Power equipment fault early warning system
Through the combination of IoT sensor network and edge computing nodes, real-time monitoring and fault warning of power equipment are achieved, solving the problems of delayed data transmission and slow response speed in traditional methods, and improving the safety and maintenance efficiency of the power system.
Patent Information
- Application Number
- CN202411032791.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-08
AI Technical Summary
Traditional power equipment monitoring and fault warning methods have problems such as high data transmission delay and slow response speed, which is difficult to meet the needs of modern power grids for real-time and reliability.
The Internet of Things sensor network, edge computing nodes, real-time data processing and analysis, as well as dynamic optimization and adaptive learning, are adopted to achieve comprehensive, real-time and accurate monitoring and fault warning of power equipment.
It significantly improves the operating safety, reliability and maintenance efficiency of the power system, shortens the time interval between fault detection and processing, improves the accuracy of fault diagnosis, reduces false alarms and missed reports, reduces operation and maintenance costs, and improves Work efficiency and satisfaction of operation and maintenance personnel.
Smart Images

Figure CN118917834A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of fault risk early warning, and in particular to a fault early warning system for electric power equipment. Background Art
[0002] With the rapid development of the electric power industry, power equipment, as a core component of the power system, plays a vital role in ensuring power supply and preventing large-scale power outages due to its safety, reliability and stability. However, traditional power equipment monitoring and fault warning methods often have many shortcomings and cannot meet the needs of modern power grids for intelligent and efficient operation and maintenance.
[0003] With the rapid development of smart grids, higher requirements are placed on real-time monitoring and fault warning of the operating status of power equipment. Traditional monitoring systems often rely on centralized data processing in the cloud, which has problems such as high data transmission delay and slow response speed. It is difficult to meet the real-time and reliability requirements of modern power grids. Therefore, it is particularly important to develop a power equipment fault warning system. Summary of the invention
[0004] The purpose of the present invention is to make up for the shortcomings of the prior art and provide a power equipment fault warning system, which integrates advanced Internet of Things technology, big data analysis, artificial intelligence algorithms and dynamic optimization mechanisms to achieve comprehensive, real-time and accurate monitoring and fault warning of power equipment, so as to significantly improve the operation safety, reliability and maintenance efficiency of the power system.
[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a power equipment fault warning system, which includes the construction of an Internet of Things sensor network, the deployment of edge computing nodes, real-time data processing and analysis, fault warning and decision support, as well as dynamic optimization and adaptive learning;
[0006] The IoT sensor network construction: highly sensitive sensors are deployed at key parts of power equipment. These sensors are wirelessly connected to the cloud and edge computing nodes through IoT technology to form a highly real-time and distributed monitoring network;
[0007] Deployment of edge computing nodes: Edge computing devices are deployed at key nodes of the monitoring network to receive raw data from sensors and perform preliminary data processing and analysis. Edge computing reduces the delay in data transmission to the cloud, enabling the system to respond immediately to abnormal conditions of the equipment.
[0008] Real-time data processing and analysis: The edge computing node uses built-in algorithms to process the received data in real time, including data cleaning, feature extraction and preliminary fault diagnosis. This process is completed locally without waiting for cloud response, which significantly improves the response speed and autonomy of the system;
[0009] Fault warning and decision support: Based on the processing results of edge computing nodes, the system can generate fault warning information in real time and send it to relevant personnel and systems through the user interface and communication interface. At the same time, the system can also provide maintenance suggestions and automatically trigger emergency response mechanisms according to the type and severity of the fault;
[0010] The dynamic optimization and adaptive learning: The system has dynamic optimization and adaptive learning capabilities, and can continuously optimize data processing algorithms and fault diagnosis models based on historical data and real-time feedback to improve the accuracy and reliability of early warning.
