Anomaly detection and method for photovoltaic substation

Abstract

The invention relates to an anomaly detection system and method for a photovoltaic substation, and the system comprises a photovoltaic power generation information collection module which is used for obtaining power generation parameter information and outputting the power generation parameter information to a convergence state monitoring module, and a support state monitoring module which is used for collecting the state information of a support structure and transmitting the state information to a data processing module. The convergence state monitoring module carries out convergence analysis on power generation parameters and outputs operation summary parameters and convergence state information, the DC-AC conversion state monitoring module carries out power stability and frequency characteristic analysis on the operation summary parameters and generates inversion operation state information, and the boost equipment operation state monitoring module collects equipment operation states. The data processing module fuses various state information, completes comprehensive analysis and generates a fault judgment result, the communication interaction module sends the judgment result to a remote platform, and the alarm module triggers audible and visual alarm when abnormal conditions are met. The method has the effect of realizing accurate identification and alarm response of fault features.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power station monitoring, and in particular to an abnormality detection method for a photovoltaic substation. Background Art

[0002] At PV substations, simple measurement tools and visual inspections are often used to detect abnormalities. One method involves inspecting the appearance and operating status of equipment, such as heat generation, to check for obvious damage or deformation. On the other hand, some monitoring systems that simply check whether PV mount parameters are tilted or cables are broken are also widely used. This method relies on the experience and responsibility of inspectors to monitor a specific parameter in real time, such as voltage or current. When that parameter exceeds the preset range, an alarm is issued. However, due to the large area and numerous equipment, manual inspection and fault diagnosis methods, which rely on experience, not only consume a lot of work experience, manpower, and time, but also make it difficult for maintenance personnel to monitor equipment in real time. This can lead to potential abnormalities not being detected in a timely manner. Summary of the Invention

[0003] In order to achieve accurate identification of fault characteristics and alarm response, the present application provides an abnormality detection method for a photovoltaic substation.

[0004] The above-mentioned invention objective of this application is achieved through the following technical solutions: A photovoltaic substation anomaly detection system includes a photovoltaic power generation information acquisition module, a bracket status monitoring module, a confluence status monitoring module, a DC / AC conversion status monitoring module, a boost equipment operation status monitoring module, a data processing module, a communication interaction module, and an alarm module, wherein: The photovoltaic power generation information acquisition module is used to obtain power generation parameter information and output the power generation parameter information to the confluence status monitoring module; The support state monitoring module is used to collect support structure state information and output the support structure state information to the data processing module; The confluence state monitoring module is used to receive the power generation parameter information, perform convergence analysis on the power generation parameter information, and obtain operation summary parameter information. The confluence state monitoring module is also used to collect confluence state information and output the confluence state information to the data processing module; The DC-AC conversion state monitoring module is connected to the bus state monitoring module, and the DC-AC conversion state monitoring module performs power stability analysis and frequency feature extraction during the power conversion process on the operation summary parameter information to obtain inverter operation state information, and outputs the inverter operation state information to the data processing module; The boost equipment operation status monitoring module is used to collect boost equipment status information and output the boost equipment status information to the data processing module; The data processing module is used to integrate the support structure status information, the confluence status information, the inverter operation status information and the boost device status information, perform comprehensive analysis, generate a fault determination result, and output the fault determination result to the communication interaction module and the alarm module; The communication interaction module is used to send the fault determination result to the remote monitoring platform; The alarm module is used to determine when the fault determination result meets the preset abnormal condition and trigger a local sound and light alarm.

[0005] By adopting the above technical solution, it is possible to centrally collect and integrate information on the support structure status, convergence status, inverter operation status, and booster equipment status in rooftop photovoltaic substations, thereby achieving unified analysis and fault identification of multi-source status information. This overcomes the existing problems of relying on manual inspections or single monitoring methods, which lead to delayed detection of abnormal conditions and incomplete diagnostic evidence. Specifically, by introducing modules such as photovoltaic module information collection, support tilt and image recognition, grounding and current balance analysis, power frequency feature extraction, and booster equipment thermal, arc, and winding status monitoring, the system can achieve comprehensive monitoring of various operating conditions within a single platform. Combined with the time alignment, feature fusion, and rule-based judgment processes set in the data processing module, it completes collaborative diagnosis and fault determination of multi-module status data. Furthermore, through the linkage mechanism of the communication interaction module and the alarm module, fault information is promptly synchronized to the remote platform and on-site audio and visual alarms are triggered. This significantly improves the intelligent management level and fault response efficiency of photovoltaic substations, effectively solving the existing anomaly detection problems that rely on experience, have delayed response, and have insufficient coverage.

[0006] In a preferred example, the present application can be further configured as follows: the photovoltaic power generation information acquisition module includes a photovoltaic component unit and a DC acquisition unit, the photovoltaic component unit is used to convert solar energy into DC electricity, and the DC acquisition unit is used to collect and analyze the DC electricity to obtain the power generation parameter information.

[0007] By adopting the above technical solution, real-time perception and accurate quantification of the power generation status of photovoltaic modules can be achieved in photovoltaic substations. The photovoltaic module unit efficiently converts solar energy into DC electricity, and the DC acquisition unit collects and analyzes multi-dimensional parameters of the generated electricity, thereby obtaining power generation parameter information including voltage, current, power, etc., ensuring that subsequent convergence status analysis and anomaly detection have an accurate and comprehensive data foundation, avoiding the problems of large delays and high errors in traditional manual recording methods, and improving the real-time and reliability of power generation unit working status monitoring.

[0008] In a preferred example, the present application can be further configured as follows: the bracket status monitoring module includes a tilt sensing unit and an image acquisition unit, the tilt sensing unit is used to obtain the tilt angle information of the bracket, the image acquisition unit is used to acquire image information of the bracket surface, and the bracket status monitoring module is used to perform correlation matching analysis on the tilt angle information and the image information within a corresponding time window to obtain the bracket structure status information, and output the bracket structure status information to the data processing module.

[0009] By adopting the above technical solution, multi-dimensional perception and intelligent identification of the structural status of rooftop photovoltaic brackets can be achieved. The tilt sensing unit monitors the angle changes of the bracket in real time, and the image acquisition unit obtains visual information such as bracket surface deformation, corrosion, and loose connections. By jointly analyzing the tilt angle information and image information within the same time window, it is possible to accurately determine whether the bracket has potential risks such as deformation, displacement or structural looseness, avoiding the problem of missed detection or misjudgment by a single sensing method, providing key structural status data support for subsequent fault diagnosis, and improving the coverage and accuracy of anomaly detection.

[0010] In a preferred example, the present application can be further configured as follows: the convergence status monitoring module includes a multi-channel electrical parameter acquisition unit and a grounding status detection unit, the multi-channel electrical parameter acquisition unit is used to collect the power generation parameter information from the each photovoltaic power generation information acquisition module, the grounding status detection unit is used to detect the grounding resistance value and insulation status in the convergence loop to obtain the grounding status detection result, the convergence status monitoring module is used to compare and analyze the power generation parameter information and judge the current balance, and perform a comprehensive evaluation based on the grounding status detection result to obtain the operation summary parameter information and the convergence status information, the operation summary parameter information is output to the DC-AC conversion status monitoring module, and the convergence status information is output to the data processing module.

[0011] By adopting the above technical solution, it is possible to realize the centralized collection and intelligent analysis of the electrical parameters of multiple strings in the photovoltaic bus link. The multi-channel electrical parameter collection unit performs a horizontal comparison of the power generation parameters of each channel, identifies the output deviation between the strings, and combines the grounding resistance value and insulation status information obtained by the grounding status detection unit to make a comprehensive judgment on possible abnormal situations such as grounding faults, loose wiring or component mismatch, thereby obtaining true and reliable operation summary parameter information and bus status information, improving the comprehensiveness of fault warning and the timeliness of response, and avoiding missed detection or false alarm of problems caused by failure of a single path.

