Wind power plant current collection line fault positioning method and system

Abstract

The invention relates to the technical field of wind power plant operation and maintenance, and discloses a wind power plant current collection line fault positioning method and system, and the method comprises the steps: setting a plurality of fault types, and constructing a fault analysis model according to all fault types; generating a real-time analysis time interval of the wind power plant current collection line, and obtaining real-time monitoring data of the wind power plant current collection line according to the real-time analysis time interval; generating an initial analysis result of the real-time monitoring data according to the fault analysis model, setting a primary positioning strategy, and collecting real-time feedback data according to the primary positioning strategy; the fault risk coefficient is generated according to the real-time feedback data, the fault positioning result is generated according to the fault risk coefficient, the accuracy and real-time performance of fault positioning are improved, and technical support is provided for safe and stable operation of the wind power plant.

Description

Technical Field

[0001] This application relates to the field of wind farm operation and maintenance technology, and in particular to a method and system for locating faults in wind farm collector lines. Background Technology

[0002] With the global trend of energy structure transitioning towards clean energy, wind farms, as a core scenario for new energy power generation, are experiencing continuous expansion in installed capacity and grid connection. As the "main artery" of power transmission in wind farms, data collection lines connect the dispersed wind turbines and substations, undertaking the crucial function of power aggregation and transmission.

[0003] In existing technologies, fault location technologies for wind farm collection lines mostly rely on a single fault model or fixed threshold judgment, which is difficult to cover the diverse fault types in the complex environment of wind farms. At the same time, data acquisition often adopts a fixed time interval mode, which can easily affect the location accuracy due to insufficient data timeliness in the fault initiation stage or during dynamic changes. In addition, the lack of a data feedback mechanism reduces the final fault location accuracy. Summary of the Invention

[0004] To address the aforementioned technical issues, this application provides a method and system for locating faults in wind farm collector lines. By constructing a fault analysis model and setting a real-time analysis time interval, initial analysis results are generated and a primary location strategy is formulated. Real-time feedback data is obtained according to the primary location strategy, and the fault risk coefficient is calculated. The fault location results are then output, improving the accuracy and real-time performance of fault location and providing technical support for the safe and stable operation of wind farms.

[0005] In some embodiments of this application, a method for locating faults in wind farm collector lines is provided, including: Define multiple fault categories and construct a fault analysis model based on all fault categories; Generate real-time analysis time intervals for wind farm collector lines, and obtain real-time monitoring data for wind farm collector lines based on these time intervals; The initial analysis results of real-time monitoring data are generated based on the fault analysis model, and a primary positioning strategy is set. Real-time feedback data is collected according to the primary positioning strategy. A fault risk coefficient is generated based on real-time feedback data, and a fault location result is generated based on the fault risk coefficient.

[0006] In some embodiments of this application, a fault analysis model is constructed based on all fault categories, including: Generate historical fault data packets for each fault category based on historical fault logs for each fault category. Generate a fault evaluation value corresponding to the fault category based on historical fault data packets; Based on the fault evaluation value, the number of associated regions for the wind farm's collection lines is set for the corresponding fault category, and multiple associated regions for the corresponding fault category are generated by combining the fault region mapping sequence for the corresponding fault category. The fault region mapping sequence includes several historical fault regions, and each historical fault region corresponds to a historical fault coefficient. The historical fault regions are arranged according to the historical fault coefficient. Historical fault features of each associated region are obtained and constructed into a training dataset. The neural network is trained based on the training dataset of all associated regions of the same fault category to obtain the initial fault diagnosis model for the corresponding fault category. The number of analysis nodes for the corresponding associated regions is set according to the historical failure coefficients, and analysis nodes for each associated region are generated and the analysis time interval for the analysis nodes is set. A fault diagnosis model is established based on the initial fault diagnosis model for all fault categories, the number of analysis nodes, and the analysis time interval.

[0007] In some embodiments of this application, constructing a fault analysis model based on all fault categories further includes: Generate historical fault sub-data packets for each analysis node in each associated region; Based on the historical fault sub-data packets, determine whether each analysis node has a correlation with other analysis nodes. If so, set it as a related node for the corresponding analysis node and determine the correlation characteristics. The association influence value between each analysis node and its corresponding associated node is calculated based on the association characteristics. Each analysis node is assigned an association mapping table based on all associated influence values. The association mapping table includes a first sub-association mapping table, a second sub-association mapping table, and a third sub-association mapping table. Each associated node in the sub-association mapping table has corresponding location information and association features. Construct a fault location sub-model for each analysis node; Construct a fault association model for each analysis node based on the association mapping table of each analysis node; A fault location model is constructed based on the fault location sub-models of all analysis nodes and the corresponding fault association models. A fault analysis model is generated based on the fault diagnosis model and the fault location model.

