Intelligent auxiliary analysis method and system for distribution network

Abstract

The invention discloses an intelligent auxiliary analysis method and system for a distribution network, and relates to the technical field of power distribution networks, and the method comprises the following steps: collecting alarm information data and operation data of each monitoring point of the power distribution network in real time; identifying the alarm information data based on a time window mechanism and filtering the same alarm information repeatedly appearing in the G time; constructing an alarm information evaluation model for the filtered alarm information data and operation data based on a BP neural network to obtain an alarm information evaluation coefficient; acquiring alarm information evaluation coefficients around each monitoring point of the power distribution network in the same time interval, and constructing a first data set; and according to the average value of the first data set of each monitoring point, performing priority ranking on the alarm events, and preferentially processing the monitoring point area with the maximum average value. The method comprehensively considers the evaluation mode of various factors, enables the alarm information evaluation to be more comprehensive and accurate, and provides a more reliable decision basis for operation and maintenance personnel.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network, and more specifically, to a distribution network intelligent auxiliary analysis method and system. Background Art

[0002] With the rapid development of the power industry, the safety and reliability of the distribution network, as a key link in the power system, directly affects all aspects of social production and life. However, traditional distribution network management methods are powerless in the face of increasingly complex grid structures and frequent failures. In recent years, thefts against distribution facilities have occurred frequently, and overloaded operation of old equipment can easily cause fires. These problems have made the anti-theft and fire prevention tasks in distribution production management particularly arduous.

[0003] In this context, the development of intelligent auxiliary analysis methods and systems for distribution networks is particularly important. The system integrates multiple technical means such as high-definition video surveillance, environmental monitoring, security alarms, and intelligent control to achieve all-round perception and monitoring of the distribution network operating environment. This system can not only effectively prevent safety hazards such as theft and fire, but also quickly locate and handle faults when they occur, improving the timeliness and accuracy of fault repair.

[0004] At the same time, with the advancement of smart grid construction, the intelligence level of distribution network is also constantly improving. As an important part of smart grid, the comprehensive auxiliary system of smart distribution network can realize multi-level networking and cross-regional monitoring of distribution network, providing strong guarantee for the stable operation of smart distribution network. By integrating advanced intelligent technologies such as image recognition, data analysis, knowledge graph, etc., the system can realize intelligent analysis and auxiliary decision-making of distribution network faults, reduce dependence on manual experience, and improve the efficiency and accuracy of fault handling. The intelligent auxiliary analysis method of distribution network realizes real-time monitoring and data collection of equipment operation status by introducing sensors, smart terminals and other equipment. These data are efficiently processed by cloud computing platform and can provide strong data support for the operation and maintenance management of distribution network. At the same time, using big data technology for data mining and analysis can reveal potential problems in power grid operation and provide scientific basis for decision-making. In addition, the introduction of artificial intelligence technology makes the fault prediction, diagnosis and optimization control of distribution network more intelligent and accurate, greatly improving the operation efficiency and reliability of power grid.

[0005] However, despite the significant technical progress made in the intelligent auxiliary analysis method of distribution network, there are still some shortcomings in practical applications. For example, in terms of fault prediction and diagnosis, due to the diversity and complexity of power grid faults, traditional technologies often find it difficult to achieve 100% accurate prediction and diagnosis. This requires us to continuously explore and innovate in algorithm optimization, model construction, etc. to improve the accuracy and reliability of intelligent auxiliary analysis.

[0006] The above disclosed technical solutions have at least the following technical problems: Traditional operation and maintenance systems often generate a large number of alarm messages, many of which are useless or repeated. These useless alarm messages not only interfere with the normal work of operation and maintenance personnel, but may also cover up truly useful alarm messages. In view of the above problems, the present invention proposes a solution. Summary of the invention

[0007] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide a distribution network intelligent assisted analysis method and system, which filters and prioritizes alarm information through intelligent means to improve the work efficiency and response speed of operation and maintenance personnel, ensure that important alarm information can be processed in time, and solve the problems raised in the above-mentioned background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions: The intelligent auxiliary analysis method and system for distribution network include the following steps: real-time collection of alarm information data and operation data of each monitoring point of the distribution network; identification of alarm information data based on a time window mechanism and filtering of the same alarm information that appears repeatedly within G time; building an alarm information evaluation model based on a BP neural network for the filtered alarm information data and operation data to obtain an alarm information evaluation coefficient; The alarm information evaluation coefficients around each monitoring point of the distribution network within the same time interval are obtained to construct a first data set; the alarm events are prioritized according to the average value of the first data set of each monitoring point, and the monitoring point area with the largest average value is processed first.

