Intelligent motor quality detection method and system based on data difference feedback

Abstract

The invention discloses a motor quality intelligent detection method and system based on data difference feedback, relates to the technical field of motor operation quality detection, and proposes the following scheme: the method comprises the steps of collecting real-time operation parameters of current, vibration and temperature of motors of the same model, and obtaining a rated parameter reference of the model; comparing the real-time operation parameters with a rated parameter reference, marking the abnormal real-time operation parameters as real-time detection data, taking each parameter item as a detection item, and generating motor detection feature information; and calculating the historical fault weight of each detection item according to the model motor fault database. A detection technology conflict optimization model is constructed through the game theory, a three-dimensional scoring system and a double-interval dynamic screening mechanism are established, high-integration weight detection items and optimal technology combinations are automatically matched, average detection steps are reduced, the resource conflict rate is reduced, and collaborative optimization of detection precision and efficiency is achieved.

Description

Technical Field

[0001] This invention relates to the field of motor operation quality detection technology, specifically to an intelligent motor quality detection method and system based on data difference feedback. Background Technology

[0002] In the field of motor manufacturing, companies typically need to periodically recall motors for quality assessment, verifying quality consistency by comparing factory baseline parameters with actual operating data. The current industry-standard technical solution relies on comparing historical databases with real-time parameters. Specifically, this involves fusing historical data and real-time parameters with fixed weights and triggering the testing process based on preset thresholds.

[0003] The aforementioned motor testing scheme uses a fixed ratio to fuse historical data with real-time parameters, resulting in insufficient statistics on low-frequency, high-loss faults. It cannot dynamically adapt to motor aging or sudden parameter anomalies, leading to biases in fault risk assessment data. These assessment biases are input into the decision-making system, and some systems lack dynamic decision-making mechanisms, resulting in a mismatch between the selection of testing technologies and resource allocation. This makes it impossible to accurately match fault risk levels, leading to a double loss in testing efficiency and accuracy. Therefore, we propose a motor quality intelligent testing method and system based on data difference feedback to solve this problem. Summary of the Invention

[0004] To address the aforementioned technical problems, this paper provides an intelligent motor quality detection method and system based on data difference feedback. This technical solution solves the problem that traditional motor detection schemes mentioned in the background technology use a fixed ratio to fuse historical data and real-time parameters, resulting in insufficient statistics on low-frequency, high-loss faults and an inability to dynamically adapt to motor aging or sudden parameter anomalies. This leads to systematic biases in fault risk assessment data, and the resulting assessment biases are input into the decision-making system. Furthermore, the lack of a dynamic decision-making mechanism results in a mismatch between the selection of detection technology and resource allocation, making it impossible to accurately match fault risk levels and causing a double loss of detection efficiency and accuracy.

[0005] To achieve the above objectives, the technical solution adopted by this invention is as follows: This invention provides an intelligent motor quality detection method based on data difference feedback, the method comprising: Collect real-time operating parameters of current, vibration and temperature of motors of the same model, and obtain the rated parameter benchmark of that model; By comparing real-time operating parameters with rated parameters, abnormal real-time operating parameters are marked as real-time detection data, each parameter item is used as a detection item, and motor detection feature information is generated. The historical fault weights of each test item are calculated based on the fault database of this type of motor. At the same time, parameter deviation weights are generated based on the motor test feature information. The historical fault weights and parameter deviation weights are weighted and merged to obtain the comprehensive weight of the test item. A mapping table between test item types and test technologies is constructed. Based on historical test data, the historical comprehensive weight of each test item is calculated, weight intervals are divided and their correspondence with test technologies is established. Combined with the frequency intervals constructed using the mean and standard deviation of the historical comprehensive weights, test technologies are dynamically screened to generate a preliminary set of test technologies. A game theory decision model is established for a preliminary set of detection technologies. Conflict optimization is performed on all detection technologies to generate the optimal sequence of detection technologies. The detection technology is executed sequentially. If a certain detection technology confirms a fault, the subsequent detection is terminated; otherwise, the remaining detection technologies are executed until the motor detection result is generated.

