Remote AI monitoring method and system for photovoltaic energy storage equipment based on management platform

Abstract

The present invention relates to the technical field of photovoltaic energy storage, and provides a remote AI monitoring method and system for photovoltaic energy storage equipment based on a management platform, which obtains the operating signal of the energy storage equipment, analyzes and processes the operating signal, and obtains signal characteristic data and signal abnormality indicators; performs trend analysis on the signal characteristic data to obtain trend analysis results, uses the trend analysis results to verify the signal abnormality indicators, and obtains abnormal verification results; inputs the abnormal verification results into a preset energy storage management system for analysis to obtain preliminary diagnostic information, and parses the preliminary diagnostic information to obtain potential faults; generates corresponding fault alarm signals according to potential faults, and generates corresponding fault handling suggestions based on the fault alarm signals. By combining artificial intelligence and data mining technology to analyze the characteristic data and abnormal indicators of the operating signal, the problems of insufficient intelligence of data analysis and insufficient fault diagnosis accuracy in practical applications are improved.

Description

Technical Field

[0001] The present application relates to the technical field of photovoltaic energy storage, and in particular to a remote AI monitoring method and system for photovoltaic energy storage equipment based on a management platform. Background Art

[0002] In recent years, as the global demand for renewable energy continues to grow, photovoltaic energy storage, as a green and environmentally friendly energy solution, has received widespread attention and application. Photovoltaic energy storage equipment can effectively convert and store solar energy into electrical energy, which not only improves energy utilization efficiency, but also provides strong support for the stability and flexibility of the power grid. In this context, how to efficiently monitor and maintain the operating status of photovoltaic energy storage equipment has become an important direction for the development of the industry.

[0003] Among the relevant technical means, the monitoring method mainly relies on local control systems and regular manual maintenance to obtain the operating data of the equipment. This method can perform basic status monitoring and fault detection on the equipment. However, with the continuous expansion and complexity of photovoltaic energy storage systems, traditional monitoring methods are stretched to the limit and it is difficult to meet the real-time, comprehensive and accurate monitoring needs. Remote monitoring systems based on management platforms have begun to be used. Through network transmission technology, centralized monitoring and data analysis of distributed energy storage equipment can be achieved, thereby improving operation and maintenance efficiency and reducing maintenance costs.

[0004] Regarding the above technical solution, although the existing remote monitoring system can realize real-time monitoring and basic fault warning of photovoltaic energy storage equipment, in actual application, there are problems such as insufficient intelligence of data analysis and insufficient accuracy of fault diagnosis, which leads to the failure to timely discover potential equipment faults or false alarms, thus affecting the reliability and service life of the equipment. Summary of the invention

[0005] In order to improve the problems of insufficient intelligence of data analysis and insufficient accuracy of fault diagnosis in actual applications, the present application provides a remote AI monitoring method and system for photovoltaic energy storage equipment based on a management platform. By introducing artificial intelligence and data mining technology, the analysis and fault diagnosis capabilities of equipment status are significantly improved, and more accurate fault handling suggestions are provided, thereby optimizing the operation and maintenance of equipment.

[0006] The present invention provides a remote AI monitoring method for photovoltaic energy storage equipment based on a management platform, comprising: acquiring an operation signal of an energy storage device, analyzing and processing the operation signal to obtain signal feature data and a signal abnormality index; performing trend analysis on the signal feature data to obtain a trend analysis result, and using the trend analysis result to verify the signal abnormality index to obtain an abnormality verification result; inputting the abnormality verification result into a preset energy storage management system for analysis to obtain preliminary diagnosis information, parsing the preliminary diagnosis information to obtain a diagnosis subcategory one and a diagnosis subcategory two; performing pattern recognition on the diagnosis subcategory one using a machine learning model to obtain a diagnosis result, and performing a comprehensive comparison between the diagnosis result and the diagnosis subcategory two to obtain a comparison result; analyzing the comparison result by using a data mining technology to obtain a comprehensive analysis result, evaluating the operation status of the energy storage device based on the comprehensive analysis result, performing a risk assessment on the operation status to obtain a potential fault; generating a corresponding fault alarm signal according to the potential fault, and generating a corresponding fault handling suggestion based on the fault alarm signal.

[0007] As a preferred solution, the steps of acquiring the operation signal of the energy storage device, analyzing and processing the operation signal, and obtaining signal characteristic data and signal abnormality indicators include: real-time monitoring of the energy storage device through a sensor array to obtain an operation signal, and using a signal processing algorithm to filter noise and extract features of the operation signal to obtain time series data and frequency analysis data; trend identification of the time series data according to a statistical analysis method to obtain trend data and seasonal components, and trend prediction of the trend data through a regression model to obtain predicted trend data; type analysis of the seasonal component using a clustering algorithm to obtain type characteristic data, and correlation analysis of the predicted trend data and the type characteristic data according to a preset association rule to obtain signal characteristic data; abnormal pattern identification of the frequency analysis data based on a machine learning method to obtain abnormal pattern 1 and abnormal pattern 2, and abnormal confirmation of the abnormal pattern 1 by a threshold judgment method to obtain confirmed abnormal data; abnormal level evaluation of the confirmed abnormal data and the abnormal pattern 2 using a logistic regression model to obtain a signal abnormality indicator.