[0009] In a preferred embodiment, the alarm information data includes an event impact coefficient, and the operating data includes a distribution network state abnormality coefficient and an environmental abnormality coefficient; the specific method for obtaining the event impact coefficient is as follows: obtain distribution network alarm information, including timestamp, alarm type, alarm content and alarm level; split the alarm information text into independent words and phrases, and convert text data in different forms into the same form; extract alarm information text features based on a preset word embedding model, including fault type, fault level, number of affected devices, and power outage scope; quantify the alarm information text features, divide the fault level into different levels, assign corresponding numerical values, and quantify features such as the number of affected devices and the power outage scope; based on the results of feature extraction and quantification, construct an event impact model to obtain the event impact coefficient; input new alarm information into the model, and calculate the event impact coefficient according to the output results of the model.

[0010] In a preferred embodiment, the specific method for obtaining the abnormal coefficient of the distribution network state is as follows: Acquire abnormal data occurring at each line node within a time interval of t; perform data cleaning, classification and quantification on the abnormal data to obtain a second data set of each line node; calculate the abnormal rate of each line node based on the data in the second data set; calculate the abnormal coefficient of each line node based on the abnormal rate, fault frequency and total fault duration and based on a preset distribution network state abnormal coefficient formula.

[0011] In a preferred embodiment, a specific method for obtaining the environmental anomaly coefficient is as follows: collecting environmental parameters in real time through sensors, the environmental parameters including relative humidity, air pressure and pollutant concentration; constructing an environmental parameter weight model based on machine learning according to the acquired environmental parameters to obtain the anomaly weights of each environmental parameter; summing the anomaly weights of each environmental parameter to obtain a comprehensive environmental anomaly coefficient.

[0012] In a preferred embodiment, the alarm information data is identified and filtered based on the time window mechanism, specifically: the size of the time window and the sliding step are set; the alarm information in the time window is stored based on a hash table, and the alarm information includes a timestamp, alarm type, alarm content and alarm level; the alarm information of each monitoring point is received in real time, and processed according to the time window mechanism, including data classification, filtering, and aggregation operations; a cache area is established in the memory or in the database, and when new alarm information is received, the alarm type and alarm content are compared with the alarm information in the cache area; if the comparison result is highly similar, it is considered to be a duplicate alarm; the identified duplicate alarm information is merged and processed, and the repeated alarm information is merged into one record; a timer mechanism is used to trigger the sliding and updating of the window, and when the end time of the window is reached, the sliding operation is performed.

[0013] In a preferred embodiment, the filtered alarm information data is used to construct an alarm information evaluation model based on a BP neural network, specifically: the number of layers of the neural network and the number of neurons in each layer are initialized, and the neural network includes an input layer, a hidden layer and an output layer; the number of nodes in the input layer is defined as the number of types of feature information data; the tanh function is selected as the activation function in the hidden layer and the output layer; the weights and biases of the neural network are randomly initialized; the cross entropy loss is selected as the loss function; the weights and biases of the neural network are updated using an Adam optimizer to minimize the loss function; during the training process, the feature information data is divided into a training set and a validation set for model training; the trained model is used as an alarm information evaluation model, and the output result of the alarm information evaluation model is used as the alarm information evaluation coefficient.

[0014] In a preferred embodiment, the specific calculation formula of the alarm information evaluation coefficient is as follows: the alarm events are prioritized according to the average value of the first data set of each monitoring point, and the monitoring point area with the largest average value is prioritized, specifically: all alarm information evaluation coefficients of each monitoring node are extracted from the first data set; all alarm information evaluation coefficients are averaged to obtain the average evaluation coefficient of each monitoring point; the monitoring points are sorted according to the size of the average evaluation coefficient, and the monitoring point with a larger average evaluation coefficient has a higher priority; a list is generated for the sorted monitoring points, and the top of the list is the monitoring point with the highest priority; according to the sorted list, the monitoring point area with the largest average evaluation coefficient is prioritized; after processing multiple high-priority monitoring points, the processing effect is evaluated, including the reduction of alarm events and the improvement of system stability; according to the evaluation results, the processing strategy is adjusted to increase attention to specific types of alarm events and optimize the evaluation model.

[0015] The technical effects and advantages of the distribution network intelligent auxiliary analysis method and system of the present invention are as follows: 1. The present invention effectively identifies and filters the same alarm information that appears repeatedly within the G time through the time window mechanism, avoids the interference of redundant information, and significantly improves the efficiency of alarm information processing. At the same time, the BP neural network is used to construct an alarm information evaluation model, and the filtered alarm information is accurately evaluated to obtain the alarm information evaluation coefficient, thereby realizing the priority sorting of alarm events. This processing method not only reduces the interference of invalid alarms, but also ensures that important alarms can be processed in a timely and prioritized manner, thereby improving the operation and maintenance efficiency and safety of the entire distribution network.