[0008] In some embodiments of this application, generating a fault evaluation value corresponding to the fault category based on historical fault data packets includes: Several fault evaluation indicators are pre-defined; Based on the historical fault data packets for each fault category, several fault evaluation indicators are generated as fault sub-evaluation values, and combined with the weight coefficients of the corresponding fault evaluation indicators, a fault evaluation value is generated. The formula for calculating the fault evaluation value is: ; Where G is the fault evaluation value, n is the fault evaluation index, gi is the fault sub-evaluation value of the i-th fault evaluation index, and ai is the weight coefficient of the i-th fault evaluation index.

[0009] In some embodiments of this application, the real-time analysis time interval for generating wind farm collector lines includes: The preset analysis time interval is set based on the comprehensive state coefficient of the wind farm collection lines in the previous monitoring period; Real-time monitoring data of preset monitoring points in each associated region are collected according to preset analysis time intervals, and real-time monitoring features are generated based on the real-time monitoring data of all preset monitoring points in the same associated region. Based on real-time monitoring characteristics, predictive periodic monitoring characteristics for the current monitoring period are generated; The predicted periodic monitoring characteristics of each associated region are compared with the preset fault characteristics of the corresponding fault category to obtain several similarity coefficients. The average similarity coefficient of all key regions of the same fault category is obtained by averaging the similarity coefficients. Fault categories with a mean similarity coefficient greater than a preset similarity coefficient threshold are set as the predicted fault categories for the current monitoring period. The analysis time intervals corresponding to the predicted fault categories are weighted, and the resulting analysis time intervals are set as the real-time analysis time intervals of several analysis nodes in the associated regions involved in the predicted fault categories. Set the preset analysis time interval to the real-time analysis time interval of several analysis nodes in the unrelated regions.

[0010] In some embodiments of this application, real-time monitoring data of the wind farm's collector lines is acquired based on real-time analysis time intervals, and initial analysis results of the real-time monitoring data are generated based on a fault analysis model, including: Real-time monitoring data at several analysis nodes in each associated region is obtained according to the real-time analysis time interval; The initial diagnostic results of real-time monitoring data at each analysis node are generated based on the fault analysis model. The analysis nodes with faults in the initial diagnostic results are selected, and the initial location results of each selected analysis node with faults are generated based on the fault analysis model. The initial location result is several associated nodes of each faulty analysis node, and the corresponding node topology graph is constructed by combining the location information and structural information of the analysis node and its corresponding associated nodes. The node topology graphs of all the analyzed nodes with faults are fused to obtain an initial fused topology graph. Based on the initial fused topology map, the undetermined region range of potential fault sources is determined, and the undetermined region range is used as the initial analysis result.

[0011] In some embodiments of this application, a primary positioning strategy is set, and real-time feedback data is collected according to the primary positioning strategy, including: Based on the range of the undetermined area in the initial analysis results, several primary positioning sub-regions are divided, and each primary positioning sub-region corresponds to a positioning priority coefficient. The primary positioning sub-regions are sorted according to the positioning priority coefficient to generate a positioning execution sequence. Each primary positioning sub-region in the positioning execution sequence is mapped to a corresponding monitoring density and a corresponding acquisition frequency. According to the monitoring density, deploy several monitoring units in each primary positioning sub-region of the positioning execution sequence; The deployed monitoring units collect real-time feedback data for the corresponding primary positioning sub-regions at the corresponding acquisition frequency.

[0012] In some embodiments of this application, a fault risk coefficient is generated based on real-time feedback data, including: Several fault risk assessment indicators are pre-defined; Based on the correlation between real-time feedback data and each fault risk assessment indicator, determine the real-time feedback data associated with each fault risk assessment indicator. The real-time feedback data associated with the same primary positioning sub-region are evaluated based on several fault risk assessment indicators to obtain the fault risk coefficient of the corresponding primary positioning sub-region. The fault risk coefficient for each primary positioning sub-region is generated sequentially.

[0013] In some embodiments of this application, generating fault location results based on fault risk coefficients includes: Pre-set risk coefficient thresholds; First-level location sub-regions with a fault risk coefficient greater than a preset risk threshold are selected as fault regions; Extract the real-time feedback data of all monitoring units in each fault area, and combine the line topology and equipment parameters of the area to construct a fault risk distribution map for the corresponding fault area; Based on the fault risk distribution map, the fault path and fault location of each fault area are determined, and the fault location result is output by combining the fault type, location information and fault level of the fault location.

