An Internet-based remote monitoring method and system for a photovoltaic power station

Abstract

The present invention relates to a method and system for remote monitoring of a photovoltaic power station based on the Internet. The method includes: obtaining in real time the total power value input by the photovoltaic power station, generating the root of a tree diagram based on the total power value, and using a preset binary mechanism to combine the current direction input in real time to sense the binary fluctuations of each photovoltaic panel in the photovoltaic power station, and simultaneously calibrating the branches of the tree diagram corresponding to each photovoltaic panel; identifying the SOC values of the respective photovoltaic panels based on the fluctuation data of the binary fluctuations; marking the obtained SOC values one by one on the branches of the tree diagram corresponding to the respective photovoltaic panels in the tree diagram, and performing a digital display process on the tree diagram, and feeding back the digitally displayed tree diagram to the terminal, so as to solve the technical problem that the current technical solution relies on the Internet of Things layout to realize the monitoring of the photovoltaic power station, resulting in high costs.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power station monitoring, and particularly to an Internet-based remote monitoring method and system for photovoltaic power stations. Background Art

[0002] With the rapid development of renewable energy, photovoltaic power stations, as an important part of clean energy, are increasing in scale and quantity. However, photovoltaic power stations are usually distributed in remote areas, and the traditional manual inspection method has problems such as low efficiency, high cost, and slow response. Therefore, developing an Internet-based remote monitoring system to achieve real-time, remote, and efficient monitoring of photovoltaic power stations is of great significance for improving the operation efficiency and maintenance convenience of photovoltaic power stations.

[0003] For example, the existing technical solution CN109301914A - A Photovoltaic Microgrid Energy Storage Control Method with SOC Optimization discloses a photovoltaic microgrid energy storage control method with SOC optimization. The specific steps are as follows: construct a photovoltaic microgrid energy storage control system; the central monitoring unit monitors the operation status of the photovoltaic microgrid; use an information collector to collect the current and voltage data of the DC bus, and transmit the collected current and voltage data to the central processing unit. The central processing unit generates a control strategy through calculation and controls the battery and supercapacitor by distinguishing the control layer area; the energy controller calculates the state of charge (SOC) values of the battery and supercapacitor, and compares and judges them with the optimal SOC range. The SOC optimization module adopts a fuzzy self-tuning strategy based on the filtering time constant, and finally completes the correction of the SOC values of the battery and supercapacitor, so that the photovoltaic microgrid system reaches a stable state.

[0004] However, the existing technical solutions rely on the Internet of Things layout to achieve the monitoring of photovoltaic power stations, that is, it is similar to setting Internet of Things monitoring devices on the line channels of each photovoltaic energy panel respectively. Although it is beneficial for subsequent SOC value calculation, the cost is high. Summary of the Invention

[0005] The main object of the present invention is to provide an Internet-based remote monitoring method and system for photovoltaic power stations, aiming to solve the technical problem that the current technical solutions rely on the Internet of Things layout to achieve the monitoring of photovoltaic power stations, resulting in high costs.

[0006] To achieve the above object, the present invention provides an Internet-based remote monitoring method for photovoltaic power stations, including the following steps:

[0007] Obtain in real time the total power value input by the photovoltaic power station, where the electronic total value is the total power generated by each power collection unit of the photovoltaic power station;

[0008] Generate the root of the tree diagram based on the total power value, and use the preset binary mechanism to combine the binary fluctuations of each photovoltaic panel in the photovoltaic power station sensed by the real-time input current direction, and simultaneously calibrate the branches of the tree diagram corresponding to each photovoltaic panel;

[0009] Based on the fluctuation data of the binary fluctuations, identify the SOC values of the respective photovoltaic panels;

[0010] Mark each obtained SOC value on the branch of the tree diagram corresponding to each photovoltaic panel in the tree diagram, and perform data display processing on the tree diagram, and feedback the tree diagram after data display processing to the terminal.

[0011] Further, the steps of using the preset binary mechanism to combine the binary fluctuations of each photovoltaic panel in the photovoltaic power station sensed by the real-time input current direction include:

[0012] Perform binary labeling on the real-time input current, and gradually reverse the labeling to the photovoltaic power station end to obtain the binary fluctuation data of the root of the tree diagram;

[0013] Identify the binary fluctuation node data when the current input by each photovoltaic panel at the photovoltaic power station end reaches the main current, and obtain the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel.

[0014] Further, the steps of identifying the SOC values of the respective photovoltaic panels based on the fluctuation data of the binary fluctuations include:

[0015] Identify the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel to obtain the SOC values of each photovoltaic panel.

