Photovoltaic parking shed current fault detection method based on Internet of Things and related device

Abstract

The invention discloses a photovoltaic parking shed current fault detection method based on the Internet of Things and a related device, and the method comprises the steps: collecting current data and temperature data in real time through Internet of Things sensors disposed on a photovoltaic panel, an inverter and a charging pile; the collected data is sent to an edge computing node for preprocessing, the preprocessing comprises filtering, standardization and resistance-temperature relation modeling, and a dynamic resistance reference curve is generated; uploading the preprocessed data to a cloud server, and analyzing the matching degree of the current waveform and the dynamic resistance reference curve in combination with a long-short-term memory model trained by historical data; in response to the fact that the current waveform abnormally fluctuates or deviates from the dynamic resistance reference curve, generating a diagnosis instruction containing a fault type and positioning information; and after the operation and maintenance center verifies the diagnosis instruction, executing remote repair or triggering an alarm, and storing the operation record to the block chain, thereby improving the recognition accuracy of the fault type and reducing the false alarm rate.

Description

Technical Field

[0001] The present application relates to the technical field of photovoltaic equipment, and in particular to a photovoltaic parking shed current fault detection method and related devices based on the Internet of Things. Background Art

[0002] Photovoltaic carports, by integrating photovoltaic power generation with parking functions, hold significant value in the field of renewable energy utilization. However, the stability of their current system directly impacts power generation efficiency and equipment lifespan. Abnormal current flow, particularly in complex environments such as high humidity and large temperature swings, can cause equipment failures or even accidents. Therefore, efficient fault detection methods are urgently needed.

[0003] At present, current fault detection in photovoltaic carports mainly relies on operation and maintenance personnel to judge faults through on-site inspection of equipment operating parameters and appearance status, or by presetting a fixed current threshold range, collecting real-time data through sensors and comparing it with the threshold. If the range is exceeded, an alarm is triggered. Manual inspections are inefficient and costly, and rely on personnel experience. In addition, the threshold judgment has a high false alarm rate. Summary of the Invention

[0004] The main technical problem solved by this application is to provide a photovoltaic parking shed current fault detection method and related devices based on the Internet of Things, so as to improve the accuracy of fault type identification and reduce the false alarm rate.

[0005] In order to solve the above technical problems, a technical solution adopted in this application is: to provide a photovoltaic carport current fault detection method based on the Internet of Things, the method comprising: real-time collection of current data and temperature data through Internet of Things sensors deployed on photovoltaic panels, inverters and charging piles; sending the collected data to the edge computing node for preprocessing, the preprocessing including filtering, standardization and resistance-temperature relationship modeling to generate a dynamic resistance reference curve; uploading the preprocessed data to the cloud server, combining the long-short term memory model trained with historical data to analyze the matching degree between the current waveform and the dynamic resistance reference curve; in response to detecting abnormal fluctuations in the current waveform or deviation from the dynamic resistance reference curve, generating a diagnostic instruction containing fault type and location information; after the operation and maintenance center verifies the diagnostic instruction, it performs remote repair or triggers an alarm, and stores the operation record to the blockchain.

[0006] The resistance-temperature relationship modeling includes fitting the exponential relationship between resistance and temperature according to the following formula: ;in, represents the predicted resistance value at temperature T; Base temperature The resistance value measured under the condition of ; k is the material coefficient, which is related to the sensitivity of the photovoltaic device material to temperature changes. is a compensation parameter based on the self-heating of the device.

[0007] Among them, constructing the long-short-term memory model includes: setting up a dual-channel mechanism in front of the input layer of the long-short-term memory model, the dual-channel mechanism includes a first channel and a second channel, the first channel inputs a historical current waveform sequence, and the second channel inputs a dynamic resistance reference curve, and data alignment of the first channel and the second channel is achieved through timestamp synchronization; the output layer is defined as five types of fault type labels, and the five types of fault type labels include: physical aging, chemical corrosion, instantaneous overload, poor contact, and normal state; and the cross-entropy loss algorithm is used to calculate the loss of the long-short-term memory model.

[0008] Among them, the cross entropy loss algorithm includes the following formula: Where Loss is the loss value, which is used to measure the difference between the LSTM model and the true label. The smaller the loss value, the more accurate the LSTM model's prediction. N is the total number of samples participating in the training. i is the i-th sample. c is the c-th fault label, where c = 1, 2, 3, 4, 5. 1 represents physical aging, 2 represents chemical corrosion, 3 represents instantaneous overload, 4 represents poor contact, and 5 represents normal state. is the c-th class true label of the i-th sample, using one-hot encoding. If sample i belongs to category c, then ,otherwise ; is the predicted probability of the long short-term memory model for the i-th sample belonging to the c-th fault label, output through the Softmax layer, The probability of taking the value of is [0,1], and the sum of all fault label probabilities of the same sample is 1.

