Safety monitoring and early warning method for power system

Abstract

The invention discloses a safety monitoring and early warning method for a power system, and relates to the technical field of power monitoring, and the method comprises the steps: obtaining a target detection line and a real-time monitoring data set; performing task disassembly on the real-time monitoring data set, and screening an identification task set; establishing a fault probability resonance model, inputting the identification task set into the fault probability resonance model to obtain the fault probability of each power device, and calculating a resonance effect value between any two devices; obtaining a transfer matrix based on the fault probability and the resonance effect value of each power device; generating a probability graph based on the transfer matrix in combination with the time sequence; predicting a fault propagation path based on the probability graph, adjusting the fault probability of the power equipment, identifying a key path and risk equipment based on the adjusted fault probability, and generating a monitoring analysis report; the response efficiency of the monitoring data of the power transmission and transformation equipment is improved; the effect of dynamic correlation analysis between faults is achieved; and the effects of fault equipment and a fault propagation path are accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power monitoring, and in particular to a safety monitoring and early warning method for an electric power system. Background Art

[0002] The safe and stable operation of power transmission and transformation equipment is crucial to ensuring the safe operation of the power grid. The existing monitoring methods for the safe operation of power transmission and transformation equipment mainly process the equipment operation data in a unified manner, which is inefficient and causes the equipment to fail to respond to faults in a timely manner. In addition, there is a lack of integrated judgment on the data generated by the equipment on the transmission and transformation lines, which leads to errors in the calculation of the probability of equipment failure, affecting the accuracy of fault warning.

[0003] The Chinese invention patent with application number 202311377850.9 provides a method for monitoring the safe operation of power transmission and transformation equipment, which obtains a target detection line, identifies the equipment on the target detection line, and determines multiple devices; obtains a real-time monitoring data set, and the cloud processor decomposes the computing tasks to obtain a decomposed task set; screens and identifies the task set and transfers it from the cloud processor to the edge device; receives data output by the edge device and the cloud processor, and uses a fault probability detection module to output multiple fault probabilities; integrates according to multiple fault probabilities to obtain a fault probability indicator, and when the fault probability indicator is greater than the preset fault probability indicator, generates a first warning signal.

[0004] However, in the prior art, multiple fault probabilities are obtained by fault probability detection based on multiple processing devices, which are all detection, analysis and early warning between single faults, and do not take into account the problem of dynamic correlation analysis between faults, which reduces the comprehensiveness of fault early warning and further reduces the accuracy of early warning. It is difficult to deal with multiple faults and the potential relationship between multiple faults, which reduces the overall safety of power grid equipment. Summary of the invention

[0005] The present application provides a safety monitoring and early warning method for a power system, which solves the problem in the prior art that single fault analysis leads to reduced early warning accuracy, and achieves the technical effect of improving early warning accuracy through dynamic correlation analysis of multiple faults. The present application provides a safety monitoring and early warning method for a power system, comprising: S100: Acquire a target detection line, the target detection line includes a plurality of power equipment, respectively acquire real-time monitoring data of each power equipment to form a real-time monitoring data set; perform task decomposition on the real-time monitoring data set to acquire a decomposition task set, and filter an identification task set from the decomposition task set; S200: Establish a fault probability resonance model, input the identification task set into the fault probability resonance model to obtain the fault probability of each power device, and calculate the resonance effect value between any two devices; S300: Obtain a transfer matrix based on the failure probability and resonance effect value of each power device; generate a probability graph based on the transfer matrix combined with the time series; predict the fault propagation path based on the probability graph, adjust the failure probability of the power device, identify the critical path and risk equipment based on the adjusted failure probability, and generate a monitoring analysis report; Among them, each row and each column in the transfer matrix represents a fault state, and the matrix elements represent the transition probability of the fault state; the probability graph uses the equipment state and fault probability at each time point as nodes and the transfer relationship as edges; the monitoring and analysis report includes equipment failure probability, fault propagation path, critical path risk and a list of high-risk equipment.

[0006] Furthermore, the resonance effect value between any two devices is calculated, including: calculating the correlation based on the electrical connection strength and physical distance between the devices, and the calculation formula is as follows:

[0007] in, is the association between device i and device j; is the electrical connection strength between device i and device j, is the physical distance between device i and device j; and They are the pre-set electrical connection strength weight coefficient and physical distance weight coefficient, both of which have a value range between [0,1]; The calculation formula of the resonance effect value is as follows:

[0008] in, is the resonance effect value of device i and device j; It is the preset resonance effect adjustment coefficient, the value range is between [0.1, 0.5], which is used to adjust the resonance effect value; is the failure probability of device i; is the failure probability of device j.

[0009] Furthermore, the calculation formula of the transition probability is as follows:

[0010] in, is the transition probability; is the global adjustment coefficient, with a value range of [0,1], which is used to control the contribution ratio of the resonance effect; is a time decay function, which describes the characteristics of the influence of device i on the failure probability of device j decaying with time;

[0011] is the decay rate, and t is the time interval between device i and device j.

