Management method for full-automatic assembly test production line of photovoltaic inverter

Abstract

The invention discloses a management method for a full-automatic assembly test production line of a photovoltaic inverter, and relates to the technical field of data processing. The method comprises the following steps: acquiring a predetermined test link of the photovoltaic inverter and a monitorable assembly parameter of an automatic assembly production line; carrying out product fault causal correlation analysis on the monitorable assembly parameters and a predetermined test link, and constructing a first fault mapping of the assembly parameters-test link; the method comprises the following steps: acquiring real-time assembly data of a photovoltaic inverter to be tested, performing fault probability prediction of a predetermined test link, and generating a test link sequence; and testing the photovoltaic inverter to be tested, dynamically optimizing the residual test link sequence, and completing the subsequent test according to the optimized test link sequence. The technical problems that in the prior art, testing is conducted according to a fixed testing process, so that flexibility is poor, and testing efficiency is affected are solved, and the technical effect of improving the testing efficiency is achieved through real-time fault prediction and dynamic optimization of the testing link sequence.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a management method for a fully automatic assembly and test production line of a photovoltaic inverter. Background Art

[0002] As a key device in a photovoltaic power generation system, a photovoltaic inverter is responsible for converting the direct current generated by photovoltaic modules into alternating current compatible with the power grid. The stability and reliability of its performance are directly related to the power generation efficiency and operation safety of the entire photovoltaic system. In order to ensure the performance and reliability of a photovoltaic inverter, strict assembly and testing must be carried out. However, traditional assembly and testing methods usually follow a fixed testing process, with poor flexibility, unable to detect potential problems in a timely manner, resulting in low testing efficiency and unable to perform targeted optimization. Therefore, how to improve the intelligent level of the automatic assembly and test production line of a photovoltaic inverter, optimize the testing process, and improve the testing efficiency has become a technical problem to be solved urgently. Summary of the Invention

[0003] This application provides a management method for a fully automatic assembly and test production line of a photovoltaic inverter, which solves the technical problem in the prior art that the flexibility is poor and the testing efficiency is affected when testing according to a fixed testing process.

[0004] This application provides a management method for a fully automatic assembly and test production line of a photovoltaic inverter, and the method includes:

[0005] Obtain the predetermined test links of the photovoltaic inverter and the monitorable assembly parameters of the automatic assembly production line; perform a causal correlation analysis of product failures on the monitorable assembly parameters and the predetermined test links to construct a first failure mapping of assembly parameters - test links; collect and obtain the real-time assembly data of the photovoltaic inverter to be tested according to the monitorable assembly parameters, predict the failure probability of the predetermined test links according to the real-time assembly data and the first failure mapping, and generate a test link sequence sorted from large to small according to the predicted failure probability; perform tests on the photovoltaic inverter to be tested according to the test link sequence, dynamically optimize the remaining test link sequence according to the second failure mapping between the first test result and the predetermined test links, and complete the subsequent tests according to the optimized test link sequence.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] First, obtain the predetermined test links of the photovoltaic inverter and the monitorable assembly parameters of the automatic assembly line. Next, conduct a causal correlation analysis of product failures between the monitorable assembly parameters and the predetermined test links to construct a first failure mapping of assembly parameters - test links. Then, collect the real-time assembly data of the photovoltaic inverter to be tested according to the monitorable assembly parameters, predict the failure probability of the predetermined test links based on the real-time assembly data and the first failure mapping, and generate a test link sequence sorted from largest to smallest according to the predicted failure probability. Finally, test the photovoltaic inverter to be tested according to the test link sequence, dynamically optimize the remaining test link sequence according to the second failure mapping between the first test result and the predetermined test links, and complete the subsequent tests according to the optimized test link sequence. This solves the technical problem in the prior art that the fixed test process leads to poor flexibility and affects the test efficiency. By predicting failures in real time and dynamically optimizing the test link sequence, the technical effect of improving the test efficiency is achieved. Description of the Drawings

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings required for the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0009] Figure 1 Schematic flowchart of a management method for a full-automatic assembly and test line of a photovoltaic inverter provided by an embodiment of the present application;

[0010] Figure 2 Schematic flowchart of dynamic optimization in a management method for a full-automatic assembly and test line of a photovoltaic inverter provided by an embodiment of the present application. Detailed Embodiments

[0011] By providing a management method for a full-automatic assembly and test line of a photovoltaic inverter, the present application solves the technical problem in the prior art that the fixed test process leads to poor flexibility and affects the test efficiency.

[0012] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0013] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0014] Examples, such as Figure 1 As shown, an embodiment of the present application provides a management method for a fully automatic assembly and test production line of a photovoltaic inverter. Among them, the method includes:

[0015] Obtain the predetermined test links of the photovoltaic inverter and the monitorable assembly parameters of the automatic assembly line.