[0006] Furthermore, a motor quality intelligent detection system based on data difference feedback is proposed to implement the motor quality intelligent detection method based on data difference feedback described above, including: A multi-source data acquisition module is used to acquire real-time operating parameters of current, vibration and temperature of motors of the same model, and obtain the rated parameter benchmark of the model. The intelligent weight calculation module is used to calculate the historical fault weight of each detection item based on the fault database of the motor model, and at the same time generate parameter deviation weight based on the motor detection feature information. The historical fault weight and parameter deviation weight are weighted and fused to obtain the comprehensive weight of the detection item. The detection technology screening module is used to construct a mapping table between detection item types and detection technologies, calculate the historical comprehensive weight of each detection item based on historical detection data, divide the weight interval and establish its correspondence with the detection technology, and dynamically screen detection technologies by combining the frequency interval constructed with the mean and standard deviation of the historical comprehensive weight, and generate a preliminary set of detection technologies. The game optimization control module is used to establish a game theory decision model for a preliminary set of detection technologies, perform conflict optimization on all detection technologies, and generate an optimal sequence of detection technologies. The motor detection execution module is used to perform detection sequentially using a sequence of detection technologies. If a certain detection technology confirms a fault, the subsequent detection is terminated; otherwise, the remaining detection technologies are executed until the motor detection result is generated.

[0007] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By quantifying the risk of low-frequency, high-loss faults using the Poisson correction formula and combining the dynamic fusion of historical fault weights and real-time parameter deviations, the problem of poor adaptability of traditional fixed-proportion fusion methods to equipment aging and sudden anomalies is solved, thereby improving the detection rate of low-frequency, high-loss faults and avoiding over-detection of low-risk parameters. 2. By constructing a detection technology conflict optimization model through game theory, and establishing a three-dimensional scoring system and a dual-interval dynamic screening mechanism, the system automatically matches high-weight detection items with the optimal technology combination, reduces the average detection steps, lowers the resource conflict rate, and achieves synergistic optimization of detection accuracy and efficiency. Attached Figure Description

[0008] Figure 1 This is a flowchart of an intelligent motor quality detection method based on data difference feedback proposed in this invention; Figure 2 This is a flowchart illustrating the acquisition of the preliminary set of detection technologies in this invention. Figure 3 This is a flowchart illustrating the process of obtaining the optimal detection technology sequence in this invention. Figure 4 This is a structural block diagram of a motor quality intelligent detection system based on data difference feedback proposed in this invention. Detailed Implementation

[0009] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0010] Reference Figure 1-3 As shown, a method for intelligent detection of motor quality based on data difference feedback is provided, the method comprising: Collect real-time operating parameters of current, vibration and temperature of motors of the same model, and obtain the rated parameter benchmark of that model; By comparing real-time operating parameters with rated parameters, abnormal real-time operating parameters are marked as real-time detection data, each parameter item is used as a detection item, and motor detection feature information is generated. Specifically, time-series data of current, vibration and temperature are collected synchronously by multiple sensors. After filtering, noise reduction and normalization preprocessing, anomaly detection is performed based on quantitative indicators: the effective value of current, the effective value of vibration velocity and the average temperature are used as core detection items. If the effective value of current exceeds the cumulative duration of the rated benchmark within a unit operating cycle and breaks through the preset threshold, it is marked as an anomaly. Finally, the quantitative values ​​and anomaly markings are integrated to generate structured feature information, which is recorded as motor detection feature information. The historical fault weights of each test item are calculated based on the fault database of this type of motor. At the same time, parameter deviation weights are generated based on the motor test feature information. The historical fault weights and parameter deviation weights are weighted and merged to obtain the comprehensive weight of the test item. A mapping table between test item types and test technologies is constructed. Based on historical test data, the historical comprehensive weight of each test item is calculated, weight intervals are divided and their correspondence with test technologies is established. Combined with the frequency intervals constructed using the mean and standard deviation of the historical comprehensive weights, test technologies are dynamically screened to generate a preliminary set of test technologies. A game theory decision model is established for a preliminary set of detection technologies. Conflict optimization is performed on all detection technologies to generate the optimal sequence of detection technologies. The detection technology is executed sequentially. If a certain detection technology confirms a fault, the subsequent detection is terminated; otherwise, the remaining detection technologies are executed until the motor detection result is generated.

