A wind farm monitoring alarm method and system based on multi-packaging point rules

Abstract

The present invention discloses a wind farm station monitoring and alarm method and system based on multi-packaging point rules, which relates to the technical field of wind power generation, including: collecting the status data and monitoring point data of each wind turbine in real time through the data acquisition module of each wind turbine, and performing preprocessing; constructing a graph of the wind turbine status data and the monitoring point data, and finding the shortest path from the monitoring point to the alarm state; constructing a bipartite graph according to the shortest path result, and performing matching optimization between the monitoring point and the alarm state; constructing a flow network, using the network flow algorithm to optimize the global data flow distribution, and comprehensively optimizing the system performance indicators; monitoring the wind turbine status in real time through the multi-packaging point rule, and pushing the alarm event message to the database. Through the shortest path calculation and bipartite graph matching optimization, the risk of false alarm and missed alarm is effectively reduced. Improve the response speed and ensure efficient and real-time data transmission. Adapt to complex operating environments and achieve high-precision monitoring and rapid fault alarm.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a wind farm monitoring and alarm method and system based on a multi-packing point rule. Background Art

[0002] As an important part of the renewable energy field, wind power generation technology has developed rapidly in recent years. With the continuous expansion of wind farm scale and the increase in the number of wind turbines, wind farm operation monitoring and fault warning have become a key issue that needs to be solved urgently. Traditional wind farm monitoring methods mainly rely on regular maintenance and on-site manual inspections. This method is not only inefficient, but also difficult to detect and handle faults in a timely manner, which may cause the wind turbine to operate in a sub-healthy state for a long time, affecting power generation efficiency and equipment life. In order to improve the operational safety and efficiency of wind farms, more and more research has begun to focus on how to achieve comprehensive monitoring of wind turbine status through real-time monitoring and intelligent early warning systems. Summary of the invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] Therefore, the technical problem solved by the present invention is that the existing technology mainly relies on a single data source, lacks multi-dimensional data fusion and comprehensive analysis, resulting in limited monitoring accuracy and fault diagnosis capabilities, making it difficult to adapt to the complex and changeable wind turbine operating environment, and there is a risk of false alarms and missed alarms. In addition, the data transmission and processing efficiency of the existing technology is low, making it difficult to achieve real-time monitoring and rapid response of large-scale wind farms.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a wind farm monitoring and alarm method based on a multi-packaging point rule, comprising:

[0006] Through the data acquisition module of each fan, the status data and monitoring point data of each fan are collected in real time and pre-processed;

[0007] Build a graph of wind turbine status data and monitoring point data to find the shortest path from the monitoring point to the alarm state;

[0008] According to the shortest path results, a bipartite graph is constructed to optimize the matching between monitoring points and alarm status;

[0009] Build a flow network, use network flow algorithms to optimize global data flow distribution, and comprehensively optimize system performance indicators;

[0010] The fan status is monitored in real time through multi-packaging point rules, and alarm event messages are pushed to the database.

[0011] As a preferred solution of the wind farm monitoring and alarm method based on the multi-packaging point rule described in the present invention, the status data of each wind turbine includes wind turbine operation status data, equipment failure data, sensor data, environmental monitoring data and operation and maintenance data;

[0012] The preprocessing includes cleaning the collected data and converting it into a unified format, then using a verification algorithm to verify the integrity and validity of the data, processing the verified abnormal values ​​using a statistical method, and finally compressing the processed data using a data compression algorithm.

[0013] As a preferred solution of the wind farm monitoring and alarm method based on the multi-packaging point rule described in the present invention, wherein: the graph for constructing wind turbine status data and monitoring point data includes nodes and edges;

[0014] The nodes include monitoring point nodes and alarm status nodes;

[0015] The edge represents the connection between the monitoring node and the alarm status node, and the weight of each edge represents the path length;

[0016] Run the Dijkstra algorithm to find the shortest path from each monitoring point to each alarm status point;

[0017] Assume there are n monitoring point nodes, denoted as M1, M2, M3...Mn;

[0018] Assume there are m alarm status nodes, denoted as A1, A2, A3...Am;

[0019] Suppose that from the monitoring point M i To alarm status point A j The path length is represented as w(M i ,A j ), the path length is normalized, and the formula is expressed as:

[0020]

[0021] Considering the interference factors in the path, the path length is expressed as:

[0022]

[0023] The formula of the Dijkstra algorithm is expressed as:

[0024]

[0025] g(x)=β 1 ·D(x)+β 2 ·H(x)+β 3 ·F(x)

[0026] in, Indicates that from the monitoring point M i To alarm status point A j The normalized path length, w(M i ,A j ) indicates that from the monitoring point M i To alarm status point A j The original path length, min(w) represents the minimum value of all path lengths, max(w) represents the maximum value of all path lengths, W(M i ,A j ) indicates that from the monitoring point M i To alarm status point A j The complex path length representation, α k represents the weight coefficient of the kth influencing factor, represents the shortest path distance to node v, is the node with the smallest current distance, Represents a slave node The length of the alternative path to node v, where v represents the current node, M i represents the i-th monitoring point node, A j represents the jth alarm state node, α k represents the weight coefficient, f k (M i ,A j ) represents the kth influencing factor, g(x) represents the comprehensive function of the influencing factors, represents the comprehensive impact from the path starting point a to the path end point b, p represents the total number of influencing factors, argmin represents the parameter with the minimum value, v∈Q represents that the node v belongs to the set Q, Q represents the set of unprocessed nodes, β 1 represents the weight of temperature, β 2 represents the weight of humidity, β 3 represents the weight of wind speed, D(x) represents the temperature at position x, H(x) represents the humidity at position x, F(x) represents the wind speed at position x, (M i ,A j ) represents the monitoring point node M i To alarm status node A j The path, Representation Node The shortest path distance, Represents a slave node The edge weight to node v;

[0027] The value range of is 0 to 1;

[0028] Shortest path distance The value range of is a non-negative real number, and the smaller the value, the shorter the path.