[0016] Further, the steps of using the preset binary mechanism to combine the binary fluctuations of each photovoltaic panel in the photovoltaic power station sensed by the real-time input current direction include:

[0017] Retrieve the preset ELC-RNN model to create a root vector for the real-time input current. Among them, a number of binary data storage nodes are trained on the root vector, and the binary data storage nodes are used to record and distinguish the current fluctuation characteristics of different photovoltaic panels in the photovoltaic power station;

[0018] Input the real-time input current data into the ELC-RNN model in chronological order, and the model dynamically updates the binary data storage nodes on the root vector according to the input current data to reflect the current states of each photovoltaic panel in the photovoltaic power station in real time;

[0019] By comparing the real-time updated binary data storage nodes with the preset binary fluctuation pattern, identify the binary fluctuations of each photovoltaic panel in the photovoltaic power station;

[0020] Associate the binary fluctuation data of each identified photovoltaic panel with the corresponding branches in the dendrogram, and calibrate the SOC value on the dendrogram branch corresponding to the correct photovoltaic panel in subsequent steps.

[0021] Furthermore, the training method of the ELC-RNN model includes:

[0022] Collect historical current data of the photovoltaic power station, including the current fluctuation conditions of each photovoltaic panel at different times and under different weather conditions, as the training data set;

[0023] Preprocess the collected historical current data, and the preprocessing includes data cleaning and normalization to eliminate noise and unify the data format;

[0024] Construct an ELC-RNN model, combining long short-term memory network and extreme learning machine;

[0025] Use the preprocessed historical current data to train the ELC-RNN model, adjust the model parameters through an iterative optimization algorithm, and accurately predict and identify the binary fluctuations of the photovoltaic panels;

[0026] During the training process, use the cross-validation method to evaluate the performance of the model;

[0027] After training is completed, save the parameters and structure of the ELC-RNN model.

[0028] The present invention also proposes an Internet-based remote monitoring system for a photovoltaic power station, including:

[0029] An acquisition unit for real-time acquisition of the total power value input by the photovoltaic power station, and the electronic total value is the total power obtained by each power collection unit of the photovoltaic power station;

[0030] A production unit for generating the root of the dendrogram based on the total power value, and using a preset binary mechanism to sense the binary fluctuations of each photovoltaic panel in the photovoltaic power station in combination with the real-time input current direction, and simultaneously calibrate the dendrogram branches corresponding to each photovoltaic panel;

[0031] An identification unit for identifying the SOC value of each of the photovoltaic panels based on the fluctuation data of the binary fluctuations;

[0032] A display unit for calibrating the obtained SOC values one by one on the dendrogram branches corresponding to each photovoltaic panel in the dendrogram, and performing data display processing on the dendrogram, and feeding back the dendrogram after data display processing to the terminal.

[0033] Furthermore, the production unit includes:

[0034] An annotation module, which is used to perform binary annotation on the real-time input current, and gradually back-propagate the annotation to the photovoltaic power station end to obtain the binary fluctuation data at the root of the tree diagram;

[0035] A production module, which is used to identify the binary fluctuation node data when the current input by each photovoltaic panel at the photovoltaic power station end reaches the main current, and obtain the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel.

[0036] Further, the identification unit includes:

[0037] An identification module, which is used to identify the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel to obtain the SOC value of each photovoltaic panel.

[0038] Further, the production unit further includes:

[0039] A model module, which is used to retrieve a preset ELC-RNN model to create a root vector for the real-time input current. Among them, a number of binary data storage nodes are trained on the root vector, and the binary data storage nodes are used to record and distinguish the current fluctuation characteristics of different photovoltaic panels in the photovoltaic power station;

[0040] A current module, which is used to input the real-time input current data into the ELC-RNN model in chronological order. The model dynamically updates the binary data storage nodes on the root vector according to the input current data to reflect the current states of each photovoltaic panel in the photovoltaic power station in real time;

[0041] An update module, which is used to identify the binary fluctuations of each photovoltaic panel in the photovoltaic power station by comparing the real-time updated binary data storage nodes with the preset binary fluctuation patterns;

[0042] An association unit, which is used to associate the identified binary fluctuation data of each photovoltaic panel with the corresponding branches in the tree diagram, and calibrate the SOC value on the branches of the tree diagram corresponding to the correct photovoltaic panel in the subsequent steps.

[0043] Further, the model module includes:

[0044] A data acquisition sub-module, which is used to collect the historical current data of the photovoltaic power station, including the current fluctuation conditions of each photovoltaic panel at different times and under different weather conditions, as the training data set;

[0045] A preprocessing sub-module, which is used to preprocess the collected historical current data. The preprocessing includes data cleaning and normalization to eliminate noise and unify the data format;

[0046] A construction sub-module, which is used to construct an ELC-RNN model by combining a long short-term memory network and an extreme learning machine;

[0047] A training sub-module, which is used to train the ELC-RNN model using the preprocessed historical current data, adjust the model parameters through an iterative optimization algorithm, and accurately predict and identify the binary fluctuations of the photovoltaic panels.

[0048] A verification sub-module, which is used to evaluate the performance of the model by using the cross-validation method during the training process.