[0012] In a preferred example, the present application can be further configured as follows: the boost equipment operation status monitoring module includes a temperature rise detection unit, an arc monitoring unit and a winding status evaluation unit, the temperature rise detection unit is used to collect temperature change information of the boost equipment under different load conditions, the arc monitoring unit is used to detect arc characteristic information generated by the boost equipment during operation, the winding status evaluation unit is used to monitor the winding status information of the transformer winding, and the boost equipment operation status monitoring module is used to comprehensively analyze the temperature change information, the arc characteristic information and the winding status information to obtain the boost equipment status information, and output the boost equipment status information to the data processing module.

[0013] By adopting the above technical solution, multi-dimensional real-time monitoring of the operating status of boost equipment in photovoltaic substations is possible. The temperature rise detection unit dynamically tracks temperature trends based on equipment load conditions, helping to promptly identify overload or heat dissipation anomalies. The arc monitoring unit captures abnormal discharge characteristics generated during operation to assist in identifying potential hazards such as poor contact and insulation degradation. The winding status assessment unit determines the risk of winding aging or local damage based on changes in winding electrical parameters. Through comprehensive analysis of these multiple types of monitoring information, the accuracy and predictive ability of boost equipment abnormality identification are effectively improved, preventing the impact of sudden equipment failures on the safe operation of photovoltaic substations.

[0014] In a preferred example, the present application can be further configured as follows: the data processing module includes a data receiving unit, a state fusion analysis unit and a fault judgment unit, the data receiving unit is used to receive various types of status information from the bracket status monitoring module, the convergence status monitoring module, the DC-AC conversion status monitoring module and the boost equipment operation status monitoring module, the state fusion analysis unit is used to perform time alignment, weight integration and feature fusion processing on the various types of status information to obtain a fusion result, the fault judgment unit is used to identify potential abnormal operating conditions based on the fusion result, and generate the fault judgment result, and the fault judgment result is output to the communication interaction module and the alarm module.

[0015] By adopting the above technical solution, it is possible to achieve unified integration and intelligent judgment of status information from multiple monitoring modules of a photovoltaic substation. The data receiving unit ensures the complete access of various status information, avoiding monitoring blind spots caused by data omissions or delays; the status fusion analysis unit aligns various status information based on a unified time base, highlights key features through a fusion weight mechanism, and improves information expression efficiency by combining feature compression and multi-dimensional modeling methods, which helps to construct more representative fusion results; the fault judgment unit accurately identifies abnormal states based on the fusion results, which can avoid the false alarm and omission problems existing in traditional judgment methods that rely on a single threshold, and realize comprehensive fault judgment across modules, multiple parameters, and multi-dimensional features, significantly improving the intelligence level and response time of photovoltaic substation anomaly detection.

[0016] In a preferred example, the present application can be further configured as follows: the communication interaction module includes a communication interface unit, a data encapsulation unit and a remote sending unit, the communication interface unit is used to establish a communication connection with the remote monitoring platform, the data encapsulation unit is used to format and encapsulate the fault judgment result to generate a standardized data message, and the remote sending unit is used to send the standardized data message to the remote monitoring platform through the communication interface unit.

[0017] By adopting the above technical solution, standardized output and remote real-time transmission of fault judgment results can be achieved. The communication interface unit ensures a stable connection between the photovoltaic substation monitoring system and the remote monitoring platform, and adapts to various communication protocols and network environments; the data encapsulation unit converts the fault judgment results into standardized data messages, enabling the remote platform to quickly parse and process them, which is conducive to unified data management and subsequent operation and maintenance response; the remote sending unit supports the efficient and reliable transmission of the encapsulated data to the remote monitoring platform, realizing remote perception and unified supervision of abnormal conditions of the substation, and improving the real-time and intelligent level of the overall monitoring system.

[0018] A photovoltaic substation anomaly detection method is applied to an anomaly detection system for a photovoltaic substation. The photovoltaic substation anomaly detection method includes: Acquiring the power generation parameter information; Performing string comparison analysis and current balance evaluation on the power generation parameter information to obtain the operation summary parameter information, and obtaining the convergence state information through the grounding state detection unit; Performing power stability analysis and frequency feature extraction on the operation summary parameter information to obtain the inverter operation status information; Acquiring the tilt angle information and the image information, performing correlation matching analysis on the tilt angle information and the image information, and obtaining the support structure status information; Acquiring the temperature change information, the discharge abnormality information, and the winding state information, and performing a comprehensive analysis on the temperature change information, the discharge abnormality information, and the winding state information to obtain the boost device state information; fusing the support structure status information, the confluence status information, the inverter operation status information, and the boost device status information to obtain a fusion result; generating the fault determination result based on the fusion result; The fault determination result is formatted and encapsulated to generate the standardized data message, and is sent to the remote monitoring platform. The fault determination result is compared with the preset abnormal condition. If the fault determination result meets the preset abnormal condition, the local sound and light alarm is triggered.

[0019] By adopting the above technical solutions, it is possible to build a multi-source state perception and fusion processing process for photovoltaic substation operation scenarios. On the basis of ensuring the comprehensive acquisition of status information of the support structure, electrical busbar, inverter operation and boost link, the time-series tracking and indicator extraction of the status of each key equipment can be realized. Furthermore, through the feature fusion mechanism under a unified time base, a correlation feature coupling model between modules is established, and combined with the abnormal rule comparison logic, potential fault situations can be efficiently identified, thereby effectively improving the accuracy of abnormality detection and timely response of photovoltaic substations under complex operating conditions, overcoming the limitations of traditional manual inspection methods such as lag, insufficient single-point monitoring coverage and reliance on experience-based judgment for abnormality identification, and contributing to the intelligent monitoring of the operating status of photovoltaic substations and fault warning management.

[0020] In a preferred example, the present application may be further configured as follows: the fusion of the support structure status information, the confluence status information, the inverter operation status information, and the boost device status information to obtain a fusion result includes: Performing data alignment processing on the support structure status information, the confluence status information, the inverter operation status information, and the boost device status information under a unified time reference to construct a time series status data set including status indicators of each module; Based on the time series state data set, the change amplitude, change rate and historical deviation trend of various state factors in the current time window are extracted to form a preliminary feature indicator set; According to the fluctuation characteristics and fault sensitivity of the various state factors in the preliminary characteristic indicator set, a fusion weight factor is assigned to construct a weighted characteristic matrix; extracting, based on the weighted feature matrix, a change consistency feature between the tilt angle information and the image information to generate a stent structure feature subset; Based on the weighted feature matrix, the string deviation index and the current balance index of each current are extracted to construct a convergence state feature subset; Based on the weighted feature matrix, extracting the frequency fluctuation index and the power stability offset index, and constructing the inverter operation feature subset; Based on the weighted feature matrix, extracting temperature rise abnormality index, discharge abnormality index and winding state abnormality index, and constructing a boost equipment feature subset; The support structure feature subset, the confluence state feature subset, the inverter operation feature subset and the boost equipment feature subset are input into a multi-dimensional state fusion model, feature compression and abnormal trend modeling are performed, and the fusion result is output.

[0021] By adopting the above technical solution, the operating status data of each key module of the photovoltaic substation can be aligned and normalized on a unified time scale, improving the accuracy and temporal consistency of cross-module data fusion. Furthermore, through the dynamic fluctuation characteristics and sensitivity assessment of characteristic factors, weight allocation and weighted fusion are achieved, extracting highly recognizable feature subsets reflecting the operating status of various equipment types, and constructing a state feature representation with clear structure and semantics. On this basis, multiple subsets are input into a multidimensional fusion model for feature compression and abnormal trend modeling. This can effectively capture the implicit coupling relationships and potential abnormal evolution patterns between different modules, thereby improving the detection sensitivity and model generalization capabilities of complex fault situations, and realizing intelligent identification and accurate early warning of abnormal conditions in photovoltaic substations.