[0029] As a preferred solution of the wind farm monitoring and alarm method based on multi-packaging point rules described in the present invention, wherein: the bipartite graph includes nodes and edges;

[0030] The nodes include monitoring point nodes and alarm status nodes, wherein the left side is the monitoring point node and the right side is the alarm status node;

[0031] The edge represents the connection between the monitoring node and the alarm state node, and the weight of each edge represents the shortest path length from the monitoring point to the alarm state;

[0032] When the path length w(M i ,A j ) is less than or equal to the set threshold T 1 When , the path is considered to be a valid path and is included in the optimization scope;

[0033] When the path length w(M i ,A j ) is greater than the set threshold T 1 , then the path is excluded and does not participate in further matching optimization;

[0034] By calculating the percentile of the historical path length data, an initial effective path setting threshold is determined, and the effective path setting threshold is dynamically adjusted in combination with the current system load and performance. The formula for the effective path setting threshold is expressed as:

[0035]

[0036] Among them, T 1 represents the dynamically adjusted effective path setting threshold, P 95 (w(M i ,A j )) represents the 95th percentile of the path length data, L represents the current system load, and L max represents the maximum load of the system, S represents the current system performance index, S max Represents the maximum performance index of the system, T e Indicates the current ambient temperature, T opt represents the optimal ambient temperature, β represents the business demand coefficient, |T e -T opt | represents the difference between the current ambient temperature and the optimal temperature, M i represents the i-th monitoring point node, A j Represents the jth alarm status node.

[0037] As a preferred solution of the wind farm monitoring and alarm method based on the multi-packaging point rule described in the present invention, the nodes include source node S, sink node T, monitoring point node M i and alarm status node A j ;

[0038] The edge capacity includes: assuming that from the source node S to the monitoring point node M i The edge capacity is c(S,M i ), from the monitoring point node M i To the alarm status node A j The edge capacity is c(M i ,A j ), from the alarm state node A j The edge capacity to the sink node T is c(A j ,T);

[0039] Construct a flow network, the formula is expressed as:

[0040]

[0041] Among them, Source:S represents the source node S, Sink:T represents the sink node T, Nodes:{M i ,A j} represents a node set; Edges:{(S,M i ),(M i ,A j ),(A j ,T)} represents the edge set;

[0042] The capacity of an edge is expressed as:

[0043] c(S,M i )=Flow capacity 1

[0044] c(M i ,A j ) = flow capacity 2

[0045] c(A j ,T)=Flow capacity3

[0046] The Edmonds-Karp algorithm is expressed as:

[0047] f(u,v)=0foralledges(u,v)

[0048] BFS(G f ,S,T)→P

[0049] c f (u,v)=c(u,v)-f(u,v)

[0050] f(u,v)=f(u,v)+Δfforeachedge(u,v)∈P

[0051] MaxFlow(S,T)=∑ v∈V f(S,v)

[0052] Among them, S represents the source node, T represents the sink node, and M i represents the i-th monitoring point node, A j represents the jth alarm state node, c(S,M i ) represents the edge capacity from the source node to the monitoring point node, c(M i ,A j ) represents the edge capacity from the monitoring point node to the alarm state node, c(A j ,T) represents the edge capacity from the alarm state node to the sink node, f(u,v) represents the flow of edge (u,v), G f represents the residual network, P represents the augmented path, u represents the starting point of the edge, c f (u,v) represents the remaining capacity of edge (u,v), Δf represents the minimum remaining capacity on the path, V represents the set of all nodes, MaxFlow(S,T) represents the maximum flow from source node S to sink node T, f(S,v) represents the flow from source node S to node v, and v represents a node in the graph.

[0053] As a preferred solution of the wind farm monitoring and alarm method based on the multi-packaging point rule described in the present invention, wherein: the real-time monitoring of the wind turbine status by the multi-packaging point rule and the pushing of the alarm event message to the database include: when the wind turbine operation status data change frequency f d Above the set threshold F 1 When the data acquisition module runs in high-frequency mode, the frequency of data acquisition is f d , otherwise, the acquisition frequency is the normal frequency f r ;

[0054] The formula for data collection frequency is:

[0055]

[0056] When the collected data quality index Q d Below the set threshold Q 1 When the data is invalid, it will be marked as invalid and collected again. The data quality index Q d Comprehensive evaluation through signal-to-noise ratio and data integrity;

[0057] The formula for data validity is expressed as:

[0058]

[0059] Among them, f c Indicates the frequency of final data collection, f d Indicates the frequency of changes in the fan operating status data, F 1 Indicates the set threshold value of data change frequency. When the data change frequency f d Above this value, the data acquisition module will operate in high frequency mode; f r represents the frequency of regular data collection, Q v The validity mark indicating the data quality, Q d Indicates the quality index of the collected data, Q 1 Indicates the set threshold of data quality. Valid indicates that the data quality is valid, and Invalid indicates that the data quality is invalid.

