Photovoltaic device safety check method and photovoltaic device safety check system
Through automated safety detection methods and systems of photovoltaic equipment, the abnormal processing and model optimization are used to use large language models to solve the problem of insufficient user interaction in photovoltaic power station management, improve detection accuracy and user experience, and reduce operation and maintenance costs.
Patent Information
- Application Number
- PCT/CN2024/101769
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2024-04-29
- Filing Date
- 2024-06-26
- Publication Date
- 2024-11-07
AI Technical Summary
The existing photovoltaic power station management methods lack effective interaction with users, resulting in the inability to resolve false alarm abnormalities in a timely and effective manner, reducing detection accuracy.
Provide an automated safety detection method and system for photovoltaic equipment, which can be displayed to users through state detection, abnormal command analysis, inquiry command, feedback processing, and execution commands sent to photovoltaic power stations, and combine large language models for model switching or self-learning. Achieve accurate detection results.
It improves the accuracy of equipment safety detection results, simplifies user input, reduces operation and maintenance costs, improves user experience and abnormal handling efficiency, and realizes convenient management of photovoltaic power stations.
Smart Images

Figure CN2024101769_07112024_PF_FP_ABST
Abstract
Description
Photovoltaic equipment safety detection method and photovoltaic equipment safety detection system Technical Field
[0001] The present invention relates to the technical field of photovoltaic equipment, and in particular to a photovoltaic equipment safety detection method and a photovoltaic equipment safety detection system. Background Art
[0002] A photovoltaic power station is a system that harnesses solar energy and utilizes specialized materials such as crystalline silicon panels, inverters, and other electronic components. It is connected to the power grid and transmits electricity to the grid. A photovoltaic power station generally consists of a solar cell array, battery packs, a battery management system (BMS), a charge and discharge controller, an inverter, an AC power distribution cabinet, and a solar tracking control system. During operation, various operating parameters, such as power, ground impedance, power generation efficiency, and power consumption, are fed back to the cloud. Remote management is achieved through cloud-based data storage and analysis. Together, the cloud and the equipment constitute the photovoltaic power station management system.
[0003] However, changes in the external environment or individual user electricity usage habits can lead to false alarms. Existing PV plant management methods lack effective interaction with users, making it difficult to promptly and effectively address these false alarms. Furthermore, PV equipment cannot be updated in real time, reducing detection accuracy. Currently, no effective solutions have been proposed to address these issues.
[0004] Summary of the Invention
[0005] A main purpose of the present invention is to overcome at least one of the above-mentioned defects, and to provide a photovoltaic equipment safety detection method that can automatically eliminate abnormalities, and can also specifically provide a guided fault solution.
[0006] A main purpose of the present invention is to overcome at least one of the above-mentioned drawbacks, and to provide a photovoltaic equipment safety detection system that can automatically eliminate abnormalities.
[0007] In order to achieve the above object, the technical solution adopted by the present invention is:
[0008] The present invention provides a photovoltaic equipment safety detection method, comprising the following processing steps:
[0009] S1: Detect the status of the photovoltaic power station and send the detection information to the cloud;
[0010] S2: confirming whether there is an abnormality based on the detection information, outputting an abnormal instruction when an abnormality exists, and parsing the abnormal instruction to form an inquiry instruction;
[0011] S3: Displaying the query instruction to the user and asking the user to provide a response;
[0012] S4: confirming, based on the reply information, whether to continue to issue an inquiry instruction to the user or to issue an execution instruction to the photovoltaic power station, wherein the photovoltaic power station takes action according to the execution instruction to eliminate the abnormality or self-checks and reports the abnormality;
[0013] S5: Feedback the execution instruction to the preprocessing port, and perform model switching or model self-learning according to the execution instruction to iteratively upgrade and obtain accurate detection results.
[0014] According to one embodiment of the present invention, the abnormal instruction is parsed and processed, and directional data matching the abnormal instruction is retrieved from a photovoltaic standard database to form a query instruction.
[0015] According to one embodiment of the present invention, in the process of retrieving the directional data matching the abnormal instruction from the photovoltaic standard database and forming the query instruction,
[0016] The retrieved directional data is read and processed through a large language model to form the query instruction.
[0017] According to one embodiment of the present invention, in the process of the large language model reading and processing the directional data to form the query instruction,
[0018] The directional data is sorted to form a plurality of interactive data, and the plurality of interactive data constitute the query instruction. When the query instruction is displayed to the user, interaction with the user is performed in sequence according to the plurality of interactive data.
[0019] According to one embodiment of the present invention, in the process of sequentially interacting with the user based on the plurality of pieces of interaction data, the interaction data is converged based on the response information fed back by the user.