[0014] In some embodiments of this application, a fault location system for wind farm collector lines is also included: The module is used to define multiple fault categories and build a fault analysis model based on all fault categories. The generation module is used to generate the real-time analysis time interval of the wind farm's collector lines and obtain the real-time monitoring data of the wind farm's collector lines based on the real-time analysis time interval. The configuration module is used to generate initial analysis results of real-time monitoring data based on the fault analysis model, and to set the primary positioning strategy and collect real-time feedback data according to the primary positioning strategy. The positioning module is used to generate a fault risk coefficient based on real-time feedback data, and then generate a fault location result based on the fault risk coefficient.

[0015] The wind farm collector line fault location method and system of this application embodiment have the following advantages compared with the prior art: By constructing a fault analysis model and setting a real-time analysis time interval, initial analysis results are generated and a primary fault location strategy is formulated. Real-time feedback data is obtained according to the primary fault location strategy, and the fault risk coefficient is calculated. The fault location results are then output, which improves the accuracy and real-time performance of fault location and provides technical support for the safe and stable operation of wind farms. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for locating faults in a wind farm collector line, as described in this application. Detailed Implementation

[0017] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate this application, but are not intended to limit the scope of this application.

[0018] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0019] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] like Figure 1 As shown in the figure, a method for locating faults in a wind farm collector line according to an embodiment of this application includes: Step S101: Define multiple fault categories and construct a fault analysis model based on all fault categories; Step S102: Generate the real-time analysis time interval of the wind farm's collection lines, and obtain the real-time monitoring data of the wind farm's collection lines based on the real-time analysis time interval; Step S103: Generate the initial analysis results of real-time monitoring data based on the fault analysis model, set the primary positioning strategy, and collect real-time feedback data according to the primary positioning strategy; Step S104: Generate a fault risk coefficient based on real-time feedback data, and generate a fault location result based on the fault risk coefficient.

[0022] In some embodiments of this application, a fault analysis model is constructed based on all fault categories, including: Generate historical fault data packets for each fault category based on historical fault logs for each fault category. Generate a fault evaluation value corresponding to the fault category based on historical fault data packets; Based on the fault evaluation value, the number of associated regions for the wind farm's collection lines is set for the corresponding fault category, and multiple associated regions for the corresponding fault category are generated by combining the fault region mapping sequence for the corresponding fault category. The fault region mapping sequence includes several historical fault regions, and each historical fault region corresponds to a historical fault coefficient. The historical fault regions are arranged according to the historical fault coefficient. Historical fault features of each associated region are obtained and constructed into a training dataset. The neural network is trained based on the training dataset of all associated regions of the same fault category to obtain the initial fault diagnosis model for the corresponding fault category. The number of analysis nodes for the corresponding associated regions is set according to the historical failure coefficients, and analysis nodes for each associated region are generated and the analysis time interval for the analysis nodes is set. A fault diagnosis model is established based on the initial fault diagnosis model for all fault categories, the number of analysis nodes, and the analysis time interval.

[0023] In this embodiment, the larger the fault evaluation value, the more associated regions there are. Based on the number of associated regions, the fault regions in the fault region mapping sequence are extracted in order to obtain the associated regions.

[0024] In this embodiment, the initial fault diagnosis model refers to the neural network trained using historical fault characteristics as training input data and the corresponding existing faults and historical fault levels as training output data.

[0025] In this embodiment, the historical fault coefficient refers to the quantitative value of the impact of the historical fault area on the overall operation of the collection line when the historical fault occurred. The higher the historical fault coefficient, the larger the number of analysis nodes, and the analysis nodes are set in combination with the historical fault nodes of the corresponding related areas.

[0026] In this embodiment, the fault diagnosis model can be used to perform targeted diagnosis of the associated areas of different fault categories. Combined with the dynamic configuration of analysis nodes, it can improve the early identification capability of various potential faults in the wind farm collection lines and provide accurate diagnostic basis for subsequent fault location.

[0027] In some embodiments of this application, constructing a fault analysis model based on all fault categories further includes: Generate historical fault sub-data packets for each analysis node in each associated region; Based on the historical fault sub-data packets, determine whether each analysis node has a correlation with other analysis nodes. If so, set it as a related node for the corresponding analysis node and determine the correlation characteristics. The association influence value between each analysis node and its corresponding associated node is calculated based on the association characteristics. Each analysis node is assigned an association mapping table based on all associated influence values. The association mapping table includes a first sub-association mapping table, a second sub-association mapping table, and a third sub-association mapping table. Each associated node in the sub-association mapping table has corresponding location information and association features. Construct a fault location sub-model for each analysis node; Construct a fault association model for each analysis node based on the association mapping table of each analysis node; A fault location model is constructed based on the fault location sub-models of all analysis nodes and the corresponding fault association models. A fault analysis model is generated based on the fault diagnosis model and the fault location model.

