Intelligent power grid fault early warning method and system for new energy power generation

Abstract

The invention discloses an intelligent power grid fault early warning method and system for new energy power generation, and relates to the technical field of power grid automatic early warning, and the method comprises the following steps: 1, collecting key operation parameters including voltage, current, temperature and the like from new energy power generation equipment in real time, forming collection data, carrying out the primary processing, carrying out the cleaning, and carrying out the fault early warning; and then key features are extracted, so that a basis is provided for subsequent fault analysis. According to the intelligent power grid fault early warning method and system for new energy power generation, various potential fault modes can be accurately identified, an early warning model is continuously optimized through a machine learning algorithm, and the early warning accuracy and reliability are improved; according to the invention, the system reduces the dependence on manual inspection and fault processing, reduces the maintenance cost of equipment and the workload of operation and maintenance personnel through the intelligent fault early warning and automatic adjustment functions, and improves the overall operation efficiency of a new energy power generation system.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic early warning of power grids, and in particular to a smart grid fault early warning method and system for renewable energy power generation. Background Art

[0002] As the global demand for new energy continues to grow, new energy power generation technologies such as wind power, solar energy, and energy storage are being used more and more widely. However, these new energy power generation systems often face complex operating environments. For example, wind power systems need to work under extreme conditions such as high wind speed and low temperature, while solar power generation systems may face challenges such as high temperature and sandstorms. Energy storage systems work in containers and may also encounter harsh environments such as high or low temperatures. These complex working conditions bring many risks to the operation of new energy power generation equipment, especially the potential risk of equipment failure, which puts higher requirements on the operation and management of new energy power generation. Traditional power grid management methods rely on manual experience and conventional monitoring systems. For high-frequency and large-data-volume new energy power generation equipment, it is difficult to conduct real-time and effective early warning and fault handling. This manual fault handling method is not only inefficient, but may also lead to increased safety hazards due to human errors. In addition, the diversity and complexity of new energy power generation systems also require special monitoring and early warning processing technologies for different types of equipment. Existing technologies are difficult to provide a unified and efficient early warning processing solution based on the characteristics and operating laws of different power generation methods. Summary of the invention

[0003] In order to solve the above technical problems, the present invention is implemented by the following technical solutions: a smart grid fault early warning method and system for renewable energy power generation, comprising the following steps: Step 1: Collect key operating parameters including voltage, current, temperature, etc. from new energy power generation equipment in real time to form collected data, and perform preliminary processing, cleaning, i.e. denoising and filtering, and then extract key features to provide a basis for subsequent fault analysis; Step 2: Use multiple warning models to perform fault analysis, including threshold warning model, trend warning model and mutation warning model; in the threshold warning model, when it is detected that the key parameter exceeds the set threshold, a threshold warning signal is immediately generated and the corresponding fault response mechanism is triggered; in the trend warning model, the system analyzes the long-term operation data of the equipment through regression analysis, predicts possible fault trends, and issues trend warning signals before the fault occurs, and triggers the corresponding fault response mechanism; in the mutation warning model, the system monitors sudden changes in the equipment's operating status, such as sudden and drastic fluctuations in current and voltage, and promptly triggers mutation warning signals and corresponding fault response mechanisms; Step 3: The system automatically selects the appropriate fault response mechanism and responds to the fault based on the warning analysis results; in the case of Level I warning, the system automatically executes the emergency shutdown procedure to protect the equipment; in the case of Level II warning, the system automatically adjusts the operating parameters of the equipment and restores the equipment to a safe operating state; in the case of Level III warning, the system continuously monitors the operating status of the equipment and prompts the operator to pay attention to the operation of the equipment.

[0004] Operators can view early warning information, equipment status and fault diagnosis reports in real time, and manually adjust the equipment's operating parameters or perform maintenance operations according to the system's prompts; when necessary, operators can also directly intervene in the equipment through the manual control interface provided by the system.

[0005] After each warning and fault response, the system feeds back relevant data to the big data analysis unit for further analysis and learning; through this closed-loop feedback mechanism, the system can continuously optimize the warning model and rules, improve the accuracy of fault warnings and the overall performance of the system.

[0006] Through the above method, the smart grid fault warning system of the present invention can quickly and accurately identify potential fault risks in a complex and changeable renewable energy power generation environment, and take effective measures in a timely manner to ensure the stable operation of the power grid and improve the safety and operation efficiency of the renewable energy power generation system.

[0007] A smart grid fault warning system for renewable energy power generation, including a data acquisition module, a data processing module, a fault warning module, a control execution module and a human-computer interaction module: The data acquisition module is used to collect data from various types of new energy power generation equipment in real time. The collected data includes the collected equipment operation data, environmental data and historical fault data, and transmits the collected data to the data processing module. The data acquisition module includes a variety of sensors and control interfaces, which can obtain the voltage, current, temperature, wind speed, light intensity and other parameters of the equipment in real time. The data is transmitted to the data processing module via wired or wireless communication to ensure the real-time and accuracy of the data; The data processing module includes a data cleaning unit, a feature extraction unit and a big data analysis unit: the data cleaning unit is used to clean the collected data, and the cleaning method includes denoising and filtering the collected data; the feature extraction unit is used to extract key features from the cleaned collected data, and the big data analysis unit analyzes the extracted key features through a pre-trained big data model; the big data model is based on a machine learning algorithm and can identify complex fault modes and improve the accuracy of early warning; The fault warning module includes multiple fault warning models, which are based on different types of warning algorithms, including threshold warning models, trend warning models and mutation warning models. The fault warning module automatically generates fault warning signals according to data analysis results and set warning rules, and manages them in different levels according to the types and severity of the fault warning signals, and forms a corresponding fault response mechanism. The fault response mechanism is divided into level I warning, level II warning and level III warning: In the case of a Level I warning, the system automatically executes the emergency shutdown procedure; in the case of a Level II warning, the system automatically adjusts the operating parameters of the equipment to restore the equipment to a safe operating state; in the case of a Level III warning, the system continuously monitors the operating status of the equipment and prompts the operator to pay attention to the operating status of the equipment; The control execution module automatically executes corresponding control operations according to the fault warning signal issued by the fault warning module, including but not limited to adjusting equipment operating parameters, switching system operating modes or triggering emergency shutdown procedures to ensure system safety; The human-computer interaction module provides a user interface for operators to view real-time data, warning information and system operating status, and can set and adjust system parameters through the interface. The user interface uses a graphical display method to simplify the operation complexity and supports multi-language switching and remote access functions.