[0009] Among them, the verification of the diagnostic instruction includes: the operation and maintenance center generates a first dynamic key containing a timestamp and embeds it into the diagnostic instruction; after the edge computing node receives the instruction, it encrypts the timestamp through a pre-stored hash function to generate a second dynamic key; if the first dynamic key matches the second dynamic key, the instruction is executed, otherwise a security alarm is triggered.

[0010] Storing operation records in the blockchain includes generating a maintenance log containing a timestamp, an operator's unique identification code, and device status based on a smart contract, and writing the log hash value into a chain code based on the Hyperledger fabric.

[0011] The method also includes: regularly calling historical data in the blockchain, and evaluating the health status of the photovoltaic system by comparing the attenuation rate of the measured resistance with the historical resistance.

[0012] To solve the above technical problems, another technical solution adopted by this application is to provide a photovoltaic carport current fault detection device based on the Internet of Things, which includes: an acquisition module, a processing module, an analysis module, a response module, and an operation and maintenance and storage module. Among them, the acquisition module: uses the Internet of Things sensors deployed on the photovoltaic panels, inverters, and charging piles to collect current data and temperature data in real time; the processing module: sends the collected data to the edge computing node for preprocessing, which includes filtering, normalization, and resistance-temperature relationship modeling to generate a dynamic resistance reference curve; the analysis module: uploads the preprocessed data to the cloud server, and analyzes the matching degree between the current waveform and the dynamic resistance reference curve based on the long-short-term memory model trained with historical data; the response module: generates a diagnostic instruction containing the fault type and location information in response to detecting abnormal fluctuations in the current waveform or deviation from the dynamic resistance reference curve; the operation and maintenance and storage module: after the operation and maintenance center verifies the diagnostic instruction, it performs remote repair or triggers an alarm, and stores the operation record in the blockchain.

[0013] In order to solve the above technical problems, another technical solution adopted in this application is: to provide a photovoltaic parking shed current fault detection device based on the Internet of Things, and the photovoltaic parking shed current fault detection device based on the Internet of Things includes: a memory and at least one processor, the memory stores instructions; at least one processor calls the instructions in the memory to enable the photovoltaic parking shed current fault detection device based on the Internet of Things to perform the steps of the photovoltaic parking shed current fault detection method based on the Internet of Things as any of the above items.

[0014] In order to solve the above technical problems, another technical solution adopted in this application is: providing a computer-readable storage medium, on which instructions are stored, and when the instructions are processed and executed, the steps of the current fault detection method of the photovoltaic carport based on the Internet of Things as any of the above items are implemented.

[0015] Different from the existing technology, the beneficial effects of this application are: by deploying multimodal IoT sensors on photovoltaic panels, inverters and charging piles, real-time collection of current, temperature and ambient humidity data can be achieved, covering the full-dimensional operating status of the current system, solving the blind spot problem of traditional single current monitoring; generating a dynamic resistance reference curve based on the resistance-temperature index relationship model to improve temperature drift adaptability; analyzing the matching degree between the current waveform and the dynamic reference curve through the cloud-based long-short-term memory model to improve the accuracy of fault type identification and reduce the false alarm rate; embedding dynamic keys in diagnostic instructions to ensure the security of instruction transmission and prevent malicious tampering; operation and maintenance records are written to the blockchain through smart contracts to achieve operation timestamps, personnel-specific identification codes and equipment status cannot be tampered with, thereby improving data traceability efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of an implementation method of a photovoltaic carport current fault detection method based on the Internet of Things in this application.

[0017] Figure 2 It is a structural framework diagram of an embodiment of a photovoltaic carport current fault detection device based on the Internet of Things in this application.

[0018] Figure 3 This is a schematic diagram of the structural framework of an implementation method of a photovoltaic carport current fault detection device based on the Internet of Things in this application. DETAILED DESCRIPTION

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. 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.

[0020] The terms "including," "having," and any variations thereof, as used in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to the process, method, product, or device.

[0021] To facilitate understanding of this embodiment, a photovoltaic carport current fault detection method based on the Internet of Things disclosed in an embodiment of the present invention is first introduced in detail. Figure 1 As shown, Figure 1 This is a flow chart of an implementation method of a photovoltaic carport current fault detection method based on the Internet of Things in this application. The method includes the following steps.

[0022] Step S11: The current data and temperature data are collected in real time through IoT sensors deployed on photovoltaic panels, inverters and charging piles.

[0023] Specifically, to effectively detect current faults during the operation of a photovoltaic carport, comprehensive and real-time operational data from related equipment is required. IoT sensors, as key components for data collection, are deployed in strategic locations such as photovoltaic panels, inverters, and charging stations.

[0024] In some specific embodiments, IoT sensors with high-precision sensing technology can be used. For example, for current data collection, current sensors based on the Hall effect principle can be used. They can accurately measure the current in the circuit and have a fast response speed, promptly capturing instantaneous changes in current. For temperature data collection, thermistor temperature sensors can be used to accurately measure temperature by sensing changes in resistance with temperature. These sensors offer excellent stability and reliability, and can operate continuously in complex environments, ensuring that the collected data truly reflects the operating status of the device.