[0060] As a preferred solution of the wind farm monitoring and alarm method based on the multi-packaging point rule of the present invention, wherein: the real-time monitoring of the wind turbine status through the multi-packaging point rule and pushing the alarm event message to the database also includes real-time monitoring of the status data of each wind turbine, when the wind turbine operating status parameter P in the status data of each wind turbine is k Exceeds the set alarm threshold T k When , the alarm is triggered;

[0061] According to the fan operating state parameter P k The degree of exceeding the threshold value divides the alarm into different levels. Alarm level L a The formula is expressed as:

[0062]

[0063] Among them, Alarm indicates the trigger mark of the alarm state, Triggered indicates the state where the alarm is triggered, NotTriggered indicates the state where the alarm is not triggered, and P k Indicates the monitored fan operation status alarm parameter value, T k Indicates the set alarm threshold, L a Indicates the alarm level. Low indicates a low-level alarm, Medium indicates a medium-level alarm, and High indicates a high-level alarm. 1 and Δ 2 They represent the set threshold increments respectively, which are used to distinguish different alarm levels;

[0064] When an alarm is triggered and the level is medium or high, the alarm message is immediately pushed to the database;

[0065] When an alarm is triggered and the level is a low-level alarm, a log is recorded and summarized and pushed.

[0066] As a wind farm monitoring and alarm system based on multiple packaging point rules described in the present invention, wherein:

[0067] The data acquisition and processing module collects the status data and monitoring point data of each fan in real time through the data acquisition module of each fan, and performs pre-processing;

[0068] The path processing module builds a graph of the fan status data and monitoring point data to find the shortest path from the monitoring point to the alarm state;

[0069] The path optimization module constructs a bipartite graph based on the shortest path results to optimize the matching between monitoring points and alarm status;

[0070] System optimization module, builds flow network, uses network flow algorithm to optimize global data flow distribution, and comprehensively optimizes system performance indicators;

[0071] The monitoring and alarm module monitors the status of the fan in real time through multi-packaging point rules and pushes alarm event messages to the database.

[0072] A computer device comprises: a memory and a processor; the memory stores a computer program, wherein the processor implements the steps of any one of the methods of the present invention when executing the computer program.

[0073] A computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of any one of the methods of the present invention.

[0074] Beneficial effects of the present invention: The wind farm monitoring and alarm method based on multi-packaging point rules provided by the present invention monitors the status of wind turbines in real time through multi-packaging point rules, integrates multi-dimensional data, and improves monitoring accuracy and fault diagnosis capabilities. Through the shortest path calculation and bipartite graph matching optimization, the risk of false alarms and missed alarms is effectively reduced. The network flow algorithm is used to optimize the global data flow distribution, improve the response speed, and ensure efficient and real-time data transmission. The system performance optimization module maximizes data transmission efficiency, adapts to complex operating environments, realizes high-precision monitoring and rapid fault alarms, and solves the problems of low monitoring accuracy, limited fault diagnosis capabilities, slow response speed, etc. in the prior art, thereby improving the safety and efficiency of wind farm operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0076] Figure 1 An overall flow chart of a wind farm monitoring and alarm method based on a multi-packaging point rule provided in the first embodiment of the present invention. DETAILED DESCRIPTION

[0077] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0078] Example 1, reference Figure 1 , is an embodiment of the present invention, and provides a wind farm monitoring and alarm method based on a multi-packaging point rule, comprising:

[0079] S1: Through the data acquisition module of each fan, the status data and monitoring point data of each fan are collected in real time and pre-processed.

[0080] The status data of each wind turbine includes wind turbine operation status data, equipment failure data, sensor data, environmental monitoring data and operation and maintenance data.

[0081] Preprocessing includes cleaning the collected data and converting it into a unified format, then using a verification algorithm to verify the integrity and validity of the data, processing the detected outliers using statistical methods, and finally compressing the processed data using a data compression algorithm.

[0082] The status data of each wind turbine includes operating status data, equipment failure data, sensor data, environmental monitoring data and operation and maintenance data. The comprehensive collection of this data can provide an all-round view of the wind turbine operating status and ensure the accuracy of monitoring and analysis.

[0083] The preprocessing step includes cleaning and format conversion of the collected data, and using verification algorithms to verify the integrity and validity of the data. Through these operations, noise and erroneous data can be effectively removed to ensure the quality and consistency of the data. The outliers detected are processed by statistical methods to avoid the impact of abnormal data on subsequent analysis. This method can not only fix data errors, but also maintain the continuity and availability of data, and improve the reliability of data analysis.

[0084] Furthermore, using data compression algorithms to compress processed data can significantly reduce the space requirements for data storage, reduce storage costs, and improve the efficiency of data transmission, allowing the system to transmit and process data more quickly. Real-time data collection and preprocessing can ensure that the latest wind turbine status information is obtained and processed in the shortest time, which helps to promptly discover and deal with potential problems and improve the response speed and emergency handling capabilities of the monitoring system.