[0008] Preferably, the data acquisition module is connected to the sensor and controller of the new energy power generation equipment, and is used to collect the key operating parameters of the new energy power generation equipment such as voltage, current, temperature, wind speed, light intensity, etc. in real time, and transmit these collected data to the data processing module through wired or wireless communication. The data acquisition module supports multiple communication protocols, including but not limited to Modbus, CAN bus, wireless sensor network (WSN) and Ethernet, to ensure the stability and compatibility of data transmission; The data acquisition module also has a data cache function. When network communication is abnormal, the collected data can be temporarily stored in the local storage unit to form cache data. After the communication is restored, the cache data is transmitted to the data processing module to ensure that the data is not lost; The data acquisition module also supports a distributed data acquisition architecture, which is used to connect new energy power generation equipment in different locations or of different types to the same fault warning system, to achieve cross-regional and cross-device data collection and centralized processing, and to improve the system's adaptability and scalability.

[0009] Preferably, the big data analysis unit in the data processing module adopts a fault pattern recognition model based on deep learning, wherein the big data analysis unit adopts a machine learning algorithm for fault pattern recognition, the algorithm includes but is not limited to support vector machine (SVM), neural network, random forest and deep learning, etc., by training the historical data and real-time data of the new energy power generation equipment, continuously optimizing the fault recognition ability of the model, the system can identify common fault modes and potential risks in equipment operation: The composition of the fault mode recognition model: Data input layer: Receives cleaned collected data, including voltage, current, temperature, wind speed, light intensity and other characteristic parameters. The data input layer is responsible for standardizing these data and converting them into a format suitable for model processing; Feature extraction layer: It consists of multiple convolutional layers that extract key features from the input data. These convolutional layers can identify abnormal features in the operation of the equipment by learning local patterns in the data. Deep learning network: includes a multi-layer feedforward neural network for pattern classification of extracted key features. Through training with a large amount of historical data, the deep learning network can learn the characteristics of different failure modes and accurately classify new data. Classification layer: Use the Softmax function to output the probability distribution of fault classification and determine the fault type corresponding to the current input data; Output layer: outputs fault identification results and passes them to the fault warning module for hierarchical management to generate corresponding fault response mechanisms; Operation steps: The key features obtained from the data cleaning unit form feature data, and the feature data is input into the fault pattern recognition model. The data input layer of the model standardizes these data to improve the training efficiency and recognition accuracy of the model; the input data is processed through the convolution layer to extract key features in the data, such as the frequency characteristics of the equipment current fluctuation, the gradient change of temperature, etc.; the deep learning network performs pattern classification on the extracted key features, identifies whether the current data belongs to a known fault mode, and outputs the classification results; based on the classification results, the fault pattern recognition model generates a fault recognition report and transmits the results to the fault response mechanism to trigger the corresponding fault warning signal to be transmitted to the control execution module; Among them, fault warning signals include threshold warning signals, trend warning signals and mutation warning signals; The fault mode recognition model uses convolutional neural networks in deep learning to extract features and classify them. Its core calculation formula is as follows:

[0010] is the output of the kth convolution kernel at position (i, j); is the weight of the convolution kernel; is the value of the input matrix; is bias; is the activation function, usually the ReLU function.

[0011] Pooling operation (taking maximum pooling as an example):

[0012] is the output after pooling.

[0013] Fully connected layer:

[0014] is the output of the fully connected layer; is the weight of the fully connected layer; is the output of the pooling layer; For bias.

[0015] Output layer:

[0016] It represents the probability that the input data belongs to the jth class.

[0017] The big data analysis unit also includes a prediction model based on time series analysis, which analyzes the time series of equipment operation data, identifies the operation rules and abnormal changes of the equipment, predicts the possible fault occurrence time and fault type in advance, and provides decision support for the fault warning module; The feature extraction unit in the data processing module extracts key fault-related features, such as current fluctuation rate, temperature gradient change, etc., from the equipment operation data based on a multi-dimensional feature extraction algorithm, providing more accurate input data for the big data analysis unit and improving the accuracy and efficiency of the fault pattern recognition model.

[0018] Preferably, the threshold warning model in the fault warning module sets warning thresholds for multiple key parameters. The warning thresholds are dynamically adjusted according to the equipment type, operating environment and historical data. When the actual operating data exceeds or falls below the set threshold, the system immediately generates a threshold warning signal. The threshold warning signal is divided into three levels. According to the level of the threshold warning signal, the system will take different response measures. The threshold warning model of the fault warning module continuously adjusts and optimizes the warning threshold through an automated learning mechanism, builds a warning optimization model, and ensures that the warning threshold can adapt to the long-term operation changes of the equipment and the dynamic changes of the external environment, thereby improving the accuracy of the warning and ensuring that the system maintains efficient warning capabilities in a constantly changing operating environment.

[0019] Preferably, the trend warning model in the fault warning module analyzes the long-term trend of the equipment operation data, predicts possible faults through regression analysis of historical data, and generates a trend warning signal before the predicted fault occurs; The trend warning model specifically adopts a prediction algorithm based on time series analysis and LSTM neural network to predict the long-term operation trend of the equipment and generate trend warning signals in advance; Trend warning signal hierarchical management: when the predicted failure probability reaches a certain threshold, the system will issue a first-level warning and recommend equipment maintenance; When the probability of failure increases further, the system issues a secondary warning and triggers the automatic protection measures of the control execution module, such as automatically adjusting the operating parameters of the equipment or entering maintenance mode to avoid more serious failures of the equipment; Based on the LSTM trend warning model, the system can predict the long-term operation trend of the equipment and provide early warning of potential failure risks.