[0008] As a preferred scheme, the steps of performing trend analysis on the signal characteristic data to obtain trend analysis results, and using the trend analysis results to verify the signal abnormality indicators to obtain abnormality verification results include: using data smoothing technology to smooth the signal characteristic data to obtain smoothed data and residual data, and performing trend prediction on the smoothed data according to time series analysis to obtain predicted data and confidence intervals; performing pattern recognition on the predicted data through pattern matching technology to obtain pattern recognition data, and using the pattern recognition data to correct the residual data to obtain corrected data; performing abnormality detection on the corrected data and the signal abnormality indicators according to a machine learning classification model to obtain trend analysis results; evaluating the trend analysis results through a decision tree model to obtain risk assessment data, and performing abnormality verification on the signal abnormality indicators according to the risk assessment data to obtain abnormality verification results.

[0009] As a preferred solution, the step of inputting the abnormality check result into a preset energy storage management system for analysis to obtain preliminary diagnostic information, parsing the preliminary diagnostic information to obtain diagnostic subcategory one and diagnostic subcategory two includes: using fuzzy logic to match the abnormality check result with rules to obtain matching results, inputting the matching results into a preset energy storage management system, in the energy storage management system, comprehensively evaluating the matching results through data association analysis to obtain evaluation results and risk warnings, performing fault prediction on the evaluation results based on model reasoning to obtain predicted fault information, and using a decision tree algorithm to classify the predicted fault information and the risk warning into fault levels to obtain preliminary diagnostic information; deconstructing the preliminary diagnostic information through logical analysis to obtain structured data and unstructured data, and using text analysis technology to perform semantic parsing on the structured data to obtain diagnostic subcategory one; and extracting content from the unstructured data using pattern recognition technology to obtain diagnostic subcategory two.

[0010] As a preferred scheme, the steps of using a machine learning model to perform pattern recognition on the diagnostic subcategory one to obtain a diagnostic result, and performing a comprehensive comparison between the diagnostic result and the diagnostic subcategory two to obtain a comparison result include: performing deep learning analysis on the diagnostic subcategory one through a neural network model to obtain a feature vector set, a probability score, and a classification result, performing feature optimization on the feature vector set to obtain an optimized feature vector, and inputting the optimized feature vector into a preset support vector machine model to perform boundary division on the probability score to obtain a division result; performing pattern matching between the division result and the diagnostic subcategory two using a preset machine learning model to obtain a matched diagnostic result; and performing a logical comparison between the diagnostic result and the diagnostic subcategory two through decision analysis to obtain a comparison result.

[0011] As a preferred scheme, the step of analyzing the comparison results by data mining technology to obtain comprehensive analysis results, evaluating the operating status of the energy storage equipment based on the comprehensive analysis results, performing risk assessment on the operating status, and obtaining potential faults includes: performing correlation detection on the comparison results according to association rule analysis to obtain correlation data and independence data, and classifying the correlation data into types by cluster analysis to obtain type clustering results; using the type clustering results as independent variables and the independence data as dependent variables to input into a preset regression model for prediction to obtain prediction results; performing in-depth mining on the prediction results by data mining technology to obtain comprehensive analysis results; performing risk assessment on the comprehensive analysis results by probability statistics methods to obtain risk scores and early warning indicators, and performing fault prediction on the risk scores and early warning indicators by early warning algorithms to obtain potential faults.

[0012] As a preferred scheme, the step of generating a corresponding fault alarm signal according to the potential fault and generating a corresponding fault handling suggestion based on the fault alarm signal includes: performing fault characterization on the potential fault through a fault detection algorithm to obtain fault type data, performing logical judgment on the fault type data using a rule engine to obtain a judgment result, inputting the judgment result into a preset fault handling system to obtain a corresponding fault alarm signal; inputting the fault alarm signal into a preset fault response system to generate a corresponding fault handling suggestion.

[0013] The present application also provides a remote AI monitoring system for photovoltaic energy storage equipment based on a management platform, including: an acquisition unit, used to acquire an operation signal of the energy storage equipment, analyze and process the operation signal, and obtain signal feature data and signal abnormality indicators; an analysis unit, used to perform trend analysis on the signal feature data to obtain a trend analysis result, and use the trend analysis result to verify the signal abnormality indicator to obtain an abnormality verification result; a parsing unit, used to input the abnormality verification result into a preset energy storage management system for analysis to obtain preliminary diagnosis information, and parse the preliminary diagnosis information to obtain a diagnosis subcategory one and a diagnosis subcategory two; a comparison unit, used to perform pattern recognition on the diagnosis subcategory one using a machine learning model to obtain a diagnosis result, and perform a comprehensive comparison between the diagnosis result and the diagnosis subcategory two to obtain a comparison result; an evaluation unit, used to analyze the comparison result by data mining technology to obtain a comprehensive analysis result, evaluate the operation status of the energy storage equipment based on the comprehensive analysis result, perform a risk assessment on the operation status, and obtain a potential fault; a generation unit, used to generate a corresponding fault alarm signal according to the potential fault, and generate a corresponding fault handling suggestion based on the fault alarm signal.

[0014] The present application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the remote AI monitoring method for photovoltaic energy storage equipment based on the management platform described above is implemented.

[0015] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor executes the remote AI monitoring method for photovoltaic energy storage equipment based on the management platform as described above.