[0085] Furthermore, through data cleaning and verification, system misjudgment and fault alarms caused by data quality problems can be avoided, ensuring the stable operation of the system. High-quality preprocessed data provides a reliable foundation for subsequent fault diagnosis, performance optimization and decision support, improving the intelligence level of the entire system and the accuracy of decision-making.

[0086] S2: Build a graph of the fan status data and monitoring point data, and find the shortest path from the monitoring point to the alarm state.

[0087] Suppose there are n monitoring point nodes, denoted as M1, M2, M3...Mn.

[0088] Assume there are m alarm status nodes, denoted as A1, A2, A3...Am.

[0089] Suppose that from the monitoring point M i To alarm status point A j The path length is represented as w(M i ,A j ), the path length is normalized, and the formula is expressed as:

[0090]

[0091] Considering the interference factors in the path, the path length is expressed as:

[0092]

[0093] The formula of Dijkstra algorithm is expressed as:

[0094]

[0095] g(x)=β 1 ·D(x)+β 2 ·H(x)+β 3 ·F(x)

[0096] in, Indicates that from the monitoring point M i To alarm status point A j The normalized path length, w(M i ,A j ) indicates that from the monitoring point Mi To alarm status point A j The original path length, min(w) represents the minimum value of all path lengths, max(w) represents the maximum value of all path lengths, W(M i ,A j ) indicates that from the monitoring point M i To alarm status point A j The complex path length representation, α k represents the weight coefficient of the kth influencing factor, represents the shortest path distance to node v, is the node with the smallest current distance, Represents a slave node The length of the alternative path to node v, where v represents the current node, M i represents the i-th monitoring point node, A j represents the jth alarm state node, α k represents the weight coefficient, f k (M i ,A j ) represents the kth influencing factor, g(x) represents the comprehensive function of the influencing factors, represents the comprehensive impact from the path starting point a to the path end point b, p represents the total number of influencing factors, argmin represents the parameter with the minimum value, v∈Q represents that the node v belongs to the set Q, Q represents the set of unprocessed nodes, β 1 represents the weight of temperature, β 2 represents the weight of humidity, β 3 represents the weight of wind speed, D(x) represents the temperature at position x, H(x) represents the humidity at position x, F(x) represents the wind speed at position x, (M i ,A j ) represents the monitoring point node M i To alarm status node A j The path, Representation Node The shortest path distance, Represents a slave node The edge weight to node v.

[0097] The value range is 0 to 1.

[0098] Shortest path distance The value range of is a non-negative real number, and the smaller the value, the shorter the path.

[0099] Furthermore, by constructing detailed wind turbine status data and monitoring point data graphs, the operating status and potential problems of the wind turbine can be fully reflected. The detailed definition of each monitoring point node and alarm status node makes fault detection more accurate. The Dijkstra algorithm is used to calculate the shortest path to ensure that the path from the monitoring point to the alarm status node is optimal, which not only improves the accuracy of detection, but also reduces the risk of false alarms and missed alarms.

[0100] Further building a data graph and calculating the shortest path can effectively reduce redundant paths and unnecessary monitoring points, thereby improving the efficiency of the entire monitoring system. The normalization of path length and comprehensive consideration of multiple factors enable the system to respond quickly to changes and update the shortest path in real time.

[0101] S3: Based on the shortest path results, a bipartite graph is constructed to optimize the matching between monitoring points and alarm status.

[0102] According to the shortest path result calculated by Dijkstra algorithm, a bipartite graph is constructed, in which the left side is the monitoring point node and the right side is the alarm state node. The weight of each edge represents the shortest path length from the monitoring point to the alarm state.

[0103] When the path length w(M i ,A j ) is less than or equal to the set threshold T 1 , the path is considered to be a valid path and is included in the optimization scope.

[0104] When the path length w(M i ,A j ) is greater than the set threshold T 1 , the path is excluded and does not participate in further matching optimization.

[0105] By calculating the percentile of historical path length data, an initial effective path setting threshold is determined. Combined with the current system load and performance, the effective path setting threshold is dynamically adjusted. The formula for the effective path setting threshold is expressed as:

[0106]

[0107] Among them, T 1 represents the dynamically adjusted effective path setting threshold, P 95 (w(M i ,A j )) represents the 95th percentile of the path length data, L represents the current system load, and L max represents the maximum load of the system, S represents the current system performance index, S max Represents the maximum performance index of the system, T e Indicates the current ambient temperature, Topt represents the optimal ambient temperature, β represents the business demand coefficient, |T e -T opt | represents the difference between the current ambient temperature and the optimal temperature, M i represents the i-th monitoring point node, A j Represents the jth alarm status node.

[0108] Furthermore, by dynamically adjusting the threshold of the effective path, the coverage of the effective path can be maximized, and the monitoring and alarm capabilities of the system can be improved. Dynamic adjustment of the threshold can also optimize the resource allocation of the system, avoid excessive load and resource waste, and improve the overall performance of the system.

[0109] Furthermore, by scientifically setting and dynamically adjusting the effective path threshold, the probability of false alarms and missed alarms can be significantly reduced, and the accuracy of fault detection and alarm can be improved. The adjustment of the effective path setting threshold can ensure that only reliable and important paths are included in the matching optimization range, avoiding unnecessary interference and false alarms.