[0020] Preferably, the mutation warning model in the fault warning module is used to detect sudden changes in the equipment operation state, such as sudden and drastic fluctuations in current and voltage. Once such changes are detected, the system will immediately trigger an early warning and record the event for further analysis; the mutation warning model uses signal processing technologies such as fast Fourier transform and wavelet transform to perform real-time monitoring and frequency domain analysis on the equipment operation signal to identify drastic changes that occur in a short period of time; The mutation warning model consists of: Data input layer: Receives real-time monitoring data, such as current, voltage, etc. The input layer performs preliminary sorting and formatting of these data; FFT transformation layer: Perform fast Fourier transform on the input time domain signal to convert the data from the time domain to the frequency domain in order to detect the frequency components in the signal; Wavelet transform layer: further analyzes the local features in the signal, especially the mutation signal in a short time, and extracts the multi-scale features of the signal through wavelet transform; Mutation detection layer: Based on the results of FFT and wavelet transform, identify the mutation points in the signal and calculate the amplitude and frequency of the mutation; Output layer: Outputs mutation detection results, i.e., mutation warning signals, and transmits the results to the fault warning module for further processing; The mutation warning model analyzes the frequency domain characteristics of real-time data to identify the mutation points of the equipment's operating status. The system determines whether the mutation is a precursor to a potential equipment failure based on the location, amplitude and frequency of the mutation point, generates a corresponding mutation warning signal, and initiates an emergency plan.

[0021] Preferably, the control execution module has two modes: automatic adjustment and manual intervention: In the automatic adjustment mode, the system automatically performs adjustment operations according to the fault warning signal, while in the manual intervention mode, the operator can manually intervene in the operation of the system through the human-computer interaction module and perform manual control operations; In the automatic adjustment mode, upon receiving the fault warning signal, the system immediately performs corresponding control operations, including but not limited to adjusting the workload of the equipment, switching the operation mode or starting the backup power supply, etc., to ensure the safe and stable operation of the system; In manual intervention mode, the system displays fault warning signals and detailed fault information on the user interface for operators to analyze and make decisions. Operators can manually adjust equipment operating parameters, delay shutdown, or trigger other control commands according to actual conditions. Through the manual operation function in manual intervention mode, operators can flexibly respond to various emergencies under special circumstances, thereby improving the system's fault tolerance and operational flexibility. The control execution module also has a remote control function. Operators can remotely access the system to monitor and intervene in the equipment operation status in real time at any location, ensuring that equipment failures can be responded to and handled in a timely manner in an emergency to avoid greater losses.

[0022] Preferably, the user interface of the human-computer interaction module can display the real-time operating status of the equipment, historical warning records and current fault warning signals, and provide fault diagnosis reports and processing suggestions. Users can adjust the equipment operating parameters or take corresponding maintenance measures according to the prompts of the interface; The user interface adopts a multi-level data display structure, which allows operators to access detailed operating information of each device from a global view, facilitating quick problem location. The user interface also integrates data visualization tools to display data changes in the form of charts, trend lines, heat maps, etc., helping operators to intuitively understand the system operation status; The human-computer interaction module supports the query and playback of historical data, so that operators can retrieve equipment operation data and warning records for a specific time period as needed, and analyze the causes of faults and the long-term trend of equipment operation; The human-computer interaction module is also equipped with intelligent fault diagnosis tools. The system automatically generates fault diagnosis reports based on real-time data and historical data. The reports include fault types, possible causes and recommended treatment measures, helping operators to quickly formulate maintenance plans and reduce the impact of equipment failures on system operations. The module also supports multi-user access rights management to ensure the security and confidentiality of system data. Different users can access and operate different functional modules of the system according to their permission levels to ensure the security and compliance of system operations.

[0023] Preferably, the data cleaning unit in the data processing module includes an abnormal data detection and processing mechanism for detecting and processing abnormal data. The abnormal data detection and processing mechanism uses a variety of algorithms to perform quality analysis on the collected data, including but not limited to statistical anomaly detection, machine learning classification model, time series analysis, etc.: The data cleaning unit uses statistical anomaly detection algorithms, such as the three sigma rule and box plot analysis, to identify and process extreme outliers in the collected data. These extreme outliers may be caused by sensor failures, communication errors, etc. The machine learning classification model automatically identifies and marks possible abnormal data by extracting and classifying the features of the collected data. It then combines historical data and equipment operating status to determine whether these abnormal data need to be corrected or deleted. Time series analysis is used to detect abnormal data in terms of time series. In particular, when the change range of an abnormal data point is significantly beyond the normal range, the system will automatically mark the abnormal data and make rational corrections based on the previous and subsequent time series data to ensure the accuracy and consistency of the input data. The data cleaning unit also includes a data interpolation function. When the system detects that the collected data is missing, it can interpolate according to historical data and time series models to restore the missing collected data to the maximum extent and ensure the integrity and reliability of subsequent analysis of the collected data.

[0024] The fault warning module in the system also has a linkage mechanism to support cross-system linkage warning: when one or more devices in the new energy power generation system fail or are abnormal, the system can trigger the warning and protection measures of the associated devices or adjacent grid nodes through the linkage mechanism: the linkage mechanism relies on the network topology information and device dependency model in the system, monitors the operating status of each device in real time, analyzes the possible path of fault propagation, and takes measures in advance to prevent the fault from expanding and the scope of impact from expanding; when the system detects an abnormality or failure of a core device, the linkage mechanism will immediately trigger the status monitoring of the associated devices, and implement preventive adjustments or shutdown operations on these devices according to the preset protection strategy to ensure the stability of the overall system; The system can also automatically generate fault isolation plans based on the early warning results, maintain the normal operation of the power grid to the maximum extent by switching power paths or adjusting power distribution, and provide operators with specific fault isolation suggestions and operation steps through the human-computer interaction module; The linkage warning function can also exchange data and coordinate operations with other adjacent or superior power grid management systems to ensure the sharing of fault information and linkage protection between different levels of the entire power grid, thereby improving the overall risk resistance of the power grid.

[0025] The system has a self-learning function, which can continuously optimize and adjust the parameters and algorithms of the fault warning model based on historical warning data and fault handling results: the system's self-learning function analyzes and models historical fault data and warning responses by integrating machine learning algorithms, identifies the shortcomings of the current warning model, and automatically adjusts the model parameters to make it more consistent with the actual operation of the equipment; The system also supports automatic model updates. When a certain amount of new data is accumulated during system operation, the model retraining process can be automatically started to incorporate the new data into the training set, thereby improving the accuracy and adaptability of the fault warning model. During the update process, the system will save the old version of the model and compare and analyze the performance of the new and old models. Through performance evaluation, it is determined whether the new model can better meet the warning requirements, ensuring that each update is carried out on the basis of system optimization. In addition, the system's self-learning function can automatically generate specific fault warning model configuration files based on different equipment types, operating environments and failure modes. These configuration files can be reused in different operating scenarios, further improving the flexibility and scalability of the system.