[0012] Furthermore, it also includes: S310: Acquire real-time monitoring data and historical monitoring analysis reports, extract fault types, and set fault levels; input the real-time monitoring data, fault types, and fault levels into a fault harmony assessment model to generate a balanced fault combination set and an unbalanced fault combination set; S320: Based on the balanced fault combination set and the unbalanced fault combination set, adjust the failure probability of the power equipment and identify the critical path and risk equipment.

[0013] Furthermore, the method further comprises: S400: Establish a probabilistic traffic flow model, take power equipment as nodes, and target detection lines as energy flow paths. Obtain energy flow speed, load change probability, and failure probability based on real-time monitoring data, analyze the failure probability distribution on the energy flow path, identify probabilistic traffic congestion points and high-incidence areas of failure, and generate a flow velocity distribution report; conduct comprehensive analysis and early warning based on the flow velocity distribution report and monitoring and analysis report.

[0014] Furthermore, the method also includes: S500: obtaining a flow velocity distribution report and a monitoring and analysis report, defining evaluation indicators, and obtaining a fault tolerance value by weighted summation based on the evaluation indicator value and the corresponding pre-set weight coefficient; pre-setting a balance bottom line value, and calculating the stability boundary of the system under multiple fault conditions based on the fault tolerance value, and then setting a redistribution mechanism to dynamically adjust the system's operating parameters and operation and maintenance strategies.

[0015] One or more technical solutions provided in this application have at least the following technical effects or advantages: By collaboratively processing data through edge devices and cloud processors, the response efficiency of monitoring data of power transmission and transformation equipment is improved; based on the fault probability resonance model, the failure probability of each power equipment and the resonance effect value between equipment can be accurately output, achieving the effect of dynamic correlation analysis between faults; setting transfer matrices and probability graphs can identify vulnerable links and high-risk periods in the power grid system, predict the fault state of equipment in the future, identify the critical paths of fault propagation and potential high-risk equipment, and generate monitoring and analysis reports, further achieving the effect of accurately predicting faulty equipment and fault propagation paths. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The present invention is a flowchart of a safety monitoring and early warning method for a power system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0017] To facilitate the understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which the present invention belongs; the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more related listed items.

[0019] Embodiment 1: Figure 1 As shown, a safety monitoring and early warning method for a power system, the method comprising: S100: Acquire a target detection line, which includes a number of power equipment, and respectively acquire real-time monitoring data of each power equipment to form a real-time monitoring data set; perform task decomposition on the real-time monitoring data set to acquire a decomposition task set, and filter an identification task set from the decomposition task set.

[0020] In some embodiments, the target detection line is locked by connecting to the power transmission and transformation monitoring system, which refers to a system platform for monitoring and operating power transmission and transformation equipment and lines in the power grid. The system collects line operation parameters and equipment status data in real time through sensor equipment and monitoring terminals distributed at various nodes of the power grid, as well as online monitoring equipment set up on the transmission line, and transmits it to the monitoring center through the on-site data acquisition unit, so as to monitor the operation status of the power grid.

[0021] In some embodiments, the target detection line is an electric energy transmission line between the target power user and the power station, that is, the target detection line is a line that transmits electric energy from the power station to the user to meet the user's electricity demand. The line is selected as the monitoring target because it is directly related to the quality and reliability of the user's power supply. By establishing a connection with the power transmission and transformation monitoring system, the target detection line is obtained to provide a basis for monitoring and analyzing the operating status of the line and equipment. All equipment contained in the target detection line is clarified as the monitoring object; the target equipment is monitored in real time for its operating data, line operating parameters and equipment status data are collected in real time, and after determining multiple devices on the target detection line, the operating parameter data of the equipment is collected in real time using sensors or monitoring terminals provided on multiple devices. The monitored operating data includes various indicators reflecting the operating status of the equipment, such as current, voltage, temperature, vibration, noise, etc., and the collected real-time data constitutes a real-time monitoring data set. The real-time monitoring data set includes monitoring data of all equipment on a target detection line.

[0022] In some embodiments, in order to improve the response efficiency of the monitoring data of the power transmission and transformation equipment, the edge device and the cloud processor are used to perform preliminary processing on the acquired data. First, the real-time monitoring data set is task-decomposed according to the task flow of the monitoring data processing, and the processing and analysis of the real-time monitoring data set are divided into multiple subtasks, such as data preprocessing, feature extraction, state evaluation, fault diagnosis, etc. The decomposed subtasks constitute a decomposition task set to obtain a decomposition task set; the multiple tasks in the obtained decomposition task set are sorted and hierarchically divided to obtain a task layer. Among them, the task flow refers to the process of streaming multiple related tasks in a specific order during the monitoring data calculation process. At the same time, a resource monitoring agent is run on the edge device, which regularly collects the hardware usage of the edge device, reads the model, main frequency and other parameters of the CPU of the edge device, and gives the computing power index; calls the system interface of the edge device to obtain the current total memory and available memory, and calculates the storage space index; monitors the CPU and memory usage of the edge device in real time, and gives the load index; summarizes the computing power index, storage space index and load index to obtain the computing resource index of the edge device to evaluate the computing power that the edge device can currently provide. Among them, the edge device is an intelligent computing platform deployed on the target detection line, which calculates and processes monitoring data for multiple devices at a close distance to improve the monitoring response efficiency.