[0016] From the historical test data of the photovoltaic inverter, obtain the predetermined test links. The predetermined test links refer to the specific operations or test items that should be performed in each link during the test of the photovoltaic inverter. These links are usually set according to the design and functional requirements of the photovoltaic inverter to ensure that it can meet basic functions such as power conversion and output. The monitorable assembly parameters refer to the data or parameters that can be monitored through sensors or devices during the automatic assembly process. For example, the installation accuracy, fastening force, welding quality, etc. of each component during the assembly process.

[0017] Furthermore, obtaining the predetermined test links of the photovoltaic inverter and the monitorable assembly parameters of the automatic assembly line includes:

[0018] Obtain the predetermined test links of the photovoltaic inverter, where the test links at least include DC input test, MPPT performance test, AC output test, electrical safety test, temperature test, overload protection test, high temperature and high humidity test, and load change test; obtain the monitorable assembly parameters of the automatic assembly line of the photovoltaic inverter, where the monitorable assembly parameters at least include welding position deviation, welding size deviation, mounting position deviation, fixing component pressure deviation, fixing component torque deviation, and component position deviation.

[0019] In the automatic assembly and test production line of photovoltaic inverters, to ensure that each inverter meets the design and quality standards, multiple test procedures are required to verify its performance. Obtain the predetermined test procedures from the historical test data of photovoltaic inverters, specifically including: DC input test (verify whether the DC power input received by the photovoltaic inverter meets the predetermined operating range to ensure the stable operation of the inverter), MPPT performance test (test whether the inverter can track and adjust to the optimal power output point in different lighting conditions), AC output test (check whether the output voltage and current after the inverter converts DC power into AC power are stable and meet the standards to ensure the quality of the output electrical energy), electrical safety test (conduct insulation resistance test, withstand voltage test, and leakage test to ensure the safety of the inverter during high-voltage operation), temperature test (monitor the temperature change of the inverter during long-term operation to ensure that the inverter can operate within the specified temperature range and prevent failures caused by overheating), overload protection test (test whether the inverter can perform normal protection under overload or abnormal operating conditions to avoid damaging the equipment), high-temperature and high-humidity test (age the inverter in a high-temperature and high-humidity environment to simulate the performance stability of the inverter during long-term operation), and load change test (simulate the load fluctuations in actual use, test the response speed and stability of the inverter, and ensure that it can still output smoothly under load changes).

[0020] During the automatic assembly of photovoltaic inverters, many key assembly parameters need to be monitored in real time to ensure that each component is correctly installed according to the standards. The monitorable assembly parameters of the automatic assembly production line refer to the data that can be collected in real time through sensors, cameras, or other monitoring devices during the automated production process; the monitorable assembly parameters include: welding position deviation (check the accuracy of the welding points to ensure that the welding positions are consistent with the design requirements and prevent poor electrical contact or failures caused by welding position deviation), welding size deviation (monitor the size of the welding area to ensure that the width and depth of the welding meet the standards and avoid poor welding quality and affecting product performance due to inconsistent sizes), component mounting position deviation (check the mounting positions of each component on the PCB board to ensure the installation accuracy and alignment of the components to avoid poor electrical contact or component loosening), fixed component pressure deviation (monitor the pressure applied when fixing the components of the photovoltaic inverter to ensure that the pressing force meets the specifications and prevent component loosening or insecure installation), fixed component torque deviation (ensure that the torque of bolts or other connectors used to fix the components meets the requirements and avoid loose connections or component damage caused by excessive or insufficient torque), and component position deviation (check whether the positions of all components, such as capacitors, resistors, ICs, etc., on the circuit board are accurate to ensure that all components match the circuit design and avoid affecting the performance of the circuit due to position deviation).

[0021] Perform a causal correlation analysis of product failures for the monitorable assembly parameters and the predetermined test procedures, and construct a first failure mapping of assembly parameters - test procedures.

[0022] By performing a correlation analysis on the data of the monitorable assembly parameters and the predetermined test procedures of the photovoltaic inverter, key factors that may cause failures in each assembly step can be found; by analyzing the causal relationship between the assembly parameters and the test procedures, a first failure mapping of assembly parameters - test procedures can be constructed, that is, to determine the impact of each assembly parameter on the test procedure results, and predict possible failure types and their occurrence probabilities.