[0026] The present invention provides a smart grid fault early warning method and system for renewable energy power generation, which has the following beneficial effects: 1. The smart grid fault warning method and system for renewable energy power generation can achieve more accurate, efficient and intelligent fault warning and processing in a complex renewable energy power generation environment, significantly improving the safety, reliability and operation efficiency of the power grid; it can still maintain strong adaptability and warning capabilities when facing different types of equipment, complex operating environments and changeable fault modes, providing a solid technical guarantee for the stable and safe operation of renewable energy power generation.

[0027] 2. The smart grid fault warning method and system for renewable energy power generation adopts a variety of warning models, including threshold warning models, trend warning models and mutation warning models, which can accurately identify various potential fault modes and continuously optimize the warning model through machine learning algorithms to improve the accuracy and reliability of warnings; reduce dependence on manual inspections and fault handling, and reduce equipment maintenance costs and the workload of operation and maintenance personnel through intelligent fault warning and automatic adjustment functions, thereby improving the overall operating efficiency of renewable energy power generation systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A schematic diagram of a process flow of a smart grid fault early warning method for renewable energy power generation according to the present invention; Figure 2 The present invention is a flowchart of a smart grid fault warning system for renewable energy power generation. DETAILED DESCRIPTION

[0029] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are provided for the purpose of illustration and description, and are not intended to be exhaustive or to limit the present invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

[0030] like Figure 1 As shown, the present invention provides a technical solution: a smart grid fault early warning method and system for renewable energy power generation, comprising the following steps: Step 1: Collect key operating parameters including voltage, current, temperature, etc. from new energy power generation equipment in real time to form collected data, and perform preliminary processing, cleaning, i.e. denoising and filtering, and then extract key features to provide a basis for subsequent fault analysis; Step 2: Use multiple warning models to analyze faults, including threshold warning, trend warning, and mutation warning. In threshold warning, when a key parameter is detected to exceed the set threshold, a threshold warning signal is immediately generated and the corresponding fault response mechanism is triggered. In trend warning, the system analyzes the long-term operation data of the equipment through regression analysis, predicts possible fault trends, and issues trend warning signals before the fault occurs, and triggers the corresponding fault response mechanism. In mutation warning, the system monitors sudden changes in the equipment's operating status, such as sudden and drastic fluctuations in current and voltage, and promptly triggers mutation warning signals and corresponding fault response mechanisms. Step 3: The system automatically selects the appropriate fault response mechanism and responds to the fault based on the warning analysis results; in the case of Level I warning, the system automatically executes the emergency shutdown procedure to protect the equipment; in the case of Level II warning, the system automatically adjusts the operating parameters of the equipment and restores the equipment to a safe operating state; in the case of Level III warning, the system continuously monitors the operating status of the equipment and prompts the operator to pay attention to the operation of the equipment.

[0031] like Figure 2 As shown, a smart grid fault warning system for renewable energy power generation includes a data acquisition module, a data processing module, a fault warning module, a control execution module and a human-computer interaction module: The data acquisition module is used to collect data from various types of new energy power generation equipment in real time. The collected data includes the collected equipment operation data, environmental data and historical fault data, and transmits the collected data to the data processing module. The data acquisition module includes a variety of sensors and control interfaces, which can obtain the equipment's voltage, current, temperature, wind speed, light intensity and other parameters in real time. The data is transmitted to the data processing module via wired or wireless communication to ensure the real-time and accuracy of the data; The data processing module includes a data cleaning unit, a feature extraction unit and a big data analysis unit: the data cleaning unit is used to clean the collected data, and the cleaning method includes denoising and filtering the collected data; the feature extraction unit is used to extract key features from the cleaned collected data, and the big data analysis unit analyzes the extracted key features through a pre-trained big data model; the big data model is based on a machine learning algorithm and can identify complex fault modes and improve the accuracy of early warning; The fault warning module includes multiple fault warning models, which are based on different types of warning algorithms, including threshold warning models, trend warning models and mutation warning models. The fault warning module automatically generates fault warning signals based on data analysis results and set warning rules, and manages them in different levels according to the type and severity of the fault warning signals, and forms a corresponding fault response mechanism. The fault response mechanism is divided into level I warning, level II warning and level III warning: In the case of a Level I warning, the system automatically executes the emergency shutdown procedure; in the case of a Level II warning, the system automatically adjusts the operating parameters of the equipment to restore the equipment to a safe operating state; in the case of a Level III warning, the system continuously monitors the operating status of the equipment and prompts the operator to pay attention to the operating status of the equipment; The control execution module automatically executes corresponding control operations according to the fault warning signal issued by the fault warning module, including but not limited to adjusting equipment operating parameters, switching system operating modes or triggering emergency shutdown procedures to ensure system safety; The human-computer interaction module provides a user interface for operators to view real-time data, warning information and system operation status, and can set and adjust system parameters through this interface. The user interface uses a graphical display method to simplify the complexity of operation, and supports multi-language switching and remote access functions.

[0032] The data acquisition module can obtain the equipment's operating data in real time. Combined with big data analysis and machine learning algorithms, it can immediately issue a warning signal when the equipment malfunctions, greatly shortening the fault response time and reducing the risk of accidents.

[0033] The data acquisition module is connected to the sensors and controllers of the new energy power generation equipment to collect the key operating parameters of the new energy power generation equipment such as voltage, current, temperature, wind speed, light intensity, etc. in real time, and transmit these collected data to the data processing module through wired or wireless communication. The data acquisition module supports multiple communication protocols, including but not limited to Modbus, CAN bus, wireless sensor network (WSN) and Ethernet, to ensure the stability and compatibility of data transmission; The data acquisition module also has a data cache function. When network communication is abnormal, the collected data can be temporarily stored in the local storage unit to form cache data. After the communication is restored, the cache data will be transmitted to the data processing module to ensure that the data is not lost. The data acquisition module also supports a distributed data acquisition architecture, which is used to connect new energy power generation equipment in different locations or of different types to the same fault warning system, to achieve cross-regional and cross-device data collection and centralized processing, and to improve the system's adaptability and scalability.

[0034] The system supports multiple communication methods and data input formats, can flexibly access various new energy power generation equipment, and adjust early warning rules and models according to different equipment types and operating conditions. It has good adaptability and scalability.