[0016] Compared with the existing technology, this application has the following beneficial effects: wide monitoring range and high diagnostic accuracy. By combining artificial intelligence and data mining technology, comprehensive monitoring and accurate diagnosis of the operating status of photovoltaic energy storage equipment are achieved. By analyzing the characteristic data and abnormal indicators of the operating signal, potential faults can be identified in advance and losses caused by equipment failure can be reduced. The use of machine learning and data mining improves the accuracy and timeliness of diagnosis, making fault handling suggestions more accurate and timely, and improving the problems of insufficient intelligence of data analysis and insufficient fault diagnosis accuracy in actual application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] 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, other drawings can be obtained based on these drawings without paying creative labor.

[0018] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0019] Figure 1 It is a flow chart of a remote AI monitoring method for photovoltaic energy storage equipment based on a management platform provided by an embodiment of the present invention;

[0020] Figure 2 It is a schematic block diagram of the structure of a remote AI monitoring system for photovoltaic energy storage equipment based on a management platform provided in an embodiment of the present invention;

[0021] Figure 3 It is a schematic block diagram of the structure of an electronic device provided by an embodiment of the present invention.

[0022] Description of reference numerals:

[0023] 10. Remote AI monitoring system for photovoltaic energy storage equipment based on management platform; 11. Acquisition unit; 12. Analysis unit; 13. Parsing unit; 14. Comparison unit; 15. Evaluation unit; 16. Generation unit; 20. Electronic device; 21. Memory; 22. Processor. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0026] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0027] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0028] The technical solution of the present invention is further described below with reference to the accompanying drawings and through specific implementation methods.

[0029] Embodiment 1:

[0030] like Figure 1 As shown, the remote AI monitoring method for photovoltaic energy storage equipment based on a management platform provided in an embodiment of the present application includes steps S100 to S600.

[0031] Step S100: Acquire the operation signal of the energy storage device, analyze and process the operation signal, and obtain signal characteristic data and signal abnormality index.

[0032] In this step, the system first collects real-time operating signals from various sensors and controllers of photovoltaic energy storage equipment. These signals include parameters such as voltage, current, temperature and equipment status. Specifically, the collected operating signals undergo preprocessing steps such as denoising, normalization and feature extraction to convert them into signal feature data. At the same time, based on historical data and normal operating parameters of the equipment, signal thresholds and models are set to identify signal abnormality indicators.

[0033] For example, if it is detected that the current fluctuation exceeds the normal range within a certain period of time, it is marked as a signal abnormality indicator.

[0034] Step S200: Perform trend analysis on the signal feature data to obtain trend analysis results, and use the trend analysis results to verify the signal anomaly indicators to obtain anomaly verification results.

[0035] In this step, the signal feature data is trend analyzed using time series analysis methods to identify long-term change patterns; specifically, moving average, exponential smoothing and other methods are applied to capture the growth, stability or downward trend of signal exceptions, and these trends are checked against the standards of the set model to see if they reflect abnormal situations, so as to generate anomaly verification results.

[0036] For example, when the signal characteristic data shows that the temperature has been rising continuously over a period of time and exceeds the set safety range, the system will verify and confirm this anomaly.

[0037] Step S300: input the abnormality check result into the preset energy storage management system for analysis to obtain preliminary diagnosis information, and parse the preliminary diagnosis information to obtain diagnosis subcategory one and diagnosis subcategory two.

[0038] In this step, the abnormality verification results are input into the intelligent analysis module of the energy storage management system, which generates preliminary diagnostic information by combining the equipment model and the intelligent consultation function. Specifically, according to the nature of the abnormality and the operation history of the equipment, the preliminary diagnostic information is refined into diagnostic subcategory one and diagnostic subcategory two, which respectively reflect different types of equipment abnormality patterns.

[0039] For example, the initial diagnostic information is displayed as "temperature is too high + current exceeds the limit". After analysis, diagnostic subcategory one is "heat dissipation failure" and diagnostic subcategory two is "current leakage".

[0040] Step S400: Use the machine learning model to perform pattern recognition on the diagnostic subcategory 1 to obtain a diagnostic result, and perform a comprehensive comparison between the diagnostic result and the diagnostic subcategory 2 to obtain a comparison result. In this step, a machine learning model, such as a neural network or a support vector machine, is applied to perform in-depth pattern recognition on the diagnostic subcategory 1 to obtain a more accurate diagnostic result; specifically, the diagnostic result output by the machine learning model is comprehensively compared with the diagnostic subcategory 2 to identify the correlation between multiple faults and form the final comparison result.

[0041] For example, through pattern recognition analysis, players identified that "heat dissipation failure" and "current leakage" originated from the same hardware problem, and thus marked them in the final comparison results.

[0042] Step S500: Analyze the comparison results by using data mining technology to obtain comprehensive analysis results, evaluate the operating status of the energy storage device based on the comprehensive analysis results, conduct risk assessment on the operating status, and obtain potential faults.

[0043] In this step, the system automatically generates comprehensive analysis results using data mining technology combined with historical data and comparison results. Specifically, through association analysis, classification and clustering technology, potential risk factors of equipment operating status are identified, and a panoramic risk assessment is conducted to ultimately identify potential faults.

[0044] For example, if comprehensive analysis results show that a specific component has behaved abnormally in multiple operations, it may identify a potential hardware failure that requires preventive maintenance.

[0045] Step S600: Generate a corresponding fault alarm signal according to the potential fault, and generate a corresponding fault handling suggestion based on the fault alarm signal.