[0028] In this embodiment, the correlation features include the degree of influence of different historical fault coefficients of the analysis node on the historical fault coefficients of the associated monitoring point, the correlation degree of fault type, the correlation degree of data transmission delay, and the co-occurrence frequency of historical faults. The degree of influence is determined by calculating the contribution rate of the analysis node to the state changes of the associated node under different states; the correlation degree of fault type is obtained by analyzing the probability that the analysis node and the associated node experience the same or similar fault types in the historical fault data; the correlation degree of data transmission delay is quantified based on the average delay time and delay fluctuation range of data transmission between the two; the co-occurrence frequency of historical faults refers to the proportion of the number of times the analysis node and the associated node experience faults in the same time period to the total number of faults. The correlation impact value is calculated by comprehensively considering the above four correlation features and using a weighted summation method.

[0029] In this embodiment, the first sub-association mapping table is constructed based on several associated nodes in the same associated region whose association influence value with the analysis node is greater than a preset third association influence value threshold. The second sub-association mapping table is constructed based on several associated nodes in the same fault category but different associated regions whose association influence value with the analysis node is greater than a preset second association influence value threshold. The third sub-association mapping table is constructed based on several associated nodes in associated regions of different fault categories whose association influence value with the analysis node is greater than a preset first association influence value threshold. The preset first association influence value threshold is less than the preset second association influence value threshold and the preset third association influence value threshold.

[0030] In this embodiment, the fault location sub-model is constructed by dividing the historical state data packets of the analysis node into multiple state sample sets according to the time series. Each state sample set contains time series data of key parameters such as voltage, current, temperature and insulation resistance under different fault types. Each state sample set corresponds to the historical state coefficients, fault types and fault occurrence probabilities of preset monitoring points.

[0031] In this embodiment, the fault association model constructs an association graph between analysis nodes based on the association mapping table of analysis nodes. It uses a graph attention layer to learn the influence weights of different associated nodes on analysis nodes. Combined with the output results of the fault location sub-model, it realizes cross-monitoring point collaborative verification and location accuracy optimization of faults of associated nodes, further improving the comprehensiveness and accuracy of fault location.

[0032] In this embodiment, a fault analysis model is obtained by constructing a fault diagnosis model and a fault location model. A basic analysis framework for different fault categories of wind farm collection lines is constructed. Then, a first-level location strategy is used to collect feedback data in depth, and finally, accurate fault location results are generated. This effectively improves the efficiency and reliability of fault investigation of wind farm collection lines and provides strong technical support for the stable operation of wind farms.

[0033] In some embodiments of this application, generating a fault evaluation value corresponding to the fault category based on historical fault data packets includes: Several fault evaluation indicators are pre-defined; Based on the historical fault data packets for each fault category, several fault evaluation indicators are generated as fault sub-evaluation values, and combined with the weight coefficients of the corresponding fault evaluation indicators, a fault evaluation value is generated. The formula for calculating the fault evaluation value is: ; Where G is the fault evaluation value, n is the fault evaluation index, gi is the fault sub-evaluation value of the i-th fault evaluation index, and ai is the weight coefficient of the i-th fault evaluation index.

[0034] In this embodiment, fault evaluation indicators include, but are not limited to, fault-affected area, fault impact degree, fault level, fault repair difficulty, and historical fault occurrence frequency. The fault-affected area is quantified by statistically analyzing the proportion of affected collector lines to the total line length in historical faults. The fault impact degree is assessed based on a comprehensive evaluation of factors such as power generation loss, downtime, and maintenance costs. The fault level is classified according to industry standards into four levels: minor, moderate, severe, and critical. The fault repair difficulty is determined by considering factors such as fault type, geographical location, and required specialized tools. The historical fault occurrence frequency is expressed as the number of times that fault category occurs per unit time.

[0035] In this embodiment, the weight coefficients of each fault evaluation index are set in advance. By using the fault sub-evaluation values ​​and corresponding weights of the above-mentioned multi-dimensional evaluation indexes, a fault evaluation value is generated, which can comprehensively and objectively reflect the severity and impact range of different fault categories, and provide a reliable basis for setting the number of associated areas in the future.