[0020] According to one embodiment of the present invention, the directional data matched with the abnormal instruction is an environmental factor or a human factor that causes the abnormal instruction to occur.
[0021] According to one embodiment of the present invention, in step S4, in the process of confirming the execution instruction to the photovoltaic power station according to the reply information,
[0022] The abnormal instruction is confirmed through the reply information, and an execution strategy is formed by searching and calling in the photovoltaic standard database and issuing the execution instruction. After receiving the execution instruction, the photovoltaic power station takes corresponding actions to resolve the abnormality.
[0023] According to one embodiment of the present invention, the photovoltaic standards database includes API call documents, product descriptions, and photovoltaic industry technical information.
[0024] According to one embodiment of the present invention, the query instruction is an output instruction based on natural language.
[0025] According to one embodiment of the present invention, the abnormal instruction includes a fault instruction, a power abnormality instruction, a power abnormality instruction, and a BMS alarm instruction.
[0026] According to one embodiment of the present invention, step S2 and step S4 are completed in the cloud, on an APP side, or at a photovoltaic power station.
[0027] According to one embodiment of the present invention, step S3 is completed on the APP side.
[0028] According to one embodiment of the present invention, when the preprocessing port is a device end, the device end obtains the execution instruction with a label to optimize and iterate the built-in model of the device end.
[0029] According to one embodiment of the present invention, when the pre-processing port is the cloud, the cloud obtains the execution instruction with the label to perform optimization iteration and secondary diagnosis on the cloud built-in model.
[0030] According to one embodiment of the present invention, when the preprocessing port is a large language model, the large language model obtains the labeled execution instructions for deep learning, generates a self-built model, and sends the self-built model to the device or the cloud.
[0031] According to one embodiment of the present invention, in step S1 , the status detection of the photovoltaic power station is arc detection and patrol inspection.
[0032] According to one embodiment of the present invention, the inspection of a photovoltaic power station includes DC insulation resistance value, fault physical location, device operating status, and battery status.
[0033] In particular, the present invention also provides a photovoltaic equipment safety detection system, which includes:
[0034] an equipment monitoring unit for detecting the status of the photovoltaic power station and recording detection information, and sending the detection information to the abnormality analysis unit;
[0035] An analysis and management unit receives the detection information, determines whether there is an abnormality based on the detection information, outputs an abnormal instruction when an abnormality exists, parses the abnormal instruction to form an inquiry instruction, and sends the inquiry instruction to the APP end;
[0036] An APP terminal is provided at the user's location, and is used to receive the query instruction and display it to the user. The user enters the reply information through the APP terminal, and the reply information is sent to the analysis management unit;
[0037] Among them, the analysis and management unit confirms whether to continue to issue an inquiry instruction to the user or to issue an execution instruction to the photovoltaic power station based on the reply information. The photovoltaic power station takes action to eliminate abnormalities or self-checks and reports abnormalities based on the execution instruction; and feeds back the execution instruction to the analysis and management unit. The analysis and management unit performs model switching or model self-learning based on the execution instruction for iterative upgrading.
[0038] According to one embodiment of the present invention, the analysis management unit includes:
[0039] An input preprocessing module is used to receive the abnormal instruction and capture the features of the abnormal instruction to form effective feature data;
[0040] A context synthesis processing module is connected to the input pre-processing module, receives the valid feature data, and retrieves and calls data matching the valid feature data to form a comprehensive input instruction;
[0041] The large language model is connected to the input preprocessing module, receives the comprehensive input instruction and processes it to form a query instruction or an execution instruction.
[0042] According to one embodiment of the present invention, the analysis and management unit further includes a memory storage module, which is used to store photovoltaic standard data and real-time interaction data. The memory storage module is connected to the context integration processing module.
[0043] According to one embodiment of the present invention, the memory storage module includes a permanent storage submodule and a flash memory submodule, the permanent storage submodule is used to store photovoltaic standard data, and the flash memory submodule is used to store real-time interactive data.
[0044] According to one embodiment of the present invention, the memory storage module includes a storage management submodule, which is connected to the large language model and is used to receive the feedback query instructions and the response information and update them in the real-time interaction data.
[0045] According to one embodiment of the present invention, the analysis and management unit further includes an output parsing module, which receives the execution instruction and parses the execution instruction to form a control instruction. The control instruction is in a machine language format after formatting and parsing. The photovoltaic power station receives the control instruction and executes it to eliminate abnormalities or self-check and report abnormalities.
[0046] According to one embodiment of the present invention, the analysis and management unit is configured in the cloud, an APP terminal or a photovoltaic power station.