[0011] Furthermore, the construction of the IoT sensor network requires a comprehensive analysis and evaluation of the power equipment to determine the key parts that may fail and need to be monitored. This is achieved by comprehensive consideration of the equipment's structure, working principle, historical fault data, and expert experience. According to the monitoring requirements of the determined key parts, a high-sensitivity sensor with corresponding measurement parameters and accuracy is selected. The performance indicators of the sensor can be expressed by the following formula: Sensitivity (S) = Δoutput / Δinput, where Δoutput represents the change in the sensor output signal, and Δinput represents the change in the measured value. The selected sensor is installed at the key part of the power equipment, and ensures that it is firmly installed and has good contact. The sensor is then debugged so that it can work normally and output accurate measurement data. The sensor is wirelessly connected to the cloud and edge computing nodes through IoT technology. The relationship between the signal transmission distance (d) and the transmission power (P) and receiving sensitivity (S) can be expressed by the following formula
[0012] Furthermore, the data collected by the sensors in the IoT sensor network are transmitted to the cloud and edge computing nodes in real time through wireless connections. In the cloud and edge computing nodes, corresponding algorithms and software are used to process and analyze the data to extract useful information. Based on the processed data, the operating status of the power equipment is monitored in real time. When the monitoring data exceeds the set threshold, an early warning signal is issued in time so that corresponding maintenance measures can be taken. The threshold (T) can be set according to the specifications and operating requirements of the equipment. The performance of the entire sensor network system is regularly evaluated and optimized, including sensor calibration, stability detection of wireless connections, and improvement of data processing algorithms. At the same time, faulty sensors are replaced and repaired in time to ensure the long-term stable operation of the system.
[0013] Furthermore, the edge computing node deployment determines the amount of data and computational complexity that the edge computing node needs to process based on the scale of the monitoring network and the data type and frequency of the sensor. Assuming that each sensor generates d bytes of data per second and there are n sensors in the network, the total data volume Dt otal = n × d bytes / second, clarify the system's latency requirements for data transmission and processing to determine the necessity of edge computing. Example expression: Assume the maximum acceptable latency is T max seconds, and the cloud processing delay is T cloud seconds, then T edge <<T cloud And T e dge+T transmission ≤T max , where T transmission It is the transmission delay of data from the edge to the cloud. When selecting edge computing devices, choose edge computing devices with sufficient computing power and storage space based on processing requirements. Consider the power consumption and cost of the equipment and choose cost-effective equipment.
[0014] Furthermore, the edge computing node deployment installs the selected edge computing device at the determined key node location, and performs hardware connection and network configuration to ensure that the device can normally access the monitoring network and establish communication with sensors and other related devices. The edge computing device receives the raw data from the sensor, and can use data cleaning and denoising preprocessing operations to improve data quality. Mean filtering is used for denoising, and the filtered data value (D filtered ) is calculated as: Among them, D i is the ith original data value, and m is the number of original data involved in the calculation.
[0015] Furthermore, the edge computing node deployment analyzes and processes the pre-processed data, extracts key features and information, and can apply statistical analysis and machine learning algorithms to use a simple linear regression model to predict data trends. The model is y=a+bx, where y is the predicted value, x is the input variable, and a and b are model parameters. According to the analysis and processing results, real-time decisions are made, and data interaction with the cloud uploads processed important data and complex data that cannot be processed locally to the cloud, while receiving strategies and model updates issued by the cloud. The relationship between data transmission delay (L) and data volume (D) and transmission speed (V) is: Regularly monitor and evaluate the performance of edge computing devices, optimize algorithms and parameters based on actual operating conditions, and maintain and upgrade equipment.
[0016] Furthermore, the real-time data processing and analysis edge computing node receives real-time data from sensors through a network interface, removes noise, missing values and outliers in the data, and makes the data more accurate and reliable. For data containing noise, median filtering can be used for processing. The value D after median filtering median ) is calculated by arranging the data in ascending and descending order, D medianAssume that the extracted features are the mean (μ) and standard deviation (σ) of the data, and the calculation methods are: Among them, x i is the i-th data value, n is the number of data, and the built-in diagnostic algorithm is used to analyze the extracted features to determine whether there is a fault. The threshold judgment method can be used, where is the extracted feature value, and the results of processing and analysis are output, including the judgment of whether there is a fault and related feature information. Local decision-making and execution make corresponding decisions locally based on the diagnostic results, such as issuing an alarm and starting a backup device, without waiting for a response from the cloud.
