A method and system for determining the root cause of failure of large wind turbines

Abstract

The present invention provides a method and system for determining the root cause of large wind turbine faults. The method comprises the following steps: obtaining initialization information for a large wind turbine generator set from wind turbine configuration data; initializing a single fault map and fault spectrum corresponding to the wind turbine model using real-time and historical wind turbine data; obtaining wind turbine information at the instant of a wind turbine fault; inputting the wind turbine information into a database, and determining the first fault of the wind turbine using a fault spectrum and a fault root cause identification algorithm; if the fault spectrum and the fault root cause identification algorithm cannot identify the fault, prompting the user to manually determine the first fault through an expert system; after the fault is resolved, the expert system, combined with on-site operation and maintenance, completes the determination, stores the result in the database, and updates the fault spectrum. The system comprises a database, an inference engine unit, and a fault spectrum unit. This system can automatically and accurately determine the first fault of a large wind turbine generator set, safely and efficiently.

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbine fault root cause determination, and in particular to a method and system for determining the root cause of a large wind turbine fault. Background Art

[0002] The root cause of wind turbine failure, also known as the initial fault, is the primary concern for users and operators when a turbine fails. It is also the root cause of all other accompanying faults and turbine downtime. It is currently widely used in the wind power industry, playing a key role in fault prediction, situational awareness, and intelligent operation and maintenance. Currently, the root cause of wind turbine failures is primarily determined manually or based simply on the chronological order of fault code triggering. This method fails to achieve intelligent automatic detection and efficient and accurate root cause determination, severely impacting subsequent operation and maintenance as well as critical statistical information. Misjudgment and incorrect statistics of initial faults increase the likelihood of future major accidents for the entire turbine.

[0003] Existing patented technologies based on fault trees and expert systems have been widely used in various fields, especially in fault diagnosis, because these technologies have inherent advantages in the field of fault diagnosis. The advantage of the fault tree is that it naturally fits the computer memory structure and has the characteristics of fast running speed. However, the brother nodes of the tree structure have different characteristics and clear levels, which have certain limitations in fault judgment. Today's computer hardware can already support more complex graph data structures and ensure the timeliness of program operation. In the wind power industry, the method of applying these technologies to the root cause determination of faults of large wind turbines is not mature enough and has many defects. Therefore, the present invention uses a method of combining fault graphs with fault chains to determine the root cause of faults. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems in the prior art, namely, that the fault root cause determination method is not intelligent enough, the determination result is not accurate enough, which affects the operation and maintenance of the wind turbine and poses a safety hazard.

[0005] To this end, a first aspect of the present invention provides a method for determining the root cause of a large wind turbine fault.

[0006] A second aspect of the present invention provides a system for determining the root cause of a large wind turbine fault.

[0007] The present invention provides a method for determining the root cause of a large wind turbine fault, comprising the following steps:

[0008] S1. Obtaining initialization information of large wind turbine generator sets through wind turbine configuration information data;

[0009] S2. Initialize the single fault map and fault spectrum corresponding to the wind turbine model using the wind turbine real-time data and historical data;

[0010] S3. When a wind turbine fails, obtain wind turbine information at the moment of failure; input the wind turbine information into the database, and determine the first failure of the wind turbine through the fault map and fault root cause identification algorithm;

[0011] S4. If the fault map and fault root cause identification algorithm cannot identify the fault, the user is prompted to manually determine the initial fault through the expert system; after the fault is resolved, the expert system combines the on-site operation and maintenance to complete the judgment, stores the result in the database and updates the fault map.

[0012] The method for determining the root cause of a large wind turbine fault according to the above technical solution of the present invention may also have the following additional technical features:

[0013] In the above technical solution, in step S3, determining the first fault of the wind turbine by using the fault map and the fault root cause identification algorithm includes the following steps:

[0014] S31, traverse the fault spectrum to determine whether there are all vertices in the fault spectrum that correspond to the actual fault and may be the root cause of the fault. If not, execute S4; if so, execute S32;

[0015] S32. Determine whether any vertex corresponding to the actual fault in S31 is the only fault that can be the root cause of the fault. If so, determine that the fault corresponding to the vertex is the root cause of the fault. If not, based on the characteristic that each fault has at least one root cause and several accompanying faults, and that the accompanying faults may include other accompanying faults that can be the root cause of the fault, use all vertices that can be the root cause of the fault as starting points and other faults as boundaries, and use the Kosaraju algorithm combined with the shrinking point coloring algorithm to extract all minimum cut sets.

