Photovoltaic transformer and control method thereof

Abstract

The invention relates to the technical field of photovoltaic power distribution, in particular to a photovoltaic transformer and a control method thereof. The knowledge graph is constructed through analysis of historical data, and the system state of the photovoltaic transformer can be accurately judged in combination with monitoring and analysis of real-time acoustic characteristic data. Different processing strategies are adopted for different system states, unnecessary intervention on normal operation is avoided, frequent abnormalities can be processed in time to guarantee stable operation of the system, meanwhile, the accidental abnormalities are continuously monitored to master the development situation of the accidental abnormalities, and the normal operation of the system is guaranteed. Therefore, the reliability of operation and the intelligent level of management of the photovoltaic transformer are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power distribution, and in particular to a photovoltaic transformer and a control method thereof. Background Art

[0002] In the actual operation of photovoltaic systems, photovoltaic transformers may face a variety of operating conditions. Traditional control methods often have difficulty responding to these complex situations accurately and promptly. On the one hand, the operating status of photovoltaic transformers is affected by many factors, such as ambient temperature, humidity, load changes, etc. These factors may cause the transformer to have varying degrees of performance fluctuations, and even cause abnormal system operation events. On the other hand, existing monitoring and control methods lack in-depth mining and effective use of historical data.

[0003] From the historical working data, photovoltaic transformers will generate various acoustic characteristic data during operation, and these acoustic characteristic data are potentially related to abnormal system operation events. Different types of system operation abnormal events, whether frequent or occasional, may have different corresponding acoustic characteristic data. However, there is currently a lack of an effective method to systematically analyze the relationship between these acoustic characteristic data and abnormal events, and to build an intelligent control strategy based on this.

[0004] Most of the existing photovoltaic transformer control methods are based on fixed thresholds or simple logical judgments, which are not flexible and accurate enough when facing complex and changeable actual operating conditions. For example, when the transformer is in an occasional state of abnormal operation, if overly aggressive control measures are adopted, it may cause unnecessary waste of resources or interfere with normal operation; and when it is in a frequent abnormal state, if the control cannot be carried out in a timely and accurate manner, it may cause the problem to deteriorate further, affect the normal operation of the entire photovoltaic power generation system, and even shorten the service life of the photovoltaic transformer.

[0005] Therefore, it is necessary to develop a control method based on historical acoustic feature data mining that can accurately judge the operating status of photovoltaic transformers and perform intelligent regulation, so as to improve the operating efficiency, reliability and safety of photovoltaic transformers, and thus improve the performance of the entire photovoltaic power generation system. Summary of the invention

[0006] The present invention overcomes the deficiencies of the prior art and provides a photovoltaic transformer and a control method thereof.

[0007] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is: The present invention discloses a control method for a photovoltaic transformer, comprising the following steps: Perform feature analysis on various system operation abnormal events that may be caused when the target photovoltaic transformer generates various historical acoustic feature data during historical operation, and obtain frequent system operation abnormal events and occasional system operation abnormal events; Construct a binary tree knowledge graph based on the historical acoustic feature data corresponding to frequent system operation abnormal events and occasional system operation abnormal events; Acquire real-time acoustic feature data generated by the target photovoltaic transformer in the real-time working process through an acoustic sensor at a preset time node, analyze the system state of the target photovoltaic transformer at the current time node according to the real-time acoustic feature data and in combination with a binary tree knowledge graph, and generate an analysis result; If the system status of the target photovoltaic transformer is in normal operation at the current time node, the target photovoltaic transformer will not be regulated; If the system state of the target photovoltaic transformer at the current preset time node is in a sporadic state of abnormal operation, the target photovoltaic transformer will not be regulated and the operation state of the target photovoltaic transformer will be continuously monitored at the next time node; If the system state of the target photovoltaic transformer at the current preset time node is in a state where abnormal operation events frequently occur, the target photovoltaic transformer is regulated.

[0008] Furthermore, the characteristics of various system operation abnormal events caused by the target photovoltaic transformer generating various historical acoustic characteristic data in the historical working process are analyzed to obtain frequent system operation abnormal events and occasional system operation abnormal events, specifically: Acquire various historical acoustic characteristic data generated by the target photovoltaic transformer in the historical working process, and count various system operation abnormal events caused by the target photovoltaic transformer generating various historical acoustic characteristic data in the historical working process according to the operation log of the target photovoltaic transformer; Perform statistical analysis on various system operation abnormal events that would be triggered when the target photovoltaic transformer generates various historical acoustic characteristic data during historical operation, and obtain the number of corresponding system operation abnormal events that would be triggered when various historical acoustic characteristic data are generated; The number of corresponding system operation abnormal events that will be triggered when various historical acoustic feature data are generated is compared with a preset value to obtain the frequency value of the corresponding system operation abnormal events that will be triggered when various historical acoustic feature data are generated; Comparing the frequency value of the corresponding system operation abnormal event that will be triggered when various historical acoustic feature data are generated with a preset frequency value threshold; System operation abnormality events whose frequency values ​​are greater than the preset frequency value threshold are marked as frequent system operation abnormality events; system operation abnormality events whose frequency values ​​are not greater than the preset frequency value threshold are marked as occasional system operation abnormality events.

[0009] Furthermore, a binary tree knowledge graph is constructed based on the historical acoustic feature data corresponding to frequent system operation abnormal events and occasional system operation abnormal events, specifically: Acquire historical acoustic feature data corresponding to each frequent system operation abnormality event, and acquire historical acoustic feature data corresponding to each occasional system operation abnormality event; Constructing a binary tree, and dividing the binary tree into two main branches, namely a first main branch and a second main branch, according to the frequent system operation abnormal events and the occasional system operation abnormal events; Iteratively splitting the first main trunk according to each frequently occurring system operation abnormality event, so as to split a plurality of frequently occurring system operation abnormality event leaf nodes on the first main trunk; wherein each frequently occurring system operation abnormality event leaf node stores characteristic information of the corresponding frequently occurring system operation abnormality event; Iteratively splitting the second main trunk according to each incidental system operation abnormality event, so as to split a plurality of incidental system operation abnormality event leaf nodes on the second main trunk; wherein each incidental system operation abnormality event leaf node stores characteristic information of the corresponding incidental system operation abnormality event; The historical acoustic feature data corresponding to each frequent system operation abnormality event is stored in the corresponding frequent system operation abnormality event leaf node; the historical acoustic feature data corresponding to each occasional system operation abnormality event is stored in the corresponding occasional system operation abnormality event leaf node; to form a binary tree knowledge graph.