[0049] A saving sub-module, which is used to save the parameters and structure of the ELC-RNN model after the training is completed.

[0050] The method and system for remote monitoring of a photovoltaic power station based on the Internet provided by the present invention have the following beneficial effects:

[0051] (1) By the recognition mechanism of the ELC electrical signal, the fluctuations of the electrical signal are deduced inversely, and there is no need to deploy an Internet of Things system at the photovoltaic power station end, reducing costs.

[0052] (2) By obtaining the total power value of the photovoltaic power station in real time and combining the binary mechanism to sense the current fluctuations of each photovoltaic panel, it is possible to achieve comprehensive, real-time, and remote monitoring of the photovoltaic power station.

[0053] (3) Using the preset ELC-RNN model to process the current input in real time, it is possible to dynamically update and identify the binary fluctuations of the photovoltaic panels. This data processing method based on deep learning can more effectively mine and utilize the operation data of the photovoltaic power station, providing strong support for subsequent fault prediction and optimized operation.

[0054] (4) By showing the SOC values of each photovoltaic panel in the form of a tree diagram, it enables the operation and maintenance personnel to intuitively and clearly understand the operation status of the photovoltaic power station. At the same time, feeding back the tree diagram processed by digital display to the terminal enables the operation and maintenance personnel to access the real-time operation data of the photovoltaic power station anytime and anywhere, greatly improving the convenience of operation and maintenance. Description of the Drawings

[0055] Figure 1 is a schematic flow chart of the method for remote monitoring of a photovoltaic power station based on the Internet in an embodiment of the present invention;

[0056] Figure 2 is a structural block diagram of the system for remote monitoring of a photovoltaic power station based on the Internet in an embodiment of the present invention;

[0057] The realization, functional features, and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the accompanying drawings. Detailed Embodiments

[0058] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0059] Referring to Figure 1 , which is a schematic flowchart of a method for remotely monitoring a photovoltaic power station based on the Internet proposed by the present invention, including the following steps:

[0060] S1. Real-time obtain the total power value input by the photovoltaic power station, where the electronic total value is the total power generated by each power collection unit of the photovoltaic power station;

[0061] S2. Generate the root of a tree diagram based on the total power value, and use a preset binary mechanism to combine the real-time input current direction to sense the binary fluctuations of each photovoltaic panel in the photovoltaic power station, and simultaneously mark the branches of the tree diagram corresponding to each photovoltaic panel;

[0062] S3. Based on the fluctuation data of the binary fluctuations, identify the SOC values of the respective photovoltaic panels;

[0063] S4. Mark the obtained SOC values on the branches of the tree diagram corresponding to each photovoltaic panel in the tree diagram, perform data display processing on the tree diagram, and feed back the tree diagram after data display processing to the terminal.

[0064] In the process of specific implementation,

[0065] In S1, the total power value input by the photovoltaic power station is obtained in real time. The total power value is actually the sum of the power generated by all power collection units (such as photovoltaic panels and battery packs) in the photovoltaic power station at the same moment. The underlying logic of this step is that by real-time monitoring and collecting the power output of the photovoltaic power station, a comprehensive and real-time overview of the power station operation status can be obtained. Specifically, a photovoltaic power station usually consists of multiple power collection units, and each unit generates different amounts of power according to factors such as light intensity and temperature environment. In order to accurately understand the power generation capacity of the entire power station, it is necessary to sum up the power of all power collection units to obtain a total power value. This value not only reflects the current power generation level of the power station, but also can serve as the basis for subsequent data analysis, status monitoring and fault warning. The key to realizing this step lies in the application of data collection and transmission technologies. Usually, a photovoltaic power station is equipped with a dedicated data collection system, which can read the power data of each power collection unit in real time and send these data to the system through a communication network. The system will process and analyze the received data to obtain the real-time total power value.

[0066] In S2, the root of the tree diagram is generated based on the power total value obtained in real time, and the preset binary mechanism is used to combine the current direction input in real time to sense the binary fluctuations of each photovoltaic panel in the photovoltaic power station. At the same time, the branches corresponding to each photovoltaic panel are marked on the tree diagram. By constructing a tree diagram structure, the states and relationships of each photovoltaic panel in the photovoltaic power station are visually represented. The root of the tree diagram represents the total power value of the entire photovoltaic power station, which reflects the overall power generation level of the power station. The branches of the tree diagram represent each photovoltaic panel, and their states and changes can be represented by binary fluctuations. To implement this step, it is necessary to use the preset binary mechanism to sense the change in the current direction of the photovoltaic panel. This mechanism can convert the change in the current direction into binary codes, which facilitates data processing and analysis. When the current direction changes, the corresponding binary code also changes, and this change can be regarded as the binary fluctuation of the photovoltaic panel. At the same time, in order to visually display the states and changes of each photovoltaic panel on the tree diagram, they need to be marked. The marking process is actually to correspond the photovoltaic panel to the branch of the tree diagram. In this way, when the state of the photovoltaic panel changes, the corresponding branch on the tree diagram also changes, thus realizing the real-time monitoring of the state of the photovoltaic power station.