[0046] In this step, the system generates corresponding fault alarm signals according to the severity and type of potential faults, and notifies users in various ways through the storage management platform; specifically, combined with the comprehensive analysis results, it provides maintenance and operation suggestions for the fault situation to guide users to make reasonable responses and maintenance.

[0047] For example, a "Cooling Fan Failure Warning" alarm signal is generated, and it is recommended to check the cooling module and fan circuit connection.

[0048] In this embodiment, the operation signals of the energy storage device are obtained, and these operation signals are analyzed and processed to extract signal feature data and signal abnormality indicators. Next, the signal feature data is trend analyzed to obtain trend analysis results, and the signal abnormality indicators are verified using the results to obtain abnormality verification results. Then, the abnormality verification results are input into the preset energy storage management system for analysis to obtain preliminary diagnostic information. After the preliminary diagnostic information is parsed, it is divided into diagnostic subcategory one and diagnostic subcategory two. Pattern recognition is performed on diagnostic subcategory one through a machine learning model to obtain a diagnostic result, and the diagnostic result is comprehensively compared with diagnostic subcategory two to obtain a comparison result. Finally, the comparison results are analyzed through data mining technology to obtain a comprehensive analysis result to evaluate the operating status of the energy storage device, and a risk assessment is performed to identify potential faults. Based on the identified potential faults, a fault alarm signal and corresponding fault handling suggestions are generated.

[0049] By combining artificial intelligence and data mining technology, comprehensive monitoring and accurate diagnosis of the operating status of photovoltaic energy storage equipment are achieved. By systematically analyzing the characteristic data and abnormal indicators of the operating signals, potential faults can be identified in advance and losses caused by equipment failures can be reduced. In addition, the use of machine learning and data mining improves the accuracy and timeliness of diagnosis, making fault handling suggestions more accurate and timely, thereby reducing the difficulty and cost of equipment operation and maintenance and improving the overall reliability and efficiency of photovoltaic energy storage systems.

[0050] Embodiment 2:

[0051] In step S100, the energy storage device is monitored in real time by a sensor array to obtain an operation signal, and the operation signal is subjected to noise filtering and feature extraction by a signal processing algorithm to obtain time series data and frequency analysis data.

[0052] The collected signals are preprocessed by using filtering algorithms to remove noise and interference signals, and extract useful feature data; specifically, the voltage, current and temperature signals are processed using Kalman filtering and wavelet transform methods to obtain accurate data at each time point.

[0053] For example, Kalman filtering is used on voltage signals to remove random noise and abnormal mutations and extract key characteristic data of voltage changes.

[0054] The trend of time series data is identified according to the statistical analysis method to obtain trend data and seasonal components. The trend data is predicted by the regression model to obtain predicted trend data.

[0055] By using time series analysis technology to decompose the data, long-term trends and cyclical changes are identified; specifically, moving average and exponential smoothing methods are used to perform trend analysis on the current signal to capture its long-term change trends and cyclical fluctuations.

[0056] For example, the moving average method is used to smooth the current data and identify monthly seasonal fluctuations and overall growth trends.

[0057] The clustering algorithm is used to perform type analysis on the seasonal component to obtain type feature data, and the correlation analysis is performed on the predicted trend data and type feature data according to the preset association rules to obtain signal feature data.

[0058] By performing K-means cluster analysis on seasonal components, different types of seasonal variation patterns are classified; specifically, the voltage data are divided into high-frequency, low-frequency and stable types, and pattern analysis is performed to identify the characteristics of each type of change.

[0059] For example, high-frequency voltage fluctuations can be classified into one category, their characteristic data can be identified, and their correlation with temperature changes can be analyzed.

[0060] Based on the machine learning method, the frequency analysis data is used to identify abnormal patterns, and abnormal pattern 1 and abnormal pattern 2 are obtained. The abnormal pattern 1 is confirmed to be abnormal through the threshold judgment method to obtain confirmed abnormal data.

[0061] By training the frequency analysis data using a support vector machine (SVM), abnormal patterns are identified; specifically, the current frequency data is classified to identify abnormal patterns whose frequencies are outside a normal range.

[0062] For example, by training the model using SVM, a pattern in which the frequency of the current abnormally increases in certain time periods is identified and marked as abnormal pattern one.

[0063] The logistic regression model is used to evaluate the abnormal level of confirmed abnormal data and abnormal pattern 2 to obtain the signal abnormality index.

[0064] By establishing a logistic regression model, different abnormal data are evaluated and the severity of the abnormality is determined; specifically, the abnormal current data is evaluated to determine its impact on the operation of the equipment and generate abnormal indicators.

[0065] For example, according to the logistic regression model evaluation, the situation where the current is outside the normal range is evaluated as moderately abnormal, and the corresponding abnormal index is generated.

[0066] In step S200, the signal feature data is smoothed using data smoothing technology to obtain smoothed data and residual data, and trend prediction is performed on the smoothed data based on time series analysis to obtain predicted data and confidence intervals.

[0067] By applying moving average and exponential smoothing methods, the signal data is smoothed to remove short-term fluctuations and noise; specifically, the voltage data is smoothed to obtain the smoothed long-term trend and short-term residual.

[0068] For example, the fluctuation of voltage data is smoothed into a trend line through exponential smoothing, and the residual data is calculated.

[0069] The predicted data is subjected to pattern recognition through pattern matching technology to obtain pattern recognition data, and the residual data is corrected using the pattern recognition data to obtain corrected data.