[0023] Furthermore, performing a causal correlation analysis of product failures for the monitorable assembly parameters and the predetermined test procedures, and constructing a first failure mapping of assembly parameters - test procedures, includes:

[0024] Randomly select a first test procedure in the predetermined test procedures; based on the historical assembly test record data of similar photovoltaic inverters, perform a causal correlation analysis of test failures on multiple assembly parameters in the monitorable assembly parameters and the first test procedure respectively, and output multiple first parameter correlation degrees; select the assembly parameters corresponding to the first parameter correlation degrees greater than the predetermined parameter correlation degree as the first associated assembly parameters, and obtain a first set of associated assembly parameters; successively obtain multiple sets of associated assembly parameters for multiple test procedures, and construct a first failure mapping of assembly parameters - test procedures.

[0025] Specifically, in a predetermined test session, randomly select one test session as the first test session, which can be any session such as DC input test, MPPT performance test, electrical safety test, etc.; then, based on the historical assembly test record data of similar photovoltaic inverters, conduct a fault causal association analysis on multiple parameters in the monitorable assembly parameters (such as welding position deviation, mounting position deviation, fixed component pressure deviation, etc.) and the selected first test session. Statistical methods such as Pearson correlation coefficient can be used to quantify the correlation between the assembly parameters and the test session. Calculate for each pair of assembly parameters and test session to obtain multiple first parameter correlation degrees. These correlation degree values reflect the role played by each assembly parameter in the process of predicting test session faults. The higher the correlation degree, the more significant the impact of the assembly parameter on the test session result; subsequently, select those assembly parameters with first parameter correlation degrees greater than a predetermined threshold as the first associated assembly parameters, and these assembly parameters will be incorporated into the first associated assembly parameter set; conduct the same fault causal association analysis on multiple test sessions in sequence to obtain multiple associated assembly parameter sets for each test session, thereby obtaining a set of assembly parameters associated with each test session; finally, organize these analysis results into a complete first fault mapping of assembly parameters - test sessions. The first fault mapping shows each test session and its related assembly parameters, and indicates the influence intensity of these assembly parameters on the test session.

[0026] Acquire the real-time assembly data of the photovoltaic inverter to be tested according to the monitorable assembly parameters, predict the fault probability of the predetermined test session based on the real-time assembly data and the first fault mapping, and generate a test session sequence sorted from largest to smallest according to the predicted fault probability.

[0027] By performing real-time monitoring on the assembly process of the photovoltaic inverter to be tested, collect the real-time assembly data of the photovoltaic inverter, including welding position deviation, component position deviation, component pressure deviation, etc. The real-time assembly data reflects the actual situation of the product during the assembly process.

[0028] After obtaining the real-time assembly data, it is necessary to combine these data with the previously constructed first fault mapping for fault probability prediction. Machine learning models (such as logistic regression, decision tree, random forest, etc.) or statistical methods (such as conditional probability calculation) can be used to predict the fault probability; for each test session, a corresponding fault probability value can be calculated, indicating the possibility of the session failing under the current assembly data; according to the fault probabilities of each test session obtained from the above calculations, sort them from largest to smallest to generate an optimized test session sequence. The sorted test session sequence ensures that the test sessions with larger fault probabilities are given priority, thereby maximizing the test efficiency.

[0029] Furthermore, the fault probability prediction for a predetermined test session based on the real-time assembly data and the first fault mapping includes:

[0030] Performing mapping combination on the real-time assembly data according to the first fault mapping to obtain multiple associated assembly data sets for multiple test sessions; pre-training a test fault predictor, where the test fault predictor includes multiple test fault prediction plugins corresponding to multiple test sessions; and using the multiple test fault prediction plugins to respectively perform fault probability prediction on the multiple associated assembly data sets and output multiple test fault probabilities for multiple test sessions.

[0031] Specifically, performing mapping combination on the real-time assembly data according to the first fault mapping, combining the real-time collected assembly data with the causal relationships in the mapping to obtain multiple associated assembly data sets for multiple test sessions. These data sets specifically reflect the relationships between each test session and the assembly parameters. For example, the associated assembly data set for electrical safety testing may include welding position deviation, electrical contact quality, etc., while the associated data set for MPPT performance testing may involve component pressure deviation, welding precision, etc. Then, based on these associated data sets, a pre-trained test fault predictor will be used. The predictor contains multiple test fault prediction plugins, each plugin corresponding to a different test session. Each plugin has learned the influence rules of each assembly parameter on the faults of a specific test session based on historical data and the relationships in the first fault mapping. Through this training, the plugin can understand the fault patterns of each test session under different assembly parameter conditions. For example, the electrical safety test fault prediction plugin will learn how to predict the fault risk of this session based on factors such as welding position deviation and electrical contact quality, while the MPPT performance test plugin will predict the fault probability based on component pressure and welding precision. When the plugins of the test fault predictor are fully trained, they can receive new associated assembly data sets in real time and process them, respectively outputting the fault probabilities of each test session. These plugins generate corresponding fault probability values according to the assembly data and the learned rules. For example, if the associated assembly data set for DC input testing shows a large welding position deviation, then the fault prediction plugin will predict a relatively high fault probability for this session, perhaps 0.4; while if the associated assembly data set for MPPT performance testing shows a small component pressure deviation, the predicted fault probability may be 0.1. Each plugin performs fault probability prediction on the corresponding test session according to its specifically learned model, thus providing a predicted fault probability for each test session.

