

5. The quality inspection method for rubber-coated dumbbells according to claim 4, characterized in that: The configuration of dynamic test parameters also includes: ,in, is the current pressure step, is the deformation rate , is the temperature change rate , is the connection status detection data, indicating the degree of looseness of the dumbbell under the current pressure, and N is the new step length of dynamic adjustment; When the deformation rate, temperature change rate or connection status detection data exceeds the set high threshold, the step size is reduced; When the deformation rate, temperature change rate and connection state detection data are all lower than a set low threshold, increasing the step size; In other cases, keep the current step size unchanged.

6. The quality inspection method for rubber-coated dumbbells according to claim 1, characterized in that: The ultrasonic flaw detector collects monitoring data of the rubber-coated part, including: The ultrasonic flaw detector is used to collect the reflected waveform of the rubber-coated part, and the amplitude, arrival time and waveform shape are extracted from the reflected waveform; The amplitude, arrival time, and waveform shape are used as processing data, and are identified and processed by a classifier to determine the loose point and loose classification information, wherein the loose classification information is used to characterize the degree of looseness; The connection status data is generated according to the loose point and the loose classification information.

7. The quality inspection method for rubber-coated dumbbells according to claim 6, characterized in that: The identification process by the classifier may also include: Collect ultrasonic flaw detection monitoring data including normal and different looseness degrees to build a sample data set; Normalizing the sample data set, and encoding the loose classification labels of the sample data set using one-hot encoding; Extract the processed sample data set to construct a training sample set and a test sample set; Construct a gradient boosting tree model, initialize model parameters, train and test the gradient boosting tree model through the training sample set and the test sample set, respectively, to obtain the classifier, which is a gradient boosting tree model after training convergence, including: The training sample set is divided into K subsets, one of which is used as a test set and the other subsets are used as training sets, the initialization model parameters are verified, and the verification performance results are judged, and the search direction and search step length are set based on the initialization model parameters as the starting point to determine multiple parameter combinations, and each group of parameter combinations is repeatedly verified to obtain K verification results, and the best parameter combination is selected; Each subset is divided into M subsets, the optimal parameter combination is configured as the current model parameters, the current model parameters are verified using the M subsets, and the optimal parameter combination is determined.

8. The quality inspection method for rubber-coated dumbbells according to claim 7, characterized in that: The initialization model parameters include: Get the range of parameters, including the number of trees, learning rate, and maximum depth; Performing minimum unit numerical segmentation based on the parameter ranges, configuring a search grid, and configuring numerical position coding according to the row and column relationship of the grid; The position encoding is obtained through a random number function, and the random parameter value is extracted to obtain the initialization model parameter.

9. A quality inspection device for rubber-coated dumbbells, characterized in that: Used to implement the quality inspection method for rubber-coated dumbbells according to any one of claims 1 to 8, the device comprises: Static parameter configuration module, used to configure static test parameters, including initial pressure value, pressure adjustment step, and pressure threshold; A static data acquisition module is used to synchronize the high-speed camera, infrared thermal imager, and ultrasonic flaw detector to collect monitoring data of the rubber-encapsulated part when executing the static test parameters, including deformation, temperature, and connection status data, and associate and map the test pressure value with the monitoring data to obtain a static test data set; A dynamic parameter configuration module is used to perform change rate screening based on the static test data set with deformation, temperature value, and connection status data as targets, and configure dynamic test parameters, wherein the dynamic test parameters include an initial test pressure value, a test quantity, an initial pressure step, and a dynamic step adjustment function; A dynamic data acquisition module is used to perform repeated loading and unloading pressure cycle tests with the initial test pressure value, test quantity, and initial pressure step length as initial configuration parameters, and to collect dynamic test data of the rubber-coated part through the high-speed camera, infrared thermal imager, and ultrasonic flaw detector; A step length analysis module, used for performing step length analysis according to the dynamic test data through the dynamic step length adjustment function to obtain a dynamic step length; A step length adjustment module, configured to adjust the initial pressure step length based on the dynamic step length, continue to perform the dynamic test, obtain dynamic test data, and construct a dynamic test data set until a target pressure threshold is reached; The detection result acquisition module is used to use the static test data set and the dynamic test data set to perform time alignment, perform consistency judgment based on the time alignment data, and determine the pressure-bearing quality detection result based on the consistency judgment result.

Images