[0036] In some embodiments of this application, the real-time analysis time interval for generating wind farm collector lines includes: The preset analysis time interval is set based on the comprehensive state coefficient of the wind farm collection lines in the previous monitoring period; Real-time monitoring data of preset monitoring points in each associated region are collected according to preset analysis time intervals, and real-time monitoring features are generated based on the real-time monitoring data of all preset monitoring points in the same associated region. Based on real-time monitoring characteristics, predictive periodic monitoring characteristics for the current monitoring period are generated; The predicted periodic monitoring characteristics of each associated region are compared with the preset fault characteristics of the corresponding fault category to obtain several similarity coefficients. The average similarity coefficient of all key regions of the same fault category is obtained by averaging the similarity coefficients. Fault categories with a mean similarity coefficient greater than a preset similarity coefficient threshold are set as the predicted fault categories for the current monitoring period. The analysis time intervals corresponding to the predicted fault categories are weighted, and the resulting analysis time intervals are set as the real-time analysis time intervals of several analysis nodes in the associated regions involved in the predicted fault categories. Set the preset analysis time interval to the real-time analysis time interval of several analysis nodes in the unrelated regions.

[0037] In this embodiment, the comprehensive state coefficient is a quantitative indicator generated based on the overall operating status of the wind farm's collection lines in the previous monitoring period. It is calculated by integrating monitoring data from various related areas, historical fault records, equipment aging degree, and environmental influencing factors, among other multi-dimensional parameters. The larger the comprehensive state coefficient, the better the overall operating status of the wind farm's collection lines in the previous monitoring period, and the larger the preset analysis time interval, and vice versa.

[0038] In this embodiment, the real-time monitoring features include key parameters such as voltage fluctuation amplitude, current peak change rate, and temperature gradient distribution at each preset monitoring point within the associated area during the monitoring period. The predicted periodic monitoring features are based on the real-time monitoring features and use time series prediction algorithms (such as LSTM neural networks) to extrapolate the parameter change trends within a preset time period in the future, generating a prediction dataset containing predicted parameter values, trend slopes, and fluctuation ranges.

[0039] In this embodiment, similarity analysis refers to calculating the cosine similarity and Euclidean distance between the predicted periodic monitoring features and the preset fault features, quantifying the degree of matching between the two. The higher the degree of matching, the larger the similarity coefficient, and vice versa.

[0040] In this embodiment, the predicted fault category refers to the fault category that may occur in the current monitoring period. The weight of the predicted fault category is obtained by converting the average similarity coefficient. The larger the average similarity coefficient, the greater the weight. The real-time analysis time interval is obtained by calculating the weight and the analysis time interval of the predicted fault category.

[0041] In this embodiment, the unrelated region refers to the region corresponding to a fault category that is not a predicted fault category.

[0042] In this embodiment, by generating predictive periodic monitoring features and performing similarity analysis with preset fault features of fault categories, the predicted fault category is obtained, and the analysis time interval is adjusted to optimize data processing efficiency while ensuring monitoring accuracy, prioritize accurate monitoring of high-risk areas, and improve the timeliness and accuracy of fault location.

[0043] In some embodiments of this application, real-time monitoring data of the wind farm's collector lines is acquired based on real-time analysis time intervals, and initial analysis results of the real-time monitoring data are generated based on a fault analysis model, including: Real-time monitoring data at several analysis nodes in each associated region is obtained according to the real-time analysis time interval; The initial diagnostic results of real-time monitoring data at each analysis node are generated based on the fault analysis model. The analysis nodes with faults in the initial diagnostic results are selected, and the initial location results of each selected analysis node with faults are generated based on the fault analysis model. The initial location result is several associated nodes of each faulty analysis node, and the corresponding node topology graph is constructed by combining the location information and structural information of the analysis node and its corresponding associated nodes. The node topology graphs of all the analyzed nodes with faults are fused to obtain an initial fused topology graph. Based on the initial fused topology map, the undetermined region range of potential fault sources is determined, and the undetermined region range is used as the initial analysis result.

[0044] In this embodiment, the initial diagnostic results include whether a fault exists or not. If a fault exists, the initial diagnostic results also include the fault level at the corresponding analysis node.

[0045] In this embodiment, during the topology fusion process, node matching and conflict verification are performed on the node topology graphs of different analysis nodes. For nodes with overlapping location information or contradictory structural relationships, the location parameters are corrected by weighted average method, and the connection weights between nodes are adjusted in combination with the magnitude of the correlation influence value to ensure that the initial fused topology graph can accurately reflect the actual distribution and interaction relationship of each associated node.

[0046] In this embodiment, the determination of the undetermined area range is based on the analysis nodes with the highest fault level ranking in the initial fusion topology map and the nodes with the highest correlation influence value ranking in the corresponding analysis nodes. Combined with the location coordinates of the nodes and the line topology, a regular or irregular area including the line segments within a preset distance range around the nodes is delineated. This provides a clear spatial analysis boundary for the implementation of the subsequent first-level positioning strategy. The preset ratio is 1 / 2, and the preset distance range refers to the range obtained by drawing a circle with the corresponding node as the center and a pre-set radius.