[0023] Traverse the tasks of each layer of the task layer, read its computing resource requirement description for each task, then obtain the source device corresponding to the task, and obtain the current computing power, storage space, and load indicators of the corresponding edge device based on the device. Compare the computing resource requirements of the task with the computing resource indicators of the edge device. If the edge device meets all the resource requirements of the task, generate identification information for the task, including the ID of the edge device that executes the task. Continue to traverse the next layer of tasks and repeat the above operations. Finally, summarize the identification of each layer of tasks to form an identification task set. Next, send each identification task in the identification task set from the cloud processor to the corresponding edge device. After the edge device has completely received the identification task, the cloud processor will delete the identification task set to ensure the computing resource utilization of the cloud processor.

[0024] Then, for the obtained task layer, each layer of tasks is identified according to the synchronization of the computing tasks, that is, whether the tasks in each layer need to depend on and wait for each other during execution, and the tasks that can be calculated synchronously and the tasks that cannot be calculated synchronously are obtained. Among them, tasks that can be calculated synchronously refer to independent computing tasks that can be performed in parallel and simultaneously during the execution process without any sequence dependency, such as data preprocessing, feature extraction and other tasks that can be run independently; tasks that cannot be calculated synchronously have dependencies and must wait for the output results of the previous tasks before they can start running. For example, some model prediction tasks can only be started after feature extraction is completed. Through synchronization identification, find out which tasks can be executed simultaneously using the parallel acceleration capabilities of edge computing, and which tasks need to be retained in the cloud processor for sequential execution.

[0025] Subsequently, the identified synchronously computable tasks are marked as suitable for edge execution based on whether they meet the edge computing resource conditions. The identification information is the metadata generated when the task layer is screened for resource matching, which is used to mark the tasks that can be assigned to the edge. The synchronously computable tasks obtained after identification form an identified task set, which is calculated in parallel by the edge device. The edge device executes the identified task set, and the cloud processor executes the tasks that cannot be calculated synchronously in the task layer, realizing cloud-edge collaborative computing, improving the processing efficiency of the monitoring data of the power transmission and transformation equipment, and then improving the efficiency of responding to abnormalities of the power transmission and transformation equipment, and ensuring the safe operation of the power transmission and transformation equipment.

[0026] S200: Establish a fault probability resonance model, input the identification task set into the fault probability resonance model to obtain the fault probability of each power device, and calculate the resonance effect value between any two devices; In some embodiments, a fault probability resonance model is established, specifically including: statistically recording the number of failures or failures of multiple devices on the target detection line, obtaining the number of failures that occur in the device per unit operating time, that is, the historical failure frequency, reflecting the reliability and health status of the device. At the same time, the time span between the current operating time and the last failure time of each device is collected to determine the improvement or deterioration of the health status of the device. The shorter the interval time, the higher the risk of device failure.

[0027] Collect historical operation data, historical fault frequency and interval time data of power transmission and transformation equipment, including current, voltage, temperature, vibration, noise and other types of data under normal and fault conditions; organize the data to form a data set for model training. Extract key features that affect the probability of failure from the original data, such as current fluctuation, voltage deviation, temperature change trend, etc. Standardize the features to improve the training efficiency and accuracy of the model. Select the long short-term memory network (LSTM) model as the basic model, and build a model architecture, including input layer, hidden layer and output layer. The input layer receives time series data, that is, identifies the task set, and the output layer is the output probability value, which represents the fault probability at the current moment. Use the sample data set to train the network model, learn the mapping relationship between input data and fault probability, set the loss function to binary cross entropy, and the optimization algorithm to Adam. Iterate multiple trainings to make the model converge and improve the prediction accuracy. LSTM or GRU units are used in the hidden layer to process time series data. Divide the sorted data set into training set, validation set and test set. Use the training set to train the model, adjust the model parameters through the back propagation algorithm, and minimize the loss function (such as mean square error or cross entropy). During the training process, use the validation set to monitor the performance of the model to prevent overfitting. Use the test set to evaluate the trained model and calculate the model's accuracy, recall rate, F1 score and other indicators. Optimize the model based on the evaluation results to obtain the final fault probability resonance model.

[0028] In some embodiments, based on the probability value output by the fault probability resonance model, the resonance effect value between any two devices is calculated, and the correlation degree is calculated. The calculation formula is as follows:

[0029] in, is the association between device i and device j; is the electrical connection strength between device i and device j, is the physical distance between device i and device j. The electrical connection strength and physical distance of each device are obtained, and the data are normalized so that they are all between [0,1], and then calculated; and They are the pre-set electrical connection strength weight coefficient and physical distance weight coefficient, both of which have a value range between [0,1]; The calculation formula of the resonance effect value is as follows:

[0030] in, is the resonance effect value of device i and device j; It is a preset resonance effect adjustment coefficient, and its value range is between [0.1, 0.5]. The resonance effect adjustment coefficient is used to adjust the resonance effect value to achieve the correction of the resonance effect value. It is calculated and statistically obtained based on historical experimental data. Its default value is 0.2. Experimental data proves that when it is between 0.1 and 0.5, it can achieve a better correction effect on the resonance effect value. The specific value needs to be designed and adjusted according to the actual situation, and this application will not go into details here. is the failure probability of device i; is the failure probability of device j.