[0016] 2. The present invention comprehensively considers multiple factors such as the distribution network state abnormality coefficient and the environmental abnormality coefficient. The distribution network state abnormality coefficient comprehensively reflects the operating status of the distribution network by calculating key indicators such as the abnormality rate, fault frequency and total fault duration of each line node; while the environmental abnormality coefficient accurately evaluates the impact of environmental factors on the operation of the distribution network by collecting environmental parameters in real time and building an environmental parameter weight model based on machine learning. This evaluation method that comprehensively considers multiple factors makes the alarm information evaluation more comprehensive and accurate, and provides a more reliable decision-making basis for operation and maintenance personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a structural schematic diagram of the distribution network intelligent auxiliary analysis method of the present invention.

[0018] Figure 2 This is a schematic diagram of the structure of the intelligent auxiliary analysis system for distribution network of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. 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.

[0020] Embodiment 1, Figure 1 The invention provides a distribution network intelligent auxiliary analysis method, which includes the following steps: Collect alarm information and operation data of each monitoring point of the distribution network in real time; Based on the time window mechanism, the alarm information data is identified and the same alarm information that appears repeatedly within G time is filtered; The alarm information evaluation model is constructed based on the BP neural network for the filtered alarm information data and operation data to obtain the alarm information evaluation coefficient; Acquire alarm information evaluation coefficients around each monitoring point of the distribution network within the same time interval to construct a first data set; According to the average value of the first data set of each monitoring point, the alarm events are prioritized, and the alarm events with the highest alarm information evaluation coefficient are processed first.

[0021] S1, real-time collection of alarm information data and operation data of each monitoring point of the distribution network; Real-time collection of alarm information data from each monitoring point of the distribution network can ensure the safety and stability of the power grid operation. By timely analyzing and processing this data, the fault point can be quickly located, and corresponding measures can be taken to repair it to prevent the fault from expanding. At the same time, this data is also an important basis for evaluating the operation status of the power grid, optimizing the power grid structure, and improving the efficiency of the power grid operation.

[0022] At present, commonly used data collection methods include wired communication (such as optical fiber communication, Ethernet communication, etc.) and wireless communication (such as Zigbee, LoRa, NB-IoT, etc.). At the same time, combined with edge computing, artificial intelligence and other technologies, real-time processing and analysis of collected data can be achieved, improving the efficiency and accuracy of data processing.

[0023] The alarm information data includes an event impact coefficient, and the operation data includes a distribution network state abnormality coefficient and an environmental abnormality coefficient.

[0024] In the management and operation of distribution networks, the event impact coefficient can be used to evaluate the potential impact of different events on the distribution network, thereby helping to formulate more effective response strategies and measures. For example, for events with a higher impact coefficient, emergency repair resources can be prioritized to ensure rapid restoration of power supply; while for events with a lower impact coefficient, more flexible scheduling and operation strategies can be adopted.

[0025] In addition, the event impact coefficient can also be used for the planning and optimization of distribution networks. Through the analysis and evaluation of historical events, the weak links and potential risk points in the distribution network can be identified, so as to formulate targeted improvement measures and optimization plans to improve the reliability and stability of the distribution network.

[0026] The specific method for obtaining the event impact coefficient is as follows: Obtain distribution network alarm information, including timestamp, alarm type, alarm content and alarm level; Split alarm information text into independent words and phrases, and convert different forms of text data into the same form; Extract alarm information text features based on the preset word embedding model, including fault type, fault level, number of affected devices, and power outage scope; Quantify the text features of alarm information, such as classifying fault levels into different levels (such as minor, general, severe, etc.) and assigning corresponding values ​​(such as 1, 2, 3, etc.). At the same time, quantify and count the features such as the number of affected devices and the scope of power outage; Based on the results of feature extraction and quantification, an event impact model is constructed to obtain the event impact coefficient; Input the new alarm information into the model and calculate the event impact coefficient based on the output results of the model.

[0027] The specific calculation formula of the event impact coefficient is as follows:

[0028] Among them, sj is the event impact coefficient, n is the number of alarm events at the monitoring point, d is the fault level, s is the number of affected devices, and t is the quantitative value of the power outage range. is the weight factor of the preset fault level, is the preset weight factor affecting the number of devices, It is the weight factor of the preset power outage range quantization value.