[0025] In some application scenarios, photovoltaic parking sheds may face different usage situations. In commercial parking lot scenarios, vehicles need to charge frequently, and charging piles are used more intensively. In this case, IoT sensors need to have high-frequency data collection capabilities to timely monitor the current and temperature changes of charging piles under different charging states. In residential parking lots, the use of photovoltaic parking sheds is relatively scattered, but they may be affected by environmental factors such as large temperature differences between day and night. This requires sensors to be able to adapt to a large temperature range and accurately collect data. In addition, in photovoltaic parking sheds in industrial plants, due to the complex surrounding electromagnetic environment, sensors must also have strong anti-interference capabilities to ensure that the collected data is not interfered with by external factors, providing a reliable basis for subsequent fault detection.

[0026] Step S12: Send the collected data to the edge computing node for preprocessing, which includes filtering, standardization, and resistance-temperature relationship modeling to generate a dynamic resistance reference curve.

[0027] Specifically, the collected current and temperature data may contain noise and other interfering factors, and data collected by different sensors may vary in terms of dimensions and numerical ranges. Filtering removes noise from the data, making it smoother and more accurate for subsequent analysis. Standardization unifies data of varying ranges and dimensions onto a standard scale, facilitating model processing and comparison.

[0028] In some specific embodiments, filtering can use a Kalman filter algorithm. Kalman filtering is an algorithm that uses a linear system state equation to optimally estimate the system state through system input and output observation data. When processing data collected from photovoltaic parking sheds, it can effectively predict and correct noise in the data, improving data reliability. For standardization, the Z-score standardization method can be used, which converts the original data into standardized data with a mean of 0 and a standard deviation of 1 by calculating the mean and standard deviation of the data.

[0029] In some application scenarios, when photovoltaic carports are exposed to high temperature and high humidity, the data collected by sensors fluctuates significantly and increases in noise. In these cases, the Kalman filter algorithm can better handle this fluctuating data and ensure effective filtering. In scenarios where multiple sensor types are used together, the Z-score normalization method can effectively standardize the data scale, making subsequent analysis more accurate, as the data collected by different sensors can vary significantly. For photovoltaic devices using novel materials, extensive experiments are conducted to determine the parameters of the resistance-temperature relationship model, generating a dynamic resistance reference curve that conforms to actual conditions. This helps to more accurately detect current faults.

[0030] In some embodiments, modeling the resistance-temperature relationship includes fitting an exponential relationship between resistance and temperature according to the following formula: ;in, represents the predicted resistance value at temperature T; Base temperature The resistance value measured under the condition of ; k is the material coefficient, which is related to the sensitivity of the photovoltaic device material to temperature changes. is a compensation parameter based on the self-heating of the device.

[0031] Specifically, the material coefficient k can be determined by collecting multiple sets of temperature-resistance data points within a controllable temperature range for a specific type of photovoltaic device in a laboratory environment, and then fitting the optimal k value through a nonlinear regression algorithm. It can be obtained through equipment thermal characteristic testing. For example, when the equipment is in normal operation, the difference between the equipment surface temperature and the ambient temperature is measured to establish a temperature compensation model.

[0032] In some specific embodiments, the nonlinear regression algorithm can use a least squares method to optimize the k value by minimizing the mean square error between the predicted resistance value and the actual measured value. During data collection, the temperature range can be set to -20°C to 80°C, with a set of data points collected every 5°C, for a total of 21 sets of data points, to ensure the applicability and accuracy of the model.

[0033] In some application scenarios, a database of material coefficient k can be established for different types of photovoltaic equipment, and the corresponding k value can be directly retrieved from the database according to the equipment model to improve modeling efficiency. It can be adjusted dynamically according to the operating power of the equipment, for example, by establishing The mapping relationship table with the device power enables more accurate temperature compensation.

[0034] Step S13: Upload the pre-processed data to the cloud server, and analyze the matching degree between the current waveform and the dynamic resistance reference curve in combination with the long short-term memory model trained with historical data.

[0035] In some specific embodiments, a dual-channel mechanism is provided before the input layer of the long-short-term memory model. The first channel inputs a historical current waveform sequence, and the second channel inputs a dynamic resistance reference curve. Data alignment between the two channels is achieved through timestamp synchronization. The output layer is defined as five types of fault labels: physical aging, chemical corrosion, transient overload, poor contact, and normal state. One-hot encoding is used, for example, [1,0,0,0,0] represents physical aging. The loss function is calculated using the following cross-entropy loss algorithm formula: Where Loss is the loss value, which is used to measure the difference between the LSTM model and the true label. The smaller the loss value, the more accurate the LSTM model's prediction. N is the total number of samples participating in the training. i is the i-th sample. c is the c-th fault label, where c = 1, 2, 3, 4, 5. 1. Physical aging, 2. Chemical corrosion, 3. Transient overload, 4. Poor contact, 5. Normal state. is the c-th class true label of the i-th sample, using one-hot encoding. If sample i belongs to category c, then ,otherwise ; is the predicted probability of the long short-term memory model for the i-th sample belonging to the c-th fault label, output through the Softmax layer, The probability of taking the value of is [0,1], and the sum of all fault label probabilities of the same sample is 1.