[0031] S300: Obtain a transfer matrix based on the failure probability and resonance effect value of each power device; generate a probability graph based on the transfer matrix combined with the time series; predict the fault propagation path based on the probability graph, adjust the failure probability of the power device, identify the critical path and risk equipment based on the adjusted failure probability, generate a monitoring analysis report and issue an early warning; Each row and column in the transfer matrix represents a fault state, and the matrix elements represent the transition probability of the fault state; the probability graph uses the device state and fault probability at each time point as nodes and the transition relationship as edges.

[0032] In some embodiments, the transfer matrix is ​​used to describe the transition probability of the power device between different fault states. It defines the probability of the device transferring from one fault state to another fault state based on the fault probability of each power device. The rows and columns of the transfer matrix represent different fault states, and the elements in the matrix represent the probability of transferring from one state to another, and also include the probability of the device transferring from a non-fault state to a fault state. The calculation formula of the transition probability is as follows:

[0033] in, is the transition probability; It is the global adjustment coefficient, and its value range is between [0,1]. It is used to control the contribution ratio of the resonance effect and prevent the probability value from crossing the boundary. is a time decay function, which describes the characteristics of the influence of device i on the failure probability of device j decaying with time. , is the decay rate, and t is the time interval between device i and device j.

[0034] During the calculation process, it is necessary to calculate according to different devices at different times. The global adjustment coefficient is pre-set according to historical experimental data. When pre-setting, the actual situation should be considered first. According to the system's requirements for the sensitivity of the resonance effect, high sensitivity can more accurately capture the mutual influence between devices, and the global adjustment coefficient tends to be a larger value (0.5-0.8); otherwise, it tends to be a smaller value (0.2-0.5); at the same time, it is adjusted according to historical operation data and equipment failure records, and the occurrence and propagation path of equipment failures in different operating states of the system are analyzed. The accuracy and timeliness of fault warning when the global adjustment coefficient is different are evaluated, so as to select an optimal global adjustment coefficient. This application does not make specific restrictions here. The transfer probability describes the transfer probability at a certain time, and the transfer matrix includes the transfer probability between different devices at different times. By integrating real-time fault probability, resonance effect between devices and time attenuation, more accurate fault propagation modeling is achieved; maintenance is deployed in advance for equipment combinations with high transfer probabilities to reduce the risk of chain failures.

[0035] For example, if the failure probability of device i is 0.2, the failure probability of device j is 0.15, the resonance effect value is 0.3, the global adjustment coefficient is set to 0.5, the decay rate is 0.1, and the time interval is 2 hours, that is, the impact of the failure of device i on device j after two hours is determined. The transfer probability is calculated based on the above parameters, and the transfer probability is 0.473, indicating that after the failure of device i, the failure probability of device j changes from 0.15 to 0.473 two hours later, reflecting the dynamic impact between the two devices.

[0036] In some embodiments, a probability graph is generated based on a time series in combination with a transfer matrix, a fault propagation path prediction is performed based on the probability graph, the failure probability of the power equipment is adjusted according to the calculated transfer probability, the critical path and risk equipment are identified based on the adjusted failure probability, and the critical path of fault propagation is identified according to the prediction result of the fault propagation path; the fault propagation probability on the critical path and the possible loss are calculated, and the path risk is evaluated; potential high-risk equipment is screened out based on the failure probability of each device and its position on the critical path; a monitoring and analysis report is generated and an early warning is issued; the monitoring and analysis report includes equipment failure probability, fault propagation path, critical path risk, a list of high-risk equipment, etc.; an early warning threshold is set according to the equipment failure probability and path risk, and the early warning mechanism is triggered when the equipment failure probability or path risk exceeds the early warning threshold.

[0037] In this embodiment, the probability graph is used to represent the fault propagation relationship between devices. The network structure of fault propagation is constructed by defining the network node as the device fault state and the edge as the transfer relationship between states. The transfer matrix focuses on describing the change law of the device fault state over time. By introducing time series information, it accurately describes the change law of the device fault state and the transition probability of the fault state between different time points. The transfer matrix with the introduction of time series can more accurately predict the fault state of the device in the future. This prediction not only takes into account the current fault state of the device, but also takes into account the impact of historical fault states and time intervals on the transfer probability.