[0047] In particular, the present invention also provides an edge computing server, comprising a memory, a processor, and a computer program stored in the memory and capable of being run by the processor, wherein the processor implements the above-described method when executing the computer program.
[0048] The present invention also provides a computer-readable storage medium having a computer program stored therein, wherein the computer program implements the above method when executed by a processor. The computer-readable storage medium is preferably a non-volatile readable storage medium.
[0049] Compared with the prior art, the advantages and beneficial effects of the photovoltaic equipment safety detection system and photovoltaic equipment safety detection method of the present invention patent application are:
[0050] In the present application, active information push inquiry and model optimization are adopted. The status of the photovoltaic power station is detected (such as arc detection), and the detection information is sent to the cloud; based on the detection information, it is confirmed whether there is an abnormality. When an abnormality exists, an abnormal instruction is output, and the abnormal instruction is parsed and processed to form an inquiry instruction; the inquiry instruction is displayed to the user and the user gives a reply information; based on the reply information, it is confirmed whether to continue to issue an inquiry instruction to the user for inquiry or to issue an execution instruction to the photovoltaic power station. The photovoltaic power station takes action according to the execution instruction to eliminate the abnormality or self-check and report the abnormality; the execution instruction is fed back to the preprocessing port, and the model is switched or the model is self-learned according to the execution instruction, so as to obtain accurate detection results through iterative upgrading, thereby achieving the purpose of convenient information interaction and model optimization, thereby achieving the technical effect of improving the accuracy of equipment safety detection results, and thus solving the technical problem that the photovoltaic power station management method in the existing technology cannot solve these false alarm abnormalities in a timely and effective manner due to the lack of effective interaction with users. At the same time, the photovoltaic equipment cannot be updated and iterated in real time, resulting in reduced detection accuracy.
[0051] The photovoltaic equipment safety detection method of the present application is based on the abnormality detection of the photovoltaic power station. When an abnormality occurs, a language-based inquiry instruction is generated, and the inquiry instruction is sent to the user for the user to confirm the status related to the abnormality, such as the status of the environment, manual power adjustment, external damage to the equipment, etc. Through effective back-and-forth interaction with the user, abnormal false alarms caused by the environment are eliminated, thereby improving the efficiency of handling abnormal conditions, saving manpower and material resources, reducing operation and maintenance costs, and effectively improving user experience.
[0052] The analysis body of the photovoltaic equipment safety detection system of the present application is set on the cloud or APP side, which simplifies the user input on the interactive side. The user only needs to respond to the inquiry instructions through voice or text input, which facilitates the information interaction between the user and the analysis and processing unit, and reduces the technical threshold requirements for users to answer questions, so that users can be more proactive in cooperating with exception handling, which not only improves the user experience, but also better realizes the management of photovoltaic power stations.
[0053] Furthermore, this application innovatively introduces a large language model (such as the ChatGPT model). The input side of the large language model uses the retrieved text-based exception instructions, and the output side uses natural language-based query instructions or execution instructions. From the front-end input to the back-end processing output, it is all based on natural language processing, which is more in line with the usage needs of traditional users. During the final specific execution, the execution instructions based on natural language are converted into a format (converted into control instructions) and then used for specific execution, which reduces the processing difficulty of this photovoltaic information processing, can better meet user needs, improve processing efficiency, and realize the automation of exception handling. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] Hereinafter, some specific embodiments of the present invention will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. In the accompanying drawings:
[0055] FIG1 is a schematic diagram of the workflow of a photovoltaic equipment safety detection method according to one embodiment of the present invention;
[0056] FIG2 is a structural block diagram of a photovoltaic equipment safety detection system according to an embodiment of the present invention;
[0057] FIG3 is a structural diagram of a computing device according to another embodiment of the present application;
[0058] FIG4 is a structural diagram of a computer-readable storage medium according to another embodiment of the present application. DETAILED DESCRIPTION
[0059] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.
[0060] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.
[0061] Example 1:
[0062] This embodiment describes a photovoltaic equipment safety detection method, including the following processing steps:
[0063] S1: Detect the status of the photovoltaic power station and send the detection information to the cloud;
[0064] S2: confirming whether there is an abnormality based on the detection information, outputting an abnormality instruction when an abnormality exists, retrieving directional data matching the abnormality instruction from a photovoltaic standard database, reading and processing the retrieved directional data using a large language model to form a query instruction;
[0065] S3: Displaying the query instruction to the user and asking the user to provide a response;
[0066] S4: confirming, based on the reply information, whether to continue to issue an inquiry instruction to the user or to issue an execution instruction to the photovoltaic power station, wherein the photovoltaic power station takes action according to the execution instruction to eliminate the abnormality or self-checks and reports the abnormality;
[0067] S5: Feedback the execution instruction to the preprocessing port, and perform model switching or model self-learning according to the execution instruction to iteratively upgrade and obtain accurate detection results.