[0022] In a preferred example, the present application may be further configured as follows: generating the fault determination result based on the fusion result further includes: Based on the fusion results, a cross-module consistency comparison is performed on the key state factors extracted from each feature subset to identify the factor combinations with abnormal state features and construct a set of candidate abnormal factors; Performing rule matching processing on the candidate abnormal factor set, searching for fault feature patterns that meet the conditions according to a preset multi-dimensional abnormality judgment rule table, and obtaining a set of fault rules that successfully matches; If the fault rule set contains multiple matching results, priority evaluation and confidence scoring are performed on the fault scenarios corresponding to each matching rule, and weighted scoring is performed in combination with the matching degree of the current state factor characteristics to filter out the fault judgment result.

[0023] By adopting this technical solution, key status factors in different modules can be horizontally compared based on the fusion results. The system can then identify possible collaborative abnormal behaviors across modules, construct factor combinations with high abnormal correlation, and further effectively map them to corresponding fault characteristic patterns by matching a preset multi-dimensional judgment rule table, thus achieving structured recognition of abnormal patterns. When multiple fault rules are met simultaneously, by introducing a priority assessment and confidence scoring mechanism, and combining the actual matching degree of the status factor characteristics to complete a weighted judgment, the accuracy and robustness of fault type screening can be improved, thereby generating more accurate and reliable fault judgment results, and enhancing the decision-making ability and response efficiency of photovoltaic substation anomaly detection.

[0024] In summary, this application includes at least one of the following beneficial technical effects: 1. It can centrally collect and integrate the support structure status information, convergence status information, inverter operation status information and boost equipment status information in the rooftop photovoltaic substation, thereby realizing unified analysis and fault identification of multi-source status information, breaking through the problem of relying on manual inspections or single monitoring methods in existing technologies, resulting in delayed detection of abnormal situations and incomplete diagnosis basis. Specifically, by introducing modules such as photovoltaic module information collection, support tilt and image recognition, grounding and current balance analysis, power frequency feature extraction, boost equipment thermal, arc and winding status monitoring, the system can realize comprehensive monitoring of various operating states within a single platform. Combined with the time alignment, feature fusion and rule judgment process set in the data processing module, it can complete the collaborative diagnosis and fault judgment of multi-module status data; then, with the help of the linkage mechanism of the communication interaction module and the alarm module, it can synchronize fault information to the remote platform in a timely manner and trigger on-site sound and light alarms, significantly improving the intelligent management level and fault response efficiency of photovoltaic substations, and effectively solving the existing abnormality detection problems that rely on experience, response lag and insufficient coverage. 2. It can build a multi-source state perception and fusion processing process for photovoltaic substation operation scenarios. On the basis of ensuring the comprehensive acquisition of status information of the support structure, electrical busbar, inverter operation and boost link, it can realize the time-series tracking and indicator extraction of the status of each key equipment. Furthermore, through the feature fusion mechanism under the unified time base, it establishes the correlation feature coupling model between modules, and combines the abnormal rule comparison logic to efficiently identify potential fault situations, thereby effectively improving the anomaly detection accuracy and response timeliness of photovoltaic substations under complex operating conditions. It overcomes the limitations of traditional manual inspection methods such as lag, insufficient single-point monitoring coverage and reliance on experience-based judgment for anomaly identification, and helps to realize intelligent monitoring of the operating status of photovoltaic substations and fault early warning management. 3. Based on the fusion results, the system can perform a horizontal comparison of key status factors in different modules, systematically identify possible collaborative abnormal behaviors across modules, construct factor combinations with high abnormal correlation, and further effectively map them to corresponding fault characteristic patterns by matching the preset multi-dimensional judgment rule table, thus achieving structured recognition of abnormal patterns. When multiple fault rules are met simultaneously, by introducing a priority assessment and confidence scoring mechanism, combined with the actual matching degree of the status factor characteristics to complete weighted judgment, the accuracy and robustness of fault type screening can be improved, thereby generating more accurate and reliable fault judgment results, and enhancing the decision-making ability and response efficiency of photovoltaic substation anomaly detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a flow chart of an abnormality detection system for a photovoltaic substation in one embodiment of the present application.

[0026] Figure 2 This is a schematic diagram of the specific structure of a photovoltaic power generation information collection module in an abnormality detection system of a photovoltaic substation in one embodiment of the present application; Figure 3 This is a schematic diagram of the specific structure of a bracket status monitoring module in an abnormality detection system of a photovoltaic substation in one embodiment of the present application; Figure 4 This is a schematic diagram of the specific structure of a confluence state monitoring module in an abnormality detection system of a photovoltaic substation in one embodiment of the present application; Figure 5 This is a schematic diagram of the specific structure of a boost equipment operating status monitoring module in an abnormality detection system of a photovoltaic substation in one embodiment of the present application; Figure 6 This is a schematic diagram of the specific structure of a data processing module in an abnormality detection system of a photovoltaic substation in one embodiment of the present application; Figure 7 This is a schematic diagram of the specific structure of a communication interaction module in an abnormality detection system of a photovoltaic substation in one embodiment of the present application; Figure 8 This is a flow chart of a method for detecting anomalies in a photovoltaic substation according to an embodiment of the present application; Figure 9 This is a flowchart for implementing step S60 in an abnormality detection system for a photovoltaic substation in one embodiment of the present application; Figure 10 This is another implementation flowchart of step S70 in an abnormality detection system for a photovoltaic substation in one embodiment of the present application.

[0027] Description of reference numerals: 1. Photovoltaic power generation information acquisition module; 2. Bracket status monitoring module; 3. Confluence status monitoring module; 4. DC-AC conversion status monitoring module; 5. Booster equipment operation status monitoring module; 6. Data processing module; 7. Communication interaction module; 8. Alarm module; 11. Photovoltaic module unit; 12. DC acquisition unit; 21. Tilt sensor unit; 22. Image acquisition unit; 31. Multi-channel electrical parameter acquisition unit; 32. Grounding status detection unit; 51. Temperature rise detection unit; 52. Arc monitoring unit; 53. Winding status evaluation unit; 61. Data receiving unit; 62. Status fusion analysis unit; 63. Fault judgment unit; 71. Communication interface unit; 72. Data encapsulation unit; 73. Remote transmission unit. DETAILED DESCRIPTION

[0028] The present application is further described in detail below with reference to the accompanying drawings.

[0029] In one embodiment, if Figure 1 As shown, the present application discloses an abnormality detection system for a photovoltaic substation, including a photovoltaic power generation information acquisition module 1, a bracket status monitoring module 2, a convergence status monitoring module 3, a DC-AC conversion status monitoring module 4, a boost equipment operation status monitoring module 5, a data processing module 6, a communication interaction module 7 and an alarm module 8, wherein: The photovoltaic power generation information acquisition module 1 is used to obtain power generation parameter information and output the power generation parameter information to the confluence status monitoring module 3; The support state monitoring module 2 is used to collect support structure state information and output the support structure state information to the data processing module 6; The confluence state monitoring module 3 is used to receive power generation parameter information, perform convergence analysis on the power generation parameter information, and obtain operation summary parameter information. The confluence state monitoring module 3 is also used to collect confluence state information and output the confluence state information to the data processing module 6; The DC-AC conversion state monitoring module 4 is connected to the confluence state monitoring module 3. The DC-AC conversion state monitoring module 4 performs power stability analysis and frequency feature extraction during the power conversion process on the operation summary parameter information to obtain inverter operation state information, and outputs the inverter operation state information to the data processing module 6; The boost equipment operation status monitoring module 5 is used to collect boost equipment status information and output the boost equipment status information to the data processing module 6; The data processing module 6 is used to integrate the support structure status information, the confluence status information, the inverter operation status information and the boost equipment status information, perform comprehensive analysis, generate a fault determination result, and output the fault determination result to the communication interaction module 7 and the alarm module 8; The communication interaction module 7 is used to send the fault determination result to the remote monitoring platform; The alarm module 8 is used to determine when the fault determination result meets the preset abnormal conditions and trigger a local sound and light alarm.