[0038] In this embodiment, the fault probability resonance effect value between devices is used to identify the vulnerable links and high-risk periods in the power grid system; the device fault state transition probability output by the transfer matrix is ​​used to predict the fault state of the device in the future, as well as the dynamic impact changes between devices; the fault propagation relationship output by the probability graph model is used to identify the critical path of fault propagation and potential high-risk devices. Improving the timeliness and accuracy of fault warnings can detect potential fault risks in advance and take preventive measures; providing strong support for fault prevention and emergency handling, and formulating more effective fault handling strategies; improving the overall safety performance of the power grid system and reducing the impact of faults on power grid operations.

[0039] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages: This application improves the response efficiency of power transmission and transformation equipment monitoring data by collaboratively processing data through edge devices and cloud processors; based on the fault probability resonance model, it can accurately output the failure probability of each power equipment and the resonance effect value between equipment, and achieve the effect of dynamic correlation analysis between faults; setting transfer matrices and probability graphs can identify vulnerable links and high-risk periods in the power grid system, predict the failure state of equipment in the future, identify the critical paths of fault propagation and potential high-risk equipment, generate monitoring and analysis reports, and further achieve the effect of accurately predicting faulty equipment and fault propagation paths.

[0040] Embodiment 2: The above embodiment realizes dynamic correlation analysis between equipment faults, but does not realize accurate analysis of different fault generation and elimination. This embodiment makes further improvements on the basis of the above content.

[0041] The method further comprises: S310: acquiring real-time monitoring data and historical monitoring analysis reports, extracting fault types, and setting fault levels; inputting the real-time monitoring data, fault types, and fault levels into a fault harmony assessment model to generate a balanced fault combination set and an unbalanced fault combination set; S320: Based on the balanced fault combination set and the unbalanced fault combination set, formulate a balanced restraint mechanism, adjust the failure probability of power equipment, and identify critical paths and risky equipment.

[0042] In some embodiments, real-time monitoring data is collected and historical monitoring and analysis reports are integrated to extract fault information recorded therein. Text mining and classification are performed on the fault descriptions in the real-time monitoring data and historical reports to identify different fault types, such as short circuit, overload, insulation aging, etc., and natural language processing technology is used for extraction, which will not be described in detail in this embodiment. Different levels are set for each fault type according to the degree of impact of the fault on equipment performance and grid safety, as well as the frequency and duration of the fault. Fault levels can be divided into minor, medium, severe, etc., and the specific classification standards need to be formulated according to actual conditions and implementation standards. For example, a short circuit fault may cause equipment damage and grid power outages, so it can be set to a severe level; while minor insulation aging may only affect the local performance of the equipment, so it can be set to a minor level.

[0043] In some embodiments, a fault harmony assessment model is established, specifically including: using the newly added fault type, fault level and other information in the real-time monitoring data set, combined with historical fault data, to analyze the coexistence status and interaction effects of different types of faults. Cleaning, denoising and standardizing the data to ensure data quality. Extracting features related to fault harmony, such as fault type, fault level, device status, timestamp, etc. Encoding and converting the features to make them suitable for model input. Selecting a machine learning or deep learning model to build a fault harmony assessment model. Using historical fault data and real-time monitoring data for model training, for example, the training data includes information such as fault type (such as short circuit, overload), fault level (such as minor, medium, severe), device status (such as normal, abnormal) and timestamp. Optimizing model parameters and improving assessment accuracy through methods such as cross-validation and grid search. Using the test set to evaluate the trained model, calculate the model's accuracy, recall rate, F1 score and other indicators. Optimizing and adjusting the model according to the evaluation results. The fault harmony assessment model outputs the degree of coordination between faults, indicating the balance and restraint relationship between faults.

[0044] In some embodiments, based on the output data of the fault harmony assessment model, the coordination degree or influence mechanism between faults is analyzed, and the output data includes the coordination degree of power equipment, fault data of power equipment and equipment combination; the equipment combination is a non-repetitive random combination of power equipment, and the number of combinations is between [2, 4]; the coordination degree refers to the degree of mutual influence between equipment combinations, and the value is between [0, 1]; the larger the coordination degree, the smaller the mutual influence between faults, and the more likely they are to cancel each other out; the lower the value, the greater the mutual influence between faults, and the more likely they are to aggravate each other. The coordination degree is comprehensively calculated based on fault type, fault level, equipment status, timestamp and mutual influence mechanism between faults, and is calculated using a weighted summation method to evaluate the coordination degree between faults; weighted summation is a well-known technical content, and this application will not describe it in detail here.

[0045] A balance threshold is pre-set to distinguish balanced fault combinations from unbalanced fault combinations. Traverse all fault combinations, and identify fault combinations with a coordination degree not less than the balance threshold as balanced fault combinations based on the comparison result between the coordination degree and the threshold. The faults in these combinations may offset each other or have a small impact, and no additional intervention is required. Fault combinations with a coordination degree less than the threshold are identified as unbalanced fault combinations. The faults in these combinations may aggravate each other or have a large impact, and intervention measures are required. The identified balanced fault combination sets and unbalanced fault combination sets are sorted out, including information such as specific equipment, specific faults, and coordination degree.