[0068] In one embodiment, while the large language model reads and processes the directional data to form the query instruction, the directional data is sorted to form a plurality of pieces of interaction data. The plurality of interaction data constitutes the query instruction. When the query instruction is presented to the user, interaction with the user is performed sequentially based on the plurality of interaction data. During the process of sequentially interacting with the user based on the plurality of interaction data, the interaction data is converged based on the response information fed back by the user.
[0069] In one embodiment, the directional data matched with the abnormal instruction is an environmental factor or a human factor that causes the abnormal instruction to occur.
[0070] In one embodiment, in the process of confirming the issuance of an execution instruction to the photovoltaic power station based on the reply information, the abnormal instruction is confirmed through the reply information, and an execution strategy is formed and the execution instruction is issued in combination with the retrieval and call in the photovoltaic standard database. After receiving the execution instruction, the photovoltaic power station performs corresponding actions to resolve the abnormality.
[0071] In one embodiment, the large language model is a ChatGPT model.
[0072] In one embodiment, the photovoltaic standards database includes API call documents, product descriptions, and photovoltaic industry technical information.
[0073] In one embodiment, the query instruction is an output instruction based on natural language, and the abnormal instruction includes a fault instruction, a power abnormality instruction, a power consumption abnormality instruction, and a BMS alarm instruction.
[0074] In one embodiment, step S2 and step S4 are completed in the cloud, APP or photovoltaic power station, and step S3 is completed in the APP. For the specific analysis and output process of the query instruction, it is preferably carried out in the cloud with strong processing capabilities, but it can also be carried out in the APP or photovoltaic power station. It just requires a larger analysis and processing capability, which can be adjusted according to the user's usage needs and the data processing performance of each end. Even the functional modules in the analysis and management unit can be reasonably split and arranged in one of the three of the APP, cloud or photovoltaic power station. In other words, the functional modules of the integrated APP, cloud or photovoltaic power station are combined to realize the processing of the analysis and management unit. This is also technically feasible.
[0075] In one embodiment, the user's response information input through the app is voice or text. This simplifies user input on the interactive side, as users only need to respond to queries via voice or text input. This facilitates information interaction between the user and the analysis and processing unit, reduces the technical threshold for users to respond to questions, and enables users to be more proactive in cooperating with exception handling, thereby improving user experience and enabling better management of photovoltaic power plants.
[0076] In one embodiment, when the preprocessing port is the device side, the device side obtains the labeled execution instruction to optimize and iterate the built-in model of the device side. By adopting including but not limited to: sending the labeled execution instruction to the device side through a large language model, the device side optimizes and iteratively upgrades the existing model by changing the device side model, adopting but not limited to: switching models or model self-learning methods to obtain better and more accurate safety detection results, so that the device can operate in a safe state. Of course, there is data interaction between the device side and the large prediction model, so as to ensure good information collection and feedback effects, and then achieve the effect of precise control and safety detection of the equipment, and ultimately avoid accidental (fire) losses of people and property.
[0077] Specifically, the arc detection and inspection arranged on the equipment side are because many photovoltaic modules are used on the equipment side, and the photovoltaic modules are connected through MCC terminals. However, they are exposed to the wild for a long time. Due to the influence of weathering, rain erosion and terrain subsidence, the terminal connections are pulled loose, resulting in arcs in the air, thereby affecting the combustion and damage to equipment and personnel safety. Therefore, the processing method of this application is adopted. First, the customer is informed of the length of the solar panel and whether the current alarm is actually generated, and customized parameter configuration is made; the inspection includes DC insulation impedance value, physical location of the fault, device operating status, battery status, and active push of warnings and optimization suggestions, which is equivalent to a power station secretary, and is pushed to the end customer; afterwards, when the abnormal situation is resolved, the relevant parameters for abnormal processing are obtained and sent to the equipment side. After receiving the parameter information, the model on the equipment side is changed, including but not limited to: switching models, model self-learning, or modifying criteria, machine learning, etc., to optimize and iteratively upgrade the built-in model on the equipment side, which can improve the accuracy of equipment safety detection and thus better meet the needs of actual use.