[0110] S4: Build a flow network, use network flow algorithms to optimize global data traffic distribution, and comprehensively optimize system performance indicators.

[0111] When constructing a flow network, the nodes include source nodes S, sink nodes T, and monitoring point nodes M. i and alarm status node A j .

[0112] Assume that from the source node S to the monitoring point node M i The edge capacity is c(S,M i ), from the monitoring point node M i To alarm status node A j The edge capacity is c(M i ,A j ), from the alarm status node A j The edge capacity to the sink node T is c(A j ,T).

[0113] Construct a flow network, the formula is expressed as:

[0114]

[0115] Among them, Source:S represents the source node S, Sink:T represents the sink node T, Nodes:{M i ,A j} represents a node set; Edges:{(S,M i ),(M i ,A j ),(A j ,T)} represents the edge set.

[0116] The capacity of an edge is expressed as:

[0117] c(S,M i )=Flow capacity 1

[0118] c(M i ,A j ) = flow capacity 2

[0119] c(A j ,T)=Flow capacity3

[0120] The Edmonds-Karp algorithm is expressed as:

[0121] f(u,v)=0foralledges(u,v)

[0122] BFS(G f ,S,T)→P

[0123] c f (u,v)=c(u,v)-f(u,v)

[0124] f(u,v)=f(u,v)+Δfforeachedge(u,v)∈P

[0125] MaxFlow(S,T)=∑ v∈V f(S,v)

[0126] Among them, S represents the source node, T represents the sink node, and M i represents the i-th monitoring point node, A j represents the jth alarm state node, c(S,M i ) represents the edge capacity from the source node to the monitoring point node, c(M i ,A j ) represents the edge capacity from the monitoring point node to the alarm state node, c(A j ,T) represents the edge capacity from the alarm state node to the sink node, f(u,v) represents the flow of edge (u,v), G f represents the residual network, P represents the augmented path, u represents the starting point of the edge, c f (u,v) represents the remaining capacity of edge (u,v), Δf represents the minimum remaining capacity on the path, V represents the set of all nodes, MaxFlow(S,T) represents the maximum flow from source node S to sink node T, f(S,v) represents the flow from source node S to node v, and v represents a node in the graph.

[0127] By building a flow network, the data flow relationship between nodes is clarified, which helps to optimize the global data flow distribution and ensure that key data can be transmitted efficiently. Using the Edmonds-Karp algorithm to calculate the maximum flow can maximize the network's flow capacity, improve data transmission efficiency, and reduce data transmission bottlenecks.

[0128] After building the flow network, the system performance indicators are comprehensively optimized through the network flow algorithm, which can effectively improve the overall performance of the system and ensure that the system can still operate stably under high load conditions. By optimizing the data flow distribution, the data transmission between the monitoring point and the alarm status node is more efficient, which helps to improve the accuracy and response speed of fault diagnosis. It ensures that the monitoring point data can be quickly transmitted to the alarm status node and the sink node, which improves the real-time and reliability of the fault alarm.

[0129] Furthermore, using network flow algorithms to optimize data flow distribution can reasonably allocate system resources, avoid unnecessary data transmission and resource waste, and improve the utilization efficiency of system resources. This ensures that the system meets monitoring and alarm requirements while reducing resource consumption and lowering system operating costs.

[0130] S5: Monitor the status of the fan in real time through multi-packaging point rules, and push alarm event messages to the database.

[0131] When the fan operating status data changes frequency f d Above the set threshold F 1 When the data acquisition module runs in high-frequency mode, the frequency of data acquisition is f d , otherwise, the acquisition frequency is the normal frequency f r .

[0132] The formula for data collection frequency is:

[0133]

[0134] When the collected data quality index Q d Below the set threshold Q 1 When the data is invalid, it will be marked as invalid and collected again. The data quality index Q d Comprehensive evaluation was performed using signal-to-noise ratio and data integrity.

[0135] The formula for data validity is expressed as:

[0136]

[0137] Among them, f c Indicates the frequency of final data collection, f d Indicates the frequency of changes in the fan operating status data, F 1Indicates the set threshold value of data change frequency. When the data change frequency f d Above this value, the data acquisition module will operate in high frequency mode; f r represents the frequency of regular data collection, Q v The validity mark indicating the data quality, Q d Indicates the quality index of the collected data, Q 1 Indicates the set threshold of data quality. Valid indicates that the data quality is valid, and Invalid indicates that the data quality is invalid.

[0138] When a key parameter P in the fan status data k Exceeds the set alarm threshold T k , triggers the alarm.

[0139]

[0140] According to the alarm parameter P k The degree of exceeding the threshold value divides the alarm into different levels. Alarm level L a The formula is expressed as:

[0141]

[0142] Among them, Alarm indicates the trigger mark of the alarm state, Triggered indicates the state where the alarm is triggered, NotTriggered indicates the state where the alarm is not triggered, and P k Indicates the monitored fan operation status alarm parameter value, T k Indicates the set alarm threshold, L a Indicates the alarm level. Low indicates a low-level alarm, Medium indicates a medium-level alarm, and High indicates a high-level alarm. 1 and Δ 2 They respectively represent the set threshold increments, which are used to distinguish different alarm levels.

[0143] When an alarm is triggered and the level is medium or high, the alarm message is immediately pushed to the database. For low-level alarms, logs are recorded and pushed in summary.