[0010] Furthermore, at a preset time node, the real-time acoustic feature data generated by the target photovoltaic transformer in the real-time working process is obtained through an acoustic sensor, and the system state of the target photovoltaic transformer at the current time node is analyzed according to the real-time acoustic feature data and combined with the binary tree knowledge graph to generate an analysis result, which is specifically: Acquiring real-time acoustic feature data generated by the target photovoltaic transformer in the real-time working process through an acoustic sensor at a preset time node; and extracting historical acoustic feature data in each leaf node from the binary tree knowledge graph; Introducing the Kendall rank correlation coefficient algorithm, and calculating the Kendall rank correlation coefficient value between the real-time acoustic feature data and the historical acoustic feature data in each leaf node based on the Kendall rank correlation coefficient algorithm; The calculated Kendall rank correlation coefficient values ​​are compared with the preset coefficient values. If the values ​​of the Kendall rank correlation coefficients are not greater than the preset coefficient values, it means that the system state of the target photovoltaic transformer at the current time node is in a normal operating state; If there is at least one Kendall rank correlation coefficient value greater than the preset coefficient value, the leaf node corresponding to the Kendall rank correlation coefficient value greater than the preset coefficient value is marked, and the feature type of the marked leaf node is determined; wherein the feature type includes an occasional system operation abnormal event leaf node and a frequent system operation abnormal event leaf node; If the characteristic type of the marked leaf node is a leaf node of frequent system operation abnormal events, it means that the system state of the target photovoltaic transformer at the current preset time node is in a state of frequent operation abnormal events; If the characteristic type of the marked leaf node is an occasional system operation abnormality event leaf node, it means that the system state of the target photovoltaic transformer at the current preset time node is in an occasional operation abnormality event state.

[0011] Furthermore, if the system state of the target photovoltaic transformer at the current preset time node is in a state where abnormal operation events frequently occur, the target photovoltaic transformer is regulated and processed, specifically: If the system state of the target photovoltaic transformer at the current preset time node is in a state where abnormal operation events frequently occur, then the real-time working parameters of the target photovoltaic transformer at the current preset time node are obtained; Genetic algorithm is introduced, the real-time working parameters of the target photovoltaic transformer are used as decision variables, and the decision variables are encoded into chromosome representation to construct the initial population. Each individual in the population represents a possible regulation scheme. Determine the key indicators related to the performance of the target photovoltaic transformer, including voltage deviation, power factor, temperature and efficiency, assign weights to each key indicator according to its importance to the normal operation of the system, multiply each indicator value by the corresponding weight and sum them up, and construct a fitness function that can comprehensively evaluate the advantages and disadvantages of the control scheme; Execute the selection operation, adopt the preset selection strategy according to the fitness value, and select excellent individuals from the current population to enter the next generation population; Perform crossover operations, exchange genes of selected individuals according to the set crossover probability, simulate the biological gene recombination process, and generate new regulation scheme individuals to increase the diversity of the population; Perform mutation operations to randomly change the genes of individuals whose mutation probability is less than the preset probability value to form new individuals; The above selection, crossover and mutation operations are continuously iterated, and the fitness of the individual is re-evaluated in each round of iteration until the maximum number of iterations is reached, and the control scheme corresponding to the best individual is screened out from the population to obtain the optimal working parameter control scheme of the target photovoltaic transformer; wherein the best individual is the individual with the highest fitness; The real-time operating parameters of the target photovoltaic transformer are regulated according to the optimal operating parameter regulation scheme.

[0012] The control method of the photovoltaic transformer further comprises the following steps: After the real-time working parameters of the target photovoltaic transformer are regulated by the optimal working parameter regulation scheme, the real-time acoustic feature data of the target photovoltaic transformer is obtained again, and the system state of the target photovoltaic transformer is analyzed according to the real-time acoustic feature data and in combination with the binary tree knowledge graph; If the target photovoltaic transformer is still in a state of frequent abnormal operation after being regulated, the target photovoltaic transformer is controlled to stop working, and an early warning message is generated and sent to a preset terminal; If the target photovoltaic transformer is in a state of occasional abnormal operation after being regulated, the operation state of the target photovoltaic transformer will continue to be monitored at the next time node; If the target photovoltaic transformer is in a normal operating state after being regulated, the instruction for continuous monitoring of the state of the target photovoltaic transformer is released.

[0013] The present invention also discloses a photovoltaic transformer, which is applied to any one of the photovoltaic transformer control methods described above. The photovoltaic transformer includes a base and a transformer body, and the transformer body is installed and fixed by the base; the transformer body is provided with heat dissipation fins; and the top of the transformer body is provided with a sleeve and a controller.

[0014] The present invention also discloses a control system for a photovoltaic transformer, the control system comprising a memory and a processor, the memory storing a control method program for the photovoltaic transformer, and when the control method program for the photovoltaic transformer is executed by the processor, any one of the control method steps for the photovoltaic transformer is implemented.

[0015] The present invention solves the technical defects existing in the background technology, and has the following beneficial effects: by constructing a knowledge graph through the analysis of historical data, combined with the monitoring and analysis of real-time acoustic feature data, the system status of the photovoltaic transformer can be accurately judged. Different processing strategies are adopted for different system states (normal, occasional abnormalities, frequent abnormalities), which not only avoids unnecessary intervention in normal operation, but also can handle frequent abnormalities in time to ensure stable operation of the system, and at the same time continuously monitors occasional abnormalities to grasp their development trend, thereby improving the reliability of photovoltaic transformer operation and the intelligent level of management. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, drawings of other embodiments can be obtained based on these drawings without paying creative work.