[0046] In some embodiments, the balance and restraint mechanism is used to identify different fault combinations, and then determine the balance and restraint factors between different faults, select corresponding intervention measures, and estimate the changing trend of the failure probability of power equipment; balance is to adjust the system operating status and fault handling strategy so that the mutual influence between faults reaches a balanced state, and restraint is to take intervention measures to restrain the aggravation trend of unbalanced fault combinations and prevent system collapse.

[0047] For each fault combination, we deeply analyze the mutual influence between faults and determine their balance and restraint factors. Balance factors refer to those factors that can offset or reduce the impact of faults. For example, some faults may suppress each other under certain conditions and will not cause system collapse. Restraint factors refer to those factors that can prevent or slow down the trend of unbalanced fault combinations, such as restraining the further development of faults by adjusting the operating status of equipment and adding redundant backups. According to the analysis results of balance and restraint factors, select corresponding intervention measures. For balanced fault combinations, observation or slight adjustment strategies can be adopted, such as regularly checking the status of equipment to ensure that no new faults occur. For unbalanced fault combinations, specific intervention measures need to be formulated, such as adjusting the operating status of equipment, adding redundant backups, and repairing or replacing equipment. Combined with real-time monitoring data and fault harmony assessment results, statistical analysis and machine learning are used to estimate the trend of changes in the failure probability of power equipment.

[0048] Specifically, big data analysis and mining technology is used to conduct in-depth analysis of real-time monitoring data and historical fault data to discover the correlation patterns and mutual influence mechanisms between faults. Balance and restraint factors are identified through methods such as association rule mining and cluster analysis. A fault simulation and simulation model is established to simulate the system operating status and fault impact under different fault combinations. Through simulation experiments, the balance and restraint relationship between faults is observed and analyzed to provide a basis for formulating intervention measures. Evaluate and make decisions on fault combinations to determine the balance and restraint factors and formulate more scientific and reasonable intervention measures. Using big data analysis and mining technology for information mining is a prior art, and this application does not make specific restrictions here.

[0049] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages: This application can extract different fault types and set reasonable levels, fully understand the fault situation, accurately output the coordination degree between faults through the fault harmony evaluation model, clearly present the balance and restraint relationship between faults, and achieve the effect of automatically clearing faults and reducing external intervention; based on the coordination degree and balance threshold, accurately identify balanced fault combinations and unbalanced fault combinations, and clarify the impact of different fault combinations on the system. Formulate a balance restraint mechanism, select appropriate intervention measures for different fault combinations, effectively adjust the failure probability of power equipment, estimate the trend of failure probability changes, and achieve more accurate fault warning.

[0050] Embodiment 3: This embodiment makes further improvements on the basis of the above contents.

[0051] The method also includes: S400: establishing a probabilistic traffic flow model, taking power equipment as nodes, taking target detection lines as energy flow paths, obtaining energy flow speed, load change probability and failure probability based on real-time monitoring data, analyzing the failure probability distribution on the energy flow path, identifying probabilistic traffic congestion points and high-incidence areas of failure, and generating a flow velocity distribution report; performing comprehensive analysis and early warning based on the flow velocity distribution report and the monitoring and analysis report.

[0052] In some embodiments, a probabilistic traffic flow model is established, specifically including: using power equipment (such as transformers, transmission lines, switchgear, etc.) as nodes in the network. Each node has specific attributes, such as equipment type, capacity, operating status, etc. The target detection line (i.e., energy flow path) is used as an edge in the network. The edge has directionality, indicating the direction of energy flow, and is accompanied by attributes such as line length, transmission capacity, load rate, etc.

[0053] Using sensors and monitoring technology, the energy flow data and load change information of the power grid system are collected in real time, including key parameters such as voltage, current, power, and load rate of each node. The collected data is cleaned, denoised, and standardized to ensure data quality. Features related to energy flow and fault probability are extracted, such as energy flow direction, flow size, load change rate, etc.

[0054] Based on real-time monitoring data, the energy flow rate on each edge is calculated by analyzing parameters such as current, voltage, and power. The probability of load changes under different time periods and conditions is analyzed using historical data and real-time monitoring data. Combined with information such as equipment status, operating environment, and historical fault records, the probability of failure of each node and edge is calculated using a probabilistic model or machine learning algorithm.

[0055] A probabilistic traffic flow model based on deep learning is established. The model is trained using historical data and real-time monitoring data. Model parameters are adjusted to improve prediction accuracy. The model is verified and used. After inputting real-time monitoring data, key parameters such as energy flow rate, load change probability, and failure probability are output.

[0056] In some embodiments, the failure probability on the energy flow path is calculated using a trained probabilistic traffic flow model. The influence of factors such as the length of the energy flow path, the load size, and the equipment status on the failure probability is considered. The failure probability distribution on the energy flow path is analyzed to identify areas and paths with higher failure probabilities. A heat map or streamline map can be generated by visualization technology to display the distribution of failure probabilities. This application does not make specific restrictions here.

[0057] According to the fault probability distribution results, the probabilistic traffic congestion points (i.e., nodes or edges with higher fault probabilities) and high-fault areas (i.e., areas with higher fault probabilities) are identified, and a flow speed distribution report is generated.