[0035] The big data analysis unit in the data processing module adopts a fault pattern recognition model based on deep learning. The big data analysis unit uses machine learning algorithms for fault pattern recognition, including but not limited to support vector machines (SVM), neural networks, random forests, and deep learning. By training the historical data and real-time data of new energy power generation equipment, the fault recognition ability of the model is continuously optimized. The system can identify common fault modes and potential risks in equipment operation: The composition of the fault mode recognition model: Data input layer: Receives cleaned collected data, including voltage, current, temperature, wind speed, light intensity and other characteristic parameters. The data input layer is responsible for standardizing these data and converting them into a format suitable for model processing; Feature extraction layer: It consists of multiple convolutional layers that extract key features from the input data. These convolutional layers can identify abnormal features in the operation of the equipment by learning local patterns in the data. Deep learning network: includes a multi-layer feedforward neural network for pattern classification of extracted key features. Through training with a large amount of historical data, the deep learning network can learn the characteristics of different failure modes and accurately classify new data. Classification layer: Use the Softmax function to output the probability distribution of fault classification and determine the fault type corresponding to the current input data; Output layer: outputs fault identification results and passes them to the fault warning module for hierarchical management to generate corresponding fault response mechanisms; Steps: Data preprocessing: The key features obtained from the data cleaning unit form feature data, which are input into the fault pattern recognition model. The data input layer of the model standardizes these data to improve the training efficiency and recognition accuracy of the model. Feature extraction: The input data is processed through the convolution layer to extract key features in the data, such as the frequency characteristics of device current fluctuations and temperature gradient changes; Pattern classification: The deep learning network performs pattern classification on the extracted key features, identifies whether the current data belongs to a known fault mode, and outputs the classification results; Fault output: Based on the classification results, the fault pattern recognition model generates a fault identification report and transmits the results to the fault response mechanism to trigger the corresponding fault warning signal to be transmitted to the control execution module; Among them, fault warning signals include threshold warning signals, trend warning signals and mutation warning signals; The fault mode recognition model uses convolutional neural networks in deep learning to extract features and classify them. Its core calculation formula is as follows:

[0036] is the output of the kth convolution kernel at position (i, j); is the weight of the convolution kernel; is the value of the input matrix; is bias; is the activation function, here it is the ReLU function.

[0037] By applying the ReLU function, when the input is a negative number, the output is not zero, but a smaller slope is output, which can avoid the "neuron death" problem in the ReLU function.

[0038] Pooling operation (taking maximum pooling as an example):

[0039] is the output after pooling.

[0040] The pooling operation is the process of downsampling the input data. It reduces the spatial dimension of the data, reduces the number of parameters and the amount of calculation, and at the same time extracts the main features to improve the robustness and generalization ability of the model.

[0041] Fully connected layer:

[0042] is the output of the fully connected layer; is the weight of the fully connected layer; is the output of the pooling layer; For bias.

[0043] Output layer:

[0044] It represents the probability that the input data belongs to the jth class.

[0045] Calculation example Assume that the input data is the current signal of a device. After the convolution operation, the extracted feature matrix is ​​h, which is then passed through the maximum pooling and fully connected layers to finally output the prediction result. In actual applications, assuming that the model predicts that the probability of a certain failure of the device is 0.85, the failure mode is identified as a potential failure of the device, and the system will generate an early warning based on the result.

[0046] Through the deep learning fault mode recognition model, the system can automatically learn and identify various failure modes of equipment and improve the accuracy of fault warning.

[0047] The big data analysis unit also includes a prediction model based on time series analysis, which analyzes the time series of equipment operation data, identifies the operation rules and abnormal changes of the equipment, predicts the possible time and type of failure in advance, and provides decision support for the fault warning module; The feature extraction unit in the data processing module extracts key fault-related features from the equipment operation data based on a multi-dimensional feature extraction algorithm, such as current fluctuation rate, temperature gradient changes, etc., to provide more accurate input data for the big data analysis unit and improve the accuracy and efficiency of the fault pattern recognition model.

[0048] The threshold warning model in the fault warning module sets the warning thresholds of multiple key parameters. The warning thresholds are dynamically adjusted according to the equipment type, operating environment and historical data. When the actual operating data exceeds or falls below the set threshold, the system immediately generates a threshold warning signal. The threshold warning signal is divided into three levels. According to the level of the threshold warning signal, the system will take different response measures. The threshold warning model of the fault warning module continuously adjusts and optimizes the warning threshold through an automated learning mechanism, builds a warning optimization model, and ensures that the warning threshold can adapt to the long-term operation changes of the equipment and the dynamic changes of the external environment, thereby improving the accuracy of the warning and ensuring that the system maintains efficient warning capabilities in a constantly changing operating environment; After the threshold warning signal is generated, the system will automatically record the current equipment status, equipment operation data and environmental parameters, and remind the operator to check the equipment through the human-computer interaction module. If necessary, the system will directly issue instructions to the control execution module to perform emergency operations such as shutdown or switching to backup power supply to prevent equipment damage or safety accidents. The composition of the early warning optimization model: Experience database: used to store system historical fault data, warning results and actual response effects, serving as basic data for model optimization; Reinforcement learning algorithm unit: including strategy network and value network. The strategy network generates a new early warning strategy according to the current system status, and the value network evaluates the warning effect of the current early warning strategy. Model update unit: based on the results of the reinforcement learning algorithm unit, update and optimize the parameters of the threshold warning model so that the threshold warning model can better adapt to the operation of the equipment; Feedback mechanism: Apply the optimized threshold warning model to system operation, and continuously collect new data feedback to further improve the model; Operation steps: During operation, the system continuously collects historical fault data, warning results, and actual response effects, and stores these data in the experience database; the reinforcement learning algorithm unit generates a new warning strategy through the policy network, evaluates the effect of the strategy through the value network, and calculates the value function of the strategy; according to the evaluation result of the value function, the model update unit adjusts and optimizes the parameters of the existing threshold warning model, and the updated threshold warning model is applied to the system again; the system puts the optimized threshold warning model into use, and continuously adjusts and improves it according to the actual operation effect to ensure that the threshold warning model can adapt to the long-term operation changes of the equipment; The introduction of reinforcement learning algorithms enables the system to automatically optimize the early warning model based on actual operating data, ensuring that the system maintains efficient fault warning capabilities under different operating conditions and improving system adaptability.