[0144] The data collection frequency is dynamically adjusted according to the frequency of changes in the fan status to ensure high-frequency data collection at critical moments, improving the accuracy and flexibility of real-time monitoring. When the frequency of data changes is low, regular frequency data collection is used to reduce resource consumption and improve system operation efficiency.

[0145] By comprehensively evaluating data quality, we can ensure that the collected data is complete and valid, and improve the accuracy of subsequent analysis and fault diagnosis. We can re-collect low-quality data to avoid monitoring errors caused by data quality issues, and improve data reliability and system stability.

[0146] Furthermore, by setting alarm thresholds and parameters, it is possible to accurately determine when to trigger an alarm, avoid false alarms and missed alarms, and improve the reliability of the system. Alarms are divided into different levels, and different response measures are taken according to the level of the alarm, which improves the emergency handling capability and efficiency of the system.

[0147] Furthermore, high-level alarms are immediately pushed to the database to ensure that relevant personnel can obtain alarm information in a timely manner, respond quickly and handle faults. Low-level alarms are logged and pushed in a summary to reduce the pressure on the system caused by frequent pushes, while ensuring that all alarm information is properly managed.

[0148] On the other hand, this embodiment also provides a wind farm monitoring and alarm system based on a multi-packaging point rule, which includes:

[0149] The data acquisition and processing module collects the status data and monitoring point data of each fan in real time through the data acquisition modules of each fan, and performs pre-processing.

[0150] The path processing module constructs a graph of the fan status data and monitoring point data and finds the shortest path from the monitoring point to the alarm state.

[0151] The path optimization module constructs a bipartite graph based on the shortest path results and optimizes the matching between monitoring points and alarm status.

[0152] The system optimization module builds a flow network, uses network flow algorithms to optimize global data flow distribution, and comprehensively optimizes system performance indicators.

[0153] The monitoring and alarm module monitors the status of the fan in real time through multi-packaging point rules and pushes alarm event messages to the database.

[0154] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0155] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0156] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0157] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0158] In Example 2, an experiment was conducted in a wind farm containing 20 wind turbines. Each wind turbine was equipped with an advanced sensor module for collecting status data and monitoring point data of each wind turbine in real time. Data collection included wind turbine operating status data, equipment failure data, sensor data, environmental monitoring data, and operation and maintenance data.

[0159] The data acquisition module collects the status data and monitoring point data of each fan in real time. In order to ensure the accuracy and effectiveness of the data, the data is preprocessed immediately after collection. The preprocessing steps include:

[0160] Construct data graph and shortest path calculation: Construct the wind turbine status data and monitoring point data into a graph. The nodes include monitoring point nodes and alarm status nodes. The weight of each edge represents the path length. Run the Dijkstra algorithm to find the shortest path from each monitoring point to each alarm status point. Suppose there are n monitoring point nodes, represented as M1, M2, .... Mn; suppose there are m alarm status nodes, represented as A1, A2, .... Am; suppose the path length from the monitoring point to the alarm status point is represented as w (Mi, Aj), and the path length is normalized.

[0161] According to the shortest path result, a bipartite graph is constructed, with monitoring point nodes on the left and alarm status nodes on the right. The valid path is determined according to the path length, and the matching optimization between monitoring points and alarm status is performed. When the path length is less than or equal to the set threshold, the path is considered to be a valid path and included in the optimization range; otherwise, the path is excluded.

[0162] Construct a flow network, where the nodes include source node S, sink node T, monitoring point node and alarm state node. Let the edge capacity from source node to monitoring point node be c(S, Mi), the edge capacity from monitoring point node to alarm state node be c(Mi, Aj), and the edge capacity from alarm state node to sink node be c(Aj, T). Use the Edmonds-Karp algorithm to calculate the maximum flow and optimize the global data flow distribution to ensure that the system performance indicators are optimal.

[0163] The fan status is monitored in real time through the multi-packaging point rule. When the frequency of the fan operation status data change is higher than the set threshold, the data acquisition module will operate in high-frequency mode, and the frequency of data collection is f d ; Otherwise, the collection frequency is the regular frequency. When the collected data quality index is lower than the set threshold, the data will be marked as invalid and re-collected. The system determines whether to trigger an alarm based on the comparison between the monitored key parameter value and the alarm threshold, and pushes the alarm message to the database according to the alarm level. The experimental results are shown in Table 1.

[0164] Table 1 Experimental data table

[0165]

[0166]

[0167] It can be seen from the table data that the present invention has shown significant innovation and advantages in many aspects. First, through the data acquisition module, the operating status data of the fan can be collected in real time, and the data acquisition frequency can be dynamically adjusted according to the frequency of data changes, thereby improving the flexibility and real-time performance of data acquisition. For example, the operating status data change frequency of fan 1 is 2.5Hz, and the system automatically adjusts the data acquisition frequency to 2.5Hz, while the data change frequency of fan 3 is lower, only 1.8Hz, and the data acquisition frequency is maintained at 1.0Hz. This dynamic adjustment mechanism avoids waste of resources and improves system efficiency.

[0168] Secondly, the data quality assessment and preprocessing steps ensure the reliability and integrity of the data. The data quality indicator (SNR) in the table shows that the data quality of all wind turbines is higher than the set threshold (such as 25), so the data is marked as valid. This step avoids monitoring errors caused by data quality issues and improves the accuracy of fault diagnosis.