[0030] Specifically, the photovoltaic power generation information acquisition module is used to obtain power generation parameter information, adopts photovoltaic module units to convert solar energy into DC power, and collects and analyzes DC power through DC acquisition units, obtains power generation parameter information of each group including voltage, current, power, etc. and outputs it to the convergence status monitoring module, and the bracket status monitoring module is used to collect bracket structure status information, including obtaining the inclination angle information of the bracket through the inclination sensor unit, collecting image information of the bracket surface through the image acquisition unit, and correlating and matching the inclination angle information with the image information in the same time window, identifying structural looseness or deflection, obtaining bracket structure status information and outputting it to the data processing module, and the convergence status monitoring module is used to receive the aforementioned power generation parameters. Information, using multiple electrical parameter acquisition units to perform group comparison and current balance evaluation on the output data of each string, identify abnormal conditions such as a current that is significantly smaller, and combine the grounding resistance value and insulation status collected by the grounding status detection unit to conduct a comprehensive evaluation of the bus circuit safety, and obtain operation summary parameter information and bus status information, wherein the operation summary parameter information is output to the DC-AC conversion status monitoring module for further analysis of the power conversion characteristics. The DC-AC conversion status monitoring module performs power stability analysis and frequency feature extraction on the operation summary parameter information, identifies operation status problems such as abnormal inverter output frequency fluctuation or power imbalance, and outputs the inverter operation status information to the data processing module. The boost equipment operation status monitoring module is used The data processing module is used to collect the status information of the boost equipment, including monitoring the temperature change information of the transformer and reactor under load changes through the temperature rise detection unit, detecting arc discharge, breakdown and other phenomena through the discharge abnormality monitoring unit, detecting the transformer winding status through the winding status evaluation unit, identifying abnormalities such as local overheating and winding aging and outputting them to the data processing module. The data processing module is used to integrate the above-mentioned support structure status information, confluence status information, inverter operation status information and boost equipment status information, perform data alignment under a unified time base, extract the characteristic indicators of each state factor through the change amplitude, rate and historical offset trend, construct a preliminary feature set, and configure the fusion weight according to the fluctuation characteristics and fault sensitivity of each factor to generate a weighted feature matrix. In one step, the consistency features between the bracket angle change and image recognition, current deviation features, frequency fluctuation features and temperature rise / discharge abnormality indicators are extracted to form four types of feature subsets, which are input into the multi-dimensional state fusion model to perform abnormal trend modeling and compression analysis, and the fusion results are output. The fault judgment result is generated by the fault judgment unit, including performing cross-module consistency comparison on the key state factors in each feature subset based on the fusion results, identifying abnormal factor combinations and constructing a candidate abnormal factor set, and then performing rule matching based on the preset multi-dimensional abnormal judgment rule table. If the match is successful, the fault rule set is obtained. If there are multiple rule triggering results, priority evaluation and confidence scoring are performed on each corresponding fault scenario, and weighted matching is performed in combination with the current state factor features.Ultimately, the fault type that best matches the current operating status is selected as the fault determination result. This fault determination result is output to the communication interaction module, where it is packaged and standardized, then sent to the remote monitoring platform for centralized supervision. It is also synchronously transmitted to the alarm module for comparison with preset abnormal conditions. If the trigger conditions are met, the local audio and visual alarm module immediately activates an indication prompt. For example, if a tilted support is accompanied by local overheating and power instability, the corresponding structural deformation + load abnormality scenario in the multidimensional rule table is triggered, and a high-priority alarm is output to assist operation and maintenance personnel in promptly identifying potential hazards on site and addressing them.

[0031] Furthermore, the remote monitoring platform refers to an information management platform deployed in the cloud or centralized control center, which is used to receive standardized fault judgment messages sent by the communication interaction module. By parsing the message content, it can realize real-time monitoring of the operating status of various equipment in the photovoltaic substation, centralized display of abnormal information and archiving management of historical data. The remote monitoring platform can also generate maintenance scheduling instructions based on the received fault judgment results, and send them to on-site operation and maintenance personnel or third-party service systems through network communication, so as to realize closed-loop management of response to abnormal events.

[0032] Further, if Figure 2 As shown, the photovoltaic power generation information collection module 1 includes a photovoltaic component unit 11 and a DC collection unit 12. The photovoltaic component unit 11 is used to convert solar energy into DC power, and the DC collection unit 12 is used to collect and analyze DC power to obtain power generation parameter information.

[0033] Specifically, the photovoltaic module unit adopts a module structure adapted to the roof installation environment, and converts the solar radiation received on the roof surface into DC power through the principle of photoelectric conversion. The photovoltaic module unit can be installed on a sloping roof or a flat roof support structure; the DC acquisition unit is used to collect and process the DC power output by the photovoltaic module unit, specifically including collecting the voltage, current and power information of each module branch, and forming power generation parameter information, wherein the voltage information is obtained in real time through the voltage sampling circuit, the current information is obtained through the current sensor installed in the module series path, and the power information is obtained through voltage and current calculation; the DC acquisition unit further normalizes the electrical parameters of each branch to eliminate the initial deviation caused by the difference in roof lighting angle, improve the accuracy of the power generation status assessment in the subsequent analysis, and finally output the power generation parameter information to the confluence status monitoring module.

[0034] Further, if Figure 3As shown, the bracket status monitoring module 2 includes a tilt sensing unit 21 and an image acquisition unit 22. The tilt sensing unit 21 is used to obtain the tilt angle information of the bracket, and the image acquisition unit 22 is used to acquire image information of the bracket surface. The bracket status monitoring module 2 is used to perform correlation matching analysis on the tilt angle information and the image information within the corresponding time window to obtain the bracket structure status information, and output the bracket structure status information to the data processing module 6.

[0035] Specifically, the tilt sensing unit is installed at the key stress-bearing part of the rooftop photovoltaic bracket, and obtains the real-time tilt angle information of each bracket node through the principle of gravity acceleration sensing, and continuously records the angle change trend at a set time interval to characterize the stability of the bracket structure and the force offset state; the image acquisition unit is fixed in an appropriate position, covering the main bracket surface area with a visual range, and obtains the current image information of the bracket by setting a shooting cycle. The image information includes possible bending, cracks or deformation features on the bracket surface; the bracket status monitoring module performs correlation matching analysis on the tilt angle information from the tilt sensing unit and the image information obtained by the image acquisition unit within a set time window, specifically including feature comparison of the angle change amplitude with the structural deformation pattern of the corresponding area in the image, and combining the structural abnormal area marked by the image recognition algorithm to determine whether the current bracket has abnormal tilt, structural looseness or force imbalance, etc., and finally generates the bracket structure status information and outputs it to the data processing module for subsequent fusion analysis.

[0036] Further, if Figure 4 As shown, the convergence state monitoring module 3 includes a multi-channel electrical parameter acquisition unit 31 and a grounding state detection unit 32. The multi-channel electrical parameter acquisition unit 31 is used to collect power generation parameter information from each photovoltaic power generation information acquisition module 1. The grounding state detection unit 32 is used to detect the grounding resistance value and insulation state in the convergence loop to obtain the grounding state detection result. The convergence state monitoring module 3 is used to compare and analyze the power generation parameter information and judge the current balance, and perform a comprehensive evaluation based on the grounding state detection result to obtain operation summary parameter information and convergence state information. The operation summary parameter information is output to the DC-AC conversion state monitoring module 4, and the convergence state information is output to the data processing module 6.

[0037] Specifically, the multi-channel electrical parameter acquisition unit 31 acquires the voltage, current, and power parameters output by each photovoltaic string in real time through a wired connection with each photovoltaic power generation information acquisition module 1, records and classifies the above parameter information at preset time intervals, and forms a comparative data set reflecting the power generation status of each string. The module calculates the string deviation index based on the degree of current difference between the strings, further determines the balance of current distribution, and identifies whether there is abnormal string deviation caused by problems such as component aging, shading, or loose wiring. At the same time, the grounding status detection unit 32 uses the grounding loop as a monitoring channel to measure the grounding resistance value of the busbar and analyzes the changes in the insulation performance between the photovoltaic module and the ground in combination with the insulation resistance measurement device. If the grounding resistance exceeds the set threshold or the insulation status shows a downward trend, it is determined that a grounding hazard exists. Finally, the busbar status monitoring module integrates the string deviation index, current balance results, and grounding status detection results to generate operation summary parameter information and busbar status information reflecting the current busbar convergence quality and electrical safety of each branch. The operation summary parameter information is output to the DC-AC conversion status monitoring module 4, and the busbar status information is output to the data processing module 6.