[0029] The distribution network status abnormality coefficient is used to quantitatively evaluate the degree to which certain status parameters or performance indicators in the distribution network deviate from normal or expected values, thereby helping operation and maintenance personnel to promptly discover and deal with potential problems.

[0030] The specific method for obtaining the abnormal coefficient of the distribution network state is as follows: Obtain the abnormal data occurring at each line node within the time interval t; Clean, classify and quantify the abnormal data to obtain the first data set of each line node; Calculate the abnormality rate of each line node according to the data in the first data set; According to the abnormality rate, fault frequency and total fault duration, the abnormality coefficient of each line node is calculated based on the preset distribution network status abnormality coefficient formula.

[0031] The specific calculation formula of the abnormality rate is as follows:

[0032] The distribution network status abnormality coefficient formula is:

[0033] In the formula, is the abnormal rate, N is the total number of abnormal events, is the duration of the abnormal event of the i-th state, is the preset weight of the i-th state abnormal event, indicating the severity of the state abnormal event, T represents the total observation time, Z represents the number of line nodes, yc is the distribution network state abnormality coefficient, f is the fault frequency, and cx is the total fault duration.

[0034] The distribution network status abnormality coefficient has the following benefits for distribution network intelligent auxiliary analysis: Early fault detection: By real-time monitoring of the abnormal coefficient of the distribution network status, potential fault risks can be quickly identified. Changes in the abnormal coefficient usually indicate abnormal working conditions of the equipment, so maintenance and repair can be carried out in advance to reduce the probability of faults.

[0035] Accurately locate faults: When the abnormal coefficient shows a specific pattern, it can help operation and maintenance personnel locate the source of the fault more accurately. This can reduce the time for troubleshooting, quickly restore power supply, and reduce power outage time.

[0036] Fault prediction: Using historical data and the changing trend of abnormal coefficients, a prediction model can be established to identify possible fault points in advance. This predictive maintenance can significantly reduce power outages caused by equipment failures.

[0037] Optimize maintenance plans: Analyzing abnormality coefficients can help the operation and maintenance department optimize maintenance and overhaul plans, arrange resources based on real-time data on equipment status, avoid failures caused by equipment aging or improper use, and improve the reliability of the overall system.

[0038] Improve operational efficiency: By analyzing the abnormal state of the distribution network, we can optimize load distribution and reduce failures caused by overload. This optimization can reduce the failure rate and improve the stability of power supply.

[0039] Reduce economic losses: A reduction in failure rate directly means a reduction in power outages, thereby reducing economic losses and user complaints caused by power outages and improving user satisfaction.

[0040] The environmental anomaly coefficient reflects the degree of abnormal changes in the environment caused by natural environmental conditions or human activities. These changes may include extreme fluctuations in climate conditions, the occurrence of natural disasters, increased emissions of pollutants, and imbalances in ecosystems. By monitoring and analyzing these environmental factors, the environmental anomaly coefficient can be calculated to assess its impact on human activities, ecosystem health, and the stable operation of specific systems (such as power systems).

[0041] The specific method for obtaining the environmental anomaly coefficient is as follows: Collect environmental parameters in real time through sensors, the environmental parameters including relative humidity, air pressure and pollutant concentration; According to the acquired environmental parameters, an environmental parameter weight model is constructed based on machine learning to obtain the abnormal weight of each environmental parameter; The anomaly weights of each environmental parameter are summed to obtain the comprehensive environmental anomaly coefficient.

[0042] The specific calculation formula of the environmental anomaly coefficient is as follows:

[0043] In the formula, hj is the environmental anomaly coefficient, sd is the relative humidity, qy is the air pressure, and nd is the pollutant concentration. is the anomaly weight of relative humidity, is the anomaly weight of the air pressure, is the anomaly weight of the pollutant concentration.

[0044] The environmental anomaly coefficient has the following benefits for intelligent auxiliary analysis of distribution networks: Real-time monitoring of environmental impacts: By monitoring climate change, temperature, humidity and other factors, its impact on distribution network equipment can be evaluated and operation and maintenance strategies can be adjusted in a timely manner.

[0045] Fault warning: Environmental anomalies such as storms and extreme temperatures can cause equipment failures. Analyzing these factors can help to provide early warnings and reduce the occurrence of failures.

[0046] Optimize equipment configuration: Optimize equipment selection and configuration based on environmental data to improve its anti-interference ability, thereby improving the reliability of the distribution network.

[0047] Improve emergency response capabilities: Understanding the impact of environmental anomalies on the distribution network can help develop more effective emergency plans and respond quickly to emergencies.