[0070] By performing pattern matching on the predicted data, parts similar to known patterns are identified; specifically, the current current trend is compared with patterns in historical data to identify the pattern category of the current signal.

[0071] For example, the current current trend is compared with the normal operating mode in the historical data, the current data is identified as belonging to the normal mode, and the residual data is corrected.

[0072] Anomaly detection is performed on the corrected data and signal anomaly indicators based on the machine learning classification model to obtain trend analysis results.

[0073] By using a random forest classifier to classify the corrected data, abnormal situations are identified; specifically, the corrected voltage data is compared with the normal operating data to detect abnormal voltage changes.

[0074] For example, through the random forest model, it is detected that the corrected voltage data has abnormal fluctuations in certain periods, and trend analysis results are generated.

[0075] The trend analysis results are evaluated through a decision tree model to obtain risk assessment data, and the signal abnormality indicators are checked for abnormality based on the risk assessment data to obtain abnormality check results.

[0076] By applying the decision tree algorithm, risk assessment is performed on the trend analysis results to identify potential risks; specifically, risk assessment is performed on abnormal voltage data to determine the risk of equipment failure caused by it.

[0077] For example, through decision tree model evaluation, it was detected that high-frequency voltage fluctuations caused equipment overheating, generating risk assessment data.

[0078] In step S300, fuzzy logic is used to match the abnormality check results to obtain matching results, and the matching results are input into a preset energy storage management system. In the energy storage management system, the matching results are comprehensively evaluated through data association analysis to obtain evaluation results and risk warnings. Fault prediction is performed on the evaluation results based on model reasoning to obtain predicted fault information. The predicted fault information and risk warnings are classified into fault levels using a decision tree algorithm to obtain preliminary diagnostic information.

[0079] The matching result is generated by matching the abnormality check result with the rule base in the energy storage management system; specifically, the abnormal voltage data is matched with the fault mode in the rule base to identify the fault type.

[0080] For example, through rule matching, it is detected that abnormal voltage fluctuation is caused by a power module failure.

[0081] The preliminary diagnostic information is deconstructed through logical analysis to obtain structured data and unstructured data. The structured data is semantically parsed using text analysis technology to obtain diagnostic subcategory one. The unstructured data is extracted using pattern recognition technology to obtain diagnostic subcategory two.

[0082] By performing logical analysis on the preliminary diagnostic information, structured data and unstructured data are separated; specifically, the voltage anomaly information is divided into specific voltage values ​​and description information.

[0083] For example, "abnormal voltage fluctuation" is divided into two parts: "voltage value fluctuation exceeds the normal range" and "causing equipment overheating".

[0084] In step S400, deep learning analysis is performed on diagnostic subcategory one through a neural network model to obtain a feature vector set, a probability score, and a classification result. Feature optimization is performed on the feature vector set to obtain an optimized feature vector. The optimized feature vector is input into a preset support vector machine model to perform boundary division on the probability score to obtain a division result.

[0085] The diagnostic subcategory one is deep-learned using a convolutional neural network (CNN) to extract feature vectors; specifically, the current fluctuation data is convolved to extract key feature vectors.

[0086] For example, the CNN model is used to extract the amplitude and frequency characteristics of current fluctuations, and then optimize the features.

[0087] The preset machine learning model is used to perform pattern matching on the segmentation results and the diagnostic subcategory two to obtain the matched diagnostic results.

[0088] The diagnostic results are pattern matched by applying a support vector machine (SVM) model to the optimized feature vectors; specifically, the optimized voltage signatures are compared with known fault patterns to generate matching results.

[0089] For example, through the SVM model, the current voltage fluctuation is matched with the known power supply failure mode to generate a diagnosis result.

[0090] The diagnostic result and the diagnostic subcategory 2 are logically compared through decision analysis to obtain the comparison result.

[0091] By applying the decision analysis method, a logical comparison is made between the diagnosis results and the second diagnosis subcategory to confirm the accuracy of the diagnosis; specifically, the voltage fault diagnosis results are compared with other fault modes to confirm the final fault type.

[0092] For example, through logical comparison, it is confirmed that voltage fluctuation is due to power module failure rather than sensor error.

[0093] In step S500, association detection is performed on the comparison result according to association rule analysis to obtain association data and independence data, and the association data is divided into types through cluster analysis to obtain type clustering results.

[0094] By comparing the results and performing association rule analysis, the correlation and independence factors are identified; specifically, the voltage failure is associated with other data and the correlation between the power failure and the temperature rise is identified.

[0095] For example, through association rule analysis, it is found that voltage fluctuations are highly correlated with temperature changes in a specific time period.

[0096] The type clustering results are used as independent variables, and the independent data are used as dependent variables to input into the preset regression model for prediction to obtain the prediction results.

[0097] By performing regression analysis on the type clustering results and independence data, future fault conditions are predicted; specifically, the voltage fluctuation type and temperature data are input into the regression model to predict future equipment failure risks.

[0098] For example, through regression analysis models, it is predicted that the frequency and amplitude of voltage fluctuations will increase in high temperature environments, and the risk of equipment failure will increase.

[0099] The prediction results are deeply mined through data mining technology to obtain comprehensive analysis results.

[0100] By applying data mining technology, the prediction results are deeply analyzed to extract valuable information and patterns; specifically, association rule mining and frequent pattern mining are used to identify the potential causes and associated factors of failures.