[0038] Further, if Figure 5 As shown, the boost equipment operation status monitoring module 5 includes a temperature rise detection unit 51, an arc monitoring unit 52 and a winding status evaluation unit 53. The temperature rise detection unit 51 is used to collect temperature change information of the boost equipment under different load conditions. The arc monitoring unit 52 is used to detect arc characteristic information generated by the boost equipment during operation. The winding status evaluation unit 53 is used to monitor the winding status information of the transformer winding. The boost equipment operation status monitoring module 5 is used to comprehensively analyze the temperature change information, arc characteristic information and winding status information to obtain the boost equipment status information, and output the boost equipment status information to the data processing module 6.

[0039] Specifically, the temperature rise detection unit 51 is arranged at the key heating parts of the boost equipment, such as the transformer oil tank, high-voltage terminal and the radiator surface, through an attached or infrared non-contact temperature sensor, to collect temperature change information under different operating loads in real time, and evaluate the temperature rise trend and thermal stability by combining the load current and ambient temperature data; the arc monitoring unit 52 collects and analyzes the high-frequency spike signals generated during the operation of the boost equipment by arranging partial discharge sensors or high-frequency transient electromagnetic wave sensors, and extracts key indicators such as pulse amplitude, repetition frequency and duration representing the arc discharge characteristics, and uses them to The winding status assessment unit 53 uses winding reactance measurement, current harmonic analysis or loop impedance change to indirectly assess the structural integrity and insulation condition of the transformer winding to determine whether there are inter-turn short circuits, winding displacement or insulation aging problems; the boost equipment operation status monitoring module 5 integrates and comprehensively judges the above-mentioned temperature change information, arc characteristic information and winding status information to form boost equipment status information that can reflect the overall operation stability and safety risks of the boost equipment, and outputs it to the data processing module 6 for subsequent abnormality analysis.

[0040] Further, if Figure 6 As shown, the data processing module 6 includes a data receiving unit 61, a state fusion analysis unit 62 and a fault judgment unit 63. The data receiving unit 61 is used to receive various state information from the bracket state monitoring module 2, the convergence state monitoring module 3, the DC-AC conversion state monitoring module 4 and the boost equipment operation state monitoring module 5. The state fusion analysis unit 62 is used to perform time alignment, weight integration and feature fusion processing on various state information to obtain a fusion result. The fault judgment unit 63 is used to identify potential abnormal operating conditions based on the fusion result and generate a fault judgment result. The fault judgment result is output to the communication interaction module 7 and the alarm module 8.

[0041] Specifically, the data receiving unit 61 establishes a data exchange channel with the support state monitoring module 2, the confluence state monitoring module 3, the DC / AC conversion state monitoring module 4, and the boost equipment operation state monitoring module 5 via wired or wireless communication, and receives the uploaded state data according to a unified sampling period, ensuring that all types of state information are aligned under the same time reference. The state fusion analysis unit 62 first synchronizes the timestamps of the received support structure state information, confluence state information, inverter operation state information, and boost equipment state information to construct a time series state data set. It then assigns fusion weights based on the historical fluctuation amplitude, fault sensitivity, and data integrity of each state factor to form a weighted feature matrix. It then performs feature compression and dimensionality reduction operations to extract fusion feature indicators that can represent the trend of operation state changes and form a fusion result. The fault determination unit 63 performs rule library comparison or feature threshold analysis on the fusion result to identify whether there are typical abnormal features such as state mutation, trend reversal, or indicator limit violation, and then determines whether to enter the abnormality determination process. If the fault trigger condition is met, the fault determination result is output and simultaneously sent to the communication interaction module 7 and the alarm module 8 for remote alarm and local prompt.

[0042] Further, if Figure 7 As shown, the communication interaction module 7 includes a communication interface unit 71, a data encapsulation unit 72 and a remote sending unit 73. The communication interface unit 71 is used to establish a communication connection with the remote monitoring platform, the data encapsulation unit 72 is used to format and encapsulate the fault judgment results to generate standardized data messages, and the remote sending unit 73 is used to send the standardized data messages to the remote monitoring platform through the communication interface unit 71.

[0043] Specifically, the communication interface unit 71 uses an Ethernet interface, a 4G / 5G communication module or a LoRa wireless communication module to establish a stable data transmission channel with the remote monitoring platform, and supports multiple communication protocols such as MQTT, Modbus TCP or HTTP to achieve compatible access; after receiving the fault judgment result from the data processing module 6, the data encapsulation unit 72 performs field parsing and encoding of the fault type, timestamp, module source and key indicator value in the result according to the preset data structure template to form a standardized data message that meets the requirements of the communication protocol; after confirming that the communication link is stable, the remote sending unit 73 calls the communication interface unit 71 to establish a session or send an instruction to push the standardized data message to the designated remote monitoring platform node, and supports the breakpoint resumption mechanism and the sending status feedback mechanism to ensure the complete transmission of key alarm information and real-time reception by the remote platform.

[0044] like Figure 8As shown, a photovoltaic substation anomaly detection method is applied to an anomaly detection system for a photovoltaic substation, characterized in that the photovoltaic substation anomaly detection method includes: S10: Acquire power generation parameter information.

[0045] Specifically, the DC acquisition unit 12 in the photovoltaic power generation information acquisition module 1 performs real-time detection on the DC power output by the rooftop photovoltaic component unit 11 to obtain the voltage, current and power data of each group of photovoltaic components, wherein the voltage information reflects the working status of the photovoltaic panel, the current information indicates the power generation load capacity, and the power information is used to analyze the output efficiency. The acquisition process can be combined with a high-frequency sampling mechanism to improve data accuracy, and support multi-channel synchronous acquisition to ensure the time consistency of the data of each group string, thereby obtaining power generation parameter information for subsequent evaluation and convergence status analysis.

[0046] S20 : ​​performing string comparison analysis and current balance evaluation on the power generation parameter information to obtain operation summary parameter information, and obtaining convergence status information through the grounding status detection unit 32 .

[0047] Specifically, in the process of performing string comparison analysis and current balance evaluation on the power generation parameter information, the photovoltaic current information of each channel collected by the multi-channel electrical parameter acquisition unit 31 is divided into branches, and based on the string current data in the same bus channel, the deviation value, mean and standard deviation between the branch currents are calculated, and the balance of the string current is judged. When the deviation exceeds the preset balance threshold, it is marked as an abnormal branch. Combined with the voltage and power calculation results of each channel, the operation summary parameter information reflecting the overall output characteristics of the current power generation side is formed; at the same time, the grounding resistance value and insulation status data of the bus circuit in each sampling period are obtained through the grounding status detection unit 32, and the trend of grounding resistance change is judged. When it is detected that the resistance value continues to decrease or the insulation performance level is reduced, it is marked as a grounding hazard or insulation degradation risk, and the bus status information representing the bus safety status is generated.

[0048] S30: Perform power stability analysis and frequency feature extraction on the operation summary parameter information to obtain inverter operation status information.

[0049] Specifically, in the process of performing power stability analysis and frequency feature extraction on the operation summary parameter information, based on the current, voltage and power data of each time node in the operation summary parameter information, the power fluctuation rate per unit time is first calculated, and the stability of the DC input power before entering the inverter link is evaluated through the maximum power difference, average power change rate and fluctuation frequency in the sliding time window; then, based on the continuous power time series signal, the fast Fourier transform is used to perform spectral analysis on the power change trend, extract the main frequency components and their energy proportion, and identify whether there are non-steady-state oscillations or periodic fluctuations; combined with the power stability parameters and frequency analysis results, the inverter operation status information that can reflect the operating quality of the inverter input side is generated.