[0017] Furthermore, the fault warning and decision support edge computing node processes and analyzes the real-time data to obtain the processing results, and compares the relevant indicators in the processing results with the preset thresholds to determine whether the fault warning conditions are met. Assuming that the fault warning indicator is I and the warning threshold is T w , then the judgment formula is: If I ≥ T w , then the fault warning is triggered and the fault warning information is generated. When the fault warning conditions are met, a warning message containing detailed information on the fault type, location, and severity is generated. The information is sent to relevant personnel and systems through the user interface and communication interface. The delay time for information sending is T d ) and the relationship between network bandwidth (B) and data volume (D) can be expressed as:
[0018] Furthermore, the fault warning and decision support queries the preset maintenance suggestion database according to the fault type and severity, and generates corresponding maintenance suggestions. Assuming that the severity is represented by a numerical value S, the value range of S is [1, 5], 1 represents mild, 5 represents severe, and different S values correspond to different maintenance suggestions M(S). If the fault severity is extremely high, the system automatically triggers the emergency response mechanism, such as emergency shutdown and starting the backup system. The triggering condition of the emergency response can be expressed as: if S≥T e , where T e When the severity threshold for triggering an emergency response is reached, the emergency response is triggered, and the entire process and results of fault warning and decision support are recorded for subsequent analysis and improvement.
[0019] Furthermore, the dynamic optimization and adaptive learning use the accuracy evaluation index to evaluate the current data processing algorithm and fault diagnosis model. The accuracy calculation formula is: Among them, TP is a true positive example, TN is a true negative example, FP is a false positive example, and FN is a false negative example. According to the results of data analysis and model evaluation, the parameters of the data processing algorithm and the fault diagnosis model are adjusted and optimized. Assuming that the model parameter is θ and the optimization objective function is L(θ), the gradient descent method can be used for optimization: Among them, α is the learning rate, is the objective function with parameters θ t The gradient at the fault location is calculated, the optimized parameters are applied to the model, the data processing algorithm and fault diagnosis model are updated, the updated model is verified using new test data, and the accuracy and reliability of the monitoring warning are improved.
[0020] Compared with the existing technology, a power equipment fault early warning system has the following beneficial effects:
[0021] 1. By integrating advanced Internet of Things technology and intelligent analysis algorithms, the present invention enables the system to monitor the operating status of power equipment in real time. Once an abnormality is detected, the system will immediately generate early warning information and notify relevant personnel. This instant response mechanism greatly shortens the time interval from fault discovery to processing, and effectively avoids the adverse effects of fault expansion on power grid operation. At the same time, the application of intelligent analysis algorithms improves the accuracy of fault diagnosis, reduces false alarms and missed alarms, and ensures that operation and maintenance personnel can accurately locate the fault point and take effective measures to repair it.
[0022] 2. The present invention greatly reduces the need for manual intervention through automated and intelligent monitoring and early warning. Operation and maintenance personnel can carry out maintenance work in a targeted manner based on the early warning information provided by the system, avoiding blind inspections and ineffective work. In addition, the system can also provide auxiliary functions such as fault analysis and maintenance suggestions to help operation and maintenance personnel complete maintenance work more efficiently, further reducing operation and maintenance costs. This intelligent operation and maintenance method not only improves work efficiency, but also reduces the labor intensity of operation and maintenance personnel, and improves overall job satisfaction and happiness. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0024] Figure 1 An operation flow chart of an electric power equipment fault early warning system. DETAILED DESCRIPTION
[0025] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0026] Embodiment 1
[0027] This embodiment describes in detail the working process of a power equipment fault warning system. First, high-sensitivity sensors, temperature sensors, oil level and oil quality sensors, vibration and acoustic sensors, insulation monitoring sensors and gas sensors are accurately deployed at key parts of the transformer. The performance indicators of the sensors can be expressed by the following formula: Sensitivity (S) = Δoutput / Δinput, where Δoutput represents the change in the sensor output signal, and Δinput represents the change in the measured value, so as to comprehensively monitor the electrical parameters of the transformer, adopt low-power and high-reliability wireless communication technology, and transmit sensor data in real time and stably. At the same time, a reasonable network topology is designed to improve data transmission efficiency and coverage.