[0067] In S3, based on the fluctuation data of the binary fluctuations mentioned above, the SOC values of each photovoltaic panel are identified. The SOC value, that is, the remaining capacity or state of the battery, is an important parameter reflecting the performance of the photovoltaic panel. By monitoring and analyzing the binary fluctuation data of the photovoltaic panel, the current state of the panel, especially its remaining capacity or SOC value, can be inferred. The binary fluctuation data reflects the changes in the current state of the panel, and these changes have a certain correlation with the SOC value of the panel. Therefore, by processing and analyzing these fluctuation data, the SOC values of each photovoltaic panel can be identified.

[0068] In S4, in the remote monitoring method of a photovoltaic power station, the SOC values of each photovoltaic panel are identified through step S3. The next key step is to mark these SOC values one by one on the corresponding branches of the tree diagram for each photovoltaic panel. The underlying logic of this step is that through the intuitive data structure of the tree diagram, the complex state information of the photovoltaic power station can be presented in a hierarchical manner, enabling the operation and maintenance personnel to quickly locate each photovoltaic panel and view its current SOC value. Specifically, the root of the tree diagram represents the total power value of the entire photovoltaic power station, which provides an overview of the overall power generation level of the power station. The branches of the tree diagram are subdivided to each photovoltaic panel. By corresponding these branches to the photovoltaic panels one by one, the SOC value of the panel can be marked on the branch. In this way, the operation and maintenance personnel can intuitively understand the current state of each photovoltaic panel by viewing the tree diagram. To further improve the readability and usability of the information, it is also necessary to perform data display processing on the tree diagram. This includes optimizing the layout of the tree diagram to make it more in line with human visual habits; color-coding or numerically annotating the SOC values to make them more easily recognizable; and adding interactive functions so that the operation and maintenance personnel can further explore the data through operations such as clicking or dragging. Finally, the processed tree diagram is fed back to the terminal, such as devices like the operation and maintenance personnel's computers, mobile phones, or tablets. In this way, no matter where the operation and maintenance personnel are, they can view the real-time status of the photovoltaic power station at any time and promptly discover and handle potential problems. The realization of this step depends on the development of modern communication technology and cloud computing technology, making the transmission and sharing of data more convenient and efficient.

[0069] In one embodiment, the steps of using a preset binary mechanism to combine with the current direction sensed in real time to sense the binary fluctuations of each photovoltaic panel in the photovoltaic power station include:

[0070] Perform binary labeling on the current input in real time and gradually back-infer the labeling to the photovoltaic power station end to obtain the binary fluctuation data at the root of the tree diagram;

[0071] Identify the binary fluctuation node data when the current input of each photovoltaic panel at the photovoltaic power station end reaches the main current to obtain the binary fluctuation data of the corresponding branches of the tree diagram for each photovoltaic panel.

[0072] Specifically, the real-time input current data is processed, converted into binary form, and annotated. This annotation is based on the direction and magnitude characteristics of the current, aiming to convert continuous current data into discrete and easily processable binary data. Then, these binary data are gradually backtracked to the photovoltaic power station end, and finally the binary fluctuation data at the root of the tree diagram are obtained. These data reflect the change of the current state of the entire photovoltaic power station. Binary annotation is a data preprocessing technology that can simplify complex data into an easily processable binary form. In the monitoring of photovoltaic power stations, this technology can help us capture and analyze current changes more effectively. By backtracking the annotation to the photovoltaic power station end step by step, we can obtain an overview of the current state of the entire power station, providing basic data for subsequent fault warning and performance evaluation.

[0073] In one embodiment, the steps of using a preset binary mechanism to combine with the real-time input current direction to sense the binary fluctuations of each photovoltaic panel in a photovoltaic power station include:

[0074] Retrieve the preset ELC-RNN model to create a root vector for the real-time input current. Among them, a number of binary data storage nodes are trained on the root vector, and the binary data storage nodes are used to record and distinguish the current fluctuation characteristics of different photovoltaic panels in the photovoltaic power station;

[0075] Input the real-time input current data into the ELC-RNN model in chronological order. The model dynamically updates the binary data storage nodes on the root vector according to the input current data to reflect the current state of each photovoltaic panel in the photovoltaic power station in real time;

[0076] By comparing the real-time updated binary data storage nodes with the preset binary fluctuation patterns, identify the binary fluctuations of each photovoltaic panel in the photovoltaic power station;

[0077] Associate the identified binary fluctuation data of each photovoltaic panel with the corresponding branches in the tree diagram, and calibrate the SOC value on the branches of the tree diagram corresponding to the correct photovoltaic panels in subsequent steps.