[0036] When training the long-short-term memory model, the input data is a historical current waveform sequence and a dynamic resistance benchmark curve, and the output is a matching score, where the matching score is a scale of [0-1]. The labeled data is manually annotated fault types, such as normal, physical aging, and chemical corrosion. The loss function uses cross-entropy loss, and the optimizer uses the Adam algorithm. The learning rate can be set to 0.001, and the number of training iterations can be set to 1000.

[0037] Step S14: In response to detecting abnormal fluctuations in the current waveform or deviations from the dynamic resistance reference curve, a diagnostic instruction including fault type and location information is generated.

[0038] Specifically, the quantization threshold for "deviation from the dynamic resistance baseline curve" is set at 5%. Specifically, if the real-time resistance value deviates by more than 5% from the resistance value at the corresponding point on the dynamic resistance baseline curve, that point is considered an outlier. This threshold, determined through analysis and verification of historical fault data, effectively reduces the false alarm rate while maintaining a high fault detection rate.

[0039] In some specific embodiments, the abnormality determination threshold can be dynamically adjusted based on the device's operating status. For example, during the device startup phase, the threshold can be appropriately relaxed to 10% to avoid false positives due to startup current surges. During stable operation, the threshold can be restored to 5% to ensure detection sensitivity.

[0040] In some application scenarios, different anomaly detection thresholds can be set for power equipment of varying importance. For critical equipment, the threshold can be set to 3% to increase detection stringency; for non-critical equipment, the threshold can be set to 7% to reduce unnecessary alarms. Furthermore, a dynamic threshold adjustment model can be established based on historical equipment operating data to automatically adjust the threshold based on the equipment's health status.

[0041] In some embodiments, the verification of the diagnostic instruction includes: the operation and maintenance center generates a first dynamic key containing a timestamp and embeds it into the diagnostic instruction; after the edge computing node receives the instruction, it encrypts the timestamp through a pre-stored hash function to generate a second dynamic key; if the first dynamic key matches the second dynamic key, the instruction is executed, otherwise a security alarm is triggered.

[0042] Specifically, the operations center generates a first dynamic key to ensure that the diagnostic instruction has not been tampered with during transmission. The inclusion of a timestamp ensures the key's timeliness, recording the exact time the instruction was generated. This first dynamic key, including the timestamp, is embedded in the diagnostic instruction and transmitted along with the instruction to the edge computing node. Upon receiving the instruction, the edge computing node extracts the timestamp from the instruction and encrypts it using a pre-stored hash function. A hash function is a specialized algorithm that converts data of any length into a fixed-length hash value, making it highly unlikely that different data will produce the same hash value. Here, the edge computing node encrypts the timestamp to generate a second dynamic key. The generated second dynamic key is then compared with the received first dynamic key. If the two match, the instruction has not been modified during transmission, and the edge computing node executes the diagnostic instruction. Otherwise, the instruction may have been tampered with, and the edge computing node triggers a security alert, notifying personnel to verify the instruction's security.

[0043] In some specific embodiments, the operations center can use the current system time as a timestamp, for example, with millisecond accuracy. Assume the current time is 10:10:10:10:10:10:100 milliseconds on October 10, 2024, and use this as the timestamp. The first dynamic key is then generated using the SHA-256 hash function. SHA-256 is a widely used hash function that generates a 256-bit hash value. After receiving the instruction, the edge computing node extracts the timestamp "10:10:10:10:100 milliseconds on October 10, 2024," and performs an encryption operation using the SHA-256 hash function to obtain the second dynamic key. If the two keys match, for example, "abcdef1234567890..." (the actual hash value is a 256-bit hexadecimal string), the instruction is executed. If they do not match, such as if the second dynamic key is "ghijkl7890123456...", a security alert is triggered.

[0044] In some application scenarios, multiple edge computing nodes are responsible for monitoring equipment in different areas of a distributed photovoltaic carport network. When a current fault is detected in a certain area and a diagnostic instruction is generated, the operation and maintenance center encrypts the instruction and sends it. Due to the complex network environment, malicious attacks may attempt to tamper with the instructions. For example, hackers may intentionally modify the diagnostic instructions to interfere with the fault repair process. In this case, if the first dynamic key in the instruction received by the edge computing node does not match the second dynamic key generated by the node, a security alarm will be triggered. Upon receiving the alarm, the operation and maintenance personnel will immediately check the network security status, trace the instruction transmission path, and check for malicious attacks. At the same time, they will resend the correct diagnostic instructions to ensure the safe and reliable operation of the photovoltaic carport fault detection and repair process.