[0028] Then, according to the actual situation of the substation and the data processing requirements, high-performance, low-power edge computing devices are selected to determine the amount of data and computational complexity that the edge computing nodes need to process. Assuming that each sensor generates d bytes of data per second and there are n sensors in the network, the total data volume D t otal = n × d bytes / second, clarify the system's latency requirements for data transmission and processing to determine the necessity of edge computing. Example expression: Assume the maximum acceptable latency is T max seconds, and the cloud processing delay is T cloud seconds, then T edge <<T cloud And T e dge+T transmission ≤T max , where T transmission It is the transmission delay of data from the edge to the cloud. Considering the scalability, reliability and maintainability of the equipment, to ensure long-term stable operation, lightweight operating systems and data processing software are deployed on edge computing nodes to realize rapid data collection, processing and analysis. At the same time, machine learning and deep learning algorithm libraries are integrated to improve the accuracy and efficiency of fault identification. The computing power of edge computing nodes is used to realize preliminary data processing and fault identification. Through localized processing, data transmission delay and bandwidth consumption are reduced, and the real-time and response speed of the system are improved.
[0029] Next, the received raw data is cleaned, denoised, and filtered to improve data quality. At the same time, the data is compressed and encoded to reduce storage space and transmission costs. Assume that the raw data is D original , the denoised data is D denoised , the denoising operation can be expressed as D denooised =f(D original ), where f represents the denoising function. Machine learning and deep learning algorithms are used to extract key features from the preprocessed data, and then classify and identify them. A fault mode library and a diagnostic rule library are established based on historical fault data and expert knowledge base to improve the accuracy and reliability of fault identification. The transformer status is diagnosed in real time, and preliminary diagnostic results and fault warning information are generated. According to the diagnostic results and fault types, specific maintenance suggestions are provided and the corresponding emergency response mechanism is triggered. At the same time, the diagnostic results and warning information are fed back to the operation and maintenance personnel and the remote monitoring center in real time.
[0030] Next, according to the urgency and potential impact of the fault, different levels of warning thresholds and response strategies are set. Color coding and sound and light alarms are used to intuitively display the health status and warning information of the transformer to determine whether the fault warning conditions are met. The fault warning index is I and the warning threshold is T. w , then the judgment formula is: If I ≥ T w , then the fault warning is triggered and the fault warning information is generated. When the fault warning conditions are met, the information is sent to the relevant personnel and systems through the user interface and communication interface. The intuitive and easy-to-use user interface and interaction method are designed to facilitate operation and maintenance personnel to view the real-time status, historical trends, and fault warning information of the transformer. Flexible alarm notification methods are provided to ensure that operation and maintenance personnel can receive warning information in a timely manner and establish a complete emergency response mechanism, including emergency plan formulation, emergency drills, and emergency resource allocation. Once the fault warning information is received, the emergency plan is immediately activated and corresponding measures are taken to reduce the impact of the fault on the operation of the power grid. At the same time, a linkage mechanism is established with the remote monitoring center to realize the rapid reporting of fault information and remote technical support.
[0031] Finally, strengthen system security protection measures to ensure that the system is protected from security threats such as cyber attacks and data leaks. Regularly back up and restore system data to ensure data security and recoverability. In the event of failures and accidents, the system can be quickly restored to normal operation and data integrity. Establish a sound operation and maintenance management system and process, including equipment inspections, maintenance records, and troubleshooting. Regularly inspect and maintain the system to ensure long-term stable and efficient operation of the system. At the same time, strengthen the training and management of operation and maintenance personnel to improve their professional skills and emergency response capabilities.