[0078] Specifically,

[0079] Retrieve the preset ELC-RNN model to create a root vector for the real-time input current:

[0080] Explanation: First, we retrieved a pre-trained ELC-RNN model. This model is specifically designed to process the current data of a photovoltaic power station and can perform deep learning and analysis on the input current data. When the real-time input current data enters the model, the model creates a root vector for it. This root vector is a multi-dimensional data representation that contains the key information of the overall current state of the photovoltaic power station.

[0081] Underlying logic: The ELC-RNN model is a deep learning model that combines the long short-term memory (LSTM) network and the encoder-decoder architecture. It can learn complex patterns from sequential data and encode and decode them effectively. In the monitoring of a photovoltaic power station, the ELC-RNN model can capture the time-series features in the current data and encode them into a root vector, providing a basis for subsequent analysis and recognition.

[0082] Several binary data storage nodes are trained on the root vector to record and distinguish the current fluctuation characteristics of different photovoltaic panels in the photovoltaic power station:

[0083] Explanation: On the root vector, we trained several binary data storage nodes. These storage nodes are like labels or indexes that can record and distinguish the current fluctuation characteristics of different photovoltaic panels in the photovoltaic power station. Each storage node is associated with a specific photovoltaic panel and stores the binary representation of the current fluctuation of that panel.

[0084] Underlying logic: The design of the binary data storage nodes is based on the uniqueness of the current fluctuations of the photovoltaic panels. Since each photovoltaic panel has its unique current fluctuation pattern, we can train the model to identify these patterns and associate them with specific storage nodes. In this way, when the model receives new current data, it can quickly match these data with the storage nodes to identify the corresponding photovoltaic panel.

[0085] The real-time input current data is input into the ELC-RNN model in chronological order, and the model dynamically updates the binary data storage nodes on the root vector according to the input current data:

[0086] Explanation: The real-time input current data is input into the ELC-RNN model in chronological order. The model dynamically updates the binary data storage nodes on the root vector according to these data. This means that as new current data is continuously input, the model will update the information in the storage nodes in real time to reflect the current current state of each photovoltaic panel in the photovoltaic power station.

[0087] Underlying Logic: The dynamic update ability of the ELC-RNN model is based on its deep learning algorithm and the memory function of the LSTM network. The LSTM network can remember the long-term dependencies in sequential data and update its internal state according to new input data. In the monitoring of photovoltaic power plants, this dynamic update ability enables the model to track and reflect the current state changes of photovoltaic panels in real time.

[0088] By comparing the real-time updated binary data storage section with the preset binary fluctuation patterns, the binary fluctuations of each photovoltaic panel in the photovoltaic power plant are identified:

[0089] Explanation: The model will compare the real-time updated binary data storage section with the preset binary fluctuation patterns. These preset patterns are established based on the current fluctuation characteristics of photovoltaic panels under normal or faulty conditions. Through comparison, the model can identify the current binary fluctuation state of each photovoltaic panel in the photovoltaic power plant.

[0090] Underlying Logic: The identification process is based on pattern matching and similarity calculation. The model calculates the similarity between the real-time updated storage section and the preset pattern, and judges the current state of the photovoltaic panel according to the similarity level. If the similarity is high, it indicates that the photovoltaic panel is in a normal or expected state; if the similarity is low, it may indicate that the photovoltaic panel has a fault or anomaly.

[0091] Associate the identified binary fluctuation data of each photovoltaic panel with the corresponding branches in the tree diagram, and calibrate the SOC value on the branch of the tree diagram corresponding to the correct photovoltaic panel in subsequent steps:

[0092] Explanation: Once the model identifies the binary fluctuation state of each photovoltaic panel in the photovoltaic power plant, it will associate these state data with the corresponding branches in the tree diagram. In this way, the operation and maintenance personnel can intuitively understand the current state of each photovoltaic panel by viewing the tree diagram. In subsequent steps, the model will also calibrate the identified SOC value on the branch of the tree diagram corresponding to the correct photovoltaic panel for the operation and maintenance personnel to conduct further monitoring and analysis.

[0093] Underlying Logic: The implementation of this step depends on the accurate identification of the photovoltaic panel state by the model and the reasonable design of the tree diagram structure. The model accurately identifies the state of the photovoltaic panel through deep learning algorithms and pattern matching techniques, and corresponds these states to the branches in the tree diagram one by one. The design of the tree diagram structure enables the operation and maintenance personnel to conveniently view and analyze the overall state of the photovoltaic power plant and the detailed information of each photovoltaic panel.