[0011] Specifically, the historical fault weights of each detection item are calculated based on the fault database of this model of motor, and parameter deviation weights are generated based on the motor detection feature information. The historical fault weights and parameter deviation weights are then weighted and fused to obtain the comprehensive weight of the detection item, which specifically includes: Based on the fault database of this model of motor, the frequency of occurrence of each fault type is statistically analyzed. Combined with the preset severity coefficient, the occurrence frequency of each fault type is analyzed by using the Poisson correction formula. It should be noted that by correcting the occurrence frequency of each fault type by using the Poisson correction formula, it is ensured that the data in the fault database can accurately reflect the real risk of low-frequency high-loss faults. Optionally, fault priority can be quantified by two dimensions: downtime and maintenance cost. Each dimension is divided into 1 to 5 points. For example, downtime exceeding 24 hours is scored as 5 points, and maintenance cost less than 1,000 yuan is scored as 1 point. Combined with preset weights, such as downtime accounting for 60% and maintenance cost accounting for 40%, a weighted comprehensive score is calculated. The comprehensive score is divided by 10 and then added by 1 to obtain the severity coefficient of the fault type, thereby achieving standardized quantification of fault priority. Calculate the probability of abnormality of the detection item under each fault type. The value is the ratio of the number of cases in which the detection item is abnormal to the total number of fault cases. The correlation degree of each detection item is calculated by accumulating the product of the frequency of occurrence of each fault type and the abnormal probability of its corresponding detection item. Then, the correlation degree of all detection items is normalized to generate the historical fault weight of the detection item. The deviation of the real-time operating parameters of each test item from the rated parameter benchmark is calculated in real time. The value is the ratio of the difference between the real-time operating parameters and the rated parameter benchmark to the rated parameter benchmark. The parameter deviation weight is generated based on this deviation. A dynamic weight fusion algorithm based on operating conditions is adopted to dynamically weight and fuse the historical fault weights and parameter deviation weights of the detection items to generate a comprehensive weight. The comprehensive weights of all detection items are then normalized so that the sum of the comprehensive weights is constant to a fixed value; for example, the fixed value can be 1. For example, in the working condition comprehensive weight fusion algorithm, the weight allocation is dynamically adjusted according to the motor operating status: motors that have been running for more than 60% of their design life are defined as old motors, and their historical fault weights and parameter deviation weights are fused at a ratio of 7:3. Motors that do not reach this threshold are new motors, and the weight allocation is 4:6. This design is based on the fact that old motors have more complete fault data, while new motors need to rely more on real-time parameter monitoring. The Poisson correction formula is: In the formula, The frequency of occurrence of each fault type, The number of times a certain fault type occurs. The total number of failures across all failure types. Number the fault type. As a smoothing factor, To adjust the coefficient, Severity coefficient; For example, A value of 1 is acceptable to ensure that low-frequency faults are not ignored. The possible value is 3, which defaults to at least one occurrence of all faults. Next, through Limit the compensation range to prevent the weight of high-frequency faults from being diluted.

[0012] Understandable =1 is a common choice in sparse data processing (refer to the smoothing strategy of the Naive Bayes classifier in "Statistical Learning Methods"). Its physical meaning is "assuming each fault occurs at least once," avoiding the zero-probability occurrence of low-frequency faults with zero weights. In motor fault detection, if low-frequency, high-loss faults are not smoothed, they will... Excluded from the key testing areas, and This ensures that it obtains a basic weight in the formula. The choice of k=3 is based on motor fault analysis. The occurrence frequency of high-frequency motor faults (such as winding insulation aging) is usually 3-5 times that of low-frequency faults. While minimizing the occurrence of low-frequency faults, this approach increases the proportion of low-frequency faults in the overall fault frequency. Therefore, it is necessary to adjust the fault frequency using an adjustment coefficient to compensate for the weight of high-frequency faults. In this embodiment, k=3 indicates that "the default base occurrence count of all faults does not exceed..." "Three times" ensures that low-frequency faults are not ignored, while also preventing the dominance of high-frequency faults from being weakened.

[0013] It is important to note the severity level. This indicates the severity of each fault type. It is a quantitative indicator reflecting the "severity" of the fault. The severity of each fault type is scored based on expert evaluation, with a value range of 1-10 (1 being the lowest risk and 10 being the highest risk).