[0017] Figure 1 This is a flow chart of the overall control method of the photovoltaic transformer; Figure 2 This is a flow chart of part of the control method of the photovoltaic transformer; Figure 3 This is a schematic diagram of the structure of the photovoltaic transformer. DETAILED DESCRIPTION

[0018] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0019] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0020] The present invention discloses a control method for a photovoltaic transformer. Figure 1 As shown, the following steps are included: S102, performing feature analysis on various system operation abnormal events that may be caused when the target photovoltaic transformer generates various historical acoustic feature data during historical operation, and obtaining frequent system operation abnormal events and occasional system operation abnormal events; S104, constructing a binary tree knowledge graph according to the historical acoustic feature data corresponding to the frequent system operation abnormality events and the occasional system operation abnormality events; S106. Acquire real-time acoustic feature data generated by the target photovoltaic transformer in the real-time working process through an acoustic sensor at a preset time node, analyze the system state of the target photovoltaic transformer at the current time node according to the real-time acoustic feature data and in combination with the binary tree knowledge graph, and generate an analysis result; S108. If the system state of the target photovoltaic transformer is in a normal operating state at the current time node, no control processing is performed on the target photovoltaic transformer; S110, if the system state of the target photovoltaic transformer at the current preset time node is in a sporadic state of abnormal operation event, the target photovoltaic transformer is not regulated and controlled, and the operation state of the target photovoltaic transformer is continuously monitored at the next time node; S112: If the system state of the target photovoltaic transformer at the current preset time node is in a state where abnormal operation events frequently occur, the target photovoltaic transformer is regulated.

[0021] Specifically, when the system is in a normal operating state, no regulation is performed because the transformer is operating normally and no intervention is required. When the system is in an occasional state of abnormal operation, no regulation is performed but continuous monitoring is performed. This is because occasional abnormalities may be temporary or have little impact, and continuous monitoring can further observe their development trends. When the system is in a state of frequent abnormal operation, regulation is performed because frequent abnormalities indicate that there are serious problems in the system and timely intervention is required to avoid more serious failures or performance degradation. By analyzing historical data to construct a knowledge graph, combined with the monitoring and analysis of real-time acoustic feature data, the system state of the photovoltaic transformer can be accurately judged. Different processing strategies are adopted for different system states (normal, occasional abnormalities, frequent abnormalities), which not only avoids unnecessary intervention in normal operation, but also can handle frequent abnormalities in time to ensure stable operation of the system. At the same time, occasional abnormalities are continuously monitored to grasp their development trends, thereby improving the reliability of photovoltaic transformer operation and the level of intelligent management.

[0022] Furthermore, the characteristic analysis of various system operation abnormal events caused by various historical acoustic characteristic data generated by the target photovoltaic transformer in the historical working process is carried out to obtain frequent system operation abnormal events and occasional system operation abnormal events, such as Figure 2 As shown, specifically: S202, obtaining various historical acoustic characteristic data generated by the target photovoltaic transformer in the historical working process, and counting various system operation abnormal events caused by the target photovoltaic transformer generating various historical acoustic characteristic data in the historical working process according to the operation log of the target photovoltaic transformer; S204, performing statistical analysis on various system operation abnormal events that would be triggered when the target photovoltaic transformer generates various historical acoustic characteristic data during historical operation, and obtaining the number of corresponding system operation abnormal events that would be triggered when the various historical acoustic characteristic data are generated; S206, performing ratio processing on the number of corresponding system operation abnormal events that will be triggered when various historical acoustic feature data are generated and a preset value, so as to obtain the frequency value of the corresponding system operation abnormal events that will be triggered when various historical acoustic feature data are generated; S208, comparing the frequency value of the corresponding system operation abnormal event that will be triggered when various historical acoustic feature data are generated with a preset frequency value threshold; S210, marking the system operation abnormality events whose frequency values ​​are greater than the preset frequency value threshold as frequent system operation abnormality events; marking the system operation abnormality events whose frequency values ​​are not greater than the preset frequency value threshold as occasional system operation abnormality events.

[0023] Specifically, first, all the historical acoustic feature data generated by the target photovoltaic transformer in the past work are obtained, which are the basis for subsequent analysis. At the same time, according to the operation log of the transformer, various system operation abnormal events accompanied by these historical acoustic feature data are counted. This step establishes the correlation between acoustic feature data and system operation abnormal events, so that possible clues to problems can be mined from historical data. The system operation abnormal events caused by each historical acoustic feature data are statistically analyzed to obtain the number of occurrences. Then this number is compared with the preset value to obtain the corresponding frequency value. This frequency value reflects the frequency of system operation abnormal events caused by specific historical acoustic feature data in the historical working process. Finally, the calculated frequency value is compared with the preset frequency value threshold. If the frequency value is greater than the preset frequency value threshold, the corresponding system operation abnormal event is marked as a frequent system operation abnormal event; if the frequency value is not greater than the preset frequency value threshold, it is marked as an occasional system operation abnormal event. This calibration method helps to distinguish abnormal events of different severity or different probability of occurrence, and provides an important basis for the subsequent construction of binary tree knowledge graph and the judgment of real-time operation status.

[0024] Through the above series of analysis and processing of historical acoustic feature data and system operation abnormal events, frequent system operation abnormal events and occasional system operation abnormal events can be accurately distinguished. This helps to build a more accurate binary tree knowledge graph, so that when analyzing the current system status of the target photovoltaic transformer based on real-time acoustic feature data, the type and severity of the abnormality can be more targeted.

[0025] Furthermore, a binary tree knowledge graph is constructed based on the historical acoustic feature data corresponding to frequent system operation abnormal events and occasional system operation abnormal events, specifically: Acquire historical acoustic feature data corresponding to each frequent system operation abnormality event, and acquire historical acoustic feature data corresponding to each occasional system operation abnormality event; Constructing a binary tree, and dividing the binary tree into two main branches, namely a first main branch and a second main branch, according to the frequent system operation abnormal events and the occasional system operation abnormal events; Iteratively splitting the first main trunk according to each frequently occurring system operation abnormality event, so as to split a plurality of frequently occurring system operation abnormality event leaf nodes on the first main trunk; wherein each frequently occurring system operation abnormality event leaf node stores characteristic information of the corresponding frequently occurring system operation abnormality event; Iteratively splitting the second main trunk according to each incidental system operation abnormality event, so as to split a plurality of incidental system operation abnormality event leaf nodes on the second main trunk; wherein each incidental system operation abnormality event leaf node stores characteristic information of the corresponding incidental system operation abnormality event; The historical acoustic feature data corresponding to each frequent system operation abnormality event is stored in the corresponding frequent system operation abnormality event leaf node; the historical acoustic feature data corresponding to each occasional system operation abnormality event is stored in the corresponding occasional system operation abnormality event leaf node; to form a binary tree knowledge graph.