[0045] Step S15: After the operation and maintenance center verifies the diagnostic instructions, it performs remote repairs or triggers an alarm and stores the operation records in the blockchain.

[0046] Specifically, after the operations center receives the diagnostic command verification results from the edge computing node and confirms that the command has not been tampered with and is authentic, it will take appropriate measures based on the fault type and severity of the diagnostic command. If the fault type is resolvable remotely, such as a software parameter error or a remote restart of some devices, the operations center will use specialized remote control software or systems to send repair instructions to the faulty device, enabling remote repair. If the fault is more serious and cannot be resolved remotely, such as hardware damage, the operations center will immediately trigger an alarm and notify relevant technicians via SMS, email, or a specialized operations management system to conduct on-site repairs. To ensure traceability and data security, the operations center will store a record of the operation on the blockchain. The operation record includes the details of the diagnostic command, verification results, the operation performed (specific remote repair steps or alarm information), the operation time, and the operator. The application of blockchain technology ensures that these operation records are encrypted and stored in a distributed ledger, making them tamper-proof and ensuring data integrity and credibility.

[0047] In some specific embodiments, suppose that an inverter in a photovoltaic carport has an abnormal current fluctuation fault, and the diagnostic instructions indicate that the fault is caused by an error in the inverter's control software parameters. After the operation and maintenance center verifies the instructions, it connects to the inverter through the remote control software and sends parameter adjustment instructions to the inverter according to the preset repair process to modify the incorrect parameters to the correct values, completing the remote repair. After the repair is completed, the operation and maintenance center will organize the operation records, including the time of the fault occurrence, the content of the diagnostic instructions, the specific instructions and execution time of the repair operation, the operator's employee number, and other information. Based on the smart contract, it generates a maintenance log containing this information and writes the log hash value into the chain code based on Hyperledger Fabric. Hyperledger Fabric is a common blockchain framework that can provide enterprise-level blockchain solutions to ensure the secure storage and efficient management of data.

[0048] In some application scenarios, such as photovoltaic carports in urban public parking lots, the high traffic volume and frequent use of photovoltaic equipment lead to a relatively high probability of failure. When a fault occurs and cannot be repaired remotely, the operations center triggers an alarm to notify technicians. For example, if a charging station experiences a serious short circuit, after the diagnostic instructions are verified, the operations center sends an alert text message and email to the technician responsible for that area, informing them of the fault's location and type. Upon receiving the notification, the technician arrives on-site with specialized tools to carry out the repair. Throughout the entire process, the operations center stores operation records on the blockchain. This not only facilitates subsequent troubleshooting and provides data support for equipment maintenance and performance optimization. For example, if analysis of operation records on the blockchain reveals frequent similar faults in charging stations in a particular area, a comprehensive inspection and upgrade of the equipment in that area can be targeted, improving the overall reliability and stability of the photovoltaic carports.

[0049] In some embodiments, storing the operation record in the blockchain includes: generating a maintenance log including a timestamp, an operator's unique identification code, and a device status based on a smart contract, and writing a hash value of the log into a chain code based on the Hyperledger Fabric.

[0050] Specifically, a smart contract is a computer program that automatically executes contract terms and is deployed on a blockchain. When operation records need to be stored, the system invokes the smart contract. Timestamps record the precise moment an operation occurred, accurate to the second or even millisecond, ensuring traceability of the sequence of operations. Operator-specific identification codes, such as employee IDs or identity verification codes, identify the operator. Equipment status provides detailed information about equipment operating parameters and fault conditions before and after the operation. This information is integrated into a maintenance log. To ensure data security and integrity, the log content is not directly stored on the blockchain. Instead, a hash value is calculated for the log. A hash value is a unique fixed-length string generated by processing the log content using a specific hash algorithm; different log contents generate different hash values. Blockchain platforms built on Hyperledger Fabric provide chaincodes for data management and storage. The calculated log hash value is written to the chaincode, enabling secure storage of operation records on the blockchain.

[0051] In some specific embodiments, suppose that the operation and maintenance personnel of a photovoltaic parking shed performs maintenance operations on a photovoltaic panel with abnormal current. After the operation is completed, the system obtains the current precise time as a timestamp, such as "2024-10-1514:30:25.123". The exclusive identification code of the operation and maintenance personnel is "YW00123", and the equipment status records the specific data of the abnormal current of the photovoltaic panel before the maintenance and the operating parameters restored to normal after the maintenance. This information is combined into a maintenance log in the format specified by the smart contract: "[Timestamp: 2024-10-1514:30:25.123, operator exclusive identification code: YW00123, equipment status: abnormal current before maintenance, specific value is...; current returns to normal after maintenance, operating parameters are...]". Then, the SHA-256 hash algorithm is used to calculate the hash value of the log to obtain a string similar to the following: "56a7f95d3c2c45f88f76d959c8d9a2b37d969358d8676a58f7c9a6f8a56d7e13".