[0032] Furthermore, the pre-training of the test fault predictor includes:

[0033] Randomly select the first test session and obtain the first associated assembly parameter set of the first test session; retrieve the historical assembly test record data of similar photovoltaic inverters based on the first associated assembly parameter set and the first test session, collect multiple sample assembly parameter sets, and count the proportion of test failures under different sample assembly parameter sets, which is set as the sample test failure probability, to obtain a sample test failure probability set; use the multiple sample assembly parameter sets and the sample test failure probability set as training data, and equally divide them into P training sets, where P is an integer greater than 20; use the sample assembly parameter set as the input and the sample test failure probability as the supervision, and use the P training sets to train a random decision forest until convergence, obtaining P first test failure prediction units, and combine them to obtain the first test failure prediction plug-in; sequentially train multiple test failure prediction plug-ins for multiple test sessions, and map and combine them to construct the test failure predictor.

[0034] Specifically, randomly select a first test session, and according to the definition of this test session, obtain the first associated assembly parameter set related to it; based on the first associated assembly parameter set and the first test session, retrieve from the historical assembly test record data of similar photovoltaic inverters to obtain multiple sample assembly parameter sets, and count the proportion of test failures in each test session under these sample assembly parameter sets. This proportion is the sample test failure probability, and then obtain a set of sample test failure probability sets for subsequent training. Next, use the obtained multiple sample assembly parameter sets and the corresponding sample test failure probability sets as training data; to improve the generalization ability of the model and avoid overfitting, equally divide these data into P training sets, where P is an integer greater than 20, to ensure the diversity and effectiveness of the training sets. Then, use the sample assembly parameter set as the input and the sample test failure probability as the supervision, and use these training data to train a random decision forest. The random decision forest is an ensemble learning method that makes predictions through the voting of multiple decision trees. The training process will continue until the model converges, that is, the parameters and weights of each decision tree reach the optimal configuration; each training set will train a first test failure prediction unit, and finally obtain P first test failure prediction units. These units are combined to form the first test failure prediction plug-in. After obtaining the first test failure prediction plug-in, next, sequentially train multiple test sessions in the same way to obtain multiple test failure prediction plug-ins corresponding to each test session. Each plug-in is trained through the sample assembly parameter set and the test failure probability set of the corresponding test session to accurately predict the failure probability of each test session. Finally, by mapping and combining all these plug-ins, a complete test failure predictor is constructed. The test failure predictor can input the associated assembly data in real time according to different test sessions and quickly and accurately predict the failure probability of the corresponding test session.

[0035] Furthermore, using the multiple test fault prediction plugins, perform fault probability prediction on the multiple associated assembly data sets respectively, including:

[0036] Randomly select the first associated assembly data set of the first test link and the first test fault prediction plugin; calculate the ratios of the multiple first associated assembly data in the first associated assembly data set to the historical maximum assembly error of the corresponding assembly parameters respectively, obtain multiple parameter deviation calculations, and calculate the mean value to obtain the first comprehensive deviation coefficient; multiply the first comprehensive deviation coefficient by P and round it to obtain the first prediction unit selection quantity, randomly select according to the first prediction unit selection quantity among the P first test fault prediction units of the first test fault prediction plugin, perform test fault prediction on the first associated assembly data set, and calculate the mean value of the prediction results, and output the first test fault probability of the first test link.