[0014] See Figure 2 As shown, the mapping table between detection item types and detection technologies is constructed. Based on historical detection data, the historical comprehensive weight of each detection item is calculated, weight intervals are divided, and their correspondence with detection technologies is established. Combined with frequency intervals constructed using the mean and standard deviation of historical comprehensive weights, detection technologies are dynamically screened to generate a preliminary set of detection technologies. Specifically, this includes: Establish a mapping table between detection item types and selectable detection technologies. The detection item types include current, vibration, and temperature. Set a detection accuracy coefficient, execution efficiency coefficient, and resource consumption coefficient for each detection technology. Optionally, each detection technology can be tested and verified in a standard fault sample library, and its correct recognition rate (TPR) and false alarm rate (FPR) can be calculated using the formula. Quantitatively assess the detection accuracy and ensure that the detection accuracy coefficient value is normalized to the range of 0-1. Optionally, the time taken to perform a single complete test on each detection technology on a standard hardware testing platform is measured by the formula. Efficiency analysis is performed, and the results are linearly mapped to the 0-1 interval to obtain standardized execution efficiency coefficients; Optionally, the equipment cost, energy consumption, and computing resources required for each detection technology can be quantitatively evaluated, with units of yuan / test, kWh, and GPU·s, respectively, using the formula... A comprehensive analysis was conducted, with the benchmark value being the maximum value among similar detection technologies, to ensure that the resource consumption coefficient value was within the effective range of 0-1. From the formula Analyze the technical score of the detection technology in the mapping table, where... Technical rating For the detection accuracy coefficient, For execution efficiency coefficient, This is the resource consumption coefficient. , and The weighting coefficients and ; Extract historical detection data sets from the fault database, calculate the historical comprehensive weight of each type of detection item, and sort them in descending order; Based on the number N of optional detection technologies corresponding to each detection item type, the sorted historical comprehensive weight range is divided into N consecutive weight intervals. At the same time, all optional detection technologies corresponding to each detection item type are sorted in descending order of technical score, and the sorting results are matched one-to-one with the weight intervals. Analyze the distribution of historical comprehensive weight values ​​for each test item type in the historical test dataset to obtain the mean and standard deviation. Construct a frequency interval with the lower limit being the mean minus 1.5 times the standard deviation and the upper limit being the mean plus the standard deviation. For real-time detection data, its comprehensive weight is compared with the weight range and frequency range of the corresponding detection item type. When the comprehensive weight is greater than the upper limit of the frequency range, the detection technology of the current and the two adjacent higher technical scores weight range is selected. When it is within the frequency range, the detection technology of the current weight range and the next lower technical score weight range is selected. When it is lower than the lower limit of the frequency range, no detection technology is selected. By integrating the real-time comparison results of all detection items, a preliminary set of detection technologies for motors is obtained.

[0015] See Figure 3 As shown, the game theory decision model for establishing a preliminary set of detection technologies performs conflict optimization on all detection technologies to generate the optimal detection technology sequence, specifically including: The initial set of detection technologies is modeled as a non-cooperative game, and the strategy space of each detection technology is defined as priority execution and non-priority execution. A conflict relationship network is constructed, and detection technology pairs with mutually exclusive resources are labeled. The independent benefits of each detection technology are calculated based on the detection accuracy coefficient, execution efficiency coefficient, and resource consumption coefficient, and a benefit matrix among the detection technologies is generated. Specifically, the payoff matrix is ​​a table that enumerates all possible strategy combinations, such as measuring only current, measuring only vibration, and measuring both current and vibration simultaneously. The payoff value of each strategy combination is calculated as follows: if a detection technology is marked as priority execution, its payoff value is the independent payoff of that detection technology minus the weighted sum of the independent payoffs of its conflicting technologies. The conflict weight is between 0.5 and 1.0. If a detection technology is not prioritized execution, its payoff value is 50% of the independent payoff of that detection technology. All strategy combinations and their corresponding payoff values ​​of detection technologies are filled into the payoff matrix. The matrix dimension is the number of strategy combinations × the number of detection technologies, and conflict markers are added to detection technology pairs that have mutually exclusive resources. The Nash equilibrium solution algorithm in non-cooperative game theory is used to calculate the optimal selection probability distribution of each detection technology, and then sort them by comprehensive weight to generate the optimal detection technology sequence. Specifically, the payoff matrix, policy space, and conflict markers are input into the Nash equilibrium solution algorithm, which outputs the optimal mixed policy probability distribution for each detection technique and generates the execution sequence of the detection techniques.