[0049] The trend warning model in the fault warning module analyzes the long-term trend of equipment operation data, predicts possible faults through regression analysis of historical data, and generates trend warning signals before the predicted fault occurs; The trend warning model specifically adopts a prediction algorithm based on time series analysis and LSTM neural network to predict the long-term operation trend of the equipment and generate trend warning signals in advance; The composition of the trend warning model: Time series data input layer: receives historical operation data, such as voltage, temperature, wind speed, etc., and inputs these data into the trend warning model in the form of time series; LSTM network layer: contains multiple LSTM units. The LSTM network is used to process time series data, can capture long-term dependencies during equipment operation, and predict future operating trends; Prediction layer: Use the fully connected layer to convert the output of the LSTM network into specific numerical predictions, such as device temperature, voltage, and other parameters at a certain point in the future; Fault threshold comparison layer: compares the prediction results with the preset thresholds to determine whether they exceed the warning range; Output layer: output prediction results and trend warning signals; Operation steps: Input the historical operation data of the equipment into the LSTM network, and keep the time series characteristics of the data in the input layer for subsequent processing; process the time series data through the LSTM network to predict the future operation status of the equipment, such as the temperature change trend and voltage fluctuation at a certain moment in the future; compare the predicted result with the preset threshold of the system. If the predicted value exceeds the threshold, the system generates a trend warning signal; based on the comparison result, the system decides whether to trigger the warning, and outputs the result to the control execution module and the human-computer interaction module to remind the operator to perform necessary maintenance or adjustments; The trend warning model uses LSTM neural network for time series prediction. The calculation formula is as follows: Calculation of LSTM unit:

[0050]

[0051]

[0052]

[0053]

[0054]

[0055] It is the output of the forget gate, which controls the forgetting of memory; is the output of the input gate, controlling the update of memory; is the output of the output gate, controlling the output at the next moment; for candidate memory; is the unit state; In hidden state.

[0056] Prediction output:

[0057] is the predicted value.

[0058] Calculation example: Predict the temperature change of a device in the next hour. The input data is the temperature time series of the past 24 hours. After LSTM model processing, the predicted temperature value in the next hour is 75℃. The threshold is set to 70℃. At this time, the prediction result exceeds the set threshold, and the system will generate a high temperature warning signal in the trend warning signal.

[0059] The trend warning model is used to analyze the trend of equipment operation data, such as the temperature increase trend and the voltage decrease trend. By establishing a regression model, the equipment operation data trend is evaluated. The system combines historical fault data and current operation status to calculate the probability of potential faults, obtain the fault probability, and generate trend warning signals based on the fault probability and possible impact range. Trend warning signals are managed in a hierarchical manner. A known fixed threshold is set in advance. When the predicted failure probability reaches the fixed threshold, the system will issue a first-level warning and recommend equipment maintenance. When the failure probability increases further, that is, the failure probability is greater than the fixed threshold, the system will issue a second-level warning and trigger the automatic protection measures of the control execution module. Automatic protection measures: automatically adjust the operating parameters of the equipment or enter maintenance mode to avoid more serious equipment failures. Through the LSTM-based trend warning model, the system can predict the long-term operating trend of the equipment and warn of potential failure risks in advance.

[0060] By performing regression analysis and trend prediction on the historical operating data of the equipment, the system can not only identify current failure risks, but also predict possible future failures, thereby taking preventive measures in advance and extending the service life of the equipment.

[0061] The mutation warning model in the fault warning module is used to detect sudden changes in the equipment's operating status, such as sudden and drastic fluctuations in current and voltage. Once such changes are detected, the system will immediately trigger an early warning and record the event for further analysis; The mutation warning model uses signal processing technologies such as fast Fourier transform and wavelet transform to conduct real-time monitoring and frequency domain analysis of equipment operation signals to identify drastic changes that occur in a short period of time; The mutation warning model consists of: Data input layer: Receives real-time monitoring data, such as current, voltage, etc. The input layer performs preliminary sorting and formatting of these data; FFT transformation layer: Perform fast Fourier transform on the input time domain signal to convert the data from the time domain to the frequency domain in order to detect the frequency components in the signal; Wavelet transform layer: further analyzes the local features in the signal, especially the mutation signal in a short time, and extracts the multi-scale features of the signal through wavelet transform; Mutation detection layer: Based on the results of FFT and wavelet transform, identify the mutation points in the signal and calculate the amplitude and frequency of the mutation; Output layer: Outputs mutation detection results, i.e., mutation warning signals, and transmits the results to the fault warning module for further processing; Steps: Real-time data input: The system receives the equipment's operating data in real time and inputs it into the mutation warning model in the form of input signals; FFT transformation: Through the FFT transformation layer, the input signal is converted from the time domain to the frequency domain, its frequency characteristics are analyzed, and the frequency changes in the signal are detected; Wavelet transform: Perform wavelet transform on frequency domain signals to identify short-term mutations in the signals and extract local features of the signals; Mutation detection: The mutation detection layer calculates the amplitude and frequency of mutations, determines whether the mutations may cause equipment failures, and generates mutation warning signals; Warning output: Output the results of mutation detection to the fault warning module to trigger the corresponding fault response mechanism; The mutation warning model analyzes the frequency domain characteristics of real-time data to identify the mutation points of the equipment's operating status. Based on the location, amplitude and frequency of the mutation points, the system determines whether the mutation is a precursor to a potential equipment failure, generates a corresponding mutation warning signal, and initiates an emergency plan. After detecting a sudden change warning signal, the system will automatically record the detailed equipment operation data of the current equipment and transmit the data to the data processing module for in-depth analysis to determine the cause of the sudden change and possible subsequent impact. The operator can view the detailed information of the sudden change event through the human-computer interaction module and make further maintenance or adjustment decisions. By combining the sudden change detection algorithm of FFT and wavelet transform, the system can capture sudden changes in equipment operation in real time and generate early warning signals in time. These improvements enable the smart grid fault warning system of the present invention to achieve more accurate, efficient and intelligent fault warning and processing in complex renewable energy power generation environments, significantly improving the safety, reliability and operating efficiency of the power grid; it can still maintain strong adaptability and warning capabilities when facing different types of equipment, complex operating environments and changeable fault modes, providing a solid technical guarantee for the stable and safe operation of renewable energy power generation.