[0048] Enhance system stability: By continuously analyzing environmental factors, we can optimize operating strategies, reduce system instability caused by environmental changes, and ensure power supply security.

[0049] S2, based on the time window mechanism, identifies the alarm information data and filters the same alarm information that appears repeatedly within G time; A time window is a specific time period that specifies the scope of data collection, processing, and storage. During this time period, the system processes the received data according to preset rules and may store some or all of the data. The time window mechanism controls the data processing flow by defining parameters such as the size, type, and sliding mode of the time window.

[0050] The time window mechanism can be divided into several types according to the type and sliding mode of the time window, among which the most common are rolling time windows and sliding time windows: Tumbling time window: The window is fixed, and each time it rolls, old data is discarded and new data is collected.

[0051] Application scenarios: Applicable to scenarios that require strict distinction between time periods, such as collecting data every hour.

[0052] Example: If the time window is set to 1 minute, the window only calculates the data within the current 1 minute, then rolls to the next 1 minute window and discards the data in the previous window.

[0053] Sliding time window: The window slides, retaining some old data and processing it together with the new data.

[0054] Application scenarios: Applicable to scenarios that require smooth transition and continuous monitoring, such as real-time traffic monitoring.

[0055] For example, if the time window is set to 5 minutes and the sliding step is 1 minute, each window will contain data from the current time and the 4 minutes before it. As time goes by, the window will slide forward, but the data from the most recent 5 minutes will always be retained.

[0056] The alarm information data is identified and filtered based on the time window mechanism, specifically: Set the time window size and sliding step size; The alarm information within the time window is stored based on a hash table, wherein the alarm information includes a timestamp, an alarm type, an alarm content and an alarm level; Receive alarm information from each monitoring point in real time and process it according to the time window mechanism, including data classification, filtering, and aggregation operations; A buffer area is established in the memory or database. When new alarm information is received, the alarm type and alarm content are compared with the alarm information in the buffer area. If the comparison results are highly similar, it is considered a duplicate alarm; Merge the identified duplicate alarm information and merge multiple duplicate alarm information into one record; Use the timer mechanism to trigger the sliding and updating of the window. When the end time of the window is reached, the sliding operation is performed.

[0057] The benefits of identifying and filtering alarm information data based on the time window mechanism are as follows: Improve alarm processing efficiency: The time window mechanism can quickly identify and filter out the same alarm information that appears repeatedly in a short period of time, thereby reducing the number of alarms that security administrators need to handle. This allows security administrators to focus more on handling truly important and unique alarms, improving overall processing efficiency.

[0058] Reduce false alarm rate: In some cases, a large number of repeated alarms may be generated due to system errors, network fluctuations or improper configuration. These alarm messages are often meaningless and may even mislead security administrators. Filtering through the time window mechanism can effectively reduce the false alarm rate and improve the accuracy of alarm information.

[0059] Optimize resource utilization: The processing of alarm information consumes a lot of computing resources and storage resources. By filtering duplicate alarms, the use of these resources can be reduced, allowing the system to run more efficiently. At the same time, this also reduces the maintenance cost of the system and improves the overall economic benefits.

[0060] Improve user experience: Frequent repeated alarms can cause distress and anxiety to users of alarm systems. By filtering repeated alarms, the psychological burden of users can be reduced and their user experience can be improved.

[0061] Enhanced security defense capabilities: After filtering out duplicate alarms, security administrators can focus more on handling truly important and urgent alarm information. This helps them discover and respond to potential security threats more quickly, thereby enhancing the system's security defense capabilities.

[0062] Facilitates data analysis and decision-making: Filtered alarm information is clearer and more accurate, which helps security administrators conduct more in-depth data analysis and decision-making. They can formulate more effective security strategies and measures based on information such as alarm type, frequency and trend.

[0063] Enhance system stability: The frequent occurrence of repeated alarms may cause unnecessary burden on the system and may even cause the system to crash. Filtering through the time window mechanism can reduce the pressure on the system and improve the stability and reliability of the system.

[0064] S3, constructing an alarm information evaluation model based on the BP neural network for the filtered alarm information data to obtain an alarm information evaluation coefficient; The BP neural network consists of an input layer, a hidden layer (can have multiple layers) and an output layer. Each neuron is connected to all neurons in the previous layer, transmitting signals through weighted sums and undergoing nonlinear transformations through activation functions. This structure enables the BP neural network to handle complex nonlinear problems.

[0065] The main functions of BP neural network include but are not limited to the following aspects: Pattern recognition: BP neural network can learn and recognize the pattern of input data and map it to the output layer, thereby realizing the classification or recognition of unknown data. This has a wide range of applications in image recognition, speech recognition and other fields.