[0169] In terms of path optimization and alarm triggering, the shortest path from the monitoring point to the alarm state was found by constructing a data graph and running the Dijkstra algorithm, and bipartite graph matching optimization was performed based on the shortest path result. The values ​​of the normalized path length in the table show that the path lengths of fans 1 to 6 are all within a reasonable range (0.30-0.55), which ensures the accuracy and timeliness of alarm triggering. For example, the normalized path length of fan 2 is 0.50, and the key parameter value is 1.3, which exceeds the set alarm threshold and triggers a high-level alarm.

[0170] Finally, the flow network construction and optimization achieved the optimization of global data flow distribution through the Edmonds-Karp algorithm, and improved the comprehensive optimization of system performance indicators. In terms of alarm message push, the system pushes high-level alarm messages to the database (such as fan 2 and fan 5) in real time according to the alarm level, while low-level alarms (such as fan 3 and fan 6) are recorded in the log and pushed in a certain period of time. This hierarchical response mechanism not only improves the response speed of the system, but also ensures the efficient management of alarm information.

[0171] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A wind farm monitoring and alarm method based on multiple packaging point rules, characterized in that: include: Through the data acquisition module of each fan, the status data and monitoring point data of each fan are collected in real time and pre-processed; Build a graph of wind turbine status data and monitoring point data to find the shortest path from the monitoring point to the alarm state; According to the shortest path results, a bipartite graph is constructed to optimize the matching between monitoring points and alarm status; Build a flow network, use network flow algorithms to optimize global data flow distribution, and comprehensively optimize system performance indicators; Monitor the status of the fan in real time through multi-packaging point rules, and push alarm event messages to the database; The graph for constructing wind turbine status data and monitoring point data includes nodes and edges; The nodes include monitoring point nodes and alarm status nodes; The edge represents the connection between the monitoring point node and the alarm status node, and the weight of each edge represents the path length; Run the Dijkstra algorithm to find the shortest path from each monitoring point to each alarm status point; Assume there are n monitoring point nodes, denoted as M1, M2, M3...Mn; Assume there are m alarm status nodes, denoted as A1, A2, A3...Am; Suppose that from the monitoring point M i To alarm status point A j The path length is represented as w(M i ,A j ), the path length is normalized, and the formula is expressed as: Considering the interference factors in the path, the path length is expressed as: The formula of the Dijkstra algorithm is expressed as: g(x)=β1·D(x)+β2H(x)+β3·F(x) in, Indicates that from the monitoring point M i To alarm status point A j The normalized path length, w(M i ,A j ) indicates that from the monitoring point M i To alarm status point A j The original path length, min(w) represents the minimum value of all path lengths, max(w) represents the maximum value of all path lengths, W(M i ,A j ) indicates that from the monitoring point M i To alarm status point A j The complex path length representation, α k represents the weight coefficient of the kth influencing factor, represents the shortest path distance to node v, is the node with the smallest current distance, Represents a slave node The length of the alternative path to node v, where v represents the current node, M i represents the i-th monitoring point node, A j represents the jth alarm state node, α k represents the weight coefficient, f k (M i ,A j ) represents the kth influencing factor, g(x) represents the comprehensive function of the influencing factors, represents the comprehensive impact from the path starting point a to the path end point b, p represents the total number of influencing factors, argmin represents the parameter taking the minimum value, v∈Q represents that the node v belongs to the set Q, Q represents the set of unprocessed nodes, β1 represents the weight of temperature, β2 represents the weight of humidity, β3 represents the weight of wind speed, D(x) represents the temperature at position x, H(x) represents the humidity at position x, F(x) represents the wind speed at position x, (M i ,A j ) represents the monitoring point node M i To alarm status node A j The path, Representation Node The shortest path distance, Represents a slave node The edge weight to node v; The value range of is 0 to 1; Shortest path distance The range of is a non-negative real number, the smaller the value, the shorter the path; The bipartite graph includes nodes and edges; The nodes include monitoring point nodes and alarm status nodes, wherein the left side is the monitoring point node and the right side is the alarm status node; The edge represents the connection between the monitoring point node and the alarm state node, and the weight of each edge represents the shortest path length from the monitoring point to the alarm state; When the path length w(M i ,A j ) is less than or equal to the set threshold T1, the path is considered to be a valid path and is included in the optimization range; When the path length w(M i ,A j ) is greater than the set threshold T1, the path is excluded and does not participate in further matching optimization; By calculating the percentile of the historical path length data, an initial effective path setting threshold is determined, and the effective path setting threshold is dynamically adjusted in combination with the current system load and performance. The formula for the effective path setting threshold is expressed as: Where T1 represents the threshold value of the dynamically adjusted effective path setting, P 95 (w(M i ,A j )) represents the 95th percentile of the path length data, L represents the current system load, and L max represents the maximum load of the system, S represents the current system performance index, S max Represents the maximum performance index of the system, T e Indicates the current ambient temperature, T opt represents the optimal ambient temperature, β represents the business demand coefficient, |T e -T opt | represents the difference between the current ambient temperature and the optimal temperature, M i represents the i-th monitoring point node, A j represents the jth alarm status node; The constructed flow network includes, node and edge capacities; The nodes include source node S, sink node T, monitoring node M i and alarm status node A j ; The edge capacity includes: assuming that from the source node S to the monitoring point node M i The edge capacity is c(S,M i ), from the monitoring point node M i To the alarm status node A j The edge capacity is c(M i ,A j ), from the alarm state node A j The edge capacity to the sink node T is c9A j ,T); Construct a flow network, the formula is expressed as: Among them, Source:S represents the source node S, Sink:T represents the sink node T, Nodes:{M i ,A j } represents a node set; Edges:{(S,M i ),(M i ,A j ),(A j ,T)} represents the edge set; The capacity of an edge is expressed as: c(S,M i ) = flow capacity 1 c(M i ,A j ) = flow capacity 2 c(A j ,T)=Flow capacity3 The Edmonds-Karp algorithm is expressed as: f(u,v)=0 for all edges(u,v) BFS(G f ,S,T)→P c f (u,v)=c(u,v)-f(u,v) f(u,v)=f(u,v)+Δf for each edge(u,v)∈P MaxFlow(S,T)=∑ v∈V f(S,v) Among them, S represents the source node, T represents the sink node, and M i represents the i-th monitoring point node, A j represents the jth alarm state node, c(S,M i ) represents the edge capacity from the source node to the monitoring point node, c(M i ,A j ) represents the edge capacity from the monitoring point node to the alarm state node, c(A j ,T) represents the edge capacity from the alarm state node to the sink node, f(u,v) represents the flow of edge (u,v), G f represents the residual network, P represents the augmented path, u represents the starting point of the edge, c f (u,v) represents the remaining capacity of edge (u,v), Δf represents the minimum remaining capacity on the path, V represents the set of all nodes, MaxFlow(S,T) represents the maximum flow from source node S to sink node T, f(S,v) represents the flow from source node S to node v, and v represents a node in the graph.