[0037] Specifically, randomly select the first associated assembly data set of the first test link and load the corresponding first test fault prediction plugin. Each test link will have a dedicated fault prediction plugin, which has obtained a set of prediction units through training and can be used to perform fault prediction on specific data. Next, calculate the ratios of the multiple first associated assembly data in the first associated assembly data set to the corresponding assembly parameters' historical maximum assembly error; specifically, for each first associated assembly data, it is necessary to find its corresponding historical maximum assembly error (for example, the maximum deviation value in historical data such as welding position deviation, mounting position deviation, etc.), and then calculate the ratio between the deviation value of the current assembly data and the historical maximum error value. This ratio can help evaluate the error size of the current assembly parameters relative to the historical data; through statistics of these ratios, obtain multiple parameter deviation calculation results; calculate the mean value of all calculated parameter deviation calculation results to obtain a first comprehensive deviation coefficient, and the first comprehensive deviation coefficient reflects the deviation degree between the current test data set and the historical data set. Based on the calculated first comprehensive deviation coefficient, multiply it by P and round it to obtain the first prediction unit selection quantity, where P is a previously defined integer value, usually greater than 20, and is used to control the number of randomly selected units. The first prediction unit selection quantity determines how many prediction units are selected from the plugin to participate in the fault prediction. According to the obtained first prediction unit selection quantity, randomly select a certain number of units from the P first test fault prediction units of the first test fault prediction plugin to perform fault prediction; each prediction unit will perform a test on the first associated assembly data set and output the fault probability of this test link. All selected prediction units will output their respective fault probabilities, and after calculating the mean value, obtain the first test fault probability of the first test link. This fault probability value reflects the possibility of a fault occurring in this test link under the current assembly parameters.

[0038] Test the photovoltaic inverter to be tested according to the described test session sequence, dynamically optimize the remaining test session sequence based on the first test result and the second fault mapping between the predetermined test sessions, and complete the subsequent tests according to the optimized test session sequence.

[0039] According to the test session sequence, preliminarily test the photovoltaic inverter to be tested in the order of decreasing fault probability. The priority order of the test sessions ensures that the sessions with higher fault probabilities are tested first, so as to detect possible faults as early as possible, optimize the test process, and reduce the test time.

[0040] After obtaining the first test result, dynamically optimize the order of the remaining test sessions by comparing the current test result with the second fault mapping between the predetermined test sessions. The second fault mapping is obtained through historical data and fault analysis in the early stage, and it provides the mutual influence and fault probability prediction relationship between each test session. By analyzing the first test result, it is possible to evaluate the influence of the current test session on the subsequent sessions, and which sessions' fault probabilities will change due to the previous test result. Specifically, when conducting the first test, assume that a certain test session (such as electrical safety test) yields a relatively high fault probability or confirms a problem. Then, according to the second fault mapping, the order of the subsequent sessions can be adjusted. For example, if the electrical safety test discovers a problem, it may be necessary to prioritize the temperature test or the MPPT performance test to further verify the cause of the fault; while if the test results of some sessions show that no major faults are involved, these test sessions can be postponed. Through dynamic optimization, it is possible to flexibly adjust the order of the subsequent test sessions according to the real-time test results and the predetermined fault mapping relationship, thereby improving the efficiency and accuracy of fault detection. Finally, complete all the subsequent test sessions according to the optimized test session sequence, and each session is executed in turn according to the optimized order, ensuring the efficiency and accuracy of the test process and being able to specifically detect and solve potential fault problems.

[0041] Furthermore, as Figure 2 shown, test the photovoltaic inverter to be tested according to the described test session sequence, dynamically optimize the remaining test session sequence based on the first test result and the second fault mapping between the predetermined test sessions, and complete the subsequent tests according to the optimized test session sequence, including:

[0042] Select the first test session in the test session sequence to test the photovoltaic inverter to be tested, and output the first test result; establish a second fault mapping between predetermined test sessions; if the first test result fails, stop subsequent tests and mark the test as unqualified; if the first test result passes, obtain the first test data, and match multiple fault compensation coefficients for multiple remaining test sessions according to the first test data and the second fault mapping; perform mapping compensation on the multiple test fault probabilities of the multiple remaining test sessions according to the multiple fault compensation coefficients, output multiple updated test fault probabilities, sort the multiple remaining test sessions in descending order of the updated test fault probabilities, generate a first optimized test session sequence, and perform subsequent tests according to the first optimized test session sequence.

[0043] Select the first test session from the test session sequence to conduct a preliminary test on the photovoltaic inverter to be tested and output the first test result. At this time, the order of the test sessions has been preliminarily arranged according to the results of the fault probability prediction to ensure that the session most likely to have a fault is tested first. Next, establish a second fault mapping. The second fault mapping defines the causal relationships between different test sessions in certain fault scenarios. For example, if the electrical safety test passes, it may mean that other sessions (such as temperature test, MPPT test) are less affected; if the electrical safety test fails, the subsequent test sessions may need to adjust their priorities based on this result.