[0066] Data mining: By training BP neural networks, useful information and knowledge can be extracted from large amounts of data, helping people better understand the patterns and trends behind the data.

[0067] Machine learning: BP neural network is an important model in the field of machine learning. It can predict future results or behaviors by learning historical data. This predictive ability makes BP neural network have broad application prospects in the fields of financial risk assessment, market forecasting, etc.

[0068] Nonlinear mapping: BP neural network can process the nonlinear mapping relationship of input data without revealing the mathematical equation describing this mapping relationship in advance. This makes BP neural network have unique advantages in solving complex nonlinear problems.

[0069] Adaptive learning: BP neural network has the ability of adaptive learning. It can automatically adjust network parameters according to changes in input data to improve network performance and accuracy. This adaptive learning ability makes BP neural network more adaptable and robust in dynamic environments.

[0070] The training process of BP neural network includes two steps: forward propagation and back propagation. In the forward propagation stage, the input data is passed through the input layer of the network to the hidden layer and output layer to calculate the output result. In the back propagation stage, according to the error between the output result and the true value, the weight and bias of each neuron are adjusted through the back propagation algorithm to reduce the error value and improve the accuracy of the network. This process will be iterated continuously until the preset stop condition is reached (such as the error is less than a certain threshold or the number of iterations reaches the upper limit).

[0071] The alarm information evaluation model is constructed based on the BP neural network for the filtered alarm information data, specifically: Initialize the number of layers of the neural network and the number of neurons in each layer. The neural network includes input layer, hidden layer and output layer; The number of nodes in the input layer is defined as the number of types of feature information data; Select tanh function as activation function in hidden layer and output layer; Randomly initialize the weights and biases of the neural network; Select cross entropy loss as the loss function; Use the Adam optimizer to update the weights and biases of the neural network to minimize the loss function; During the training process, the feature information data is divided into a training set and a validation set for model training; The trained model is used as the alarm information evaluation model, and the output result of the alarm information evaluation model is used as the alarm information evaluation coefficient.

[0072] The specific calculation formula of the alarm information evaluation coefficient is as follows:

[0073] In the formula, is the alarm information evaluation coefficient, yc is the distribution network status abnormality coefficient, hj is the environmental abnormality coefficient, sj is the event impact coefficient, is the preset weight factor of the abnormal coefficient of the distribution network status, is the preset environmental anomaly coefficient weight factor, It is the preset event impact coefficient weight factor.

[0074] S4, obtaining alarm information evaluation coefficients around each monitoring point of the distribution network within the same time interval to construct a first data set; Select a fixed time interval, such as hourly, daily, or weekly. This time interval should be short enough to capture the real-time changes of the distribution network monitoring points, but not too short to avoid excessive data volume. Design the structure of the data set, including fields such as time interval, monitoring point number, alarm information (such as alarm type, level, time, etc.), and evaluation coefficient.

[0075] Data entry: Enter the calculated evaluation coefficients and related information into the data set. This can be achieved by writing a program or using a database management tool.

[0076] Data verification: Verify the entered data to ensure its accuracy and completeness. This includes checking whether the data conforms to the expected format, whether there are duplicate or missing values, etc.

[0077] S5, according to the average value of the first data set of each monitoring point, the alarm events are prioritized, and the monitoring point area with the largest average value is processed first.

[0078] The method of prioritizing the alarm events according to the average value of the first data set of each monitoring point, and giving priority to the monitoring point area with the largest average value, is as follows: Extract all alarm information evaluation coefficients of each monitoring node from the first data set; The evaluation coefficients of all alarm information are averaged to obtain the average evaluation coefficient of each monitoring point; The monitoring points are sorted according to the size of the average evaluation coefficient. The monitoring point with a larger average evaluation coefficient has a higher priority. Generate a list of the sorted monitoring points, with the top of the list being the monitoring point with the highest priority; According to the sorted list, the monitoring point area with the largest average evaluation coefficient is processed first; After processing multiple high-priority monitoring points, evaluate the processing results, including the reduction of alarm events and the improvement of system stability; Based on the evaluation results, adjust the processing strategy, increase attention to specific types of alarm events, and optimize the evaluation model.

[0079] Alarm Type: Different types of alarms may represent different levels of urgency and risk. For example, a fire alarm is usually of higher urgency than a simple equipment failure alarm.

[0080] Abnormality of environmental parameters: If the alarm event is related to abnormalities in environmental parameters, the abnormality of these parameters (such as the degree to which temperature, humidity, smoke concentration, etc. exceed preset thresholds) can be used as part of the evaluation coefficient.