[0016] Specifically, the construction of a three-dimensional benefit evaluation system, which quantifies the benefit value of different combinations of detection technologies using detection accuracy coefficients, execution efficiency coefficients, and resource consumption coefficients, includes: With parameters Each detection technology is designated as a number, and the detection accuracy coefficient is used as the benefit of detection accuracy. ; Time consumption of each detection technology And screen out the time-consuming one among all detection technologies. From the formula Analyze the benefits of execution efficiency, among which, For execution efficiency coefficient, For the sake of execution efficiency and benefits; Obtain the number of detection technologies in the preliminary detection technology set that share the same resources as the current detection technology. From the formula Analyze resource consumption penalties, among which, This is the resource consumption coefficient. Penalty for resource consumption; Normalize the gains in detection accuracy, execution efficiency, and resource consumption penalties respectively, using the formula... Analyze the independent revenue value of current detection technologies, among which , is the independent revenue value. y and z are weighting coefficients.

[0017] Specifically, the detection technology is executed sequentially. If a detection technology confirms a fault, the subsequent detection is terminated; otherwise, the remaining detection technologies are executed until a motor detection result is generated. This includes: The detection technology is used to execute a sequence of tests on the motor, and the detection progress and intermediate results are monitored in real time. If a clear fault characteristic is detected, the subsequent detection process will be terminated immediately, and the current fault diagnosis result will be output. If no clear fault is detected, it is determined whether supplementary testing is needed based on the preset confidence threshold. When supplementary testing is required, the execution order of the remaining testing technologies is dynamically adjusted, prioritizing the testing technologies that are strongly correlated with the detected abnormal features; After all testing techniques have been performed, the test results are fused and analyzed to generate the final motor quality test results.

[0018] See Figure 4 As shown, this solution proposes a motor quality intelligent detection system based on data difference feedback, used to implement the aforementioned motor quality intelligent detection method based on data difference feedback, including: A multi-source data acquisition module is used to acquire real-time operating parameters of current, vibration and temperature of motors of the same model, and obtain the rated parameter benchmark of the model. The intelligent weight calculation module is used to calculate the historical fault weight of each detection item based on the fault database of the motor model, and at the same time generate parameter deviation weight based on the motor detection feature information. The historical fault weight and parameter deviation weight are weighted and fused to obtain the comprehensive weight of the detection item. The detection technology screening module is used to construct a mapping table between detection item types and detection technologies, calculate the historical comprehensive weight of each detection item based on historical detection data, divide the weight interval and establish its correspondence with the detection technology, and dynamically screen detection technologies by combining the frequency interval constructed with the mean and standard deviation of the historical comprehensive weight, and generate a preliminary set of detection technologies. The game optimization control module is used to establish a game theory decision model for a preliminary set of detection technologies, perform conflict optimization on all detection technologies, and generate an optimal sequence of detection technologies. The motor detection execution module is used to perform detection sequentially using a sequence of detection technologies. If a certain detection technology confirms a fault, the subsequent detection is terminated; otherwise, the remaining detection technologies are executed until the motor detection result is generated.

[0019] The intelligent weight calculation module specifically includes: The fault frequency analysis unit is used to statistically analyze the occurrence frequency of each fault type based on the fault database of this model of motor, and analyze the occurrence frequency of each fault type by combining the preset severity coefficient and the Poisson correction formula. The real-time deviation calculation unit is used to calculate the deviation of the real-time operating parameters of each detection item from the rated parameter benchmark in real time. The value is the ratio of the difference between the real-time operating parameters and the rated parameter benchmark to the rated parameter benchmark, and the parameter deviation weight is generated based on the deviation. The dynamic fusion decision unit is used to dynamically weight and fuse the historical fault weights and parameter deviation weights of the detection items using the working condition dynamic weight fusion algorithm to generate a comprehensive weight. The comprehensive weights of all detection items are then normalized so that the sum of the comprehensive weights remains constant at a fixed value.