[0050] S40: Obtaining tilt angle information and image information, performing correlation matching analysis on the tilt angle information and the image information, and obtaining support structure status information.

[0051] Specifically, in the process of obtaining the tilt angle information and image information, the posture change trajectory of the current bracket within the monitoring period is obtained based on the tilt angle change value output in real time by the tilt sensing unit 21; at the same time, the image acquisition unit 22 is used to obtain the bracket surface image data under the corresponding timestamp, and the image edge extraction and structural contour recognition operations are performed to extract the position, shape and deformation characteristics of the key structural parts of the bracket; on this basis, the tilt angle information and image information are correlated and matched and analyzed, and through time synchronization correction and spatial consistency comparison, it is determined whether the tilt angle change trend is consistent with the structural deformation area. If there is an abnormal correspondence (such as a sudden change in the tilt angle but the image does not detect a structural offset, or the image detects deformation but the tilt angle is stable), it is marked as a structural abnormality feature; finally, the information reflecting the bracket stability and deformation risk in the matching analysis results is fused to generate the bracket structure status information.

[0052] S50: Acquire temperature change information, discharge abnormality information, and winding status information, and perform comprehensive analysis on the temperature change information, discharge abnormality information, and winding status information to obtain boosting device status information.

[0053] Specifically, in the process of obtaining temperature change information, the temperature data of the boost equipment under various operating load conditions is collected in real time through the temperature rise detection unit 51, and the temperature change rate is trend analyzed to determine whether there is abnormal temperature rise or overheating; in the process of obtaining discharge abnormality information, the arc monitoring unit 52 is used to monitor whether the boost equipment generates high-frequency arc signals or local discharge phenomena during operation, and the frequency and duration of abnormal discharge events are identified; in the process of obtaining winding status information, the winding status evaluation unit 53 is used to detect the electrical parameters and structural status of the transformer winding, such as winding resistance changes, insulation impedance fluctuations, etc., to analyze its operating reliability and aging trend; then, the above three types of information are aligned according to a unified time base, and feature cross-analysis and risk indicator quantification processing are performed to identify the coordinated change relationship between multi-source features, comprehensively evaluate the health level and potential risks of the boost equipment under the current operating state, and finally generate the boost equipment status information.

[0054] S60: Fusion of the support structure status information, the confluence status information, the inverter operation status information, and the boost device status information to obtain a fusion result.

[0055] Specifically, data alignment processing is performed on the support structure status information, confluence status information, inverter operation status information and boost equipment status information under a unified time base to construct a time series status data set containing the key status indicators of each module; based on the time series status data set, the change amplitude, change rate and historical offset trend of various status factors in the current monitoring period are extracted to generate a preliminary feature indicator set; according to the sensitivity and fluctuation characteristics of the status factors in the feature indicator set to fault signs, differentiated fusion weights are set to construct a weighted feature matrix; based on the weighted feature matrix, the consistency features of the support inclination angle and image texture change, the joint features of the current string deviation and the grounding state, the synchronous fluctuation features of power and frequency, and the composite abnormal features between temperature rise, discharge and winding are extracted respectively to form the support structure feature subset, the confluence state feature subset, the inverter operation feature subset and the boost equipment feature subset respectively; each feature subset is input into the multi-dimensional state fusion processing logic to perform feature compression and trend modeling, and finally the fusion result reflecting the global characteristics of the current operating conditions is output.

[0056] S70: Generate a fault determination result based on the fusion result.

[0057] Specifically, based on the fusion results, cross-module consistency comparison processing is performed on the key state factors in the support structure feature subset, confluence state feature subset, inverter operation feature subset and boost equipment feature subset, and factor combinations with synchronous fluctuations, trend mutations or out-of-limit behaviors are identified to construct a set of candidate abnormal factors; rule matching operations are performed on the candidate abnormal factor set, and according to the preset multi-dimensional abnormal judgment rule table, whether there is a fault feature pattern that matches the abnormal factor combination is searched in turn. If there are multiple matches, the fault scenarios corresponding to each rule are prioritized and confidence scored, and the fault type that best matches the current state is selected through a weighted scoring strategy, and the corresponding fault judgment result is finally output.

[0058] S80: Format and encapsulate the fault determination result to generate a standardized data message, and send it to the remote monitoring platform. Compare the fault determination result with the preset abnormal condition. If the fault determination result meets the preset abnormal condition, a local sound and light alarm is triggered.

[0059] Specifically, the data processing module outputs the fault judgment result to the communication interaction module, and the data encapsulation unit in the communication interaction module standardizes the timestamp, fault type, impact range and corresponding status parameters in the fault judgment result according to the preset communication protocol format to form a structured data message; after the remote sending unit receives the standardized data message, it establishes a communication connection with the remote monitoring platform through the communication interface unit, and transmits the standardized data message to the remote monitoring platform for remote alarm and information archiving; at the same time, the data processing module synchronously sends the fault judgment result to the alarm module, and the alarm module compares the fault judgment result with the preset abnormal conditions stored locally item by item. If the abnormal type, duration or fluctuation amplitude contained in the judgment result meets the corresponding alarm threshold condition, the alarm module controls the local sound and light alarm device to send an abnormal prompt signal to prompt the maintenance personnel to respond on site.

[0060] In one embodiment, if Figure 9 As shown, in step S60, the support structure status information, the confluence status information, the inverter operation status information and the boost device status information are integrated to obtain the integration result, including: S601: Perform data alignment processing on the support structure status information, the confluence status information, the inverter operation status information and the boost device status information under a unified time base, and construct a time series status data set containing the status indicators of each module.

[0061] Specifically, the state parameter items and corresponding acquisition timestamps contained in the support structure state information, confluence state information, inverter operation state information and boost equipment state information are extracted respectively, all state parameters are aligned according to the timestamps, and a unified time window division method is used to construct a time interval of fixed length. If a module has multiple data points in the current time window, the state parameters in the time period are subjected to sliding average or median filtering to enhance the steady-state characteristics. If a module has missing data in the current time window, the latest valid value is used to fill in the data or linear interpolation is used to complete it to ensure data synchronization and integrity. After completing the above alignment and processing operations, the various state parameters are summarized in turn according to the module classification to form a time series state data set indexed by time series, including support structure indicators, confluence characteristic indicators, inverter fluctuation indicators and boost operation indicators.

[0062] S602: Based on the time series state data set, extract the change amplitude, change rate and historical deviation trend of various state factors in the current time window to form a preliminary feature indicator set.

[0063] Specifically, for each type of state factor in the time series state data set, its original numerical sequence is extracted in the current time window, and the sliding interval difference algorithm is used to calculate the numerical change amplitude between consecutive time points to obtain the change amplitude index; based on the numerical difference and time interval of adjacent data points, the first-order difference or linear regression slope method is used to calculate the change rate per unit time to form a change rate index; the mean, maximum, minimum and standard deviation of the state factor in the current time window and multiple historical windows are combined to calculate the historical deviation trend index by comparing and analyzing the statistical deviation degree; the above-mentioned variation amplitude index, variation rate index and historical deviation trend index are classified in turn according to the state factor to form a complete preliminary feature index set. S603: According to the fluctuation characteristics and fault sensitivity of various state factors in the preliminary characteristic indicator set, fusion weight factors are assigned and a weighted characteristic matrix is ​​constructed.

[0064] Specifically, for each type of state factor included in the preliminary characteristic indicator set, the corresponding fault sensitivity score is first calculated based on the variation range, response delay and abnormal trigger frequency of the state factor in known fault scenarios in the historical monitoring data; then the fluctuation range, change frequency and standard deviation of the state factor during the normal operating cycle are statistically analyzed to obtain numerical parameters characterizing the fluctuation characteristics; according to the preset fusion weight configuration rules, the fault sensitivity score and the fluctuation characteristic parameters are normalized, and a weighted integration operation is performed to generate the corresponding fusion weight factor; then the fusion weight factor is used to weight the characteristic values ​​of each type of state factor in the preliminary characteristic indicator set, and a weighted feature matrix with the state factor type as the dimension and the time series as the span is constructed.