[0101] For example, through data mining, it was found that the aging of power modules is the main cause of frequent voltage fluctuations.

[0102] The comprehensive analysis results are evaluated for risk through probability statistics to obtain risk scores and early warning indicators. The risk scores and early warning indicators are predicted through early warning algorithms to obtain potential failures.

[0103] By applying probabilistic statistical methods, the comprehensive analysis results are quantitatively evaluated to identify the risk level of the equipment; specifically, Bayesian networks and Markov chains are used to model and evaluate risks and generate early warning indicators.

[0104] For example, through risk assessment, voltage fluctuation warnings are generated and power module failures are predicted in the future.

[0105] In step S600, the potential fault is characterized by a fault detection algorithm to obtain fault type data, and a rule engine is used to perform logical judgment on the fault type data to obtain a judgment result, which is input into a preset fault handling system to obtain a corresponding fault alarm signal.

[0106] By applying the fault detection algorithm, potential faults are analyzed and characterized in detail; specifically, the decision tree and support vector machine models are used to judge the fault type and generate specific fault type data.

[0107] For example, through the fault detection algorithm, it is confirmed that the voltage fluctuation is caused by the power module failure, and a power module failure alarm signal is generated.

[0108] The fault alarm signal is input into the preset fault response system to generate corresponding fault handling suggestions.

[0109] By transmitting the fault alarm signal to the fault response system, detailed processing suggestions and operation steps are generated; specifically, troubleshooting and maintenance suggestions are provided based on the fault type and equipment status.

[0110] For example, a "power module failure" alarm is generated and a module replacement and system check are recommended.

[0111] In this embodiment, the energy storage device is monitored in real time through a sensor array, and the signal is preprocessed by a signal processing algorithm to remove noise and extract features; statistical analysis and regression models are used to analyze and predict data trends and identify potential fault modes; clustering and association rule analysis are used to classify and analyze the correlation of data to identify abnormal signals and fault modes; machine learning and deep learning techniques are used to perform pattern recognition and evaluation on abnormal data to generate diagnostic results and risk assessment data; data mining techniques are used to deeply analyze the prediction results, extract valuable information, and perform fault prediction and characterization; finally, a fault detection and response system is used to generate detailed fault alarm signals and processing suggestions. Combining a variety of data analysis and machine learning methods, comprehensive monitoring and intelligent diagnosis of the operating status of energy storage equipment is achieved, the accuracy and timeliness of fault detection are improved, the losses caused by equipment failures are reduced, and the reliability and operating efficiency of the system are improved.

[0112] Embodiment 3:

[0113] like Figure 2 As shown, the present application also provides a remote AI monitoring system 10 for photovoltaic energy storage equipment based on a management platform, including an acquisition unit 11, an analysis unit 12, a parsing unit 13, a comparison unit 14, an evaluation unit 15 and a generation unit 16.

[0114] The acquisition unit 11 is mainly used to acquire the operation signal of the energy storage device, analyze and process the operation signal, and obtain signal characteristic data and signal abnormality indicators.

[0115] The analysis unit 12 is mainly used to perform trend analysis on the signal feature data to obtain trend analysis results, and to verify the signal anomaly indicators using the trend analysis results to obtain anomaly verification results.

[0116] The analysis unit 13 is mainly used to input the abnormality verification result into the preset energy storage management system for analysis to obtain preliminary diagnosis information, analyze the preliminary diagnosis information to obtain diagnosis subcategory one and diagnosis subcategory two.

[0117] The comparison unit 14 is mainly used to perform pattern recognition on the diagnostic subcategory one using a machine learning model to obtain a diagnostic result, and to perform a comprehensive comparison between the diagnostic result and the diagnostic subcategory two to obtain a comparison result.

[0118] The evaluation unit 15 is mainly used to analyze the comparison results through data mining technology to obtain comprehensive analysis results, evaluate the operating status of the energy storage device based on the comprehensive analysis results, conduct risk assessment on the operating status, and obtain potential faults.

[0119] The generating unit 16 is used to generate a corresponding fault alarm signal according to the potential fault, and generate a corresponding fault handling suggestion based on the fault alarm signal.

[0120] In this embodiment, the operation signal of the energy storage device is acquired in real time by combining the sensor array, and the advanced signal processing algorithm is used to filter noise and extract features to ensure the accuracy and reliability of the data. The acquisition unit 11 transmits the processed signal data to the analysis unit 12 for time series analysis and frequency analysis to identify trends and seasonal changes. The parsing unit 13 further analyzes the abnormal signal in detail, generates diagnostic information, and refines it into specific diagnostic subcategories. The comparison unit 14 uses a machine learning model to perform pattern matching and logical comparison on the diagnostic results to verify their accuracy. The evaluation unit 15 applies data mining technology to perform in-depth analysis of the comparison results, identify potential risks and generate a comprehensive evaluation report. The generation unit 16 generates corresponding fault alarm signals and processing suggestions based on the risk assessment results to ensure early warning and efficient response to faults. Through this systematic AI monitoring, all-round monitoring and intelligent diagnosis of photovoltaic energy storage equipment are achieved, and the safety and stability of equipment operation are improved.

[0121] It should be noted that technicians in the relevant technical field can clearly understand that for the convenience and conciseness of description, the specific working process of the system and each unit described above can refer to the corresponding process in the embodiment of the remote AI monitoring method for photovoltaic energy storage equipment based on the management platform in the aforementioned embodiment 1, and will not be repeated here.