[0020] The detection technology screening module specifically includes: The technology mapping table unit is used to establish a mapping relationship table between the detection item type and the optional detection technology. The detection item types include current, vibration and temperature, and a detection accuracy coefficient, execution efficiency coefficient and resource consumption coefficient are set for each detection technology. The dynamic filtering unit compares the comprehensive weight of real-time detection data with the weight range and frequency range of the corresponding detection item type. When the comprehensive weight is greater than the upper limit of the frequency range, the detection technology of the current and the two adjacent higher technical score weight ranges is selected. When it is within the frequency range, the detection technology of the current weight range and the next lower technical score weight range is selected. When it is lower than the lower limit of the frequency range, no detection technology is selected.

[0021] The game optimization control module specifically includes: The conflict network construction unit is used to model the initial set of detection technologies as a non-cooperative game, define the strategy space of each detection technology as priority execution and non-priority execution, construct a conflict relationship network, and label detection technology pairs with mutually exclusive resources; The revenue matrix generation unit is used to calculate the independent revenue of each detection technology based on the detection accuracy coefficient, execution efficiency coefficient, and resource consumption coefficient, and to generate the revenue matrix among the detection technologies. The Nash equilibrium solving unit is used to calculate the optimal selection probability distribution of each detection technology using the Nash equilibrium solving algorithm in non-cooperative games, and sort them by comprehensive weights to generate the optimal detection technology sequence.

[0022] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for intelligent detection of motor quality based on data difference feedback, characterized in that, The method includes: Collect real-time operating parameters of current, vibration and temperature of motors of the same model, and obtain the rated parameter benchmark of that model; By comparing real-time operating parameters with rated parameters, abnormal real-time operating parameters are marked as real-time detection data, each parameter item is used as a detection item, and motor detection feature information is generated. The historical fault weights of each test item are calculated based on the fault database of this type of motor. At the same time, parameter deviation weights are generated based on the motor test feature information. The historical fault weights and parameter deviation weights are weighted and merged to obtain the comprehensive weight of the test item. A mapping table between test item types and test technologies is constructed. Based on historical test data, the historical comprehensive weight of each test item is calculated, weight intervals are divided and their correspondence with test technologies is established. Combined with the frequency intervals constructed using the mean and standard deviation of the historical comprehensive weights, test technologies are dynamically screened to generate a preliminary set of test technologies. A game theory decision model is established for a preliminary set of detection technologies. Conflict optimization is performed on all detection technologies to generate the optimal sequence of detection technologies. The detection technology is executed sequentially. If a certain detection technology confirms a fault, the subsequent detection is terminated; otherwise, the remaining detection technologies are executed until the motor detection result is generated.

2. The method according to claim 1, characterized in that, The process involves calculating the historical fault weights of each detection item based on the fault database of this motor model, generating parameter deviation weights based on motor detection feature information, and then weighting and fusing the historical fault weights and parameter deviation weights to obtain the comprehensive weight of the detection item. Specifically, this includes: Based on the fault database of this model of motor, the frequency of occurrence of each fault type is statistically analyzed. Combined with the preset severity coefficient, the occurrence frequency of each fault type is analyzed by the Poisson correction formula. Calculate the probability of abnormality of the detection item under each fault type. The value is the ratio of the number of cases in which the detection item is abnormal to the total number of fault cases. The correlation degree of each detection item is calculated by accumulating the product of the frequency of occurrence of each fault type and the abnormal probability of its corresponding detection item. Then, the correlation degree of all detection items is normalized to generate the historical fault weight of the detection item. The deviation of the real-time operating parameters of each test item from the rated parameter benchmark is calculated in real time. The value is the ratio of the difference between the real-time operating parameters and the rated parameter benchmark to the rated parameter benchmark. The parameter deviation weight is generated based on this deviation. The working condition dynamic weight fusion algorithm is adopted to dynamically weight and fuse the historical fault weights and parameter deviation weights of the detection items to generate a comprehensive weight. The comprehensive weights of all detection items are normalized so that the sum of the comprehensive weights is constant to a fixed value. The Poisson correction formula is: In the formula, The frequency of occurrence of each fault type, The number of times a certain fault type occurs. The total number of failures across all failure types. Number the fault type. As a smoothing factor, To adjust the coefficient, This represents the severity level.