[0026] Specifically, a binary tree structure is constructed, and the binary tree is divided into two main branches, namely the first main branch and the second main branch, according to the frequent system operation abnormal events and the occasional system operation abnormal events. This division method lays a framework foundation for the subsequent processing of different types of abnormal events. For the first main branch, iterative splitting is performed according to each frequent system operation abnormal event, so that a number of frequent system operation abnormal event leaf nodes are generated on the main branch. Each leaf node stores the characteristic information of the corresponding frequent system operation abnormal event, which enables different frequent abnormal events to have a clear position and identification in the binary tree structure. Similarly, for the second main branch, iterative splitting is performed according to each occasional system operation abnormal event to obtain a number of occasional system operation abnormal event leaf nodes, and each leaf node also stores the characteristic information of the corresponding occasional system operation abnormal event. Finally, the historical acoustic feature data that triggers each frequent system operation abnormal event is stored in the corresponding frequent system operation abnormal event leaf node, and the same is true for the occasional system operation abnormal event. In this way, a binary tree knowledge graph is completely constructed, and different types of abnormal events and their related acoustic feature data are organized in an orderly structure.

[0027] By constructing such a binary tree knowledge graph, frequent system operation abnormal events and occasional system operation abnormal events and their corresponding historical acoustic feature data can be efficiently and orderly organized. When monitoring the real-time operating status of the target photovoltaic transformer, the real-time acoustic feature data can be quickly searched and matched in the binary tree knowledge graph to accurately determine whether the current operating status is a frequent abnormality, an occasional abnormality, or a normal state. This structure helps to improve analysis efficiency and accuracy, reduce misjudgment, and thus more accurately take corresponding measures according to different operating states, such as timely regulation and processing of frequent abnormal states, continuous monitoring of occasional abnormal states, etc., to ensure the stable operation of photovoltaic transformers.

[0028] Furthermore, at a preset time node, the real-time acoustic feature data generated by the target photovoltaic transformer in the real-time working process is obtained through an acoustic sensor, and the system state of the target photovoltaic transformer at the current time node is analyzed according to the real-time acoustic feature data and combined with the binary tree knowledge graph to generate an analysis result, which is specifically: Acquiring real-time acoustic feature data generated by the target photovoltaic transformer in the real-time working process through an acoustic sensor at a preset time node; and extracting historical acoustic feature data in each leaf node from the binary tree knowledge graph; Introducing the Kendall rank correlation coefficient algorithm, and calculating the Kendall rank correlation coefficient value between the real-time acoustic feature data and the historical acoustic feature data in each leaf node based on the Kendall rank correlation coefficient algorithm; The calculated Kendall rank correlation coefficient values ​​are compared with the preset coefficient values. If the values ​​of the Kendall rank correlation coefficients are not greater than the preset coefficient values, it means that the system state of the target photovoltaic transformer at the current time node is in a normal operating state; If there is at least one Kendall rank correlation coefficient value greater than the preset coefficient value, the leaf node corresponding to the Kendall rank correlation coefficient value greater than the preset coefficient value is marked, and the feature type of the marked leaf node is determined; wherein the feature type includes an occasional system operation abnormal event leaf node and a frequent system operation abnormal event leaf node; If the characteristic type of the marked leaf node is a leaf node of frequent system operation abnormal events, it means that the system state of the target photovoltaic transformer at the current preset time node is in a state of frequent operation abnormal events; If the characteristic type of the marked leaf node is an occasional system operation abnormality event leaf node, it means that the system state of the target photovoltaic transformer at the current preset time node is in an occasional operation abnormality event state.

[0029] Specifically, at a preset time point, the acoustic sensor is used to collect the real-time acoustic feature data generated by the target photovoltaic transformer when it is working in real time. At the same time, the historical acoustic feature data stored in each leaf node is extracted from the pre-built binary tree knowledge graph. This step provides a data basis for subsequent correlation analysis. The real-time data reflects the current state, and the historical data is the basis for judgment.

[0030] The Kendall rank correlation coefficient algorithm is introduced to calculate the Kendall rank correlation coefficient value between the real-time acoustic feature data and the historical acoustic feature data in each leaf node. The Kendall rank correlation coefficient is a statistic used to measure the correlation between two variables. Here, it is used to measure the similarity between the real-time acoustic features and the historical acoustic features. The calculated Kendall rank correlation coefficient value is compared with the preset coefficient value. If all Kendall rank correlation coefficient values ​​are not greater than the preset coefficient value, it means that the real-time acoustic feature data has no strong correlation with any historical acoustic feature data that may indicate an abnormality, so it can be determined that the system state of the target photovoltaic transformer at the current time node is in normal operation.

[0031] If there is at least one Kendall rank correlation coefficient value greater than the preset coefficient value, the corresponding leaf node is marked. Then the feature type of the marked leaf node is determined. If it is a frequent system operation abnormality event leaf node, it means that the current real-time acoustic feature data has a strong correlation with the acoustic features related to the frequent abnormalities in history, and the system state of the target photovoltaic transformer at the current preset time node is in a frequent operation abnormality event state. If it is an occasional system operation abnormality event leaf node, it means that the system state of the target photovoltaic transformer at the current preset time node is in an occasional operation abnormality event state.

[0032] Through this analysis method based on the Kendall rank correlation coefficient algorithm combined with the binary tree knowledge graph, the system status of the target photovoltaic transformer at the current time node can be accurately analyzed. It can effectively distinguish between the normal state, the occasional state of abnormal operation events, and the frequent state of abnormal operation events. This precise state judgment provides a basis for taking different response measures for photovoltaic transformers in the future. For example, no intervention is required under normal conditions, continuous monitoring can be carried out under occasional abnormal conditions, and timely regulation and control can be carried out under frequent abnormal conditions, thereby improving the accuracy and effectiveness of photovoltaic transformer operation management, ensuring its stable operation and improving the reliability of the entire photovoltaic power generation system.