[0047] In this embodiment, by constructing an initial fusion topology map, the scattered fault information of the analysis nodes can be integrated into a systematic regional fault association network, laying a spatial analysis foundation for subsequent accurate positioning, effectively avoiding the limitations of single-node analysis, and improving the comprehensiveness and reliability of the initial analysis results.

[0048] In some embodiments of this application, a primary positioning strategy is set, and real-time feedback data is collected according to the primary positioning strategy, including: Based on the range of the undetermined area in the initial analysis results, several primary positioning sub-regions are divided, and each primary positioning sub-region corresponds to a positioning priority coefficient. The primary positioning sub-regions are sorted according to the positioning priority coefficient to generate a positioning execution sequence. Each primary positioning sub-region in the positioning execution sequence is mapped to a corresponding monitoring density and a corresponding acquisition frequency. According to the monitoring density, deploy several monitoring units in each primary positioning sub-region of the positioning execution sequence; The deployed monitoring units collect real-time feedback data for the corresponding primary positioning sub-regions at the corresponding acquisition frequency.

[0049] In this embodiment, the location priority coefficient is determined by comprehensively considering the fault level of the analysis node with the highest fault level in the sub-region, the sum of associated impact values, and the distance to the center of the region to be determined. The higher the fault level, the greater the sum of associated impact values, and the closer the distance, the larger the location priority coefficient, and vice versa.

[0050] In this embodiment, the monitoring density refers to the number of monitoring units deployed within the area of ​​the primary positioning sub-region. The analysis node is also a monitoring unit. Its value is positively correlated with the positioning priority coefficient of the primary positioning sub-region. That is, the higher the positioning priority coefficient, the higher the monitoring density and the more monitoring units there are, so as to achieve refined data collection in high-risk areas. When deploying monitoring units, it is necessary to combine the terrain features, route orientation and existing analysis node distribution of the primary positioning sub-region to ensure the accuracy and completeness of real-time feedback data.

[0051] In this embodiment, the acquisition frequency is set according to the fault development trend prediction results of the sub-region. For sub-regions with higher fault levels or faster changes in fault characteristics, a higher acquisition frequency is used to ensure that the dynamic process of fault evolution can be captured. For sub-regions with lower risk, the acquisition frequency can be appropriately reduced to reduce data redundancy.

[0052] In this embodiment, the real-time feedback data includes, but is not limited to, key electrical parameters such as line current, voltage, temperature, partial discharge, and insulation resistance, as well as external influencing factors such as ambient temperature and humidity, wind speed and direction, providing multi-dimensional real-time data support for the accurate location of the subsequent fault source.

[0053] In some embodiments of this application, a fault risk coefficient is generated based on real-time feedback data, including: Several fault risk assessment indicators are pre-defined; Based on the correlation between real-time feedback data and each fault risk assessment indicator, determine the real-time feedback data associated with each fault risk assessment indicator. The real-time feedback data associated with the same primary positioning sub-region are evaluated based on several fault risk assessment indicators to obtain the fault risk coefficient of the corresponding primary positioning sub-region. The fault risk coefficient for each primary positioning sub-region is generated sequentially.

[0054] In this embodiment, the fault risk assessment indicators include parameters such as parameter deviation, trend change rate, and abnormal fluctuation frequency. The parameter deviation is quantified by calculating the degree of deviation between each key parameter in the real-time feedback data and the corresponding normal operating threshold range. The trend change rate is calculated by the slope of parameter change at the continuous acquisition node. The abnormal fluctuation frequency is the number of times the parameter value exceeds the preset fluctuation range per unit time. The larger the evaluation value of the fault risk assessment indicator, the higher the fault risk of the corresponding sub-region. After obtaining the evaluation value of each fault risk assessment indicator, the weighting method is used to calculate the fault risk coefficient of the first-level positioning sub-region.

[0055] In this embodiment, by conducting further fault risk assessment based on the initial diagnostic results of the fault analysis model, the real-time operating status of the primary positioning sub-region can be transformed into quantifiable fault risk indicators, providing a key basis for subsequent accurate positioning.

[0056] In some embodiments of this application, generating fault location results based on fault risk coefficients includes: Pre-set risk coefficient thresholds; First-level location sub-regions with a fault risk coefficient greater than a preset risk threshold are selected as fault regions; Extract the real-time feedback data of all monitoring units in each fault area, and combine the line topology and equipment parameters of the area to construct a fault risk distribution map for the corresponding fault area; Based on the fault risk distribution map, the fault path and fault location of each fault area are determined, and the fault location result is output by combining the fault type, location information and fault level of the fault location.