[0062] By introducing the above-mentioned models and algorithms, the smart grid fault warning system of the present invention has achieved high intelligence and precision in data processing, fault identification and warning generation, enabling the system to accurately process and analyze complex renewable energy power generation data, provide real-time, efficient and intelligent fault warning and response solutions, and ensure the safe and stable operation of the power grid.

[0063] The control execution module has two modes: automatic adjustment and manual intervention: In the automatic adjustment mode, the system automatically performs adjustment operations according to the fault warning signal, while in the manual intervention mode, the operator can manually intervene in the operation of the system through the human-computer interaction module and perform manual control operations; In the automatic adjustment mode, upon receiving the fault warning signal, the system immediately performs corresponding control operations, including but not limited to adjusting the workload of the equipment, switching the operation mode or starting the backup power supply, etc., to ensure the safe and stable operation of the system; In the manual intervention mode, the system will display the fault warning signal and detailed fault information on the user interface for the operator to analyze and make decisions. The operator can manually adjust the equipment operating parameters, delay shutdown or trigger other control commands according to the actual situation. Through the manual operation function in the manual intervention mode, the operator can flexibly respond to various emergencies under special circumstances, improve the system's fault tolerance and operational flexibility; the control execution module also has a remote control function. The operator can remotely access the system to monitor and intervene in the equipment operation status in real time at any location, ensuring that the equipment failure can be responded to and handled in time in an emergency to avoid greater losses.

[0064] The control execution module automatically adjusts the operating parameters of the equipment or executes the emergency shutdown procedure according to the early warning signal, reducing the reliance on manual operation and the possibility of human error. At the same time, the human-computer interaction module provides operators with an intuitive user interface, making system operation easier. The user interface of the human-computer interaction module can display the real-time operating status of the equipment, historical warning records and current fault warning signals, and provide fault diagnosis reports and processing suggestions. Users can adjust the equipment operating parameters or take corresponding maintenance measures according to the prompts on the interface; the user interface adopts a multi-level data display structure, so that operators can access the detailed operating information of each device from a global view, which is convenient for quickly locating problems. The user interface also integrates data visualization tools to display data changes in the form of charts, trend lines, heat maps, etc., to help operators intuitively understand the system operation status; The human-computer interaction module supports the query and playback of historical data, so that operators can retrieve equipment operation data and warning records for a specific period of time as needed, and analyze the causes of faults and the long-term trends of equipment operation. The human-computer interaction module is also equipped with intelligent fault diagnosis tools. The system automatically generates fault diagnosis reports based on real-time and historical data. The reports include fault types, possible causes, and recommended treatment measures, helping operators to quickly formulate maintenance plans and reduce the impact of equipment failures on system operations. The module also supports multi-user access rights management to ensure the security and confidentiality of system data. Different users can access and operate different functional modules of the system according to their permission levels to ensure the security and compliance of system operations.

[0065] The data cleaning unit in the data processing module includes an abnormal data detection and processing mechanism for detecting and processing abnormal data. The abnormal data detection and processing mechanism uses a variety of algorithms to perform quality analysis on the collected data, including but not limited to statistical anomaly detection, machine learning classification models, time series analysis, etc.: The data cleaning unit uses statistical anomaly detection algorithms, such as the three sigma rule and box plot analysis, to identify and process extreme outliers in the collected data. These extreme outliers may be caused by sensor failures, communication errors, etc. The machine learning classification model automatically identifies and marks possible abnormal data by extracting and classifying the features of the collected data, and then combines historical data and equipment operating status to determine whether these abnormal data need to be corrected or deleted. Time series analysis is used to detect abnormal data in terms of time series. In particular, when the change range of an abnormal data point is significantly beyond the normal range, the system will automatically mark the abnormal data and make rational corrections based on the previous and subsequent time series data to ensure the accuracy and consistency of the input data. The data cleaning unit also includes a data interpolation function. When the system detects that the collected data is missing, it can interpolate according to historical data and time series models to restore the missing collected data to the maximum extent and ensure the integrity and reliability of subsequent analysis of the collected data.

[0066] The system uses a variety of early warning models, including threshold warning models, trend warning models and mutation warning models, which can accurately identify various potential failure modes and continuously optimize the early warning models through machine learning algorithms to improve the accuracy and reliability of the early warning.

[0067] The system reduces reliance on manual inspections and fault handling. Through intelligent fault warning and automatic adjustment functions, it reduces equipment maintenance costs and the workload of operation and maintenance personnel, and improves the overall operating efficiency of the new energy power generation system.

[0068] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. A smart grid fault early warning method for renewable energy power generation, characterized in that: The steps include: Step 1: Collect key operating parameters from renewable energy power generation equipment in real time, generate collected data, perform preliminary processing, and then extract key features; Step 2: Use multiple early warning models to perform fault analysis, including threshold early warning model, trend early warning model and mutation early warning model; In the threshold warning model, when it is detected that the key parameters exceed the set threshold, a threshold warning signal is immediately generated and the corresponding fault response mechanism is triggered; In the trend warning model, the system analyzes the long-term operation data of the equipment through regression analysis, predicts possible fault trends, and issues trend warning signals before the fault occurs, and triggers the corresponding fault response mechanism; in the mutation warning model, the system monitors sudden changes in the equipment operation status, triggers mutation warning signals in time, and triggers the corresponding fault response mechanism; Step 3: The system automatically selects the appropriate fault response mechanism and responds to the fault based on the warning analysis results; in the case of level I warning, the system automatically executes the emergency shutdown procedure; In the case of Level II warning, the system automatically adjusts the operating parameters of the equipment and restores the equipment to a safe operating state; in the case of Level III warning, the system continuously monitors the operating status of the equipment and prompts the operator to pay attention to the operating status of the equipment.