[0122] Embodiment 4:

[0123] like Figure 3 As shown, the present application also provides an electronic device 20, including a memory 21 and a processor 22, the memory 21 stores a computer program that can be run on the processor 22, and when the processor 22 executes the computer program, the remote AI monitoring method of photovoltaic energy storage equipment based on the management platform of Example 1 is implemented.

[0124] In this embodiment, by integrating efficient storage and processing units in the electronic device 20 and cooperating with advanced computer programs, remote AI monitoring of energy storage equipment is achieved. The computer program built into the memory 21 contains multiple algorithm modules, including signal processing, data analysis, machine learning, and risk assessment. When executing the computer program, the processor 22 can acquire and process the operating signals of the energy storage device in real time, and generate detailed diagnostic reports and fault warnings. Through this integrated design, the electronic device 20 can not only realize remote monitoring, but also provide immediate processing suggestions under abnormal circumstances, which significantly improves the operating efficiency and reliability of the photovoltaic energy storage system.

[0125] Embodiment 5:

[0126] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor executes the remote AI monitoring method for photovoltaic energy storage equipment based on a management platform as in Example 1.

[0127] In this embodiment, the computer program pre-stored on the computer-readable storage medium enables the processor to efficiently execute the remote AI monitoring method of the photovoltaic energy storage device. The computer program covers multiple links such as signal acquisition, data processing, trend analysis, machine learning model application and fault assessment. When the processor runs the program, it can process the operating signals from the energy storage device in real time, and perform abnormality detection and diagnostic analysis. In this way, the computer-readable storage medium gives ordinary electronic devices powerful intelligent monitoring and diagnostic capabilities, so that the equipment can quickly identify and respond to potential faults when facing a complex operating environment, thereby improving the overall stability and safety of the system.

[0128] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A remote AI monitoring method for photovoltaic energy storage equipment based on a management platform, characterized in that: include: Acquire the operation signal of the energy storage device, analyze and process the operation signal, and obtain signal characteristic data and signal abnormality indicators: monitor the energy storage device in real time through the sensor array to obtain the operation signal, use the signal processing algorithm to filter out noise and extract features of the operation signal, and obtain time series data and frequency analysis data; Performing trend identification on the time series data according to a statistical analysis method to obtain trend data and seasonal components, and performing trend prediction on the trend data through a regression model to obtain predicted trend data; Performing type analysis on the seasonal component using a clustering algorithm to obtain type feature data, and performing correlation analysis on the predicted trend data and the type feature data according to a preset association rule to obtain signal feature data; Based on the machine learning method, the frequency analysis data is subjected to abnormal pattern recognition to obtain abnormal pattern 1 and abnormal pattern 2, and the abnormal pattern 1 is subjected to abnormal confirmation by the threshold judgment method to obtain confirmed abnormal data; the confirmed abnormal data and the abnormal pattern 2 are subjected to abnormal level assessment by the logistic regression model to obtain a signal abnormality index; Performing trend analysis on the signal characteristic data to obtain a trend analysis result, and using the trend analysis result to verify the signal abnormality indicator to obtain an abnormality verification result; Inputting the abnormality check result into a preset energy storage management system for analysis to obtain preliminary diagnostic information, parsing the preliminary diagnostic information to obtain diagnostic subcategory one and diagnostic subcategory two; Using a machine learning model to perform pattern recognition on the diagnostic subcategory one to obtain a diagnostic result, and performing a comprehensive comparison between the diagnostic result and the diagnostic subcategory two to obtain a comparison result; Analyze the comparison results by data mining technology to obtain comprehensive analysis results, evaluate the operating status of the energy storage device based on the comprehensive analysis results, perform risk assessment on the operating status, and obtain potential faults; A corresponding fault alarm signal is generated according to the potential fault, and a corresponding fault handling suggestion is generated based on the fault alarm signal.

2. The remote AI monitoring method for photovoltaic energy storage equipment based on a management platform according to claim 1 is characterized in that: The step of performing trend analysis on the signal characteristic data to obtain trend analysis results, and using the trend analysis results to verify the signal abnormality index to obtain abnormality verification results comprises: Smoothing the signal characteristic data using data smoothing technology to obtain smoothed data and residual data, and performing trend prediction on the smoothed data according to time series analysis to obtain predicted data and confidence intervals; Performing pattern recognition on the predicted data by pattern matching technology to obtain pattern recognition data, and correcting the residual data by using the pattern recognition data to obtain corrected data; Performing anomaly detection on the corrected data and the signal anomaly indicator according to a machine learning classification model to obtain a trend analysis result; The trend analysis result is evaluated by a decision tree model to obtain risk assessment data, and the signal abnormality indicator is abnormally checked according to the risk assessment data to obtain an abnormality check result.

3. The remote AI monitoring method for photovoltaic energy storage equipment based on a management platform according to claim 1 is characterized in that: The step of inputting the abnormality check result into a preset energy storage management system for analysis to obtain preliminary diagnostic information, parsing the preliminary diagnostic information to obtain diagnostic subcategory 1 and diagnostic subcategory 2 includes: Using fuzzy logic to match the abnormality check results, obtain matching results, input the matching results into a preset energy storage management system, in which the matching results are comprehensively evaluated through data association analysis to obtain evaluation results and risk prompts, and based on model reasoning, the evaluation results are predicted to obtain predicted fault information, and the predicted fault information and the risk prompts are classified into fault levels using a decision tree algorithm to obtain preliminary diagnosis information; The preliminary diagnostic information is deconstructed through logical analysis to obtain structured data and unstructured data. The structured data is semantically parsed using text analysis technology to obtain diagnostic subcategory one. The unstructured data is content extracted using pattern recognition technology to obtain diagnostic subcategory two.