[0016] S33. Use the fuzzy contrast algorithm to compare the minimum cut set with the basic single fault graph stored in the database, compare the feature set, logic set and degree set, and obtain the minimum cut set with the highest similarity value. The vertex corresponding to the minimum cut set is used as the root cause of the fault.

[0017] In the above technical solution, S33 includes the following steps:

[0018] S331, traverse all single fault graphs in the database;

[0019] S332, comparing the minimum cut set obtained in S32 with the single fault graph, and calculating a similarity value;

[0020] S333. Determine whether the similarity value meets the requirements. If so, determine the root cause of the fault. If not, execute S4.

[0021] In the above technical solution, the comparison method in S332 is:

[0022] y = f(x_1,x_2,x_3)

[0023] Among them, y represents the final similarity value, x_1, x_2, and x_3 represent the feature set, logical set, and degree set respectively.

[0024] In any of the above technical solutions, the fault map is a map composed of interconnected single fault maps corresponding to all possible faults of a certain model of wind turbine, which is called the fault map of this model of wind turbine;

[0025] The single fault graph is a directed graph structure corresponding to a single fault of a wind turbine, consisting of data objects and data relationships;

[0026] Among them, a data object is a collection of data elements with the same or similar characteristics, called a vertex. For a large wind turbine, each fault code it belongs to is a vertex in the fault map. Faults without fault codes are also vertices of the fault map. A data relationship is the connectivity relationship between any two data objects Va and Vb.

[0027] In the above technical solution, in the fault map, each vertex in the map consists of a single fault, has an independent fault code in the main control system, and belongs to a specific fan component; every two associated vertices are composed of a specific data relationship.

[0028] In the above technical solution, in the fault graph, each fault that occurs is composed of a subset of the graph, which is implemented in the system by a customized single data graph structure, including customized data relationships, data objects and other basic data structures.

[0029] In the above technical solution, the fault map contains all possible faults and accompanying faults of the wind turbine. Each fault and accompanying fault of the wind turbine can be identified through the fault map. If the identification fails, the fault map needs to be updated and identified with the help of human beings.

[0030] In any of the above technical solutions, when executing step S3, after preliminarily determining that the fan has initially failed, it is necessary to determine whether the failure has ended. If it has ended, the determination result is output; if it has not ended, S3 is repeated.

[0031] The present invention further provides a system for determining the root cause of a large wind turbine fault, which is applied to a method for determining the root cause of a large wind turbine fault as described in any one of the above technical solutions, comprising:

[0032] Database: used to store wind turbine data and fault maps, and provide event sources for the inference engine unit and fault map unit;

[0033] Inference engine unit: used to execute the single fault graph generation algorithm, fault spectrum generation algorithm and fault root cause identification algorithm, and determine the fault root cause through the fault spectrum when a fault is triggered;

[0034] Fault map unit: Initialize the configuration through the configuration table corresponding to the expert and fan model, and dynamically generate and update it based on the database and fan data.

[0035] In summary, due to the adoption of the above technical features, the beneficial effects of the present invention are:

[0036] The fault root cause determination method provided by the present invention can automatically and accurately determine the first fault of a large wind turbine generator set, providing a good foundation for wind turbine generator set fault analysis and intelligent operation and maintenance, saving a large amount of manpower, material and financial resources, and ensuring timeliness, safety and efficiency.

[0037] Additional aspects and advantages of the invention will become apparent from the description which follows, or may be learned by practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments with reference to the accompanying drawings, in which:

[0039] Figure 1 This is a flowchart of a method for determining the root cause of a large wind turbine fault according to an embodiment of the present invention;

[0040] Figure 2 This is a flow chart of a fault root cause determination algorithm in a method for determining the root cause of a fault of a large wind turbine according to one embodiment of the present invention;

[0041] Figure 3 This is a flow chart of a fuzzy comparison algorithm in a method for determining the root cause of a large wind turbine fault according to one embodiment of the present invention;

[0042] Figure 4 The present invention is an embodiment of a large-scale wind turbine fault root cause determination system execution logic diagram. DETAILED DESCRIPTION

[0043] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0044] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0045] Refer to the following Figures 1 to 4 A method and system for determining the root cause of a large wind turbine fault according to some embodiments of the present invention will be described.

[0046] Some embodiments of the present application provide a method for determining the root cause of a large wind turbine failure.