[0033] Furthermore, if the system state of the target photovoltaic transformer at the current preset time node is in a state where abnormal operation events frequently occur, the target photovoltaic transformer is regulated and processed, specifically: If the system state of the target photovoltaic transformer at the current preset time node is in a state where abnormal operation events frequently occur, then the real-time working parameters of the target photovoltaic transformer at the current preset time node are obtained; Genetic algorithm is introduced, the real-time working parameters of the target photovoltaic transformer are used as decision variables, and the decision variables are encoded into chromosome representation to construct the initial population. Each individual in the population represents a possible regulation scheme. Determine the key indicators related to the performance of the target photovoltaic transformer, including voltage deviation, power factor, temperature and efficiency, assign weights to each key indicator according to its importance to the normal operation of the system, multiply each indicator value by the corresponding weight and sum them up, and construct a fitness function that can comprehensively evaluate the advantages and disadvantages of the control scheme; Execute the selection operation, adopt the preset selection strategy according to the fitness value, select excellent individuals from the current population to enter the next generation population, and ensure that the excellent regulation scheme genes can be passed on; Perform crossover operations, exchange genes of selected individuals according to the set crossover probability, simulate the biological gene recombination process, and generate new regulation scheme individuals to increase the diversity of the population; Perform mutation operations to randomly change the genes of individuals whose mutation probability is less than the preset probability value to form new individuals; The above selection, crossover and mutation operations are continuously iterated, and the fitness of the individual is re-evaluated in each round of iteration until the maximum number of iterations is reached, and the control scheme corresponding to the best individual is screened out from the population to obtain the optimal working parameter control scheme of the target photovoltaic transformer; wherein the best individual is the individual with the highest fitness; The real-time operating parameters of the target photovoltaic transformer are regulated according to the optimal operating parameter regulation scheme.

[0034] Specifically, when it is determined that the target photovoltaic transformer is in a state of frequent abnormal operation events at the current preset time node, its real-time working parameters (such as input voltage, output voltage, working temperature, working humidity, etc.) at this moment are first obtained. These working parameters contain various information about the current operating state of the transformer and are the basis for the subsequent search of control schemes. Genetic algorithms are introduced to take real-time working parameters as decision variables. The chromosome representation is formed by encoding, and the initial population is constructed. Each individual in the population represents a possible control scheme. This step is the starting point for optimization search using genetic algorithms, which simulates the initial state of the biological population. Each individual (control scheme) has potential optimization possibilities. Determine the key indicators related to the performance of the target photovoltaic transformer, such as voltage deviation, power factor, temperature, and efficiency. According to the importance of these indicators to the normal operation of the system, weights are assigned, and the values ​​of each indicator are multiplied by the corresponding weights and then summed to construct a fitness function. This function can comprehensively evaluate the pros and cons of the control scheme. The higher the fitness, the more the control scheme can meet the requirements of the normal operation of the transformer. Perform the selection operation, and select excellent individuals from the current population according to the fitness value using the preset selection strategy to enter the next generation population. This operation is similar to natural selection in biological evolution, ensuring that excellent control scheme genes (expressed as excellent characteristics of individuals in genetic algorithms) can be inherited and the population can develop in a better direction. According to the set crossover probability, the selected individuals are exchanged for genes to simulate the biological gene recombination process. In this way, new control scheme individuals are generated, which increases the diversity of the population and helps to discover more potential excellent control schemes. Genes in individuals with mutation probabilities less than the preset probability value are randomly changed to form new individuals. The mutation operation can introduce new gene combinations to prevent the population from falling into local optimal solutions, further explore the entire solution space, and increase the possibility of finding the global optimal control scheme. The above selection, crossover and mutation operations are continuously iterated, and the individual fitness is re-evaluated in each round of iteration until the maximum number of iterations is reached. Finally, the control scheme corresponding to the optimal individual (the individual with the highest fitness) is selected from the population to obtain the optimal working parameter control scheme of the target photovoltaic transformer. According to the obtained optimal working parameter control scheme, the real-time working parameters of the target photovoltaic transformer are regulated and processed, thereby improving the operating state of the transformer and restoring it from the frequently abnormal state to the normal or near-normal operating state.

[0035] Through this genetic algorithm-based control processing method, when the target photovoltaic transformer is in a state of frequent abnormal operation events, the optimal working parameter control scheme can be constructed and searched according to its current real-time working parameters. By constructing a fitness function to accurately evaluate the pros and cons of the control scheme, and then through iterative optimization of selection, crossover and mutation operations, the local optimal solution is effectively avoided and the possible control scheme space is fully explored. The optimal control scheme finally obtained can accurately control the real-time working parameters of the target photovoltaic transformer, thereby improving its operating state, improving the performance of the photovoltaic transformer, reducing the adverse effects caused by frequent abnormalities, and ensuring the stable operation of the photovoltaic power generation system.

[0036] The control method of the photovoltaic transformer further comprises the following steps: After the real-time working parameters of the target photovoltaic transformer are regulated by the optimal working parameter regulation scheme, the real-time acoustic feature data of the target photovoltaic transformer is obtained again, and the system state of the target photovoltaic transformer is analyzed according to the real-time acoustic feature data and in combination with the binary tree knowledge graph; If the target photovoltaic transformer is still in a state of frequent abnormal operation after being regulated, the target photovoltaic transformer is controlled to stop working, and an early warning message is generated and sent to a preset terminal; If the target photovoltaic transformer is in a state of occasional abnormal operation after being regulated, the operation state of the target photovoltaic transformer will continue to be monitored at the next time node; If the target photovoltaic transformer is in a normal operating state after being regulated, the instruction for continuous monitoring of the state of the target photovoltaic transformer is released.