[0065] S604: Extracting the change consistency features between the tilt angle information and the image information based on the weighted feature matrix to generate a support structure feature subset.

[0066] Specifically, the feature columns corresponding to the tilt angle information and image information in the weighted feature matrix are analyzed synchronously. First, the structural characterization parameters such as edge contours, deformation areas or occlusion ratios in the tilt angle change sequence and image information are extracted based on a unified time window. Then, the consistency of the change direction, change amplitude and abnormal jump position of the two at each time point is calculated to construct a change consistency measurement index; then, the consistency measurement index is jointly modeled with the tilt angle change rate and the image structure feature change rate to form a stent structure state factor for identifying structural abnormality trends; finally, the above state factors are combined into a feature vector as a stent structure feature subset.

[0067] S605: Based on the weighted feature matrix, extract the string deviation index and current balance index of each current, and construct a convergence state feature subset.

[0068] Specifically, the feature columns corresponding to the current data of each channel are extracted from the weighted feature matrix. First, based on the current values ​​of each string group in each time window, its mean and standard deviation are calculated, and the deviation between the current of each string group and the overall mean is further calculated to form a string deviation index. Then, the current of each string group is normalized, and the amplitude difference between its maximum and minimum values ​​in the current window is analyzed. Combined with the current distribution trend of the previous time window, the balance degree of the current distribution is evaluated to obtain the current balance index. Finally, the string deviation index and the current balance index are combined to form a feature vector that reflects the abnormal trend of convergence, and the convergence state feature subset is constructed.

[0069] S606: Based on the weighted feature matrix, extract the frequency fluctuation index and the power stability offset index, and construct an inverter operation feature subset.

[0070] Specifically, based on the characteristic dimension corresponding to the inverter operating state in the weighted feature matrix, the output frequency sequence and output power sequence in each time window are extracted, and continuous difference processing is performed on the frequency sequence. The rate of change of frequency between adjacent time points is calculated, and the fluctuation range in multiple windows before and after is combined to form a frequency fluctuation index; trend fitting is performed on the output power sequence to evaluate its degree of deviation from the ideal output power curve, and the power stability offset index is constructed by combining the instantaneous mutation amplitude and the long-period offset amplitude; the frequency fluctuation index and the power stability offset index are combined into a feature vector for describing the changes in the inverter operating conditions, and an inverter operation feature subset is constructed.

[0071] S607: Based on the weighted feature matrix, extract the temperature rise abnormality index, discharge abnormality index and winding state abnormality index to construct a boost equipment feature subset.

[0072] Specifically, based on the feature dimensions related to the boost equipment status in the weighted feature matrix, the historical data sequence representing the temperature change of the equipment is extracted, the temperature rise rate and temperature difference distribution of each monitoring point in the current time window are calculated, and combined with the rated load conditions of the equipment, it is determined whether there is a temperature rise anomaly, and the temperature rise anomaly index is obtained; a time-frequency joint analysis is performed on the feature information provided by the arc monitoring unit to identify abnormal discharge signals such as voltage disturbances, transient spikes or continuous high-frequency discharges, and form a discharge anomaly index; the status information of the transformer winding is statistically compared with trends to identify the abnormal coupling of the winding resistance change amplitude, temperature and insulation parameters, and generate a winding status anomaly index; the temperature rise anomaly index, discharge anomaly index and winding status anomaly index are combined according to the feature expression structure to construct a boost equipment feature subset.

[0073] S608: Input the support structure feature subset, the confluence state feature subset, the inverter operation feature subset, and the boost equipment feature subset into the multi-dimensional state fusion model, perform feature compression and abnormal trend modeling, and output the fusion result.

[0074] Specifically, all feature vectors in the support structure feature subset, confluence state feature subset, inverter operation feature subset and boost equipment feature subset are uniformly normalized to ensure that the numerical ranges of various state factors are comparable, and then input into the multidimensional state fusion model; through the feature compression structure set in the model, principal component extraction and dimensionality reduction are performed on the high-dimensional feature information to retain the dominant factors reflecting key state changes; in the compressed feature space, according to the time change trajectory and spatial distribution pattern of each state factor, the time series modeling mechanism is used to construct the abnormal trend evolution path; the abnormal trend is multi-dimensionally fitted and dynamically scored, and the overall state deviation degree is comprehensively evaluated, and the fusion result is finally output.

[0075] In one embodiment, if Figure 10 As shown, in step S70, that is, generating a fault determination result based on the fusion result, it also includes: S701: Based on the fusion results, perform cross-module consistency comparison on the key state factors extracted from each feature subset, identify the factor combination with state abnormality features, and construct a set of candidate abnormal factors.

[0076] Specifically, the key state factors in the support structure feature subset, confluence state feature subset, inverter operation feature subset and boost equipment feature subset are cross-mapped according to a unified timestamp, and the abnormal fluctuation performance of each factor in the current time window is extracted respectively, including characteristic indicators such as mutation amplitude, change rate reversal and over-threshold deviation amplitude, to construct a multi-module factor state comparison matrix; a consistency comparison analysis is performed on the state comparison matrix to identify the state factor combination that simultaneously shows abnormal trends or correlated change patterns in multiple modules, and it is determined that the factor combination has potential correlated abnormal characteristics; the above-mentioned factor combinations with correlated state abnormal performance are constructed as a candidate abnormal factor set.

[0077] S702: Execute rule matching processing on the candidate abnormal factor set, retrieve the fault feature pattern that meets the conditions according to the preset multi-dimensional abnormal judgment rule table, and obtain the fault rule set that matches successfully.

[0078] Specifically, each state factor combination included in the candidate abnormal factor set is matched one by one with each fault feature pattern in the preset multi-dimensional abnormal judgment rule table. The fault feature pattern includes abnormal indicator items, indicator combination relationships, time series change conditions and equipment component correspondences defined for different types of faults; during the matching process, the candidate factor combination is searched in sequence according to the set logical judgment order to see whether it meets various abnormal judgment conditions defined in a certain feature pattern, including whether the state factors exceed the corresponding judgment threshold at the same time, whether fluctuation reversal occurs synchronously within a given time range, and whether it complies with the corresponding equipment structure dependency logic; if the match is successful, the fault rule number, fault type and related information corresponding to the feature pattern are included in the fault rule set.

[0079] S703: If the fault rule set contains multiple matching results, priority evaluation and confidence scoring are performed on the fault scenarios corresponding to the matching rules, and weighted scoring is performed in combination with the matching degree of the current state factor characteristics to filter out the fault judgment results.

[0080] Specifically, first, for the fault scenario corresponding to each matching rule in the fault rule set, metadata such as fault type, related equipment location and historical occurrence frequency are extracted, and combined with the preset fault severity level and handling urgency index to construct a multi-dimensional priority evaluation parameter set; then, based on the abnormal amplitude and fluctuation trend of the state factor in each feature subset and the degree of compliance with the various judgment conditions in the matching rule, the state factor matching score is calculated; further, the priority evaluation parameters and the state factor matching score are fused according to the set weights to obtain a comprehensive confidence score for each matching result; finally, all scores are sorted, and the fault type, abnormal module and specific location corresponding to the matching rule with the highest comprehensive score and confidence exceeding the judgment threshold are selected as the final output fault judgment result.