[0032] Embodiment 2
[0033] This embodiment further describes the working process of a power equipment fault warning system based on the first embodiment. First, a highly sensitive vibration sensor and temperature sensor are selected. The vibration sensor can capture tiny vibration changes during the operation of the bearing, and the temperature sensor is used to monitor the operating temperature of the bearing. The combination of the two can fully reflect the health status of the bearing. The vibration sensor is installed at an appropriate position on the generator bearing seat to ensure that the vibration signal of the bearing can be accurately captured. The temperature sensor is installed close to the bearing housing to obtain the most accurate temperature data. The LoRa and NB-IoT low-power wide area network technologies are used to realize wireless connection between the sensor and the cloud and edge computing nodes to ensure the stability and real-time performance of data transmission.
[0034] Then, a high-performance edge computing server is deployed in the wind farm control room. The server has powerful data processing capabilities and storage space, which can meet the needs of real-time data processing and fault warning. The edge computing server is equipped with a high-speed network interface and a multi-core processor, as well as sufficient storage space to ensure the efficiency and stability of data processing. Customized data processing and analysis software is installed on the edge computing server, including signal processing algorithms, fault diagnosis models and emergency response mechanisms.
[0035] Next, the edge computing node is used to adopt FFT advanced signal processing technology to perform spectrum analysis on the vibration signal. For the vibration signal x(n), its discrete Fourier transform (DFT) can be expressed as: Fast Fourier Transform (FFT) is an algorithm that quickly calculates DFT, extracts key eigenvalues that reflect the bearing status, monitors the bearing temperature data at the same time, and conducts a comprehensive analysis with the vibration signal to determine whether the bearing is overheating or abnormally worn. Based on the extracted eigenvalues and the preset health assessment model, the health status of the bearing is evaluated in real time to identify potential failure modes.
[0036] When the bearing health status assessment results are abnormal, the system automatically generates fault warning information, including fault type, location, severity and preliminary diagnostic suggestions, and sends the warning information to the wind farm's operation and maintenance team and relevant managers through the user interface and communication interface. According to the fault type and severity, the system provides detailed maintenance suggestions and automatically triggers the emergency response mechanism to remotely adjust the generator load, start the backup generator and send shutdown maintenance instructions.
[0037] Finally, the system uses machine learning algorithms to continuously learn and optimize historical fault data. Assuming that a machine learning algorithm is used to train the bearing fault diagnosis model, the loss function is L(θ), where θ is the parameter of the model. The parameters are updated through the gradient descent optimization algorithm. The formula is: Among them, α is the learning rate, is the loss function with parameters θ t The accuracy and reliability of the bearing fault diagnosis model are continuously improved by adjusting the gradient at the edge, and the performance and data processing algorithms of the edge computing nodes are regularly evaluated. The system is tuned and upgraded according to the evaluation results. According to the actual operating conditions and environmental changes of the wind farm, the sensor parameters, data processing algorithms and fault diagnosis models are automatically adjusted to adapt to the fault warning needs under different conditions.
[0038] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.
Claims
1. A power equipment fault early warning system, characterized in that: The system includes the construction of IoT sensor networks, edge computing node deployment, real-time data processing and analysis, fault warning and decision support, as well as dynamic optimization and adaptive learning; The IoT sensor network is constructed by deploying highly sensitive sensors at key locations of power equipment. These sensors are wirelessly connected to the cloud and edge computing nodes through IoT technology to form a highly real-time and distributed monitoring network. The edge computing node deployment: Edge computing devices are deployed at key nodes of the monitoring network, which are responsible for receiving raw data from sensors and performing preliminary data processing and analysis. Edge computing reduces the delay in data transmission to the cloud, enabling the system to respond immediately to abnormal conditions of the equipment. Real-time data processing and analysis: The edge computing node uses built-in algorithms to process the received data in real time, including data cleaning, feature extraction and preliminary fault diagnosis. This process is completed locally without waiting for cloud response, which significantly improves the response speed and autonomy of the system. The fault warning and decision support: Based on the processing results of the edge computing nodes, the system can generate fault warning information in real time and send it to relevant personnel and systems through the user interface and communication interface. At the same time, the system can also provide maintenance suggestions and automatically trigger emergency response mechanisms according to the type and severity of the fault. The dynamic optimization and adaptive learning: The system has dynamic optimization and adaptive learning capabilities, and can continuously optimize data processing algorithms and fault diagnosis models based on historical data and real-time feedback to improve the accuracy and reliability of early warning.