[0037] Specifically, after the real-time working parameters of the target photovoltaic transformer are regulated and processed through the optimal working parameter control scheme, its real-time acoustic feature data is obtained again, and its system status is analyzed in combination with the binary tree knowledge graph. This step is to verify the effectiveness of the control measures. Through the same analysis method as before (based on acoustic feature data and binary tree knowledge graph), the operating status of the transformer after regulation can be accurately judged. If the target photovoltaic transformer is still in a state of frequent abnormal operation after the control process, it indicates that the previous control scheme has failed to effectively solve the problem. At this time, the target photovoltaic transformer is controlled to stop working to avoid possible more serious faults or safety hazards. At the same time, early warning information is generated and sent to the preset terminal, so that relevant personnel can be informed of the situation in time and take further measures, such as on-site inspection and maintenance.

[0038] If the abnormal operation event occurs occasionally after the regulation, it means that the regulation measures have a certain effect, but the transformer has not yet fully recovered to normal. In this case, continue to monitor the operating status of the target PV transformer at the next time node to detect possible problems in a timely manner.

[0039] When the target photovoltaic transformer is in normal operation after regulation, the continuous monitoring instruction is released. This is because the transformer has resumed normal operation and no additional continuous monitoring is required, thus saving monitoring resources and reducing unnecessary operations.

[0040] Through the re-analysis of the state after regulation and the corresponding treatment measures, the effect of the regulation plan can be fully evaluated. For the situation that is still in frequent abnormality after regulation, the transformer is stopped in time and an early warning is issued to ensure the safety of the system; for the situation that turns into occasional abnormality, continued monitoring can prevent the problem from worsening; for the situation that returns to normal, the monitoring command is released to improve the utilization efficiency of monitoring resources. Overall, the management level of the operating status of photovoltaic transformers is improved, ensuring the stable and safe operation of photovoltaic transformers and the entire photovoltaic power generation system.

[0041] In addition, the control method further comprises the following steps: During the operation of the target photovoltaic transformer, the frequency response characteristics of the target photovoltaic transformer are obtained at a number of preset time nodes; Based on discrete wavelet transform, multi-layer decomposition operation is performed on the frequency response characteristic data of each time node to obtain wavelet coefficients at different scales; For each time node, the energy value of the wavelet coefficients at each scale is calculated according to the preset energy calculation rules; and a data structure is constructed to store the energy values ​​corresponding to different scales at each time node; With the scale as the horizontal coordinate and the energy values ​​corresponding to different time nodes as the vertical coordinate, an energy distribution curve diagram of the target photovoltaic transformer within a preset time period is generated; Acquire a preset working task of a target photovoltaic transformer within a preset time period, and acquire an energy distribution threshold range of the target photovoltaic transformer within the preset time period according to the preset working task; Determine an upper energy distribution threshold and a lower energy distribution threshold according to the energy distribution threshold range of the target photovoltaic transformer within a preset time period; Mapping the upper energy distribution threshold and the lower energy distribution threshold in the energy distribution curve diagram to define a normal working area and an abnormal working area in the energy distribution curve diagram; In the energy distribution curve diagram, calculating the curve length of the energy distribution curve in the abnormal working area, and calculating the curve length of the energy distribution curve in the normal working area; The curve length of the energy distribution curve in the abnormal working area is processed by ratio with the curve length of the energy distribution curve in the normal working area to obtain the energy distribution area curve length ratio; The energy distribution area curve length ratio is compared with a preset ratio; if the energy distribution area curve length ratio is greater than the preset ratio, the target photovoltaic transformer is controlled to stop working, and an early warning message is generated and sent to a preset terminal.

[0042] It should be noted that the energy distribution area curve length ratio is obtained by calculating the ratio of the curve length of the energy distribution curve in the abnormal working area to the curve length in the normal working area. It can quantitatively reflect the degree to which the working state of the target photovoltaic transformer deviates from the normal state within the preset time period from a new perspective. The larger the energy distribution area curve length ratio, the higher the proportion of the energy distribution curve in the abnormal area, which means that the target photovoltaic transformer is more likely to be in an abnormal working state within the preset time period or the duration of the abnormal state is relatively longer; conversely, the smaller the energy distribution area curve length ratio, the more inclined the transformer is to a normal working state within the preset time period. This parameter provides a unique quantitative basis for the state assessment and fault diagnosis of photovoltaic transformers.

[0043] It should be noted that during the operation of the target photovoltaic transformer, its frequency response characteristic data is obtained according to the preset time nodes. Then, the discrete wavelet transform is used to perform multi-layer decomposition operations to obtain wavelet coefficients at different scales. For each time node, the energy values ​​of these wavelet coefficients are calculated according to the preset rules, and a data structure is constructed to store the energy values ​​corresponding to different scales at each time node. This series of operations is to deeply mine and organize the original frequency response characteristic data. The discrete wavelet transform is helpful to analyze data from multiple scales, and the energy value calculation provides more representative data for subsequent analysis. With the scale as the horizontal axis and the energy values ​​corresponding to different time nodes as the vertical axis, an energy distribution curve is generated. According to the preset working tasks of the target photovoltaic transformer, the energy distribution threshold range is determined, and then the upper and lower energy distribution thresholds are obtained, and mapped to the curve to define the normal and abnormal working areas. This construction method makes it possible to intuitively observe the working state area of ​​the transformer from the figure, and to link the abstract energy distribution with the working state. In the energy distribution curve, the curve lengths of the abnormal and normal working areas are calculated respectively, and then the ratio of the two is calculated to obtain the energy distribution area curve length ratio. This ratio combines the distribution of the energy distribution curve in different working areas and quantifies the working state characteristics of the transformer within a preset time period from a new perspective. The energy distribution area curve length ratio is compared with the preset ratio. If it is greater than the preset ratio, it means that the transformer is in a serious abnormal working area. At this time, the transformer is controlled to stop working and generate an early warning message to the preset terminal. This step realizes the effective monitoring of the transformer working state, timely discovers potential serious problems and takes measures to avoid possible more serious failures or damage, and ensures the safe and stable operation of the photovoltaic transformer and the entire photovoltaic system.

[0044] In summary, through multi-step processing of the frequency response characteristic data of the target photovoltaic transformer, the energy distribution curve is constructed and the working area is defined, and the special ratio is calculated and compared with the preset value, the quantitative evaluation and monitoring of the working state of the photovoltaic transformer is realized. It can timely detect the serious abnormal working state of the transformer, automatically stop the transformer and issue an early warning, effectively improving the safety and reliability of the operation of the photovoltaic transformer, reducing the losses caused by faults, and ensuring the stable operation of the entire photovoltaic power generation system.