[0047] like Figures 1 to 4 As shown, the first embodiment of the present invention provides a method for determining the root cause of a large wind turbine fault, comprising the following steps:

[0048] S1. Obtaining initialization information of large wind turbine generator sets through wind turbine configuration information data;

[0049] S2. Initialize the single fault map and fault spectrum corresponding to the wind turbine model using the wind turbine real-time data and historical data;

[0050] S3. When a wind turbine fails, the wind turbine information is obtained at the moment of failure. The wind turbine information includes fault code, brake level, shutdown code, etc. The wind turbine information is input into the database, and the first fault of the wind turbine is determined by the fault map and fault root cause identification algorithm. The process is repeated during the fault continuation stage, and the fault map is updated in real time.

[0051] In step S3, the root cause of each comprehensive fault of the wind turbine is determined by only considering the two situations where the root cause occurs or does not occur. The entire fault process is monitored through the determination algorithm to analyze the final fault cause. Determining the first fault of the wind turbine through the fault map and the fault root cause identification algorithm includes the following steps:

[0052] S31, traverse the fault spectrum to determine whether there are all vertices in the fault spectrum that correspond to the actual fault and may be the root cause of the fault. If not, execute S4; if so, execute S32;

[0053] S32. Determine whether any vertex corresponding to the actual fault in S31 is the only fault that can be the root cause of the fault. If so, determine that the fault corresponding to the vertex is the root cause of the fault. If not, based on the characteristic that each fault has at least one root cause and several accompanying faults, and that the accompanying faults may include other accompanying faults that can be the root cause of the fault, use all vertices that can be the root cause of the fault as starting points and other faults as boundaries, and use the Kosaraju algorithm combined with the shrinking point coloring algorithm to extract all minimum cut sets.

[0054] S33. Use the fuzzy contrast algorithm to compare the minimum cut set with the basic single fault graph stored in the database, compare the feature set, logical set and degree set, and obtain the minimum cut set with the highest similarity value. The vertex corresponding to the minimum cut set is used as the root cause of the fault. After the results are obtained, they are confirmed by experts and the single fault graph and fault spectrum are updated.

[0055] S33 includes the following steps:

[0056] S331, traverse all single fault graphs in the database;

[0057] S332, comparing the minimum cut set obtained in S32 with the single fault graph, and calculating a similarity value;

[0058] The comparison method in S332 is: the fault graphs generated by the same fault event are similar and at least some subsets are identical. The similarity value is calculated by comparing the feature set, logic set, and degree set. The calculation method is as follows:

[0059] y = f(x_1,x_2,x_3);

[0060] Among them, y represents the final similarity value, x_1, x_2, and x_3 represent the feature set, logical set, and degree set respectively.

[0061] S333. Determine whether the similarity value meets the requirements. If so, determine the root cause of the fault. If not, execute S4.

[0062] S4. If the fault map and fault root cause identification algorithm cannot identify the fault, the human-computer interaction interface generates a report, prompting the user to manually determine the initial fault through the expert system; after the fault is resolved, the expert system combines the on-site operation and maintenance to complete the judgment, stores the result in the database and updates the fault map.

[0063] The fault spectrum is a spectrum composed of interconnected single fault diagrams corresponding to all possible faults of a certain model of fan, which is called the fault spectrum of this model of fan.

[0064] The fault graph has the following characteristics: it consists of multiple data objects and data relationships; it is composed of a finite number of single fault graphs; all faults of a certain model of wind turbine can be mapped to a subset of a certain fault graph; it is a directed graph; each model of wind turbine corresponds to a separate fault graph; it uses dynamic storage management; each sub-single fault graph in the fault graph is not necessarily connected;

[0065] The fault map generation process is as follows:

[0066] At the end of each fault, the root cause of the single fault is determined, the single fault map is stored in the database, and the new single fault map is connected to update the total fault map.

[0067] The single fault graph is a directed graph structure corresponding to a single fault of a wind turbine, consisting of data objects and data relationships;

[0068] The single fault graph has the following properties: vertex set, data relationship set corresponding to the vertex, fault root cause determined by human, fault root cause determined by the inference engine, accuracy probability of the fault root cause determined by the inference engine, fault snapshot data, whether the fault root cause can be determined by the program, fault root cause related components, feature set, logic set, and degree set.

[0069] The single fault graph has the following characteristics: contains zero, one or more data relationships; contains vertices consisting of one or more data objects; each single fault graph corresponds to a certain fault of a certain model of fan; directed graph structure; contains the root cause of the fault and other accompanying faults; since accompanying faults may not occur, a certain fault of a certain model of fan may correspond to multiple single fault graphs; may not contain accompanying faults, and only consist of a single fault vertex corresponding to the root cause of the fault; each graph corresponds to a root cause of the fault that causes the system fault; contains a snapshot of this fault; contains the root cause of the fault determined manually; the feature set is obtained based on the fault code, duration, shutdown code, and associated components of each vertex; the logical set is obtained based on the degree of each vertex, each connected set, and the data relationship between vertices; the degree set is obtained based on the number of vertices, fault duration, fault repair time, number of faults that can be used as root causes, and braking level.