[0094] In one embodiment, the training method of the ELC-RNN model includes:

[0095] Collect the historical current data of the photovoltaic power station, including the current fluctuation conditions of each photovoltaic panel at different times and under different weather conditions, as the training data set;

[0096] Preprocess the collected historical current data. The preprocessing includes data cleaning and normalization to eliminate noise and unify the data format;

[0097] Construct an ELC-RNN model, combining the long short-term memory network and the extreme learning machine;

[0098] Use the preprocessed historical current data to train the ELC-RNN model, and adjust the model parameters through an iterative optimization algorithm to accurately predict and identify the binary fluctuations of the photovoltaic panels;

[0099] During the training process, adopt the cross-validation method to evaluate the performance of the model;

[0100] After the training is completed, save the parameters and structure of the ELC-RNN model.

[0101] Specifically,

[0102] Collect historical current data of the photovoltaic power station: This step is the basis for model training. It is necessary to collect the current fluctuation conditions of each photovoltaic panel in the photovoltaic power station at different times (such as daily, monthly, and yearly) and different weather conditions (such as sunny, cloudy, rainy, etc.). These data will be used as the training dataset for subsequent training of the ELC-RNN model. Preprocess the collected historical current data: Preprocessing is a crucial step before data analysis and modeling. It includes data cleaning and normalization. The purpose of data cleaning is to eliminate noise and outliers in the data to ensure the accuracy and reliability of the data. Normalization is to unify the data into the same format and range so that the model can process and learn more effectively. Build the ELC-RNN model: The ELC-RNN model is a deep learning model that combines the long short-term memory (LSTM) network and the extreme learning machine (ELM). The LSTM network can handle long-term dependencies in sequential data and is suitable for time series analysis of the current data of the photovoltaic power station. The ELM is a fast and effective single-hidden-layer feedforward neural network, suitable for classification and regression tasks of large-scale data. Combining LSTM and ELM can make full use of their advantages and improve the prediction and recognition ability of the model. Use the preprocessed historical current data to train the ELC-RNN model: Training is the process of model learning. By using iterative optimization algorithms (such as gradient descent method) to adjust the model parameters, the model can accurately predict and recognize the binary fluctuations of the photovoltaic panels. During the training process, the model will learn the features and patterns in the current data and associate them with the binary fluctuations of the photovoltaic panels. During the training process, use the cross-validation method to evaluate the performance of the model: Cross-validation is a method for evaluating the performance of the model. It divides the dataset into a training set and a validation set, trains the model using the training set, and then tests the performance of the model using the validation set. Through cross-validation, the performance of the model on different data subsets can be evaluated, thus ensuring the generalization ability and stability of the model. After training, save the parameters and structure of the ELC-RNN model: After training, it is necessary to save the parameters and structure of the ELC-RNN model for use in subsequent monitoring and analysis of the photovoltaic power station. The saved model can be loaded into the monitoring system to predict and recognize the real-time input current data, so as to realize the real-time monitoring of the state of the photovoltaic power station and fault warning.

[0103] Reference appendix Figure 2 A remote monitoring system for a photovoltaic power station based on the Internet proposed by the present invention includes:

[0104] An acquisition unit for real-time acquisition of the total power value input by the photovoltaic power station, where the electronic total value is the total power obtained by each power acquisition unit of the photovoltaic power station;

[0105] A production unit for generating the root of a tree diagram based on the total power value, and using a preset binary mechanism to combine the binary fluctuations of each photovoltaic panel in a photovoltaic power station with the real-time input current direction, and simultaneously calibrating the branches of the tree diagram corresponding to each photovoltaic panel;

[0106] An identification unit for identifying the SOC values of the respective photovoltaic panels based on the fluctuation data of the binary fluctuations;

[0107] A display unit for marking the obtained SOC values one by one on the branches of the tree diagram corresponding to each photovoltaic panel in the tree diagram, and performing data display processing on the tree diagram, and feeding back the data-displayed tree diagram to the terminal.

[0108] Among them, the production unit includes:

[0109] A marking module for performing binary marking on the real-time input current, and gradually backtracking the marking to the photovoltaic power station end to obtain the binary fluctuation data of the root of the tree diagram;

[0110] A production module for identifying the binary fluctuation node data when the current of each photovoltaic panel at the photovoltaic power station end is input to the main current, and obtaining the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel.

[0111] The identification unit includes:

[0112] An identification module for identifying the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel to obtain the SOC values of each photovoltaic panel.

[0113] The production unit further includes:

[0114] A model module for invoking a preset ELC-RNN model to create a root vector for the real-time input current, wherein a number of binary data storage sections are trained on the root vector, and the binary data storage sections are used to record and distinguish the current fluctuation characteristics of different photovoltaic panels in the photovoltaic power station;

[0115] A current module for inputting the real-time input current data into the ELC-RNN model in chronological order, and the model dynamically updates the binary data storage sections on the root vector according to the input current data to reflect the current states of each photovoltaic panel in the photovoltaic power station in real time;

[0116] An update module for identifying the binary fluctuations of each photovoltaic panel in the photovoltaic power station by comparing the real-time updated binary data storage sections with the preset binary fluctuation patterns;

[0117] An association unit, which is used to associate the identified binary fluctuation data of each photovoltaic panel with the corresponding branches in the tree diagram, and calibrate the SOC value on the branch of the tree diagram corresponding to the correct photovoltaic panel in the subsequent steps.