[0081] Historical data and trend analysis: By analyzing historical alarm data and current trends, the potential impact and development direction of alarm events can be predicted, thereby adjusting the assessment coefficient.

[0082] Scope of impact and potential losses: Assess the scope of possible impact of the alarm event and the potential losses that may be caused, which is also an important factor in determining the assessment coefficient.

[0083] After calculating the evaluation coefficient of each alarm event, the alarm events can be prioritized according to the coefficient. Generally speaking, the alarm event with a higher evaluation coefficient has a higher priority and needs to be processed first.

[0084] The alarm events with the largest alarm information evaluation coefficient are processed first, specifically: Immediate response: For alarm events with the largest assessment coefficient, the emergency plan should be activated immediately and relevant personnel should be dispatched to the scene for processing.

[0085] Resource allocation: Ensure that there are sufficient resources (such as personnel, equipment, materials, etc.) to handle the alarm event. If necessary, resources from other departments can be deployed for support.

[0086] On-site handling: Take appropriate measures according to the on-site situation, including but not limited to evacuating personnel, isolating dangerous sources, repairing faulty equipment, etc.

[0087] Follow-up: After the alarm incident is initially handled, follow-up tracking and evaluation should be carried out to ensure that the problem is completely resolved and to prevent similar incidents from happening again.

[0088] Embodiment 2, Figure 2 The intelligent auxiliary analysis system for distribution network of the present invention includes a data collection module, a sliding window processing module, an alarm information evaluation module, an evaluation coefficient processing module and an alarm information comprehensive analysis module; Data collection module: used to collect alarm information data and operation data of each monitoring point of the distribution network in real time; Sliding window processing module: used to identify alarm information data based on the time window mechanism and filter the same alarm information that appears repeatedly within G time; Alarm information evaluation module: used to construct an alarm information evaluation model based on the BP neural network for the filtered alarm information data to obtain the alarm information evaluation coefficient; Evaluation coefficient processing module: used to obtain the alarm information evaluation coefficients around each monitoring point of the distribution network within the same time interval and construct a first data set; Comprehensive analysis module: used to prioritize alarm events according to the average value of the first data set of each monitoring point, and give priority to the monitoring point area with the largest average value.

[0089] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0090] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0091] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0092] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0093] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0094] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A distribution network intelligent auxiliary analysis method, characterized in that: The following steps are involved: Collect alarm information and operation data of each monitoring point of the distribution network in real time; Based on the time window mechanism, the alarm information data is identified and the same alarm information that appears repeatedly within G time is filtered; The alarm information evaluation model is constructed based on the BP neural network for the filtered alarm information data and operation data to obtain the alarm information evaluation coefficient; Acquire alarm information evaluation coefficients around each monitoring point of the distribution network within the same time interval to construct a first data set; According to the average value of the first data set of each monitoring point, the alarm events are prioritized, and the monitoring point area with the largest average value is processed first.

2. The distribution network intelligent auxiliary analysis method according to claim 1, characterized in that: The alarm information data includes the event impact coefficient, and the operation data includes the distribution network state abnormality coefficient and the environment abnormality coefficient; The specific method for obtaining the event impact coefficient is as follows: Obtain distribution network alarm information, including timestamp, alarm type, alarm content and alarm level; Split alarm information text into independent words and phrases, and convert different forms of text data into the same form; Extract alarm information text features based on the preset word embedding model, including fault type, fault level, number of affected devices, and power outage scope; Quantify the text features of the alarm information, classify the fault levels into different levels, assign corresponding values, and quantify and count the features such as the number of affected devices and the scope of power outage; Based on the results of feature extraction and quantification, an event impact model is constructed to obtain the event impact coefficient; Input the new alarm information into the model and calculate the event impact coefficient based on the output results of the model.

3. The distribution network intelligent auxiliary analysis method according to claim 2, characterized in that: The specific method for obtaining the abnormal coefficient of the distribution network state is as follows: Obtain the abnormal data occurring at each line node within the time interval t; Clean, classify and quantify the abnormal data to obtain the second data set of each line node; Calculating the abnormality rate of each line node according to the data in the second data set; According to the abnormality rate, fault frequency and total fault duration, the abnormality coefficient of each line node is calculated based on the preset distribution network status abnormality coefficient formula.

4. The distribution network intelligent auxiliary analysis method according to claim 3 is characterized in that: The specific method for obtaining the environmental anomaly coefficient is as follows: Collect environmental parameters in real time through sensors, the environmental parameters including relative humidity, air pressure and pollutant concentration; According to the acquired environmental parameters, an environmental parameter weight model is constructed based on machine learning to obtain the abnormal weight of each environmental parameter; The anomaly weights of each environmental parameter are summed to obtain the comprehensive environmental anomaly coefficient.