[0070] The single fault diagram generation process includes the following steps:

[0071] Manually initialize the minimum failure mode, that is, the basic event that can cause a certain type of large wind turbine system failure and shutdown, and configure the corresponding attributes;

[0072] Each time a fault occurs, the first triggered fault is used as the starting point, and the inference engine is used to find the direct cause event from the top event step by step, and connect each accompanying fault;

[0073] When the fault persists, repeat the above steps to continuously update the single fault diagram.

[0074] A data object is a collection of data elements with the same or similar characteristics, called a vertex. For a large wind turbine, each fault code is a vertex in the fault map. Faults without fault codes are also considered vertices in the fault map.

[0075] Each vertex has the characteristics of an abstract data structure, inherited from the "basic vertex data structure", and is scalable;

[0076] The basic vertex data structure contains at least the following attributes: fault code, fault code occurrence time, fault code duration, shutdown code, brake level, fault code associated components (one or more), component strongly connected vertex set, component weakly connected vertex set, component strongly connected component set, fault duration, fault repair time, vertex degree, vertex out-degree, vertex in-degree, starting vertex flag, vertex type, and whether it can be used as the root cause of the fault.

[0077] A data relationship is a connectivity relationship between any two data objects Va and Vb.

[0078] The data structure of the data relationship includes the following attributes: starting vertex, arrival vertex, connection direction, typical interval time distribution between vertices, vertex causal connection relationship, normal distribution of typical interval time between vertices, mean, variance, maximum, minimum value of typical interval time between vertices, and set of reasons for data relationship connectivity;

[0079] In the fault map, each vertex in the map consists of a single fault, has an independent fault code in the main control system, belongs to a specific fan component, and is implemented by a custom data object structure in the system; every two associated vertices are composed of a specific data relationship, which is implemented by a custom data relationship structure in the system.

[0080] In the fault graph, each fault is composed of a subset of the graph, which is implemented in the system by a custom single data graph structure, including custom data relationships, data objects and other basic data structures.

[0081] The fault map contains all possible faults and accompanying faults of the wind turbine. Each fault and accompanying fault of the wind turbine can be identified through the fault map. If the identification fails, the fault map needs to be updated and identified with the help of human beings.

[0082] The second embodiment of the present invention proposes a system for determining the root cause of a large wind turbine fault, and based on the first embodiment, Figures 1 to 4 As shown, including:

[0083] Database: used to store wind turbine data and fault maps, providing an event source for the inference engine unit and fault map unit; the database stores up to 30 years of second-level historical data of the wind turbine, real-time data of the wind turbine, historical fault snapshots, wind turbine parameters, wind turbine equipment information, and other content needed to generate fault maps.

[0084] The inference engine unit is used to execute the single fault graph generation algorithm, the fault spectrum generation algorithm, and the fault root cause identification algorithm. When a fault is triggered, the root cause is determined by the fault spectrum. The inference engine unit primarily uses the Kosaraju algorithm combined with the shrinkage point coloring algorithm to determine the fault cut set corresponding to the fault spectrum of the current fault. It then uses the fuzzy comparison algorithm to compare historical fault spectra to determine the root cause of the fault. If the fault cannot be determined, manual determination is required. After the determination is completed, the fault spectrum is updated in the database.

[0085] Fault map unit: Initialize the configuration through the configuration table corresponding to the expert and fan model, and dynamically generate and update it based on the database and fan data.

[0086] During use, experts configure basic fault maps through the human-machine interface and monitor wind turbine data; when the inference engine unit is unable to determine the root cause of the fault, experts make a judgment.

[0087] The database is the foundation of the entire system, providing the event source for the inference engine and fault maps. It stores basic wind turbine data, such as fault codes, wind speed, and other operational data. It includes all the real-time data required to generate fault maps, as well as 30 years of historical, second-level data.

[0088] In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any suitable manner in any one or more embodiments or examples.