[0044] If the first test result fails, that is, the test result indicates a fault in this link, then the subsequent tests will stop immediately, and the unqualified test results will be marked. If the first test result passes, then the first test data will be obtained, including all relevant data extracted from this test link, such as voltage, current, temperature, etc.; according to the first test data, it is matched with the second fault mapping to obtain multiple fault compensation coefficients for multiple remaining test links. These compensation coefficients represent how the fault probabilities of other test links should be adjusted according to the results of the first test link. By applying these compensation coefficients to the subsequent test links, the test fault probabilities of each test link can be mapped and compensated. For example, if some latent faults are revealed in the first test link, the fault probabilities of the subsequent test links may increase or decrease accordingly, reflecting the actual fault risks. Sort the multiple remaining test links in descending order according to the updated test fault probabilities after compensation to generate the first optimized test link sequence. This optimized sequence is generated based on the latest fault prediction data and test results, ensuring that the test priority order can maximize the test efficiency and accurately detect potential problems. Finally, according to the generated first optimized test link sequence, the subsequent tests will continue. All the remaining test links will be executed in the optimized order, ensuring that the test process is not only efficient and accurate, but also can dynamically adjust the priority according to the fault probability of each link, maximize the accuracy of fault diagnosis, reduce unnecessary tests, and ensure that the quality and reliability of the PV inverter meet the standards.

[0045] Furthermore, establish a second fault mapping between the predetermined test links, including:

[0046] According to the historical test record data of the same type of PV inverter, conduct a correlation analysis on the first test link and other test links. If there is no correlation, there is no fault compensation relationship, and the fault compensation coefficient is 0; if there is a correlation, according to the historical test record data, conduct a test data - fault enhancement probability analysis on the first test link and the associated test links respectively, and output multiple sets of associated test data intervals and multiple sets of fault compensation coefficients; construct a second fault mapping based on the multiple associated test links, multiple sets of associated test data intervals, and multiple sets of fault compensation coefficients.

[0047] Specifically, based on the historical test record data of similar photovoltaic inverters, analyze the correlation between the first test link and other test links. The historical test record data usually includes the results of different test links under different test conditions, and these results can reveal the possible interdependent relationships between each test link; determine whether there is a correlation between the first test link and other test links; if there is no correlation, it means that the result of this test link will not affect the test results of other links. Therefore, in subsequent fault prediction, the fault compensation coefficient will be set to 0. In other words, when there is no direct relationship between test links, the fault probability of subsequent links will not be affected by the result of the previous link. If there is a correlation, that is, when the result of the first test link affects the results of other links, a more in-depth analysis will be carried out; based on the historical test record data, conduct test data - fault enhancement probability analysis, which aims to reveal the influence degree of the result of a certain test link (such as passing or failing the test) on the fault probability of subsequent test links. Specifically, analyze how different test data values (such as different data intervals of voltage, current, temperature, etc.) of the first test link in the historical test data affect the probability of faults occurring in subsequent test links, and these influences will be obtained by calculating the fault enhancement probability, that is, the change amount of the fault probability of each test link under the influence of the first test result. Based on this analysis, output multiple sets of associated test data intervals, and these data interval sets represent the possible fault ranges under different test data values (such as temperature intervals, voltage intervals, etc.). At the same time, multiple sets of fault compensation coefficients will be output, and these coefficients represent the influence degree of the first test result on the fault probability of each associated test link. For example, if the first test link shows a certain fault type, then the fault probability of subsequent test links may increase due to this fault mode, and the compensation coefficient will reflect this increment. After completing these analyses, construct a second fault mapping based on multiple associated test links, multiple sets of associated test data intervals, and multiple sets of fault compensation coefficients. The core goal of the second fault mapping is to transmit the influence of the result of the first test link to subsequent test links by quantifying the correlation. The second fault mapping can provide more accurate fault prediction for subsequent tests, enabling the fault probability to be dynamically adjusted based on the actual result of the previous test link, thereby optimizing the test process and discovering potential problems in advance. Through the established second fault mapping, the system can effectively compensate and adjust the fault probability of each test link according to the first test result during the subsequent test process, thereby ensuring that the test process is more flexible and accurate, and can effectively improve the accuracy of fault diagnosis, reducing ineffective tests and resource waste.

[0048] Furthermore, complete the subsequent tests according to the optimized test link sequence, including:

[0049] Establish a third fault mapping between the completed test links and the remaining test links; according to the completed test data and the third fault mapping, perform test fault probability compensation on the remaining test links, and iteratively update the optimized test link sequence until the subsequent tests are completed.