5. The distribution network intelligent auxiliary analysis method according to claim 4, characterized in that: The alarm information data is identified and filtered based on the time window mechanism, specifically: Set the time window size and sliding step size; The alarm information within the time window is stored based on a hash table, wherein the alarm information includes a timestamp, an alarm type, an alarm content and an alarm level; Receive alarm information from each monitoring point in real time and process it according to the time window mechanism, including data classification, filtering, and aggregation operations; A buffer area is established in the memory or database. When new alarm information is received, the alarm type and alarm content are compared with the alarm information in the buffer area. If the comparison results are highly similar, it is considered a duplicate alarm; Merge the identified duplicate alarm information and merge multiple duplicate alarm information into one record; Use the timer mechanism to trigger the sliding and updating of the window. When the end time of the window is reached, the sliding operation is performed.

6. The distribution network intelligent auxiliary analysis method according to claim 5, characterized in that: The alarm information evaluation model is constructed based on the BP neural network for the filtered alarm information data, specifically: Initialize the number of layers of the neural network and the number of neurons in each layer. The neural network includes input layer, hidden layer and output layer; The number of nodes in the input layer is defined as the number of types of feature information data; Select tanh function as activation function in hidden layer and output layer; Randomly initialize the weights and biases of the neural network; Select cross entropy loss as the loss function; Use the Adam optimizer to update the weights and biases of the neural network to minimize the loss function; During the training process, the feature information data is divided into a training set and a validation set for model training; The trained model is used as the alarm information evaluation model, and the output result of the alarm information evaluation model is used as the alarm information evaluation coefficient.

7. The distribution network intelligent auxiliary analysis method according to claim 6, characterized in that: The specific calculation formula of the alarm information evaluation coefficient is as follows: The alarm events are prioritized according to the average value of the first data set of each monitoring point, and the monitoring point area with the largest average value is processed first, specifically: Extract all alarm information evaluation coefficients of each monitoring node from the first data set; The evaluation coefficients of all alarm information are averaged to obtain the average evaluation coefficient of each monitoring point; The monitoring points are sorted according to the size of the average evaluation coefficient. The monitoring point with a larger average evaluation coefficient has a higher priority. Generate a list of the sorted monitoring points, with the top of the list being the monitoring point with the highest priority; According to the sorted list, the monitoring point area with the largest average evaluation coefficient is processed first; After processing multiple high-priority monitoring points, evaluate the processing results, including the reduction of alarm events and the improvement of system stability; Based on the evaluation results, adjust the processing strategy, increase attention to specific types of alarm events, and optimize the evaluation model.

8. The distribution network intelligent auxiliary analysis method according to claim 7, characterized in that: The specific calculation formula of the distribution network status abnormality coefficient is as follows: The specific calculation formula of the abnormality rate is as follows: The distribution network status abnormality coefficient formula is: In the formula, is the abnormal rate, N is the total number of abnormal events, is the duration of the abnormal event of the i-th state, is the preset weight of the i-th state abnormal event, indicating the severity of the state abnormal event, T represents the total observation time, Z represents the number of line nodes, yc is the distribution network state abnormality coefficient, f is the fault frequency, and cx is the total fault duration.

9. The distribution network intelligent auxiliary analysis method according to claim 8, characterized in that: The specific calculation formula of the environmental anomaly coefficient is as follows: In the formula, hj is the environmental anomaly coefficient, sd is the relative humidity, qy is the air pressure, and nd is the pollutant concentration. is the anomaly weight of relative humidity, is the anomaly weight of the air pressure, is the anomaly weight of the pollutant concentration.

10. A system using the distribution network intelligent auxiliary analysis method according to any one of claims 1 to 9 comprises a data collection module, a sliding window processing module, an alarm information evaluation module and an alarm information comprehensive analysis module; Data collection module: used to collect alarm information data and operation data of each monitoring point of the distribution network in real time; Sliding window processing module: used to identify alarm information data based on the time window mechanism and filter the same alarm information that appears repeatedly within G time; Alarm information evaluation module: used to construct an alarm information evaluation model based on the BP neural network for the filtered alarm information data to obtain the alarm information evaluation coefficient; Evaluation coefficient processing module: used to obtain the alarm information evaluation coefficients around each monitoring point of the distribution network within the same time interval and construct a first data set; Comprehensive analysis module: used to prioritize alarm events according to the average value of the first data set of each monitoring point, and give priority to the monitoring point area with the largest average value.

Images