[0052] Finally, the hash value is written into the chain code based on the Hyperledger Fabric through the smart contract to complete the storage of the operation record.

[0053] In some application scenarios, the operation and maintenance of large-scale photovoltaic power plants involves numerous photovoltaic devices and frequent operations. For example, during regular inspections and maintenance of a group of inverters, each operation generates a corresponding operation record. Maintenance logs generated by smart contracts can record detailed information such as the time of each inspection, the operator, and changes in inverter performance parameters before and after the inspection. After the log hash value is written to the Hyperledger Fabric chaincode, these operation records are securely stored. When the maintenance history of a specific inverter is subsequently needed, the corresponding hash value can be quickly retrieved through the blockchain and the operation record can be traced back based on the hash value. This is crucial for the full lifecycle management of equipment, fault tracing, and accountability. If a serious fault occurs on an inverter, reviewing the operation record on the blockchain can clearly understand the previous maintenance situation and determine whether improper operation or untimely maintenance was the problem, providing a strong basis for troubleshooting and improving the operation and maintenance process.

[0054] In some embodiments, the method further includes: periodically calling historical data in the blockchain, and evaluating the health status of the photovoltaic system by comparing the attenuation rate of the measured resistance with the historical resistance.

[0055] Specifically, historical data is regularly retrieved from the blockchain. This data contains resistance measurements of the PV system at different points in time. The measured resistance is the actual resistance value at the current moment, calculated using data collected by sensors deployed on the PV panels, inverters, and charging stations. The decay rate is typically calculated using the formula: decay rate = (historical resistance - measured resistance) / historical resistance × 100%. By analyzing this decay rate, the health of the PV system can be assessed. If the decay rate is within a reasonable range, the PV system is operating relatively stably. If the decay rate exceeds the normal range, it may indicate potential issues with the PV system, such as equipment aging or increased corrosion.

[0056] In some specific embodiments, historical data from the blockchain is regularly retrieved once a week. Assume that the historical resistance value of a photovoltaic panel was 10Ω a week ago, and the measured resistance value this week is 10.3Ω. According to the above decay rate formula, the decay rate = (10-10.3) / 10 × 100% = -3%. By analyzing the historical data and decay rates of a large number of similar photovoltaic panels, it is determined that the normal decay rate range for this type of photovoltaic panel is between -5% and 5%. Since the calculated decay rate of -3% is within the normal range, it can be preliminarily determined that the photovoltaic panel is currently in a healthy operating state.

[0057] In some application scenarios, large-scale photovoltaic power plants utilize a vast number of photovoltaic devices. Regularly assessing the health of the photovoltaic system can promptly identify potential faults, preemptively schedule maintenance, and prevent large-scale failures. For example, if analysis reveals that the degradation rates of multiple photovoltaic panels in a certain area are approaching or exceeding the upper limit of the normal range, this may indicate common issues in that area, such as abnormal lighting conditions or increased environmental corrosion. Based on these assessment results, operations and maintenance personnel can conduct focused inspections of that area and promptly implement protective measures or replace aging equipment to ensure stable operation of the entire photovoltaic system, improve power generation efficiency, and reduce operations and maintenance costs.

[0058] In the above solution, multimodal IoT sensors are deployed on photovoltaic panels, inverters and charging piles to collect current, temperature and ambient humidity data in real time, covering the full-dimensional operating status of the current system and solving the blind spot problem of traditional single current monitoring; a dynamic resistance reference curve is generated based on the resistance-temperature index relationship model to improve temperature drift adaptability; the matching degree between the current waveform and the dynamic reference curve is analyzed through the cloud-based long-short-term memory model to improve the accuracy of fault type identification and reduce the false alarm rate; dynamic keys are embedded in diagnostic instructions to ensure the security of instruction transmission and prevent malicious tampering; operation and maintenance records are written to the blockchain through smart contracts to achieve operation timestamps, personnel-specific identification codes and equipment status are tamper-proof and recorded, improving data traceability efficiency.

[0059] See also Figure 2 , Figure 2 This is a schematic diagram of the structural framework of an embodiment of the photovoltaic carport current fault detection device based on the Internet of Things. Figure 2 As shown, the IoT-based photovoltaic carport current fault detection device 20 includes: an acquisition module 21, a processing module 22, an analysis module 23, a response module 24, and an operation and maintenance and storage module 25. The acquisition module 21 collects current and temperature data in real time through IoT sensors deployed on photovoltaic panels, inverters, and charging piles. The processing module 22 sends the collected data to the edge computing node for preprocessing, which includes filtering, normalization, and resistance-temperature relationship modeling to generate a dynamic resistance reference curve. The analysis module 23 uploads the preprocessed data to the cloud server and analyzes the match between the current waveform and the dynamic resistance reference curve using a long-short-term memory model trained with historical data. The response module 24 generates a diagnostic instruction containing fault type and location information in response to detecting abnormal fluctuations in the current waveform or deviation from the dynamic resistance reference curve. The operation and maintenance and storage module 25 verifies the diagnostic instruction and executes remote repairs or triggers an alarm, storing the operation record in the blockchain.