[0057] In this embodiment, the fault location result includes a complete fault location result containing fault trend, fault location, fault location coordinates, fault type and risk level.

[0058] In this embodiment, the preset risk threshold is set based on the safety operation standards of wind farm collection lines and historical fault handling experience.

[0059] In this embodiment, the distribution map includes several colors, and the darker the color of the node, the higher the fault level.

[0060] In this embodiment, when outputting the fault location results, fault handling suggestions are generated simultaneously, including the priority repair order, the required spare parts models, and on-site safety precautions, further improving the pertinence and efficiency of fault handling for wind farm collection lines.

[0061] In some embodiments of this application, a fault location system for wind farm collector lines is also included: The module is used to define multiple fault categories and build a fault analysis model based on all fault categories. The generation module is used to generate the real-time analysis time interval of the wind farm's collector lines and obtain the real-time monitoring data of the wind farm's collector lines based on the real-time analysis time interval. The configuration module is used to generate initial analysis results of real-time monitoring data based on the fault analysis model, and to set the primary positioning strategy and collect real-time feedback data according to the primary positioning strategy. The positioning module is used to generate a fault risk coefficient based on real-time feedback data, and then generate a fault location result based on the fault risk coefficient.

[0062] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of this application, and these improvements and substitutions should also be considered within the scope of protection of this application.

Claims

1. A method for fault location of a power collection line of a wind farm, characterized in that, The method comprises the following steps: setting a plurality of fault categories, and constructing a fault analysis model according to all the fault categories; generating a real-time analysis time interval of a power collection line of a wind farm, and acquiring real-time monitoring data of the power collection line of the wind farm according to the real-time analysis time interval; generating an initial analysis result of the real-time monitoring data according to the fault analysis model, setting a first positioning strategy, and collecting real-time feedback data according to the first positioning strategy; generating a fault risk coefficient according to the real-time feedback data, and generating a fault positioning result according to the fault risk coefficient.

2. The method of claim 1, wherein, The fault analysis model is constructed according to all the fault categories, which comprises the following steps: generating a historical fault data packet of each fault category based on a historical fault log of each fault category; generating a fault evaluation value of a corresponding fault category according to the historical fault data packet; setting a number of associated regions of the corresponding fault category for the power collection line of the wind farm according to the fault evaluation value, and generating a plurality of associated regions of the corresponding fault category in combination with a fault region mapping sequence of the corresponding fault category; The fault region mapping sequence comprises a plurality of historical fault regions, and each historical fault region corresponds to a historical fault coefficient, and the historical fault regions are arranged according to the historical fault coefficients. acquiring historical fault features of each associated region and constructing a training data set, and performing neural network training according to the training data set of all the associated regions of the same fault category to obtain an initial fault diagnosis model of the corresponding fault category; setting a number of analysis nodes of the corresponding associated region according to the historical fault coefficient, generating an analysis node of each associated region, and setting an analysis time interval of the analysis node; establishing a fault diagnosis model according to the initial fault diagnosis model of all the fault categories, the number of analysis nodes, and the analysis time interval.

3. The method of claim 2, wherein, The fault analysis model is constructed according to all the fault categories, which further comprises the following steps: generating a historical fault sub-data packet of each analysis node of each associated region; determining whether each analysis node has an associated relationship with other analysis nodes according to the historical fault sub-data packet, and if yes, setting an associated node of the corresponding analysis node and determining an associated feature; calculating an associated influence value of each analysis node and the corresponding associated node according to the associated feature; setting an associated mapping table of each analysis node according to all the associated influence values, wherein the associated mapping table comprises a first sub-associated mapping table, a second sub-associated mapping table, and a third sub-associated mapping table, and each sub-associated mapping table comprises position information and an associated feature of the associated node; constructing a fault positioning sub-model of each analysis node; constructing a fault association model of each analysis node according to the associated mapping table of each analysis node; constructing a fault positioning model according to the fault positioning sub-models of all the analysis nodes and the corresponding fault association models; generating the fault analysis model according to the fault diagnosis model and the fault positioning model.

4. The method of claim 3, wherein, The fault evaluation value of the corresponding fault category is generated according to the historical fault data packet, which comprises the following steps: pre-setting a plurality of fault evaluation indexes; generating a fault sub-evaluation value of each fault evaluation index based on the historical fault data packet of each fault category, and generating the fault evaluation value in combination with a weight coefficient of the corresponding fault evaluation index; The calculation formula of the fault evaluation value is: ； Wherein, G is a fault evaluation value, n is a fault evaluation index, gi is a fault sub-evaluation value of the i th fault evaluation index, and ai is a weight coefficient of the i th fault evaluation index.