2. A smart grid fault warning system for renewable energy power generation, characterized in that: Including data acquisition module, data processing module, fault warning module, control execution module and human-computer interaction module: The data acquisition module is used to collect data from various types of new energy power generation equipment in real time, the collected data including the collected equipment operation data, environmental data and historical fault data, and transmit the collected data to the data processing module; The data processing module includes a data cleaning unit, a feature extraction unit and a big data analysis unit: the data cleaning unit is used to clean the collected data, and the cleaning method includes denoising and filtering the collected data; the feature extraction unit is used to extract key features from the cleaned collected data, and the big data analysis unit analyzes the extracted key features through a pre-trained big data model; the big data model is based on a machine learning algorithm and can identify complex fault modes and improve the accuracy of early warning; The fault warning module includes multiple fault warning models, which are based on different types of warning algorithms, including threshold warning models, trend warning models and mutation warning models. The fault warning module automatically generates fault warning signals according to data analysis results and set warning rules, and manages them in different levels according to the types and severity of the fault warning signals, and forms a corresponding fault response mechanism. The control execution module automatically executes corresponding control operations according to the fault warning signal issued by the fault warning module, including but not limited to adjusting equipment operating parameters, switching system operating modes or triggering emergency shutdown procedures; The human-computer interaction module provides a user interface for operators to view real-time data, warning information and system operating status, and can set and adjust system parameters through the interface. The user interface uses a graphical display method to simplify the operation complexity and supports multi-language switching and remote access functions.

3. The smart grid fault early warning system for renewable energy power generation according to claim 2, characterized in that: The data acquisition module is connected to the sensors and controllers of the new energy power generation equipment, and is used to collect the key operating parameters of the new energy power generation equipment in real time, and transmit these collected data to the data processing module through wireless communication. The data acquisition module supports multiple communication protocols; The data acquisition module also has a data cache function. When network communication is abnormal, the collected data can be temporarily stored in a local storage unit to form cache data. After the communication is restored, the cache data is transmitted to the data processing module. The data acquisition module also supports a distributed data acquisition architecture, which is used to connect new energy power generation equipment at different locations or of different types to the same fault warning system.

4. The smart grid fault early warning system for renewable energy power generation according to claim 3 is characterized in that: The big data analysis unit in the data processing module adopts a fault pattern recognition model based on deep learning, wherein the big data analysis unit adopts a machine learning algorithm to perform fault pattern recognition; The big data analysis unit also includes a prediction model based on time series analysis, which identifies the operation rules and abnormal changes of the equipment by analyzing the time series of the equipment operation data, and predicts the possible fault occurrence time and fault type in advance; The feature extraction unit in the data processing module extracts key features related to faults from the equipment operation data based on a multi-dimensional feature extraction algorithm.

5. The smart grid fault early warning system for renewable energy power generation according to claim 4, characterized in that: The threshold warning model in the fault warning module sets the warning thresholds of multiple key parameters. The warning thresholds are dynamically adjusted according to the equipment type, operating environment and historical data. When the actual operating data exceeds or falls below the set threshold, the system immediately generates a threshold warning signal. The threshold warning signal is divided into three levels. According to the level of the threshold warning signal, the system will take different response measures. The threshold warning model of the fault warning module builds a warning optimization model through an automated learning mechanism.

6. The smart grid fault early warning system for renewable energy power generation according to claim 5, characterized in that: The trend warning model in the fault warning module analyzes the long-term trend of the equipment operation data, predicts possible faults through regression analysis of historical data, and generates a trend warning signal before the predicted fault occurs; Trend warning signal hierarchical management: when the predicted failure probability reaches a certain threshold, the system will issue a first-level warning and recommend equipment maintenance; When the probability of failure increases further, the system issues a secondary warning and triggers the automatic protection measures of the control execution module, such as automatically adjusting the operating parameters of the equipment or entering maintenance mode to avoid more serious failures of the equipment.

7. The smart grid fault early warning system for renewable energy power generation according to claim 6, characterized in that: The mutation warning model in the fault warning module is used to detect sudden changes in the equipment operation status; the mutation warning model uses fast Fourier transform and wavelet transform signal processing technology to monitor the equipment operation signal in real time and perform frequency domain analysis to identify drastic changes that occur in a short time; the mutation warning model analyzes the frequency domain characteristics of real-time data to identify the mutation point of the equipment operation status. The system determines whether the mutation is a precursor to a potential equipment failure based on the position, amplitude and frequency of the mutation point, generates a corresponding mutation warning signal, and initiates an emergency plan.

8. The smart grid fault early warning system for renewable energy power generation according to claim 7, characterized in that: The control execution module has two modes: automatic adjustment and manual intervention: In automatic adjustment mode, the system automatically performs adjustment operations according to the fault warning signal. In manual intervention mode, the operator can manually intervene in the operation of the system through the human-computer interaction module and perform manual control operations. In automatic adjustment mode, the system immediately performs corresponding control operations after receiving the fault warning signal. In manual intervention mode, the system displays the fault warning signal and detailed fault information on the user interface for the operator to analyze and make decisions. The operator can manually adjust the equipment operating parameters and delay shutdown according to the actual situation.

9. The smart grid fault early warning system for renewable energy power generation according to claim 8, characterized in that: The user interface of the human-computer interaction module can display the real-time operating status of the equipment, historical warning records and current fault warning signals, and provide fault diagnosis reports and processing suggestions; The user interface adopts a multi-level data display structure, which allows operators to access detailed operating information of each device from a global view; the human-computer interaction module supports the query and playback of historical data, which allows operators to retrieve equipment operating data and warning records for a specific time period, and analyze the causes of faults and the long-term trend of equipment operation; The human-computer interaction module is also equipped with intelligent fault diagnosis tools. The system automatically generates fault diagnosis reports based on real-time and historical data. The reports include fault types, possible causes and recommended treatment measures. The module also supports multi-user access rights management for the security and confidentiality of system data. Different users can access and operate different functional modules of the system according to their permission levels.

10. A smart grid fault early warning system for renewable energy power generation according to claim 9, characterized in that: The data cleaning unit in the data processing module includes an abnormal data detection and processing mechanism for detecting and processing abnormal data. The abnormal data detection and processing mechanism uses multiple algorithms to perform quality analysis on the collected data: The data cleaning unit uses statistical anomaly detection algorithms to identify and process extreme outliers in the collected data; Time series analysis is used to detect time series anomalies of abnormal data. In particular, when the change range of an abnormal data point is obviously beyond the normal range, the system will automatically mark the abnormal data and make rational corrections based on the previous and subsequent time series data. The data cleaning unit also includes a data interpolation function. When the system detects that the collected data is missing, it can interpolate according to historical data and time series models to restore the missing collected data to the maximum extent and ensure the integrity and reliability of subsequent analysis of the collected data.

Images