[0078] In one embodiment, when the pre-processing port is the cloud, the cloud obtains the tagged execution instruction to optimize, iterate and perform secondary diagnosis on the cloud-based built-in model. When the pre-processing port is the cloud, such as a cloud BMS, the cloud receives a tagged execution command, and the cloud optimizes and iterates the built-in model. The methods used include but are not limited to machine learning methods, so that the built-in model in the cloud is optimized, iterated and upgraded, and a better model is generated. Of course, there is a two-way information interaction function between the cloud and the large language model, that is, an information interaction function, so as to ensure good information collection and feedback effects. For example: based on voltage, current, temperature detection, charge and discharge rate and other prior conditions, a cloud BMS is generated.
[0079] In one embodiment, when the preprocessing port is a large language model, the large language model obtains the labeled execution instructions for deep learning, generates a self-built model, and sends the self-built model to the device or cloud. By feeding all information back to the large language model, then directly generating a model through the large prediction model, and sending the model to the device or cloud for direct use, the cloud or device does not need to perform other model learning or training operations.
[0080] In one implementation, edge computing devices are deployed on both the cloud and device sides. "Cloud-edge-end" refers to the combination of cloud computing, edge computing, and terminal computing, forming a new distributed computing architecture. In this architecture, cloud computing provides high-performance, high-reliability, and highly scalable computing and storage capabilities, edge computing pushes data processing and analysis to the edge of the network, and terminal computing performs local computing and data collection. The addition of edge computing devices can accelerate data processing capabilities and improve efficiency, while avoiding cloud computing overload that could affect overall operational efficiency.
[0081] Taking the handling of battery temperature rise anomalies in the background technology as an example, after receiving an abnormality instruction, it is retrieved from the database to obtain visible factors that cause the anomaly rather than internal equipment operating failures, which may include external ambient temperature, foreign objects around the battery that cause temperature rise, reflective irradiation of the battery, etc. Therefore, the large language model intends to issue multiple query instructions such as "Is the local temperature rising?" and "Is there any foreign object around the battery?" The query instructions are sent to the user's app and the app is awakened. If the user answers "yes" to "Is the local temperature rising?", the user is asked "how many degrees has it risen?" If the temperature has indeed risen suddenly, it is determined that the battery temperature has risen due to the environment, and the query ends. However, if the user answers "no", another query instruction is issued, "Is there any foreign object around the battery?" Then, based on the user's answer of "yes" or "no", the next query instruction is issued in a targeted manner. For example, if there is a foreign object, the user is guided to remove the foreign object, eliminating the abnormality from the environment. This cannot be conveniently achieved through interaction in the existing technology. Furthermore, if the cause of the rapid battery heating cannot be determined after eliminating all environmental anomalies, the comprehensive self-inspection of the photovoltaic power station or manual intervention is initiated, or an execution instruction is issued to shut down certain circuits for further analysis and confirmation of the cause of the anomaly. However, there are many implementation methods for how to use equipment self-inspection to troubleshoot anomalies in the existing technology, but this does not involve the protection scope of this protection point for interactively eliminating environmental factors in this application. Therefore, the equipment self-inspection to troubleshoot anomalies will not be elaborated on.
[0082] Taking the example of "reduced solar panel power generation," environmental factors that may cause this situation include "obstructions around the solar panel" or "fragmentation of the solar panel." Analysis of the anomaly can pinpoint the problem to a specific solar panel, allowing for more specific inquiries such as "Please check if there are any obstructions around solar panel number X" or "Please check if solar panel number X is damaged." Based on the user's feedback, the cause of the anomaly can be determined and resolved.
[0083] It can be seen that through the interaction between the APP and the user, information can be obtained from the user in reverse, which can supplement the abnormal solution. It can facilitate the method to accurately give suggestions to the user and formulate solutions more accurately.
[0084] In addition, based on detection information, it is possible to learn and record users' electricity usage habits, so anomalies may also be caused by sudden changes in electricity usage. In this operating condition, it is impossible to determine whether this is a temporary change in electricity usage in the short term (such as a decrease in electricity usage due to a business trip) or a long-term change in electricity usage (such as the introduction of a new appliance in the home). Therefore, an inquiry instruction can be issued to the user to determine the cause of the change in electricity usage. After the specific cause is determined, the situation can be recorded. For example, if the user's electricity usage is only temporary, then the abnormal alarm can be directly canceled. If it is a long-term electricity meter, by issuing execution instructions, the load usage forecast can be adjusted, the electricity usage habit record can be updated, and the battery storage capacity of the photovoltaic power station can be increased.