[0058] In some embodiments, the flow rate distribution report includes a probability distribution diagram of faults on the energy flow path, detailed information on probable traffic congestion points and high-prone areas of faults, and monitoring and preventive measures for these areas. Comprehensive analysis and early warning are performed based on the flow rate distribution report and the monitoring and analysis report.

[0059] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages: This application establishes a probabilistic traffic flow model that can accurately calculate the energy flow speed, load change probability and failure probability, clearly present the failure probability distribution on the energy flow path, intuitively display the failure probability distribution, accurately identify probabilistic traffic congestion points and high-incidence areas of failure, reduce the risk of failures, and further improve the stability and reliability of the power grid system.

[0060] Embodiment 4: This embodiment makes further improvements on the basis of the above contents.

[0061] The method also includes: S500: obtaining a flow velocity distribution report and a monitoring and analysis report, defining evaluation indicators, and calculating fault tolerance values ​​under different faults; presetting a balance bottom line value, and calculating the stability boundary of the system under multiple faults according to the fault tolerance value, and then setting a redistribution mechanism to dynamically adjust the system's operating parameters and operation and maintenance strategies.

[0062] In some embodiments, the evaluation indicators should be able to comprehensively and objectively quantify the system's ability to maintain normal operation under equipment failure conditions. They should be based on the system's actual operating data, failure history records, and related performance indicators to ensure that they can accurately reflect the system's fault tolerance and stability.

[0063] In some embodiments, the evaluation indicators include but are not limited to: load recovery rate, fault isolation time, energy loss rate and system recovery time; load recovery rate refers to the proportion of the system that can quickly restore the load to a normal level after a fault occurs. It reflects the system's ability to recover after a fault; fault isolation time refers to the time required from the occurrence of a fault to the successful isolation of the fault area by the system, which is used to reflect the system's response speed and isolation ability to the fault; energy loss rate refers to the ratio of the energy lost by the system during the fault to the energy during normal operation, which is used to reflect the impact of the fault on the energy flow of the system; system recovery time refers to the time required from the occurrence of a fault to the complete recovery of the system to normal operation, which is used to reflect the system's fault tolerance and recovery speed.

[0064] In some embodiments, according to the evaluation index, the corresponding weight coefficient is pre-set, and the fault tolerance value is obtained by weighted summation according to the actual index value and the corresponding weight coefficient; according to the fault tolerance value, the stability boundary of the system under continuous multiple faults is calculated, and the balance bottom line value is the minimum value used to evaluate the stability of the system, which is set according to historical data, and the present application does not impose specific restrictions on this. Specifically, it includes: using simulation technology or actual test data to simulate the system operation status under different fault scenarios; considering the situation of continuous multiple faults, that is, after a fault occurs, the system may fail again before it is fully recovered. In each fault scenario, the stability performance of the system is analyzed, including the changes in key parameters such as voltage, current, and power. The stability boundary refers to the critical point where the fault tolerance value of the power system is lower than the balance bottom line value under continuous multiple faults. Select indicators that can reflect the stability boundary of the system, such as the maximum allowable deviation of key parameters, the shortest time before the system collapse, etc.; used to quantify the stability boundary of the system under continuous multiple faults. According to the simulation results and the stability boundary indicators, the stability boundary of the system under continuous multiple faults is calculated.

[0065] In some embodiments, the redistribution mechanism refers to a mechanism that ensures that the system can maintain normal operation and optimize performance by replanning the energy flow path, adjusting the load distribution, etc. when the system fails or the load changes. According to the fault location and load changes, the energy flow path is dynamically adjusted to avoid the fault area or overload area. According to the load capacity and operating status of each node, the load distribution is dynamically adjusted to ensure the load balance of the system. When a fault occurs, the backup equipment is automatically or manually enabled to replace the faulty equipment to maintain the normal operation of the system.

[0066] According to the fault identification results, the redistribution mechanism is triggered to start replanning the energy flow path and adjusting the load distribution. According to the results of the redistribution mechanism, the system operating parameters, such as voltage, current, power, etc., are dynamically adjusted to ensure stable operation of the system. Targeted operation and maintenance strategies are formulated, such as strengthening equipment inspections, replacing aging equipment in advance, optimizing load distribution, etc., to improve the system's operation and maintenance efficiency and reliability. The implementation effect of the redistribution mechanism is evaluated, and the redistribution mechanism and operation and maintenance strategy are adjusted based on the evaluation results to achieve continuous optimization.

[0067] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages: This application accurately quantifies the system's ability to maintain normal operation under equipment failure conditions by defining evaluation indicators and calculating fault tolerance values; accurately calculates the system's stability boundary under continuous multiple failures. The reallocation mechanism can dynamically adjust the system's operating parameters and operation and maintenance strategies, effectively respond to system failures and load changes, ensure stable system operation and optimize performance.