2. According to claim 1, a power equipment fault early warning system is characterized in that: The construction of the IoT sensor network requires a comprehensive analysis and evaluation of the power equipment to determine the key parts that may fail and need to be monitored. This is achieved by comprehensive consideration of the equipment's structure, working principle, historical fault data, and expert experience. According to the monitoring requirements of the determined key parts, a high-sensitivity sensor with corresponding measurement parameters and accuracy is selected. The performance index of the sensor can be expressed by the following formula: Sensitivity (S) = Δoutput / Δinput, where Δoutput represents the change in the sensor output signal, and Δinput represents the change in the measured value. The selected sensor is installed at the key part of the power equipment, and ensures that it is firmly installed and has good contact. Then the sensor is debugged so that it can work normally and output accurate measurement data. Through the IoT technology, the sensor is wirelessly connected to the cloud and edge computing nodes. The relationship between the signal transmission distance (d) and the transmission power (P) and the receiving sensitivity (S) can be expressed by the following formula 3. According to claim 2, a power equipment fault early warning system is characterized in that: The data collected by the sensors in the IoT sensor network are transmitted to the cloud and edge computing nodes in real time through wireless connections. In the cloud and edge computing nodes, corresponding algorithms and software are used to process and analyze the data to extract useful information. Based on the processed data, the operating status of the power equipment is monitored in real time. When the monitoring data exceeds the set threshold, an early warning signal is issued in time so that corresponding maintenance measures can be taken. The threshold (T) can be set according to the specifications and operating requirements of the equipment. The performance of the entire sensor network system is regularly evaluated and optimized, including sensor calibration, wireless connection stability detection, and data processing algorithm improvement. At the same time, faulty sensors are replaced and repaired in time to ensure the long-term stable operation of the system.
4. According to claim 1, a power equipment fault early warning system is characterized in that: The edge computing node deployment determines the amount of data and computational complexity that the edge computing node needs to process based on the scale of the monitoring network and the data type and frequency of the sensor. Assume that each sensor generates d bytes of data per second and there are n sensors in the network, then the total data volume D t otal = n × d bytes / second, clarify the system's latency requirements for data transmission and processing to determine the necessity of edge computing. Example expression: Assume the maximum acceptable latency is T max seconds, and the cloud processing delay is T cloud seconds, then T edge <<T cloud And T e dge+T transmission ≤T max , where T transmission It is the transmission delay of data from the edge to the cloud. When selecting edge computing devices, choose edge computing devices with sufficient computing power and storage space based on processing requirements. Consider the power consumption and cost of the equipment and choose cost-effective equipment.
5. According to claim 4, a power equipment fault early warning system is characterized in that: The edge computing node deployment installs the selected edge computing device at the determined key node location, and performs hardware connection and network configuration to ensure that the device can normally access the monitoring network and establish communication with sensors and other related devices. The edge computing device receives the raw data from the sensor, and can use data cleaning and denoising preprocessing operations to improve data quality. Mean filtering is used for denoising. The filtered data value D filtered ) is calculated as: Among them, D i is the ith original data value, and m is the number of original data involved in the calculation.