[0081] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0082] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. 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 various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. An abnormality detection system for a photovoltaic substation, characterized in that: The abnormality detection system of the photovoltaic substation comprises a photovoltaic power generation information acquisition module (1), a bracket state monitoring module (2), a confluence state monitoring module (3), a DC-AC conversion state monitoring module (4), a boost equipment operation state monitoring module (5), a data processing module (6), a communication interaction module (7) and an alarm module (8), wherein: The photovoltaic power generation information acquisition module (1) is used to obtain power generation parameter information and output the power generation parameter information to the confluence state monitoring module (3); The support state monitoring module (2) is used to collect support structure state information and output the support structure state information to the data processing module (6); The confluence state monitoring module (3) is used to receive the power generation parameter information, perform convergence analysis on the power generation parameter information, and obtain operation summary parameter information. The confluence state monitoring module (3) is also used to collect confluence state information and output the confluence state information to the data processing module (6); The DC-AC conversion state monitoring module (4) is connected to the convergence state monitoring module (3), and the DC-AC conversion state monitoring module (4) performs power stability analysis and frequency feature extraction in the electric energy conversion process on the operation summary parameter information to obtain inverter operation state information, and outputs the inverter operation state information to the data processing module (6); The boosting equipment operating status monitoring module (5) is used to collect boosting equipment status information and output the boosting equipment status information to the data processing module (6); The data processing module (6) is used to integrate the support structure status information, the confluence status information, the inverter operation status information and the boost device status information, perform comprehensive analysis, generate a fault determination result, and output the fault determination result to the communication interaction module (7) and the alarm module (8); The communication interaction module (7) is used to send the fault determination result to a remote monitoring platform; The alarm module (8) is used to determine that when the fault determination result meets a preset abnormal condition, a local sound and light alarm is triggered.

2. The abnormality detection system for a photovoltaic substation according to claim 1, characterized in that: The photovoltaic power generation information acquisition module (1) comprises a photovoltaic assembly unit (11) and a direct current acquisition unit (12); the photovoltaic assembly unit (11) is used to convert solar energy into direct current electric energy; and the direct current acquisition unit (12) is used to collect and analyze the direct current electric energy to obtain the power generation parameter information.

3. The abnormality detection system for a photovoltaic substation according to claim 2, characterized in that: The support state monitoring module (2) comprises an inclination sensing unit (21) and an image acquisition unit (22), wherein the inclination sensing unit (21) is used to obtain inclination angle information of the support, and the image acquisition unit (22) is used to acquire image information of the support surface. The support state monitoring module (2) is used to perform correlation matching analysis on the inclination angle information and the image information within a corresponding time window to obtain the support structure state information, and output the support structure state information to the data processing module (6).

4. A photovoltaic substation anomaly detection system according to claim 3, wherein the convergence state monitoring module (3) includes a multi-channel electrical parameter acquisition unit (31) and a grounding state detection unit (32), wherein the multi-channel electrical parameter acquisition unit (31) is used to acquire the power generation parameter information from each photovoltaic power generation information acquisition module (1), and the grounding state detection unit (32) is used to detect the grounding resistance value and insulation state in the convergence loop to obtain a grounding state detection result, and the convergence state monitoring module (3) is used to compare and analyze the power generation parameter information and judge the current balance, and perform a comprehensive evaluation based on the grounding state detection result to obtain the operation summary parameter information and the convergence state information, and the operation summary parameter information is output to the DC-AC conversion state monitoring module (4), and the convergence state information is output to the data processing module (6).

5. According to the abnormality detection system of a photovoltaic substation according to claim 4, the boost equipment operation status monitoring module (5) includes a temperature rise detection unit (51), an arc monitoring unit (52) and a winding status evaluation unit (53), the temperature rise detection unit (51) is used to collect temperature change information of the boost equipment under different load conditions, the arc monitoring unit (52) is used to detect arc characteristic information generated by the boost equipment during operation, and the winding status evaluation unit (53) is used to monitor the winding status information of the transformer winding. The boost equipment operation status monitoring module (5) is used to comprehensively analyze the temperature change information, the arc characteristic information and the winding status information to obtain the boost equipment status information, and output the boost equipment status information to the data processing module (6).

6. According to the abnormality detection system of a photovoltaic substation according to claim 5, the data processing module (6) includes a data receiving unit (61), a state fusion analysis unit (62) and a fault judgment unit (63), the data receiving unit (61) is used to receive various types of status information from the bracket status monitoring module (2), the convergence status monitoring module (3), the DC-AC conversion status monitoring module (4) and the boost equipment operation status monitoring module (5), the state fusion analysis unit (62) is used to perform time alignment, weight integration and feature fusion processing on the various types of status information to obtain a fusion result, the fault judgment unit (63) is used to identify potential abnormal operating conditions based on the fusion result and generate the fault judgment result, and the fault judgment result is output to the communication interaction module (7) and the alarm module (8).

7. An abnormality detection system for a photovoltaic substation according to claim 6, wherein the communication interaction module (7) includes a communication interface unit (71), a data encapsulation unit (72) and a remote sending unit (73), wherein the communication interface unit (71) is used to establish a communication connection with the remote monitoring platform, the data encapsulation unit (72) is used to format and encapsulate the fault judgment result to generate a standardized data message, and the remote sending unit (73) is used to send the standardized data message to the remote monitoring platform through the communication interface unit (71).

8. A photovoltaic substation anomaly detection method, applied to a photovoltaic substation anomaly detection system according to any one of claims 1 to 7, characterized in that: The abnormality detection method of a photovoltaic substation includes: Acquiring the power generation parameter information; Performing group string comparison analysis and current balance evaluation on the power generation parameter information to obtain the operation summary parameter information, and obtaining the convergence state information through the grounding state detection unit (32); Performing power stability analysis and frequency feature extraction on the operation summary parameter information to obtain the inverter operation status information; Acquiring the tilt angle information and the image information, performing correlation matching analysis on the tilt angle information and the image information, and obtaining the support structure status information; Acquiring the temperature change information, the discharge abnormality information, and the winding state information, and performing a comprehensive analysis on the temperature change information, the discharge abnormality information, and the winding state information to obtain the boost device state information; fusing the support structure status information, the confluence status information, the inverter operation status information, and the boost device status information to obtain a fusion result; generating the fault determination result based on the fusion result; The fault determination result is formatted and encapsulated to generate the standardized data message, and is sent to the remote monitoring platform. The fault determination result is compared with the preset abnormal condition. If the fault determination result meets the preset abnormal condition, the local sound and light alarm is triggered.

9. The abnormality detection method for a photovoltaic substation according to claim 8, characterized in that: The fusing of the support structure status information, the confluence status information, the inverter operation status information, and the boost device status information to obtain a fusion result includes: Performing data alignment processing on the support structure status information, the confluence status information, the inverter operation status information, and the boost device status information under a unified time reference to construct a time series status data set including status indicators of each module; Based on the time series state data set, the change amplitude, change rate and historical deviation trend of various state factors in the current time window are extracted to form a preliminary feature indicator set; According to the fluctuation characteristics and fault sensitivity of the various state factors in the preliminary characteristic indicator set, a fusion weight factor is assigned to construct a weighted characteristic matrix; extracting, based on the weighted feature matrix, a change consistency feature between the tilt angle information and the image information to generate a stent structure feature subset; Based on the weighted feature matrix, the string deviation index and the current balance index of each current are extracted to construct a convergence state feature subset; Based on the weighted feature matrix, extracting the frequency fluctuation index and the power stability offset index, and constructing the inverter operation feature subset; Based on the weighted feature matrix, extracting temperature rise abnormality index, discharge abnormality index and winding state abnormality index, and constructing a boost equipment feature subset; The support structure feature subset, the confluence state feature subset, the inverter operation feature subset and the boost equipment feature subset are input into a multi-dimensional state fusion model, feature compression and abnormal trend modeling are performed, and the fusion result is output.

10. The abnormality detection method for a photovoltaic substation according to claim 8, characterized in that: Generating the fault determination result based on the fusion result further includes: Based on the fusion results, a cross-module consistency comparison is performed on the key state factors extracted from each feature subset to identify the factor combinations with abnormal state features and construct a set of candidate abnormal factors; Performing rule matching processing on the candidate abnormal factor set, searching for fault feature patterns that meet the conditions according to a preset multi-dimensional abnormality judgment rule table, and obtaining a set of fault rules that successfully matches; If the fault rule set contains multiple matching results, priority evaluation and confidence scoring are performed on the fault scenarios corresponding to each matching rule, and weighted scoring is performed in combination with the matching degree of the current state factor characteristics to filter out the fault judgment result.

Images