[0085] From the above specific cases, it can be seen that the photovoltaic equipment safety detection method of this embodiment is based on the abnormality detection of the photovoltaic power station. When an abnormality occurs, a language-based inquiry instruction is generated, and the inquiry instruction is sent to the user for the user to confirm the status related to the abnormality, such as the environment, manual power adjustment, external damage to the equipment, etc. Through effective interaction with the user, false alarms caused by abnormalities caused by the environment are eliminated, thereby improving the efficiency of handling abnormal conditions, saving manpower and material resources, reducing operation and maintenance costs, and effectively improving user experience.
[0086] Example 2:
[0087] This embodiment provides a photovoltaic equipment safety detection system, as shown in FIG2 , which includes:
[0088] an equipment monitoring unit for detecting the status of the photovoltaic power station and recording detection information, and sending the detection information to the abnormality analysis unit;
[0089] An analysis and management unit receives the detection information, determines whether there is an abnormality based on the detection information, outputs an abnormal instruction when an abnormality exists, parses the abnormal instruction to form an inquiry instruction, and sends the inquiry instruction to the APP end;
[0090] An APP terminal is provided at the user's location, and is used to receive the query instruction and display it to the user. The user enters the reply information through the APP terminal, and the reply information is sent to the analysis management unit;
[0091] Among them, the analysis and management unit confirms whether to continue to issue an inquiry instruction to the user or issue an execution instruction to the photovoltaic power station based on the reply information. The photovoltaic power station takes action to eliminate abnormalities or self-check and report abnormalities according to the execution instruction; and feeds back the execution instruction to the preprocessing port, and performs model switching or model self-learning according to the execution instruction to iteratively upgrade and obtain accurate detection results.
[0092] In one embodiment, the analysis management unit includes:
[0093] An input preprocessing module is used to receive the abnormal instruction and capture the features of the abnormal instruction to form effective feature data;
[0094] A context synthesis processing module is connected to the input pre-processing module, receives the valid feature data, and retrieves and calls data matching the valid feature data to form a comprehensive input instruction;
[0095] The large language model is connected to the input preprocessing module, receives the comprehensive input instruction and processes it to form a query instruction or an execution instruction.
[0096] In this embodiment, the large language model used is the currently popular ChatGPT model. The ChatGPT model is a natural language processing tool driven by artificial intelligence technology. It can conduct conversations by understanding and learning human language, and can also interact according to the context of the chat. Therefore, the context comprehensive processing module of this embodiment outputs the comprehensive input instructions to the non-word single word model to form an output instruction based on natural language. This ChatGPT model can be trained using an open source algorithm to obtain a model or deployed after fine-tuning an open source model. Taking site opening as an example, fine-tune can convert the site opening process into a series of conversations, and then use these conversations as training data to fine-tune the large language model to make it more suitable for site opening scenarios.
[0097] In one embodiment, the analysis and management unit further includes a memory storage module, which is used to store photovoltaic standard data and real-time interaction data. The memory storage module is connected to the context integration processing module.
[0098] Specifically, the memory storage module includes a permanent storage submodule and a flash memory submodule. The permanent storage submodule is used to store photovoltaic standard data, and the flash memory submodule is used to store real-time interaction data. The memory storage module includes a storage management submodule, which is connected to the large language model and is used to receive the feedback query instructions and the response information and update them in the real-time interaction data.
[0099] In one embodiment, the analysis and management unit also includes an output parsing module, which receives the execution instruction and parses the execution instruction to form a control instruction. The control instruction is in a machine language format after formatting and parsing. The photovoltaic power station receives the control instruction and executes it to eliminate abnormalities or self-check and report abnormalities.
[0100] In one embodiment, the analysis and management unit is configured in the cloud, on the APP side, or at the photovoltaic power station. In this embodiment, the analysis and management unit is preferably set in the cloud, but it can also be configured in the terminal or photovoltaic power station, and can be adjusted according to the user's usage requirements and the data processing performance of each terminal. Even the functional modules in the analysis and management unit can be reasonably divided and respectively arranged in one of the three terminals, the APP side, the cloud side, or the photovoltaic power station. In other words, the processing of the analysis and management unit can be realized by combining the functional modules of the APP side, the cloud side, or the photovoltaic power station. This is also technically feasible.
[0101] The analysis body of the photovoltaic equipment safety detection system of this embodiment is set on the cloud or APP side, which simplifies the user input on the interactive side (APP side). The user only needs to respond to the query instructions through voice or text input, which facilitates the information interaction between the user and the analysis and processing unit, reduces the technical threshold requirements for users to answer questions, and enables users to be more proactive in cooperating with exception handling, which not only improves the user experience, but also better realizes the management of photovoltaic power stations.