[0050] First, establish a third fault mapping between the completed test links and the remaining test links. After preliminary testing, the completed test links provide important reference information for subsequent tests. The third fault mapping, based on the completed test data and known historical data, describes the causal relationship between the completed test links and the subsequent test links. The third fault mapping can reveal the impact of the results of the already tested links (such as electrical safety tests, temperature tests, etc.) on the remaining test links (such as MPPT performance tests, load change tests, etc.). For example, if a fault is found in the photovoltaic inverter under a certain specific test condition in the completed test link, the third fault mapping can help predict how this fault will affect the subsequent test links and provide a compensation mechanism for the subsequent test fault probability. Next, according to the completed test data and the third fault mapping, perform test fault probability compensation on the remaining test links. The completed test links provide information about the actual performance of the photovoltaic inverter under specific conditions. By analyzing the test data, the impact of this link on the subsequent test links can be obtained. For example, if the first test link (such as an electrical safety test) fails, the fault probability of the remaining test links (such as temperature tests, MPPT performance tests, etc.) can be compensated according to the third fault mapping, and the fault risk prediction of these links can be adjusted. According to these compensated fault probability data, the optimized test link sequence will be iteratively updated. After each compensation, the priorities and test orders of the subsequent test links will be re-ordered to ensure that the links with higher fault probabilities are executed first. As each test link is completed, the optimized test link sequence will be continuously iteratively adjusted to timely detect potential faults and maximize test efficiency. This process will continue until all the remaining test links are tested. Through this mechanism of dynamic optimization and compensation, the test order can be adjusted in real time and the fault probability can be compensated, thus ensuring the efficiency and accuracy of the entire test process, avoiding unnecessary repeated tests, reducing test time and resource consumption. Finally, all the remaining test links will be tested according to the dynamically optimized and iteratively updated test order to ensure that the performance of the photovoltaic inverter can be comprehensively and accurately evaluated in the end, timely discover and handle possible fault problems, and guarantee the quality and reliability of the product.

[0051] In summary, the embodiments of the present application have at least the following technical effects:

[0052] First, obtain the predetermined test links of the photovoltaic inverter and the monitorable assembly parameters of the automatic assembly line. Next, conduct a causal correlation analysis of product failures between the monitorable assembly parameters and the predetermined test links to construct a first failure mapping of assembly parameters - test links. Then, collect the real-time assembly data of the photovoltaic inverter to be tested according to the monitorable assembly parameters, predict the failure probability of the predetermined test links based on the real-time assembly data and the first failure mapping, and generate a test link sequence sorted from largest to smallest according to the predicted failure probability. Finally, test the photovoltaic inverter to be tested according to the test link sequence, dynamically optimize the remaining test link sequence according to the second failure mapping between the first test result and the predetermined test links, and complete the subsequent tests according to the optimized test link sequence. This solves the technical problem in the prior art that the flexibility is poor and the test efficiency is affected when testing according to a fixed test process. By predicting failures in real time and dynamically optimizing the order of test links, the technical effect of improving the test efficiency is achieved.

[0053] It should be noted that the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above description of specific embodiments of this specification has been made. The processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0054] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0055] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A management method for a fully automatic assembly and test production line of a photovoltaic inverter, characterized in that, The method includes: Obtaining the predetermined test links of a photovoltaic inverter and the monitorable assembly parameters of an automatic assembly line; Conducting product fault causal correlation analysis on the monitorable assembly parameters and the predetermined test links, and constructing a first fault mapping of assembly parameters - test links; Collecting and obtaining the real-time assembly data of the photovoltaic inverter to be tested according to the monitorable assembly parameters, predicting the fault probability of the predetermined test links based on the real-time assembly data and the first fault mapping, and generating a test link sequence sorted from largest to smallest according to the predicted fault probability; Testing the photovoltaic inverter to be tested according to the test link sequence, dynamically optimizing the remaining test link sequence according to the second fault mapping between the first test result and the predetermined test links, and completing the subsequent tests according to the optimized test link sequence.

2. The management method for a fully automatic assembly and testing production line of a photovoltaic inverter according to claim 1, wherein Obtaining the predetermined test links of a photovoltaic inverter and the monitorable assembly parameters of an automatic assembly line includes: Obtaining the predetermined test links of a photovoltaic inverter, where the test links at least include DC input test, MPPT performance test, AC output test, electrical safety test, temperature test, overload protection test, high temperature and high humidity test, and load change test; Obtaining the monitorable assembly parameters of the automatic assembly line of the photovoltaic inverter, where the monitorable assembly parameters at least include welding position deviation, welding size deviation, mounting position deviation, fixing component pressure deviation, fixing component torque deviation, and component position deviation.

3. A management method for a fully automatic assembly and testing production line of a photovoltaic inverter, characterized in that, Conducting product fault causal correlation analysis on the monitorable assembly parameters and the predetermined test links, and constructing a first fault mapping of assembly parameters - test links includes: Randomly selecting a first test link in the predetermined test links; Conducting test fault causal correlation analysis on multiple assembly parameters in the monitorable assembly parameters and the first test link respectively according to the historical assembly test record data of similar photovoltaic inverters, and outputting multiple first parameter correlation degrees; Selecting the assembly parameters corresponding to the first parameter correlation degrees greater than the predetermined parameter correlation degree as the first associated assembly parameters, and obtaining a first associated assembly parameter set; Successively obtaining multiple associated assembly parameter sets for multiple test links, and constructing a first fault mapping of assembly parameters - test links.