3. The method according to claim 2, characterized in that, The mapping table between detection item types and detection technologies is constructed. Based on historical detection data, the historical comprehensive weight of each detection item is calculated, weight intervals are divided, and their correspondence with detection technologies is established. Combined with frequency intervals constructed using the mean and standard deviation of historical comprehensive weights, detection technologies are dynamically screened to generate a preliminary set of detection technologies, specifically including: Establish a mapping table between detection item types and selectable detection technologies. The detection item types include current, vibration, and temperature. Set a detection accuracy coefficient, execution efficiency coefficient, and resource consumption coefficient for each detection technology. From the formula Analyze the technical score of the detection technology in the mapping table, where... Technical rating For the detection accuracy coefficient, For execution efficiency coefficient, This is the resource consumption coefficient. , and The weighting coefficients and ; Extract historical detection data sets from the fault database, calculate the historical comprehensive weight of each type of detection item, and sort them in descending order; Based on the number N of optional detection technologies corresponding to each detection item type, the sorted historical comprehensive weight range is divided into N consecutive weight intervals. At the same time, all optional detection technologies corresponding to each detection item type are sorted in descending order of technical score, and the sorting results are matched one-to-one with the weight intervals. Analyze the distribution of historical comprehensive weight values ​​for each test item type in the historical test dataset to obtain the mean and standard deviation. Construct a frequency interval with the lower limit being the mean minus 1.5 times the standard deviation and the upper limit being the mean plus the standard deviation. For real-time detection data, its comprehensive weight is compared with the weight range and frequency range of the corresponding detection item type. When the comprehensive weight is greater than the upper limit of the frequency range, the detection technology of the current and the two adjacent higher technical scores weight range is selected. When it is within the frequency range, the detection technology of the current weight range and the next lower technical score weight range is selected. When it is lower than the lower limit of the frequency range, no detection technology is selected. By integrating the real-time comparison results of all detection items, a preliminary set of detection technologies for motors is obtained.

4. The method according to claim 3, characterized in that, The game-theoretic decision-making model for establishing a preliminary set of detection technologies performs conflict optimization on all detection technologies to generate the optimal sequence of detection technologies, specifically including: The initial set of detection technologies is modeled as a non-cooperative game, and the strategy space of each detection technology is defined as priority execution and non-priority execution. A conflict relationship network is constructed, and detection technology pairs with mutually exclusive resources are labeled. The independent benefits of each detection technology are calculated based on the detection accuracy coefficient, execution efficiency coefficient, and resource consumption coefficient, and a benefit matrix among the detection technologies is generated. The Nash equilibrium algorithm in non-cooperative game theory is used to calculate the optimal selection probability distribution of each detection technology, and the optimal detection technology sequence is generated by combining the comprehensive weights.

5. The method according to claim 4, characterized in that, The aforementioned three-dimensional benefit evaluation system quantifies the benefit value of different combinations of detection technologies using detection accuracy coefficients, execution efficiency coefficients, and resource consumption coefficients. Specifically, it includes: With parameters Each detection technology is designated as a number, and the detection accuracy coefficient is used as the benefit of detection accuracy. Time consumption of each detection technology And screen out the time-consuming one among all detection technologies. From the formula Analyze the benefits of execution efficiency, among which, For execution efficiency coefficient, For the sake of execution efficiency and benefits; Obtain the number of detection technologies in the preliminary detection technology set that share the same resources as the current detection technology. From the formula Analyze resource consumption penalties, among which, This is the resource consumption coefficient. Penalty for resource consumption; Normalize the gains in detection accuracy, execution efficiency, and resource consumption penalties respectively, using the formula... Analyze the independent revenue value of current detection technologies, among which , is the independent revenue value. y and z are weighting coefficients.

6. The method according to claim 1, characterized in that, The detection technology is executed sequentially. If a detection technology confirms a fault, the subsequent detection is terminated; otherwise, the remaining detection technologies are executed until a motor detection result is generated. Specifically, this includes: The detection technology is used to execute a sequence of tests on the motor, and the detection progress and intermediate results are monitored in real time. If a clear fault characteristic is detected, the subsequent detection process will be terminated immediately, and the current fault diagnosis result will be output. If no clear fault is detected, it is determined whether supplementary testing is needed based on the preset confidence threshold. When supplementary testing is required, the execution order of the remaining testing technologies is dynamically adjusted, prioritizing the testing technologies that are strongly correlated with the detected abnormal features; After all testing techniques have been performed, the test results are fused and analyzed to generate the final motor quality test results.