[0102] In addition, this application innovatively introduces a large language model (such as the ChatGPT model). The input side of the large language model uses the retrieved text-based exception instructions, and the output side is a natural language-based query instruction or execution instruction. From the front-end input to the back-end processing output, it is all based on natural language processing, which is more in line with the needs of traditional users. During the final specific execution, the execution instruction based on natural language is converted into a format (converted into a control instruction) and then used for specific execution, which reduces the processing difficulty of this photovoltaic information processing, can better meet user needs, improve processing efficiency, and realize the automation of exception handling.
[0103] Based on the detailed description of the specific embodiments of the present application in conjunction with the accompanying drawings below, those skilled in the art will become more aware of the above and other objects, advantages and features of the present application.
[0104] An embodiment of the present application also provides a computing device, referring to Figure 3, the computing device includes a memory 1120, a processor 1110 and a computer program stored in the memory 1120 and capable of being run by the processor 1110, the computer program is stored in the space 1130 for program code in the memory 1120, and when the computer program is executed by the processor 1110, it is implemented to execute any one of the method steps 1131 according to the present invention.
[0105] The embodiment of the present application further provides a computer-readable storage medium. Referring to FIG4 , the computer-readable storage medium includes a storage unit for program code, wherein the storage unit is provided with a program 1131 ′ for executing the method steps according to the present invention, and the program is executed by a processor.
[0106] The present application also provides a computer program product comprising instructions. When the computer program product is run on a computer, the computer is caused to execute the steps of the method according to the present invention.
[0107] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When software is used for implementation, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer loads and executes the computer program instructions, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrations. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).
[0108] Professionals should also be further aware that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0109] Those skilled in the art will appreciate that all or part of the steps in the above-described embodiment method can be performed by instructing a processor through a program, and the program can be stored in a computer-readable storage medium, which is a non-transitory medium, such as a random access memory, a read-only memory, a flash memory, a hard disk, a solid-state drive, a magnetic tape, a floppy disk, an optical disc, and any combination thereof.
[0110] In particular, the present invention also provides a computing device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method described above when executing the computer program.
[0111] The present invention also provides a computer-readable storage medium having a computer program stored therein, wherein the computer program implements the above method when executed by a processor. The computer-readable storage medium is preferably a non-volatile readable storage medium.
[0112] The above embodiments are only for illustrating the technical concept and features of the present invention. Their purpose is to enable people familiar with this technology to understand the contents of the present invention and implement them accordingly. They are not intended to limit the scope of protection of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be included in the scope of protection of the present invention.
Claims
1. A photovoltaic equipment safety detection method, characterized in that: The processing steps include: S1: Detect the status of the photovoltaic power station and send the detection information to the cloud; S2: confirming whether there is an abnormality according to the detection information, outputting an abnormal instruction when an abnormality exists, and parsing and processing the abnormal instruction to form an inquiry instruction; S3: Displaying the inquiry instruction to the user and asking the user to give a reply; S4: confirming whether to continue to issue an inquiry instruction to the user or to issue an execution instruction to the photovoltaic power station according to the reply information, and the photovoltaic power station takes action according to the execution instruction to eliminate the abnormality or self-checks and reports the abnormality; S5: Feedback the execution instruction to the preprocessing port, and perform model switching or model self-learning according to the execution instruction to iteratively upgrade and obtain accurate detection results.
2. The photovoltaic equipment safety detection method according to claim 1, characterized in that: The abnormal instruction is parsed and processed, and the directional data matching the abnormal instruction is retrieved from the photovoltaic standard database to form an inquiry instruction. The photovoltaic standard database includes API call documents, product descriptions, and photovoltaic industry technical information.
3. The photovoltaic equipment safety detection method according to claim 2, characterized in that: In the process of retrieving the directional data matching the abnormal instruction from the photovoltaic standard database and forming an inquiry instruction, The retrieved directional data is read and processed through a large language model to form the query instruction. The directional data matched with the abnormal instruction is the environmental factor or human factor that causes the abnormal instruction. The query instruction is an output instruction based on natural language.
4. The photovoltaic equipment safety detection method according to claim 3, characterized in that: In the process of the large language model reading and processing the directional data to form the query instruction, The directional data is sorted to form a plurality of interactive data, and the plurality of interactive data constitute the inquiry instruction. When the inquiry instruction is displayed to the user, the interaction with the user is performed in sequence according to the plurality of interactive data, and the interactive data is converged according to the response information fed back by the user.
5. The photovoltaic equipment safety detection method according to claim 1, characterized in that: In step S4, in the process of confirming the sending of the execution instruction to the photovoltaic power station according to the reply information, The abnormal instruction is confirmed through the reply information, and an execution strategy is formed by searching and calling in the photovoltaic standard database and issuing the execution instruction. After receiving the execution instruction, the photovoltaic power station performs corresponding actions to resolve the abnormality.