[0118] The model module includes:

[0119] A data acquisition sub-module, which is used to collect the historical current data of the photovoltaic power station, including the current fluctuation conditions of each photovoltaic panel at different times and under different weather conditions, as the training data set;

[0120] A preprocessing sub-module, which is used to preprocess the collected historical current data. The preprocessing includes data cleaning and normalization to eliminate noise and unify the data format;

[0121] A construction sub-module, which is used to construct an ELC-RNN model, combining a long short-term memory network and an extreme learning machine;

[0122] A training sub-module, which is used to train the ELC-RNN model using the preprocessed historical current data, and adjust the model parameters through an iterative optimization algorithm to accurately predict and identify the binary fluctuations of the photovoltaic panels;

[0123] A verification sub-module, which is used to evaluate the performance of the model by using a cross-validation method during the training process;

[0124] A saving sub-module, which is used to save the parameters and structure of the ELC-RNN model after the training is completed

[0125] In summary, the total power value input by the photovoltaic power station is obtained in real time, the root of the tree diagram is generated based on the total power value, and the binary fluctuations of each photovoltaic panel in the photovoltaic power station are sensed by using a preset binary mechanism in combination with the real-time input current direction, and at the same time, the branches of the tree diagram corresponding to each photovoltaic panel are calibrated; based on the fluctuation data of the binary fluctuations, the SOC values of the respective photovoltaic panels are identified; each obtained SOC value is calibrated on the branch of the tree diagram corresponding to each photovoltaic panel in the tree diagram, and the tree diagram is subjected to data display processing, and the tree diagram subjected to data display processing is fed back to the terminal to solve the technical problem that the current technical solution relies on the Internet of Things layout to realize the monitoring of the photovoltaic power station, resulting in high costs.

[0126] It should be noted that in this text, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the phrase "comprising an..." does not exclude the presence of additional identical elements in the process, apparatus, article, or method comprising such element.

[0127] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A remote monitoring method for a photovoltaic power station based on the Internet, characterized in that, The steps include: Obtain in real time the total power value input by the photovoltaic power station, where the total power value is the total power generated by each power collection unit of the photovoltaic power station; Generate the root of the tree diagram based on the total power value, and use a preset binary mechanism to combine the current direction input in real time to sense the binary fluctuations of each photovoltaic panel in the photovoltaic power station, and at the same time mark the branches of the tree diagram corresponding to each photovoltaic panel; Based on the fluctuation data of the binary fluctuations, identify the SOC values of the respective photovoltaic panels; Mark the obtained SOC values on the branches of the tree diagram corresponding to each photovoltaic panel in the tree diagram, and perform data display processing on the tree diagram, and feedback the data-displayed tree diagram to the terminal; Invoke the preset ELC-RNN model to create a root vector for the current input in real time. Among them, a number of binary data storage nodes are trained on the root vector, and the binary data storage nodes are used to record and distinguish the current fluctuation characteristics of different photovoltaic panels in the photovoltaic power station. The current fluctuation characteristic is to convert the change of the current direction into a binary code. When the current direction changes, the corresponding binary code also changes. This change can be regarded as the binary fluctuation of the photovoltaic panel; at the same time, in order to intuitively display the states and changes of each photovoltaic panel on the tree diagram, it is necessary to calibrate the states and changes of the respective photovoltaic panels; Input the current data input in real time into the ELC-RNN model in chronological order, and the model dynamically updates the binary data storage nodes on the root vector according to the input current data to reflect the current states of each photovoltaic panel in the photovoltaic power station in real time; By comparing the binary data storage nodes updated in real time with the preset binary fluctuation patterns, identify the binary fluctuations of each photovoltaic panel in the photovoltaic power station; these preset binary fluctuation patterns are established based on the current fluctuation characteristics of the photovoltaic panels under normal or faulty conditions; through comparison, the model can identify the current binary fluctuation states of each photovoltaic panel in the photovoltaic power station; Associate the identified binary fluctuation data of each photovoltaic panel with the corresponding branches in the tree diagram, and in subsequent steps, mark the SOC value on the branch of the tree diagram corresponding to the correct photovoltaic panel; among them, The training method of the ELC-RNN model includes: Collect the historical current data of the photovoltaic power station, including the current fluctuation conditions of each photovoltaic panel at different times and under different weather conditions, as the training data set; Preprocess the collected historical current data, and the preprocessing includes data cleaning and normalization to eliminate noise and unify the data format; Construct an ELC-RNN model, combining long short-term memory network and extreme learning machine; Use the preprocessed historical current data to train the ELC-RNN model, and adjust the model parameters through an iterative optimization algorithm to accurately predict and identify the binary fluctuations of the photovoltaic panels; During the training process, use the cross-validation method to evaluate the performance of the model; After the training is completed, save the parameters and structure of the ELC-RNN model.