5. The method of claim 4, wherein, The real-time analysis time interval of the power collection line of the wind farm is generated, including: According to the comprehensive state coefficient of the power collection line of the wind farm in the last monitoring period, a preset analysis time interval is set; Real-time monitoring data of preset monitoring points in each associated area is collected according to the preset analysis time interval, and real-time monitoring features are generated according to the real-time monitoring data of all preset monitoring points in the same associated area; According to the real-time monitoring features, a prediction period monitoring feature of the current monitoring period is generated; The prediction period monitoring feature of each associated area is analyzed for similarity with the preset fault feature of the fault category corresponding to the associated area, and a plurality of similarity coefficients are obtained; The similarity coefficients of all key areas of the same fault category are processed by averaging to obtain a similarity coefficient average value; The fault category with a similarity coefficient average value greater than a preset similarity coefficient threshold value is set as a prediction fault category of the current monitoring period; The analysis time interval corresponding to the prediction fault category is processed by weight, and the analysis time interval obtained by weight processing is set as the real-time analysis time interval of a plurality of analysis nodes of the associated area involved in the prediction fault category; The preset analysis time interval is set as the real-time analysis time interval of a plurality of analysis nodes of an uninvolved associated area.

6. The method of claim 5, wherein, Real-time monitoring data of the power collection line of the wind farm is obtained according to the real-time analysis time interval, and an initial analysis result of the real-time monitoring data is generated according to the fault analysis model, including: Real-time monitoring data at a plurality of analysis nodes in each associated area is obtained according to the real-time analysis time interval; Based on the fault analysis model, an initial diagnosis result of the real-time monitoring data at each analysis node is generated, and analysis nodes with an initial diagnosis result of existing faults are screened out, and an initial positioning result of each analysis node with existing faults is generated based on the fault analysis model; The initial positioning result is a plurality of associated nodes of each analysis node with existing faults, and a corresponding node topology graph is constructed in combination with the position information and structure information of the analysis node and the corresponding associated node; Topology fusion is performed according to the node topology graphs of all analysis nodes with existing faults to obtain an initial fusion topology graph; The range of the pending area of the potential fault source is determined based on the initial fusion topology graph, and the range of the pending area is taken as the initial analysis result.

7. The method of claim 6, wherein, A first positioning strategy is set, and real-time feedback data is collected according to the first positioning strategy, including: According to the range of the pending area in the initial analysis result, a plurality of first positioning sub-areas are divided, and each first positioning sub-area corresponds to a positioning priority coefficient; The first positioning sub-areas are sorted according to the positioning priority coefficients to generate a positioning execution sequence, and each first positioning sub-area in the positioning execution sequence is mapped with a corresponding monitoring density and a corresponding collection frequency; A plurality of monitoring units are deployed in each first positioning sub-area in the positioning execution sequence according to the monitoring density; Real-time feedback data of the corresponding first positioning sub-area is collected by the deployed monitoring unit according to the corresponding collection frequency.

8. The method of claim 7, wherein, A fault risk coefficient is generated according to the real-time feedback data, including: A plurality of fault risk evaluation indexes are preset; Based on the correlation between real-time feedback data and each fault risk evaluation index, the real-time feedback data associated with each fault risk evaluation index is determined; Based on the plurality of fault risk evaluation indexes, the associated real-time feedback data in the same first positioning sub-region is evaluated to obtain a fault risk coefficient of the corresponding first positioning sub-region; The fault risk coefficients of each first positioning sub-region are generated in sequence.

9. The method of claim 8, wherein, According to the fault risk coefficient, a fault positioning result is generated, including: A risk coefficient threshold is preset; Filter out the first positioning sub-region with a fault risk coefficient greater than the preset risk threshold as a fault area; Extract the real-time feedback data of all monitoring units in each fault area, combine the line topology structure and equipment parameters of the area, and construct a fault risk distribution map of the corresponding fault area; Based on the fault risk distribution map, the fault trend and fault point of each fault area are determined, and combined with the fault type, location information and fault level of the fault point, the fault positioning result is output.

10. A wind farm collection line fault location system, characterized in that, Including: A construction module is used to set up a plurality of fault categories and construct a fault analysis model according to all fault categories; A generation module is used to generate a real-time analysis time interval of the wind farm collection line and obtain real-time monitoring data of the wind farm collection line according to the real-time analysis time interval; A setting module is used to generate an initial analysis result of the real-time monitoring data according to the fault analysis model, set a first positioning strategy, and collect real-time feedback data according to the first positioning strategy; A positioning module is used to generate a fault risk coefficient according to the real-time feedback data and generate a fault positioning result according to the fault risk coefficient.

Images