4. The management method for a fully automatic assembly and test production line of a photovoltaic inverter according to claim 3, wherein, Predicting the fault probability of the predetermined test links based on the real-time assembly data and the first fault mapping includes: Performing mapping combination on the real-time assembly data according to the first fault mapping to obtain multiple associated assembly data sets for multiple test links; Pre-training a test fault predictor, where the test fault predictor includes multiple test fault prediction plugins corresponding to multiple test links; Using the multiple test fault prediction plugins to respectively predict the fault probability of the multiple associated assembly data sets, and outputting multiple test fault probabilities for multiple test links.

5. A management method for a fully automatic assembly and testing production line of a photovoltaic inverter, characterized in that, Pre-training a test fault predictor includes: Randomly selecting a first test link and obtaining the first associated assembly parameter set of the first test link; Retrieve the historical assembly test record data of similar photovoltaic inverters based on the first associated assembly parameter set and the first test session, collect multiple sample assembly parameter sets, and count the proportion of test failures under different sample assembly parameter sets, which is set as the sample test failure probability, to obtain the sample test failure probability set; Use the multiple sample assembly parameter sets and the sample test failure probability set as training data, and divide them equally into P training sets, where P is an integer greater than 20; Use the sample assembly parameter set as the input and the sample test failure probability as the supervision, and use the P training sets to train the random decision forest until convergence, obtaining P first test failure prediction units, and combining them to obtain the first test failure prediction plug-in; Train multiple test failure prediction plug-ins for multiple test sessions in sequence, and map and combine them to construct the test failure predictor.

6. The management method for a fully automatic assembly and test production line of a photovoltaic inverter according to claim 5, wherein, Use the multiple test failure prediction plug-ins to respectively predict the failure probability of the multiple associated assembly data sets, including: Randomly select the first associated assembly data set of the first test session and the first test failure prediction plug-in; Calculate the ratio of the historical maximum assembly error between multiple first associated assembly data and the corresponding assembly parameters in the first associated assembly data set respectively, obtain multiple parameter deviation calculations, and calculate the mean value to obtain the first comprehensive deviation coefficient; Multiply the first comprehensive deviation coefficient by P and round it to obtain the first prediction unit selection quantity, randomly select from the P first test failure prediction units of the first test failure prediction plug-in according to the first prediction unit selection quantity, predict the test failure of the first associated assembly data set, and calculate the mean value of the prediction results, and output the first test failure probability of the first test session.

7. A management method for a fully automatic assembly and testing production line of a photovoltaic inverter, characterized in that, Test the photovoltaic inverter to be tested according to the test session sequence, dynamically optimize the remaining test session sequence according to the second fault mapping between the first test result and the predetermined test session, and complete the subsequent test according to the optimized test session sequence, including: Select the first test session in the test session sequence, test the photovoltaic inverter to be tested, and output the first test result; Establish the second fault mapping between the predetermined test sessions; If the first test result fails, stop the subsequent test and mark it as unqualified; If the first test result passes, obtain the first test data, and match multiple fault compensation coefficients of multiple remaining test sessions according to the first test data and the second fault mapping; Perform mapping compensation on the multiple test failure probabilities of the multiple remaining test sessions according to the multiple fault compensation coefficients, output multiple updated test failure probabilities, sort the multiple remaining test sessions in descending order of the updated test failure probabilities, generate the first optimized test session sequence, and perform the subsequent test according to the first optimized test session sequence.

8. A management method for a fully automatic assembly and test production line of a photovoltaic inverter, characterized in that, Establish the second fault mapping between the predetermined test sessions, including: According to the historical test record data of similar photovoltaic inverters, perform correlation analysis on the first test session and other test sessions. If there is no correlation, there is no fault compensation relationship, and the fault compensation coefficient is 0; If there is a correlation, according to the historical test record data, perform test data - fault enhancement probability analysis on the first test link and the associated test link respectively, and output multiple associated test data interval sets and multiple fault compensation coefficient sets; Construct a second fault mapping based on multiple associated test links, multiple associated test data interval sets, and multiple fault compensation coefficient sets.

9. The management method for a fully automatic assembly and testing production line of a photovoltaic inverter according to claim 8, wherein Complete the subsequent tests according to the optimized test link sequence, including: Establish a third fault mapping between the completed test links and the remaining test links; Perform test fault probability compensation on the remaining test links according to the completed test data and the third fault mapping, and iteratively update the optimized test link sequence until the subsequent tests are completed.

Images