7. A motor quality intelligent detection system based on data difference feedback, characterized in that, A method for implementing a motor quality intelligent detection method based on data difference feedback as described in any one of claims 1-6 includes: A multi-source data acquisition module is used to acquire real-time operating parameters of current, vibration and temperature of motors of the same model, and obtain the rated parameter benchmark of the model. The intelligent weight calculation module is used to calculate the historical fault weight of each detection item based on the fault database of the motor model, and at the same time generate parameter deviation weight based on the motor detection feature information. The historical fault weight and parameter deviation weight are weighted and fused to obtain the comprehensive weight of the detection item. The detection technology screening module is used to construct a mapping table between detection item types and detection technologies, calculate the historical comprehensive weight of each detection item based on historical detection data, divide the weight interval and establish its correspondence with the detection technology, and dynamically screen detection technologies by combining the frequency interval constructed with the mean and standard deviation of the historical comprehensive weight, and generate a preliminary set of detection technologies. The game optimization control module is used to establish a game theory decision model for a preliminary set of detection technologies, perform conflict optimization on all detection technologies, and generate an optimal sequence of detection technologies. The motor detection execution module is used to perform detection sequentially using a sequence of detection technologies. If a certain detection technology confirms a fault, the subsequent detection is terminated; otherwise, the remaining detection technologies are executed until the motor detection result is generated.

8. The intelligent motor quality detection system based on data difference feedback according to claim 7, characterized in that, The intelligent weight calculation module specifically includes: The fault frequency analysis unit is used to statistically analyze the occurrence frequency of each fault type based on the fault database of this model of motor, and analyze the occurrence frequency of each fault type by combining the preset severity coefficient and the Poisson correction formula. The real-time deviation calculation unit is used to calculate the deviation of the real-time operating parameters of each detection item from the rated parameter benchmark in real time. The value is the ratio of the difference between the real-time operating parameters and the rated parameter benchmark to the rated parameter benchmark, and the parameter deviation weight is generated based on the deviation. The dynamic fusion decision unit is used to dynamically weight and fuse the historical fault weights and parameter deviation weights of the detection items using the working condition dynamic weight fusion algorithm to generate a comprehensive weight. The comprehensive weights of all detection items are then normalized so that the sum of the comprehensive weights remains constant at a fixed value.

9. The intelligent motor quality detection system based on data difference feedback according to claim 7, characterized in that, The detection technology screening module specifically includes: The technology mapping table unit is used to establish a mapping relationship table between the detection item type and the optional detection technology. The detection item types include current, vibration and temperature, and a detection accuracy coefficient, execution efficiency coefficient and resource consumption coefficient are set for each detection technology. The dynamic filtering unit compares the comprehensive weight of real-time detection data with the weight range and frequency range of the corresponding detection item type. When the comprehensive weight is greater than the upper limit of the frequency range, the detection technology of the current and the two adjacent higher technical score weight ranges is selected. When it is within the frequency range, the detection technology of the current weight range and the next lower technical score weight range is selected. When it is lower than the lower limit of the frequency range, no detection technology is selected.

10. The intelligent motor quality detection system based on data difference feedback according to claim 7, characterized in that, The game optimization control module specifically includes: The conflict network construction unit is used to model the initial set of detection technologies as a non-cooperative game, define the strategy space of each detection technology as priority execution and non-priority execution, construct a conflict relationship network, and label detection technology pairs with mutually exclusive resources; The revenue matrix generation unit is used to calculate the independent revenue of each detection technology based on the detection accuracy coefficient, execution efficiency coefficient, and resource consumption coefficient, and to generate the revenue matrix among the detection technologies. The Nash equilibrium solving unit is used to calculate the optimal selection probability distribution of each detection technology using the Nash equilibrium solving algorithm in non-cooperative games, and sort them by comprehensive weights to generate the optimal detection technology sequence.

Images