6. According to claim 5, a power equipment fault early warning system is characterized in that: The edge computing node deployment analyzes and processes the pre-processed data, extracts key features and information, and can apply statistical analysis and machine learning algorithms to predict data trends using a simple linear regression model. The model is y=a+bx, where y is the predicted value, x is the input variable, and a and b are model parameters. According to the analysis and processing results, real-time decisions are made, and data interaction with the cloud uploads processed important data and complex data that cannot be processed locally to the cloud, while receiving strategies and model updates issued by the cloud. The relationship between data transmission delay (L) and data volume (D) and transmission speed (V) is: Regularly monitor and evaluate the performance of edge computing devices, optimize algorithms and parameters based on actual operating conditions, and maintain and upgrade equipment.
7. The power equipment fault early warning system according to claim 1, characterized in that: The real-time data processing and analysis edge computing node receives real-time data from sensors through a network interface, removes noise, missing values and abnormal values in the data, and makes the data more accurate and reliable. For data containing noise, median filtering can be used for processing. The value after median filtering (D median ) is calculated by arranging the data in ascending and descending order, D median Assume that the extracted features are the mean (μ) and standard deviation (σ) of the data, and the calculation methods are: Among them, x i is the i-th data value, n is the number of data, and the built-in diagnostic algorithm is used to analyze the extracted features to determine whether there is a fault. The threshold judgment method can be used, then there may be a fault. Where is the extracted feature value, and the results of processing and analysis are output, including the judgment of whether there is a fault and related feature information. Local decision-making and execution make corresponding decisions locally based on the diagnostic results, such as issuing an alarm and starting backup equipment, without waiting for a response from the cloud.
8. The power equipment failure early warning system according to claim 1, characterized in that: The fault warning and decision support edge computing node processes and analyzes the real-time data, obtains the processing results, and compares the relevant indicators in the processing results with the preset thresholds to determine whether the fault warning conditions are met. Assume that the fault warning indicator is I and the warning threshold is T w , then the judgment formula is: If I ≥ T w , then the fault warning is triggered and the fault warning information is generated. When the fault warning conditions are met, a warning message containing detailed information on the fault type, location, and severity is generated. The information is sent to relevant personnel and systems through the user interface and communication interface. The delay time for information sending (T d ) and the relationship between network bandwidth (B) and data volume (D) can be expressed as:
9. The power equipment failure early warning system according to claim 8, characterized in that: The fault warning and decision support system queries the preset maintenance suggestion database according to the fault type and severity, and generates corresponding maintenance suggestions. Assuming that the severity is represented by a numerical value S, the value range of S is [1, 5], 1 represents slight, 5 represents severe, and different S values correspond to different maintenance suggestions M(S). If the fault severity is extremely high, the system automatically triggers the emergency response mechanism, such as emergency shutdown and startup of the backup system. The triggering condition of the emergency response can be expressed as: if S≥T e , where T e When the severity threshold for triggering an emergency response is reached, the emergency response is triggered, and the entire process and results of fault warning and decision support are recorded for subsequent analysis and improvement.
10. The power equipment failure early warning system according to claim 1, characterized in that: The dynamic optimization and adaptive learning use the accuracy evaluation index to evaluate the current data processing algorithm and fault diagnosis model. The accuracy calculation formula is: Among them, TP is a true positive example, TN is a true negative example, FP is a false positive example, and FN is a false negative example. According to the results of data analysis and model evaluation, the parameters of the data processing algorithm and the fault diagnosis model are adjusted and optimized. Assuming that the model parameter is θ and the optimization objective function is L(θ), the gradient descent method can be used for optimization: Among them, α is the learning rate, is the objective function with parameters θ t The gradient at the fault location is calculated, the optimized parameters are applied to the model, the data processing algorithm and fault diagnosis model are updated, the updated model is verified using new test data, and the accuracy and reliability of the monitoring warning are improved.
Citation Information
Cited By
Fault self-checking method and device for lighting system
CN119172894A
Intelligent remote supervision early warning method and system based on Internet of Things
CN119204353A
Intelligent remote supervision and early warning method and system based on Internet of Things
CN119204353B
Power line carrier communication fault detection system
CN119232201A
A power line carrier communication fault detection system
CN119232201B