[0089] Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for determining the root cause of a large wind turbine fault, characterized in that: The following steps are involved: S1. Obtaining initialization information of large wind turbine generator sets through wind turbine configuration information data; S2. Initialize the single fault map and fault spectrum corresponding to the wind turbine model using the wind turbine real-time data and historical data; S3. When a wind turbine fails, obtain wind turbine information at the moment of failure; input the wind turbine information into the database, and determine the first failure of the wind turbine through the fault map and fault root cause identification algorithm; S4. If the fault map and the fault root cause identification algorithm cannot identify the fault, the user is prompted to manually determine the first fault through the expert system; After the fault is resolved, the expert system, combined with on-site operation and maintenance, completes the judgment, stores the results in the database and updates the fault map; In step S3, determining the first fault of the wind turbine by using the fault map and the fault root cause identification algorithm includes the following steps: S31, traverse the fault spectrum to determine whether there are all vertices in the fault spectrum that correspond to the actual fault and may be the root cause of the fault. If not, execute S4; if so, execute S32; S32. Determine whether any vertex corresponding to the actual fault in S31 is the only fault that can be the root cause of the fault. If so, determine that the fault corresponding to the vertex is the root cause of the fault. If not, based on the characteristic that each fault has at least one root cause and several accompanying faults, and that the accompanying faults may include other accompanying faults that can be the root cause of the fault, use all vertices that can be the root cause of the fault as starting points and other faults as boundaries, and use the Kosaraju algorithm combined with the shrinking point coloring algorithm to extract all minimum cut sets. S33. Using a fuzzy contrast algorithm, the minimum cut set is compared with the basic single fault graph stored in the database, and the feature set, logic set, and degree set are compared to obtain the minimum cut set with the highest similarity value. The vertex corresponding to the minimum cut set is used as the root cause of the fault. Among them, the fault map is a map composed of interconnected single fault maps corresponding to all possible faults of a certain model of wind turbine, which is called the fault map of this model of wind turbine; The single fault graph is a directed graph structure corresponding to a single fault of a wind turbine, consisting of data objects and data relationships; A data object is a collection of data elements with the same or similar characteristics, called a vertex. For a large wind turbine, each fault code is a vertex in the fault map. Faults without fault codes are also considered vertices in the fault map. A data relationship is the connectivity between any two data objects. The feature set is obtained based on the fault code, duration, shutdown code, and associated components of each vertex; the logical set is obtained based on the degree of each vertex, each connected set, and the data relationship between vertices; the degree set is obtained based on the number of vertices, fault duration, fault repair time, number of faults that can be the root cause of the fault, and braking level.

2. A method for determining the root cause of a large wind turbine fault according to claim 1, characterized in that: S33 includes the following steps: S331, traverse all single fault graphs in the database; S332, comparing the minimum cut set obtained in S32 with the single fault graph, and calculating a similarity value; S333. Determine whether the similarity value meets the requirements. If so, determine the root cause of the fault. If not, execute S4.

3. The method for determining the root cause of a large wind turbine fault according to claim 2, wherein: The comparison method in S332 is: y = f(x_1,x_2,x_3) Among them, y represents the final similarity value, x_1, x_2, and x_3 represent the feature set, logical set, and degree set respectively.

4. The method for determining the root cause of a large wind turbine fault according to claim 1, wherein: In the fault map, each vertex in the map consists of a single fault, has an independent fault code in the main control system, and belongs to a specific component of a wind turbine; and every two associated vertices are composed of a specific data relationship.

5. The method for determining the root cause of a large wind turbine fault according to claim 1, characterized in that: In the fault graph, each fault is composed of a subset of the graph, which is implemented in the system by a custom single data graph structure, including custom data relationships, data objects and other basic data structures.

6. The method for determining the root cause of a large wind turbine fault according to claim 1, wherein: The fault map contains all possible faults and accompanying faults that may occur in the wind turbine. Each fault and accompanying fault that occurs in the wind turbine is identified through the fault map. If the identification fails, the fault map needs to be updated and identified with the help of human beings.

7. A method for determining the root cause of a large wind turbine fault according to any one of claims 1 to 3, characterized in that: When executing step S3, after preliminarily determining that the fan has failed for the first time, it is necessary to determine whether the failure has ended. If it has ended, the determination result is output; if it has not ended, S3 is repeated.

8. A system for determining the root cause of a large wind turbine fault, characterized in that: A method for determining a root cause of a large wind turbine fault as claimed in any one of claims 1 to 7, comprising: Database: used to store wind turbine data and fault maps, and provide event sources for the inference engine unit and fault map unit; Inference engine unit: used to execute the single fault graph generation algorithm, fault spectrum generation algorithm and fault root cause identification algorithm, and determine the fault root cause through the fault spectrum when a fault is triggered; Fault map unit: Initialize the configuration through the configuration table corresponding to the expert and fan model, and dynamically generate and update it based on the database and fan data.

Images