[0045] In addition, the control method further comprises the following steps: The state transition probability value of the target photovoltaic transformer when working under various working environment parameter conditions is obtained through the big data network; Taking the working environment parameter as an input node and the state transition probability value as an output node, constructing a graph neural network according to the input node and the output node; Determining the connection relationship between the working environment parameter input node and the state transition probability value output node in the graph neural network, and constructing a conditional transition probability matrix according to the connection relationship between the working environment parameter input node and the state transition probability value output node; Constructing a state transition probability prediction model, and embedding the conditional transfer probability matrix into the state transition probability prediction model; During the operation of the target photovoltaic transformer, real-time working environment parameters of the target photovoltaic transformer are obtained at a preset time node, and the real-time working environment parameters are imported into a state transition probability prediction model for prediction; Through prediction, the state transition probability value of the target photovoltaic transformer under the real-time working environment parameter conditions is obtained; If the state transition probability value of the target photovoltaic transformer under the real-time working environment parameter conditions is greater than the preset probability value, the continuous state monitoring instruction for the target photovoltaic transformer is restarted.

[0046] It should be noted that the state transition probability value of the target photovoltaic transformer under different working environment parameters is first obtained through the big data network, and then the graph neural network is constructed based on this, with the working environment parameters set as input nodes and the state transition probability value set as output nodes. This step lays the data and structural foundation for subsequent analysis, and uses the characteristics of the graph neural network to process the input-output relationship. After determining the connection relationship between the input and output nodes in the graph neural network, the conditional transition probability matrix is ​​constructed and embedded in the state transition probability prediction model. The conditional transition probability matrix reflects the probability transition under different input-output relationships. After being embedded in the model, the model can have the ability to predict the state transition probability based on environmental parameters. When the target photovoltaic transformer is working, the real-time working environment parameters are obtained and imported into the prediction model to obtain the real-time state transition probability value. If the value is greater than the preset probability value, the state continuous monitoring instruction is restarted. This enables the monitoring strategy to be adjusted in time according to the real-time environmental conditions, and to focus on possible state transition situations. By constructing the graph neural network, the conditional transition probability matrix and the state transition probability prediction model, the state transition probability of the target photovoltaic transformer based on its working environment parameters is predicted. The decision on whether to restart the status continuous monitoring instruction is made based on the prediction results, which improves the accuracy and timeliness of photovoltaic transformer status monitoring and ensures its stable operation.

[0047] The present invention also discloses a photovoltaic transformer, which is applied to any of the photovoltaic transformer control methods described above. Figure 3 As shown, the photovoltaic transformer includes a base 401 and a transformer body 402, and the transformer body is installed and fixed through the base; the transformer body is provided with heat dissipation fins 403; and the top of the transformer body is provided with a sleeve 404 and a controller 405.

[0048] The present invention also discloses a control system for a photovoltaic transformer, the control system comprising a memory and a processor, the memory storing a control method program for the photovoltaic transformer, and when the control method program for the photovoltaic transformer is executed by the processor, any one of the control method steps for the photovoltaic transformer is implemented.

[0049] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A control method for a photovoltaic transformer, characterized in that: The following steps are involved: Perform feature analysis on various system operation abnormal events that may be caused when the target photovoltaic transformer generates various historical acoustic feature data during historical operation, and obtain frequent system operation abnormal events and occasional system operation abnormal events; Construct a binary tree knowledge graph based on the historical acoustic feature data corresponding to frequent system operation abnormal events and occasional system operation abnormal events; Acquire real-time acoustic feature data generated by the target photovoltaic transformer in the real-time working process through an acoustic sensor at a preset time node, analyze the system state of the target photovoltaic transformer at the current time node according to the real-time acoustic feature data and in combination with a binary tree knowledge graph, and generate an analysis result; If the system status of the target photovoltaic transformer is in normal operation at the current time node, the target photovoltaic transformer will not be regulated; If the system state of the target photovoltaic transformer at the current preset time node is in a sporadic state of abnormal operation, the target photovoltaic transformer will not be regulated and the operation state of the target photovoltaic transformer will be continuously monitored at the next time node; If the system state of the target photovoltaic transformer at the current preset time node is in a state where abnormal operation events frequently occur, the target photovoltaic transformer is regulated.

2. A photovoltaic transformer control method according to claim 1, characterized in that: The characteristic analysis of various system operation abnormal events caused by various historical acoustic characteristic data generated by the target photovoltaic transformer in the historical working process is carried out to obtain frequent system operation abnormal events and occasional system operation abnormal events, specifically: Acquire various historical acoustic characteristic data generated by the target photovoltaic transformer in the historical working process, and count various system operation abnormal events caused by the target photovoltaic transformer generating various historical acoustic characteristic data in the historical working process according to the operation log of the target photovoltaic transformer; Perform statistical analysis on various system operation abnormal events that would be triggered when the target photovoltaic transformer generates various historical acoustic characteristic data during historical operation, and obtain the number of corresponding system operation abnormal events that would be triggered when various historical acoustic characteristic data are generated; The number of corresponding system operation abnormal events that will be triggered when various historical acoustic feature data are generated is compared with a preset value to obtain the frequency value of the corresponding system operation abnormal events that will be triggered when various historical acoustic feature data are generated; Comparing the frequency value of the corresponding system operation abnormal event that will be triggered when various historical acoustic feature data are generated with a preset frequency value threshold; A system operation abnormality event whose frequency value is greater than a preset frequency value threshold is marked as a frequent system operation abnormality event; The system operation abnormality events whose frequency values ​​are not greater than the preset frequency value threshold are marked as occasional system operation abnormality events.