6. The photovoltaic equipment safety detection method according to claim 1, characterized in that: The abnormal instructions include fault instructions, power abnormal instructions, power consumption abnormal instructions, and BMS alarm instructions.
7. The photovoltaic equipment safety detection method according to claim 1, characterized in that: Step S2 and step S4 are completed on the cloud, APP or photovoltaic power station, and step S3 is completed on the APP.
8. The photovoltaic equipment safety detection method according to claim 1, characterized in that: When the preprocessing port is a device end, the device end obtains the execution instruction with a label to optimize and iterate the built-in model of the device end.
9. The photovoltaic equipment safety detection method according to claim 1, characterized in that: When the preprocessing port is the cloud, the cloud obtains the execution instruction with the label to perform optimization iteration and secondary diagnosis on the cloud built-in model.
10. The photovoltaic equipment safety detection method according to claim 1, characterized in that: When the preprocessing port is a large language model, the large language model obtains the execution instruction with labels for deep learning, generates a self-built model, and sends the self-built model to the device or the cloud.
11. The photovoltaic equipment safety detection method according to claim 1, characterized in that: In step S1, the state detection of the photovoltaic power station is arc detection and inspection.
12. The photovoltaic equipment safety detection method according to claim 11, characterized in that: Inspections of photovoltaic power stations include DC insulation impedance value, physical location of faults, device operating status, and battery status.
13. A photovoltaic equipment safety detection system, characterized in that: include: The equipment monitoring unit detects the status of the photovoltaic power station and records the detection information. Send to the analysis management unit; An analysis management unit receives the detection information, confirms whether there is an abnormality according to the detection information, outputs an abnormal instruction when an abnormality exists, parses the abnormal instruction to form an inquiry instruction, and sends the inquiry instruction to the APP end; An APP end is provided at the user's place, and is used to receive the inquiry instruction and display it to the user. The user inputs the reply information through the APP end, and the reply information is sent to the analysis management unit; The analysis management unit determines whether to continue to issue an inquiry instruction to the user or to issue an execution instruction to the photovoltaic power station according to the reply information, and the photovoltaic power station takes action to eliminate the abnormality or self-checks and reports the abnormality according to the execution instruction; The execution instruction is fed back to the preprocessing port, and model switching or model self-learning is performed according to the execution instruction to iteratively upgrade and obtain accurate detection results.
14. The photovoltaic equipment safety detection system according to claim 13, characterized in that: The analysis management unit comprises: An input preprocessing module is used to receive the abnormal instruction and capture the features of the abnormal instruction to form effective feature data; A context comprehensive processing module, connected to the input preprocessing module, receives the valid feature data, and retrieves and calls data matching the valid feature data to form a comprehensive input instruction; The large language model is connected to the input preprocessing module, receives the comprehensive input instruction and processes it to form a query instruction or an execution instruction.
15. The photovoltaic equipment safety detection system according to claim 14, characterized in that: The analysis management unit also includes a memory storage module, which is used to store photovoltaic standard data and real-time interactive data. The memory storage module is connected to the context comprehensive processing module.
16. The photovoltaic equipment safety detection system according to claim 15, characterized in that: The memory storage module includes a permanent storage submodule and a flash memory submodule. The permanent storage submodule is used to store photovoltaic standard data, and the flash memory submodule is used to store real-time interactive data.
17. The photovoltaic equipment safety detection system according to claim 16, characterized in that: The memory storage module includes a storage management submodule, which is connected to the large language model and is used to receive the feedback query instructions and the response information and update them in the real-time interaction data.
18. The photovoltaic equipment safety detection system according to claim 13, characterized in that: The analysis management unit also includes an output parsing module, which receives the execution instruction and parses the execution instruction to form a control instruction. The control instruction is in a machine language format after formatting and parsing. The photovoltaic power station receives and executes the control instruction to eliminate abnormalities or self-check and report abnormalities.
19. The photovoltaic equipment safety detection system according to any one of claims 13 to 18, characterized in that: The analysis and management unit is configured in the cloud, APP or photovoltaic power station.
Citation Information
Patent Citations
Distributed photovoltaic power station monitoring device, system and method
CN108449051A
Photovoltaic power station monitoring method and system
CN111404482A
System and method for monitoring distributed photovoltaic power station
CN112736959A
Photovoltaic power station fault detection method based on fault knowledge graph
CN114765451A
System and method for performance analysis and classification of losses for solar power systems
US20160190984A1
Cited By
Lightweight end-to-end defect detection method based on hanging rail type photovoltaic cleaning robot
CN120411091A