[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A safety monitoring and early warning method for a power system, characterized in that: include: S100: Acquire a target detection line, where the target detection line includes a plurality of power devices, and respectively acquire real-time monitoring data of each power device to form a real-time monitoring data set; Decomposing the real-time monitoring data set into tasks, obtaining a decomposed task set, and filtering an identified task set from the decomposed task set; S200: Establish a fault probability resonance model, input the identification task set into the fault probability resonance model to obtain the fault probability of each power device, and calculate the resonance effect value between any two devices; S300: obtaining a transfer matrix based on the failure probability and resonance effect value of each power device; Generate probability graph based on transition matrix combined with time series; Predict fault propagation paths based on probability graphs, adjust the fault probability of power equipment, identify critical paths and risk equipment based on the adjusted fault probability, and generate monitoring and analysis reports; Wherein, each row and each column in the transfer matrix represents a fault state, and the matrix elements represent the transition probability of the fault state; The probability graph uses the equipment status and failure probability at each time point as nodes and the transfer relationship as edges; the monitoring and analysis report includes equipment failure probability, failure propagation path, critical path risk and a list of high-risk equipment.

2. A safety monitoring and early warning method for a power system according to claim 1, characterized in that: Calculate the resonance effect value between any two devices, including: calculate the correlation based on the electrical connection strength and physical distance between the devices, the calculation formula is as follows: in, is the association between device i and device j; is the electrical connection strength between device i and device j, is the physical distance between device i and device j; and They are the pre-set electrical connection strength weight coefficient and physical distance weight coefficient, both of which have a value range between [0,1]; The calculation formula of the resonance effect value is as follows: in, is the resonance effect value of device i and device j; It is the preset resonance effect adjustment coefficient, the value range is between [0.1, 0.5], which is used to adjust the resonance effect value; is the failure probability of device i; is the failure probability of device j.

3. A safety monitoring and early warning method for a power system as claimed in claim 2, characterized in that: The calculation formula of transition probability is as follows: in, is the transition probability; is the global adjustment coefficient, with a value range of [0,1], which is used to control the contribution ratio of the resonance effect; is a time decay function, which describes the characteristics of the influence of device i on the failure probability of device j decaying with time; is the decay rate, and t is the time interval between device i and device j.

4. A safety monitoring and early warning method for a power system according to claim 1, characterized in that: In step S300, it also includes: S310: Acquire real-time monitoring data and historical monitoring analysis reports, extract fault types, and set fault levels; input the real-time monitoring data, fault types, and fault levels into a fault harmony assessment model to generate a balanced fault combination set and an unbalanced fault combination set; S320: Based on the balanced fault combination set and the unbalanced fault combination set, adjust the failure probability of the power equipment and identify the critical path and risk equipment.

5. A safety monitoring and early warning method for a power system as claimed in claim 4, characterized in that: Generate a balanced fault combination set and an unbalanced fault combination set, including: based on the output data of the fault harmony evaluation model, compare the coordination degree with the balance threshold according to a preset balance threshold, mark the device combination with a coordination degree not less than the balance threshold as a balanced fault combination, and add it to the balanced fault combination set; mark the device combination with a coordination degree less than the balance threshold as an unbalanced fault combination, and add it to the unbalanced fault combination set.

6. A safety monitoring and early warning method for a power system as claimed in claim 5, characterized in that: The output data includes power equipment, fault data of power equipment and coordination degree of equipment combination; the equipment combination is a non-repetitive random combination of power equipment, and the number of combinations is between [2, 4]; the coordination degree refers to the degree of mutual influence between equipment combinations, and the value is between [0, 1].

7. A safety monitoring and early warning method for a power system according to claim 1, characterized in that: The method further comprises: S400: Establish a probabilistic traffic flow model, take power equipment as nodes, and target detection lines as energy flow paths. Obtain energy flow speed, load change probability, and failure probability based on real-time monitoring data, analyze the failure probability distribution on the energy flow path, identify probabilistic traffic congestion points and high-incidence areas of failure, and generate a flow velocity distribution report; conduct comprehensive analysis and early warning based on the flow velocity distribution report and monitoring and analysis report.

8. A safety monitoring and early warning method for a power system as claimed in claim 7, characterized in that: The flow rate distribution report includes a failure probability distribution diagram on the energy flow path, probabilistic traffic congestion points, detailed information on high-probability failure areas, and preventive measures recommendations.

9. A safety monitoring and early warning method for a power system according to claim 7, characterized in that: The method also includes: S500: obtaining a flow velocity distribution report and a monitoring and analysis report, defining evaluation indicators, and obtaining a fault tolerance value by weighted summation based on the evaluation indicator value and the corresponding pre-set weight coefficient; pre-setting a balance bottom line value, and calculating the stability boundary of the system under multiple fault conditions based on the fault tolerance value, and then setting a redistribution mechanism to dynamically adjust the system's operating parameters and operation and maintenance strategies.

10. A safety monitoring and early warning method for a power system according to claim 9, characterized in that: The evaluation indicators include load recovery rate, fault isolation time, energy loss rate and system recovery time; the stability boundary refers to the critical point where the fault tolerance value of the power system is lower than the balance bottom line value under multiple fault conditions; the balance bottom line value is the minimum value used to evaluate system stability.

Images