2. The method for remotely monitoring a photovoltaic power station based on the Internet according to claim 1, characterized in that, Steps for using a preset binary mechanism to combine with the binary fluctuations of each photovoltaic panel in a photovoltaic power station by sensing the current direction of real-time input, including: Perform binary labeling on the real-time input current, and gradually reverse the labeling to the photovoltaic power station end to obtain the binary fluctuation data at the root of the tree diagram; Identify the binary fluctuation node data when the current of each photovoltaic panel at the photovoltaic power station end is input into the main current, and obtain the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel.

3. The remote monitoring method for a photovoltaic power station based on the Internet according to claim 2, wherein Steps for identifying the SOC values of the above-mentioned photovoltaic panels based on the fluctuation data of the binary fluctuations, including: Identify the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel to obtain the SOC values of each photovoltaic panel.

4. A remote monitoring system for a photovoltaic power station based on the Internet, characterized in that, Including: An acquisition unit for real-time acquisition of the total power value input by the photovoltaic power station, where the total power value is the total power obtained by each power acquisition unit of the photovoltaic power station; A production unit for generating the root of the tree diagram based on the total power value, and using a preset binary mechanism to combine with the binary fluctuations of each photovoltaic panel in the photovoltaic power station by sensing the current direction of real-time input, and at the same time calibrating the branches of the tree diagram corresponding to each photovoltaic panel; An identification unit for identifying the SOC values of the above-mentioned photovoltaic panels based on the fluctuation data of the binary fluctuations; A display unit for marking the obtained SOC values one by one on the branches of the tree diagram corresponding to each photovoltaic panel in the tree diagram, and performing data display processing on the tree diagram, and feeding back the data-displayed tree diagram to the terminal; A model module for calling a preset ELC-RNN model to create a root vector for the real-time input current. Among them, several binary data storage nodes are trained on the root vector, and the binary data storage nodes are used to record and distinguish the current fluctuation characteristics of different photovoltaic panels in the photovoltaic power station. The current fluctuation characteristics are to convert the change in the current direction into binary codes. When the current direction changes, the corresponding binary codes also change, and this change can be regarded as the binary fluctuation of the photovoltaic panel; at the same time, in order to intuitively display the states and changes of each photovoltaic panel on the tree diagram, it is necessary to calibrate the states and changes of each photovoltaic panel; A current module for inputting the real-time input current data into the ELC-RNN model in chronological order, and the model dynamically updates the binary data storage nodes on the root vector according to the input current data to reflect the current states of each photovoltaic panel in the photovoltaic power station in real time; An update module for identifying the binary fluctuations of each photovoltaic panel in the photovoltaic power station by comparing the real-time updated binary data storage nodes with the preset binary fluctuation patterns; these preset binary fluctuation patterns are established based on the current fluctuation characteristics of the photovoltaic panels under normal or faulty conditions; through comparison, the model can identify the current binary fluctuation states of each photovoltaic panel in the photovoltaic power station; An association unit is used to associate the identified binary fluctuation data of each photovoltaic panel with the corresponding branches in the tree diagram, and calibrate the SOC value on the branch of the tree diagram corresponding to the correct photovoltaic panel in subsequent steps; among them, The model module includes: A data acquisition sub-module is used to collect the historical current data of the photovoltaic power station, including the current fluctuation conditions of each photovoltaic panel at different times and under different weather conditions, as the training data set; A preprocessing sub-module is used to preprocess the collected historical current data. The preprocessing includes data cleaning and normalization to eliminate noise and unify the data format; A construction sub-module is used to construct an ELC-RNN model, combining a long short-term memory network and an extreme learning machine; A training sub-module is used to train the ELC-RNN model using the preprocessed historical current data, and adjust the model parameters through an iterative optimization algorithm to accurately predict and identify the binary fluctuations of the photovoltaic panels; A verification sub-module is used to evaluate the performance of the model using a cross-validation method during the training process; A saving sub-module is used to save the parameters and structure of the ELC-RNN model after the training is completed.

5. The remote monitoring system for a photovoltaic power station based on the Internet according to claim 4, characterized in that The production unit includes: A labeling module is used to perform binary labeling on the real-time input current, and gradually back-infer the labeling to the photovoltaic power station end to obtain the binary fluctuation data at the root of the tree diagram; A production module is used to identify the binary fluctuation node data when the current input of each photovoltaic panel at the photovoltaic power station end reaches the main current, and obtain the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel.

6. The remote monitoring system for a photovoltaic power station based on the Internet according to claim 5, characterized in that, The identification unit includes: An identification module is used to identify the binary fluctuation data of the branches of the tree diagram corresponding to each photovoltaic panel to obtain the SOC value of each photovoltaic panel.

Images