4. The remote AI monitoring method for photovoltaic energy storage equipment based on a management platform according to claim 1 is characterized in that: The step of using a machine learning model to perform pattern recognition on the diagnostic subcategory 1 to obtain a diagnostic result, and performing a comprehensive comparison between the diagnostic result and the diagnostic subcategory 2 to obtain a comparison result comprises: Performing deep learning analysis on the diagnostic subcategory one through a neural network model to obtain a feature vector set, a probability score, and a classification result, performing feature optimization on the feature vector set to obtain an optimized feature vector, and inputting the optimized feature vector into a preset support vector machine model to perform boundary division on the probability score to obtain a division result; Using a preset machine learning model to perform pattern matching on the division result and the second diagnostic subcategory to obtain a matched diagnostic result; The diagnostic result and the second diagnostic subcategory are logically compared through decision analysis to obtain a comparison result.

5. The remote AI monitoring method for photovoltaic energy storage equipment based on a management platform according to claim 1 is characterized in that: The step of analyzing the comparison results by using data mining technology to obtain a comprehensive analysis result, evaluating the operating status of the energy storage device based on the comprehensive analysis result, performing risk assessment on the operating status, and obtaining potential faults includes: Performing correlation detection on the comparison results according to association rule analysis to obtain correlation data and independence data, and classifying the correlation data into types through cluster analysis to obtain type clustering results; The type clustering result is used as an independent variable, and the independent data is used as a dependent variable to input into a preset regression model for prediction to obtain a prediction result; Perform in-depth mining on the prediction results through data mining technology to obtain comprehensive analysis results; The comprehensive analysis results are evaluated for risk using a probability statistics method to obtain risk scores and early warning indicators, and the risk scores and early warning indicators are predicted for failures using an early warning algorithm to obtain potential failures.

6. The remote AI monitoring method for photovoltaic energy storage equipment based on a management platform according to claim 1 is characterized in that: The step of generating a corresponding fault alarm signal according to the potential fault, and generating a corresponding fault handling suggestion based on the fault alarm signal, comprises: The potential fault is characterized by a fault detection algorithm to obtain fault type data, a rule engine is used to perform logical judgment on the fault type data to obtain a judgment result, and the judgment result is input into a preset fault processing system to obtain a corresponding fault alarm signal; The fault alarm signal is input into a preset fault response system to generate corresponding fault handling suggestions.

7. A remote AI monitoring system for photovoltaic energy storage equipment based on a management platform, characterized in that: include: An acquisition unit is used to acquire an operation signal of an energy storage device, analyze and process the operation signal, and obtain signal characteristic data and signal abnormality indicators: the energy storage device is monitored in real time by a sensor array to obtain an operation signal, and the operation signal is subjected to noise filtering and feature extraction by a signal processing algorithm to obtain time series data and frequency analysis data; Performing trend identification on the time series data according to a statistical analysis method to obtain trend data and seasonal components, and performing trend prediction on the trend data through a regression model to obtain predicted trend data; Performing type analysis on the seasonal component using a clustering algorithm to obtain type feature data, and performing correlation analysis on the predicted trend data and the type feature data according to a preset association rule to obtain signal feature data; Based on the machine learning method, the frequency analysis data is subjected to abnormal pattern recognition to obtain abnormal pattern 1 and abnormal pattern 2, and the abnormal pattern 1 is subjected to abnormal confirmation by the threshold judgment method to obtain confirmed abnormal data; the confirmed abnormal data and the abnormal pattern 2 are subjected to abnormal level assessment by the logistic regression model to obtain a signal abnormality index; An analysis unit, configured to perform a trend analysis on the signal feature data to obtain a trend analysis result, and to verify the signal abnormality indicator using the trend analysis result to obtain an abnormality verification result; A parsing unit, configured to input the abnormality check result into a preset energy storage management system for analysis to obtain preliminary diagnostic information, and parse the preliminary diagnostic information to obtain a diagnostic subcategory 1 and a diagnostic subcategory 2; a comparison unit, configured to perform pattern recognition on the diagnosis subcategory 1 by using a machine learning model to obtain a diagnosis result, and to perform a comprehensive comparison between the diagnosis result and the diagnosis subcategory 2 to obtain a comparison result; An evaluation unit, configured to analyze the comparison result by using data mining technology to obtain a comprehensive analysis result, evaluate the operating state of the energy storage device based on the comprehensive analysis result, perform risk assessment on the operating state, and obtain potential faults; A generating unit is used to generate a corresponding fault alarm signal according to the potential fault, and generate a corresponding fault handling suggestion based on the fault alarm signal.

8. An electronic device, characterized in that: It includes a memory and a processor, the memory stores a computer program that can be run on the processor, and the processor implements the remote AI monitoring method for photovoltaic energy storage equipment based on a management platform as described in any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the processor executes the remote AI monitoring method for photovoltaic energy storage equipment based on a management platform as described in any one of claims 1 to 6.

Images