2. The wind farm monitoring and alarm method based on multiple packaging point rules according to claim 1 is characterized in that: The status data of each wind turbine includes wind turbine operation status data, equipment failure data, sensor data, environmental monitoring data and operation and maintenance data; The preprocessing includes cleaning the collected data and converting it into a unified format, then using a verification algorithm to verify the integrity and validity of the data, processing the verified abnormal values ​​using a statistical method, and finally compressing the processed data using a data compression algorithm.

3. The wind farm monitoring and alarm method based on multiple packaging point rules as claimed in claim 2 is characterized in that: The real-time monitoring of the fan status through the multi-packaging point rule and the pushing of the alarm event message to the database include: when the frequency of the fan operation status data change f d When the frequency is higher than the set threshold value F1, the data acquisition module will operate in high-frequency mode, and the frequency of data acquisition is f d , otherwise, the acquisition frequency is the normal frequency f r ; The formula for data collection frequency is: When the collected data quality index Q d When the data is lower than the set threshold Q1, it will be marked as invalid and collected again. The data quality indicator Q d Comprehensive evaluation through signal-to-noise ratio and data integrity; The formula for data validity is expressed as: Among them, f c Indicates the frequency of final data collection, f d Indicates the frequency of the fan operating status data change. F1 indicates the set threshold of the data change frequency. When the data change frequency f d Above this value, the data acquisition module will operate in high frequency mode; f r represents the frequency of regular data collection, Q v The validity mark indicating the data quality, Q d It indicates the quality index of the collected data, Q1 indicates the set threshold of the data quality, Valid indicates the mark of valid data quality, and Invalid indicates the mark of invalid data quality.

4. The wind farm monitoring and alarm method based on multiple packaging point rules as claimed in claim 3 is characterized in that: The real-time monitoring of the status of the fans by the multi-packaging point rule and pushing the alarm event message to the database also includes real-time monitoring of the status data of each fan, when the fan operation status parameter P in the status data of each fan is k Exceeds the set alarm threshold T k When , the alarm is triggered; According to the fan operating state parameter P k The degree of exceeding the threshold value divides the alarm into different levels. Alarm level L a The formula is expressed as: Among them, Alarm indicates the trigger mark of the alarm state, Triggered indicates the state where the alarm is triggered, NotTriggered indicates the state where the alarm is not triggered, and P k Indicates the monitored fan operation status alarm parameter value, T k Indicates the set alarm threshold, L a Indicates the alarm level. Low indicates a low-level alarm, Medium indicates a medium-level alarm, and High indicates a high-level alarm. Δ1 and Δ2 indicate the set threshold increments, which are used to distinguish different alarm levels. When an alarm is triggered and the level is medium or high, the alarm message is immediately pushed to the database; When an alarm is triggered and the level is a low-level alarm, a log is recorded and summarized and pushed.

5. A wind farm monitoring and alarm system based on multiple packaging point rules using the method according to any one of claims 1 to 4, characterized in that: The data acquisition and processing module collects the status data and monitoring point data of each fan in real time through the data acquisition module of each fan, and performs pre-processing; The path processing module builds a graph of the fan status data and monitoring point data to find the shortest path from the monitoring point to the alarm state; The path optimization module constructs a bipartite graph based on the shortest path results to optimize the matching between monitoring points and alarm status; System optimization module, builds flow network, uses network flow algorithm to optimize global data flow distribution, and comprehensively optimizes system performance indicators; The monitoring and alarm module monitors the status of the fan in real time through multi-packaging point rules and pushes alarm event messages to the database.

Images