[0060] In some embodiments, the processing module 22 performs resistance-temperature relationship modeling, including fitting an exponential relationship between resistance and temperature according to the following formula: ;in, represents the predicted resistance value at temperature T; Base temperature The resistance value measured under the condition of ; k is the material coefficient, which is related to the sensitivity of the photovoltaic device material to temperature changes. is a compensation parameter based on the self-heating of the device.

[0061] In some embodiments, constructing a long-short-term memory model includes: setting a dual-channel mechanism in front of the input layer of the long-short-term memory model, the dual-channel mechanism includes a first channel and a second channel, the first channel inputs a historical current waveform sequence, and the second channel inputs a dynamic resistance reference curve, and data alignment of the first channel and the second channel is achieved through timestamp synchronization; the output layer is defined as five types of fault type labels, and the five types of fault type labels include: physical aging, chemical corrosion, instantaneous overload, poor contact, and normal state; and a cross-entropy loss algorithm is used to calculate the loss of the long-short-term memory model.

[0062] In some embodiments, the cross entropy loss algorithm used by the processing module 22 includes the following formula: Where Loss is the loss value, which is used to measure the difference between the LSTM model and the true label. The smaller the loss value, the more accurate the LSTM model's prediction. N is the total number of samples participating in the training. i is the i-th sample. c is the c-th fault label, where c = 1, 2, 3, 4, 5. 1 represents physical aging, 2 represents chemical corrosion, 3 represents instantaneous overload, 4 represents poor contact, and 5 represents normal state. is the c-th class true label of the i-th sample, using one-hot encoding. If sample i belongs to category c, then ,otherwise ; is the predicted probability of the long short-term memory model for the i-th sample belonging to the c-th fault label, output through the Softmax layer, The probability of taking the value of is [0,1], and the sum of all fault label probabilities of the same sample is 1.

[0063] In some embodiments, the operation and maintenance and storage module 25 performs verification of diagnostic instructions, including: the operation and maintenance center generates a first dynamic key containing a timestamp and embeds the diagnostic instruction; after the edge computing node receives the instruction, it encrypts the timestamp through a pre-stored hash function to generate a second dynamic key; if the first dynamic key matches the second dynamic key, the instruction is executed, otherwise a security alarm is triggered.

[0064] In some embodiments, the operation and storage module 25 stores the operation record in the blockchain, including: generating a maintenance log including a timestamp, an operator's unique identification code, and a device status based on a smart contract, and writing the log hash value into a chain code based on the Hyperledger fabric.

[0065] In some embodiments, the system further includes: an evaluation module for periodically calling historical data in the blockchain and evaluating the health status of the photovoltaic system by comparing the attenuation rate of the measured resistance with the historical resistance.

[0066] In the above solution, multimodal IoT sensors are deployed on photovoltaic panels, inverters and charging piles to collect current, temperature and ambient humidity data in real time, covering the full-dimensional operating status of the current system and solving the blind spot problem of traditional single current monitoring; a dynamic resistance reference curve is generated based on the resistance-temperature index relationship model to improve temperature drift adaptability; the matching degree between the current waveform and the dynamic reference curve is analyzed through the cloud-based long-short-term memory model to improve the accuracy of fault type identification and reduce the false alarm rate; dynamic keys are embedded in diagnostic instructions to ensure the security of instruction transmission and prevent malicious tampering; operation and maintenance records are written to the blockchain through smart contracts to achieve operation timestamps, personnel-specific identification codes and equipment status are tamper-proof and recorded, improving data traceability efficiency.

[0067] Figure 3This is a schematic diagram of the structural framework of an embodiment of the photovoltaic carport current fault detection device based on the Internet of Things of this application. The photovoltaic carport current fault detection device 30 based on the Internet of Things may have relatively large differences due to different configurations or performances, and may include one or more processors 31 and memory 32. The processor 31 can be configured to communicate with the memory 32, and execute a series of instruction operations in the memory on the photovoltaic carport current fault detection device based on the Internet of Things to implement the steps of the photovoltaic carport current fault detection method based on the Internet of Things. Those skilled in the art will understand that Figure 3 The structure of the IoT-based photovoltaic carport current fault detection device shown does not constitute a limitation of the IoT-based photovoltaic carport current fault detection device provided by the present invention, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0068] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of a photovoltaic carport current fault detection method based on the Internet of Things.