3. A photovoltaic transformer control method according to claim 2, characterized in that: A binary tree knowledge graph is constructed based on the historical acoustic feature data corresponding to frequent system operation abnormal events and occasional system operation abnormal events, specifically: Acquire historical acoustic feature data corresponding to each frequent system operation abnormality event, and acquire historical acoustic feature data corresponding to each occasional system operation abnormality event; Constructing a binary tree, and dividing the binary tree into two main branches, namely a first main branch and a second main branch, according to the frequent system operation abnormal events and the occasional system operation abnormal events; Iteratively splitting the first main trunk according to each frequently occurring system operation abnormality event, so as to split a plurality of frequently occurring system operation abnormality event leaf nodes on the first main trunk; wherein each frequently occurring system operation abnormality event leaf node stores characteristic information of the corresponding frequently occurring system operation abnormality event; Iteratively splitting the second main trunk according to each incidental system operation abnormality event, so as to split a plurality of incidental system operation abnormality event leaf nodes on the second main trunk; wherein each incidental system operation abnormality event leaf node stores characteristic information of the corresponding incidental system operation abnormality event; The historical acoustic feature data corresponding to each frequent system operation abnormality event is stored in the corresponding frequent system operation abnormality event leaf node; the historical acoustic feature data corresponding to each occasional system operation abnormality event is stored in the corresponding occasional system operation abnormality event leaf node; to form a binary tree knowledge graph.

4. A photovoltaic transformer control method according to claim 1, characterized in that: At a preset time node, the real-time acoustic feature data generated by the target photovoltaic transformer in the real-time working process is obtained through an acoustic sensor. According to the real-time acoustic feature data and combined with the binary tree knowledge graph, the system state of the target photovoltaic transformer at the current time node is analyzed to generate an analysis result, which is specifically: Acquiring real-time acoustic feature data generated by the target photovoltaic transformer in the real-time working process through an acoustic sensor at a preset time node; and extracting historical acoustic feature data in each leaf node from the binary tree knowledge graph; Introducing the Kendall rank correlation coefficient algorithm, and calculating the Kendall rank correlation coefficient value between the real-time acoustic feature data and the historical acoustic feature data in each leaf node based on the Kendall rank correlation coefficient algorithm; The calculated Kendall rank correlation coefficient values ​​are compared with the preset coefficient values. If the values ​​of the Kendall rank correlation coefficients are not greater than the preset coefficient values, it means that the system state of the target photovoltaic transformer at the current time node is in a normal operating state; If there is at least one Kendall rank correlation coefficient value greater than the preset coefficient value, the leaf node corresponding to the Kendall rank correlation coefficient value greater than the preset coefficient value is marked, and the feature type of the marked leaf node is determined; wherein the feature type includes an occasional system operation abnormal event leaf node and a frequent system operation abnormal event leaf node; If the characteristic type of the marked leaf node is a leaf node of frequent system operation abnormal events, it means that the system state of the target photovoltaic transformer at the current preset time node is in a state of frequent operation abnormal events; If the characteristic type of the marked leaf node is an occasional system operation abnormality event leaf node, it means that the system state of the target photovoltaic transformer at the current preset time node is in an occasional operation abnormality event state.

5. A photovoltaic transformer control method according to claim 1, characterized in that: If the system state of the target photovoltaic transformer at the current preset time node is in a state of frequent abnormal operation events, the target photovoltaic transformer is regulated and processed, specifically: If the system state of the target photovoltaic transformer at the current preset time node is in a state where abnormal operation events frequently occur, then the real-time working parameters of the target photovoltaic transformer at the current preset time node are obtained; Genetic algorithm is introduced, the real-time working parameters of the target photovoltaic transformer are used as decision variables, and the decision variables are encoded into chromosome representation to construct the initial population. Each individual in the population represents a possible regulation scheme. Determine the key indicators related to the performance of the target photovoltaic transformer, including voltage deviation, power factor, temperature and efficiency, assign weights to each key indicator according to its importance to the normal operation of the system, multiply each indicator value by the corresponding weight and sum them up, and construct a fitness function that can comprehensively evaluate the advantages and disadvantages of the control scheme; Execute the selection operation, adopt the preset selection strategy according to the fitness value, and select excellent individuals from the current population to enter the next generation population; Perform crossover operations, exchange genes of selected individuals according to the set crossover probability, simulate the biological gene recombination process, and generate new regulation scheme individuals to increase the diversity of the population; Perform mutation operations to randomly change the genes of individuals whose mutation probability is less than the preset probability value to form new individuals; The above selection, crossover and mutation operations are continuously iterated, and the fitness of the individual is re-evaluated in each round of iteration until the maximum number of iterations is reached, and the control scheme corresponding to the best individual is screened out from the population to obtain the optimal working parameter control scheme of the target photovoltaic transformer; wherein the best individual is the individual with the highest fitness; The real-time operating parameters of the target photovoltaic transformer are regulated according to the optimal operating parameter regulation scheme.

6. A photovoltaic transformer control method according to claim 5, characterized in that: The control method of the photovoltaic transformer further comprises the following steps: After the real-time working parameters of the target photovoltaic transformer are regulated by the optimal working parameter regulation scheme, the real-time acoustic feature data of the target photovoltaic transformer is obtained again, and the system state of the target photovoltaic transformer is analyzed according to the real-time acoustic feature data and in combination with the binary tree knowledge graph; If the target photovoltaic transformer is still in a state of frequent abnormal operation after being regulated, the target photovoltaic transformer is controlled to stop working, and an early warning message is generated and sent to a preset terminal; If the target photovoltaic transformer is in a state of occasional abnormal operation after being regulated, the operation state of the target photovoltaic transformer will continue to be monitored at the next time node; If the target photovoltaic transformer is in a normal operating state after being regulated, the instruction for continuous monitoring of the state of the target photovoltaic transformer is released.

7. A photovoltaic transformer, applied to a control method of a photovoltaic transformer according to any one of claims 1 to 6, characterized in that: The photovoltaic transformer comprises a base and a transformer body, and the transformer body is installed and fixed by the base; the transformer body is provided with heat dissipation fins; and the top of the transformer body is provided with a bushing and a controller.

8. A control system for a photovoltaic transformer, characterized in that: The control system includes a memory and a processor, wherein a control method program for a photovoltaic transformer is stored in the memory, and when the control method program for the photovoltaic transformer is executed by the processor, the control method steps for the photovoltaic transformer according to any one of claims 1 to 6 are implemented.

Images