[0069] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0070] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0071] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A photovoltaic carport current fault detection method based on the Internet of Things, characterized in that: The method comprises: IoT sensors deployed on photovoltaic panels, inverters, and charging piles collect current and temperature data in real time; The collected data is sent to the edge computing node for preprocessing, which includes filtering, normalization, and resistance-temperature relationship modeling to generate a dynamic resistance reference curve; The pre-processed data is uploaded to the cloud server, and the matching degree between the current waveform and the dynamic resistance reference curve is analyzed by combining the long short-term memory model trained with historical data; In response to detecting abnormal fluctuations in the current waveform or deviation from the dynamic resistance reference curve, generating a diagnostic instruction including fault type and location information; After the operation and maintenance center verifies the diagnostic instructions, it performs remote repairs or triggers an alarm and stores the operation records in the blockchain.

2. The photovoltaic carport current fault detection method based on the Internet of Things according to claim 1 is characterized in that: The resistance-temperature relationship modeling includes: The exponential relationship between resistance and temperature is fitted according to the following formula: ;in, represents the predicted resistance value at temperature T; Base temperature The resistance value measured under the condition of ; k is the material coefficient, which is related to the sensitivity of the photovoltaic device material to temperature changes. is a compensation parameter based on the self-heating of the device.

3. The photovoltaic carport current fault detection method based on the Internet of Things according to claim 1 is characterized in that: Constructing the long short-term memory model includes: A dual-channel mechanism is provided before the input layer of the long short-term memory model, the dual-channel mechanism including a first channel and a second channel, the first channel inputting a historical current waveform sequence, the second channel inputting the dynamic resistance reference curve, and data alignment of the first channel and the second channel is achieved through timestamp synchronization; The output layer is defined as five types of fault type labels, including physical aging, chemical corrosion, instantaneous overload, poor contact, and normal state; The cross entropy loss algorithm is used to calculate the loss of the long short-term memory model.

4. The photovoltaic carport current fault detection method based on the Internet of Things according to claim 3 is characterized in that: The cross entropy loss algorithm includes the following formula: Where Loss is the loss value, which is used to measure the difference between the LSTM model and the true label. The smaller the loss value, the more accurate the LSTM model's prediction. N is the total number of samples participating in the training. i is the i-th sample. c is the c-th fault label, where c = 1, 2, 3, 4, 5. 1 represents physical aging, 2 represents chemical corrosion, 3 represents instantaneous overload, 4 represents poor contact, and 5 represents normal state. is the c-th class true label of the i-th sample, using one-hot encoding. If sample i belongs to category c, then ,otherwise ; is the predicted probability of the long short-term memory model for the i-th sample belonging to the c-th fault label, output through the Softmax layer, The probability of taking the value of is [0,1], and the sum of all fault label probabilities of the same sample is 1.

5. The photovoltaic carport current fault detection method based on the Internet of Things according to claim 1 is characterized in that: The verification of the diagnostic instruction includes: The operation and maintenance center generates a first dynamic key including a timestamp and embeds the key into a diagnostic instruction; After receiving the instruction, the edge computing node encrypts the timestamp using a pre-stored hash function to generate a second dynamic key; If the first dynamic key matches the second dynamic key, the instruction is executed, otherwise a security alarm is triggered.

6. The photovoltaic carport current fault detection method based on the Internet of Things according to any one of claims 1 to 5, characterized in that: Storing the operation record in the blockchain includes: generating a maintenance log containing a timestamp, an operator's unique identification code, and a device status based on a smart contract, and writing the log hash value into a chain code based on the Hyperledger Fabric.

7. The photovoltaic carport current fault detection method based on the Internet of Things according to claim 6 is characterized in that: The method further includes: regularly calling historical data in the blockchain and evaluating the health status of the photovoltaic system by comparing the decay rate of the measured resistance with the decay rate of the historical resistance.

8. A photovoltaic carport current fault detection device based on the Internet of Things, characterized in that: The photovoltaic parking shed current fault detection device based on the Internet of Things includes: Acquisition module: Through IoT sensors deployed on photovoltaic panels, inverters and charging piles, current data and temperature data are collected in real time; Processing module: Sends the collected data to the edge computing node for preprocessing, which includes filtering, normalization, and resistance-temperature relationship modeling to generate a dynamic resistance reference curve; Analysis module: uploads the pre-processed data to the cloud server and analyzes the matching degree between the current waveform and the dynamic resistance reference curve by combining the long short-term memory model trained with historical data; A response module: in response to detecting abnormal fluctuation of the current waveform or deviation from the dynamic resistance reference curve, generating a diagnostic instruction including fault type and location information; Operation and maintenance and storage module: After the operation and maintenance center verifies the diagnostic instructions, it performs remote repairs or triggers alarms and stores the operation records in the blockchain.

9. A photovoltaic carport current fault detection device based on the Internet of Things, characterized in that: The photovoltaic carport current fault detection device based on the Internet of Things includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory to enable the Internet of Things-based photovoltaic carport current fault detection device to perform the steps of the Internet of Things-based photovoltaic carport current fault detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instruction is processed and executed, the steps of the photovoltaic carport current fault detection method based on the Internet of Things as described in any one of claims 1 to 7 are implemented.

Images