Operation inspection method, system and equipment for wind turbine generator equipment and medium

Abstract

The invention relates to an operation inspection method, system, equipment and medium for wind turbine generator equipment, and the operation inspection method for the wind turbine generator equipment comprises the steps: obtaining operation inspection problem information, input by a user to a wind turbine generator operation inspection model, of equipment to be operated and inspected, determining an operation inspection problem sequence corresponding to the operation inspection problem information according to the operation inspection problem information, according to the operation inspection question sequence, operation inspection answers corresponding to the operation inspection question information can be accurately and effectively determined, the normalization and safety of operation inspection work of the wind turbine generator set are improved, the intelligent level of fault processing is enhanced, operation and maintenance personnel are rapidly assisted in knowledge query, and therefore the decision-making efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbine generator set equipment operation and inspection, and in particular to an operation and inspection method, system, equipment and medium for wind turbine generator set equipment. Background Art

[0002] In the context of today's energy structure transformation and digitalization, wind power, as a clean and renewable energy form, is developing rapidly and occupying an increasingly important position. With the increase in wind power installed capacity, its operation and maintenance pressure is also gradually increasing. It is urgent to quickly improve the operation and maintenance level to match the scale of wind power operation. Due to the cost, it is impossible to arrange a large number of operation and maintenance personnel, and the level of wind power operation and maintenance informatization is still insufficient, and the maintenance efficiency is low. Summary of the invention

[0003] In order to solve the problems existing in the prior art, the present invention provides an operation and inspection method, system, equipment and medium for wind turbine equipment.

[0004] A first aspect of the present invention provides a method for operating and inspecting a wind turbine generator set, the method comprising:

[0005] Obtaining the operation and inspection problem information of the equipment to be inspected input by the user into the wind turbine unit operation and inspection model;

[0006] Determine, according to the operation and inspection problem information, an operation and inspection problem sequence corresponding to the operation and inspection problem information;

[0007] According to the operation inspection question sequence, an operation inspection answer corresponding to the operation inspection question information is determined.

[0008] Optionally, determining, according to the operation inspection question sequence, an operation inspection answer corresponding to the operation inspection question information includes:

[0009] According to the operation and inspection problem sequence, searching for target triple data corresponding to the operation and inspection problem sequence from a preset query statement database;

[0010] Arrange the entity labels and relationship types in the target triple data according to preset query statement rules to obtain a target query statement corresponding to the operation and inspection problem sequence;

[0011] According to the target query statement, an operation and inspection answer corresponding to the operation and inspection question information is obtained.

[0012] Optionally, searching, according to the operation and inspection problem sequence, a preset query statement database for target triple data corresponding to the operation and inspection problem sequence includes:

[0013] According to the operation and inspection problem sequence, searching for a plurality of triplet data to be used from a preset query statement database;

[0014] Calculating the similarity between each standby triplet data and the operation and inspection problem sequence to obtain a plurality of standby similarities;

[0015] The standby triple data corresponding to the maximum similarity among the multiple standby similarities is used as the target triple data corresponding to the operation and inspection question sequence.

[0016] Optionally, obtaining an operation and inspection answer corresponding to the operation and inspection question information according to the target query statement includes:

[0017] According to the target query statement, the operation and inspection answer corresponding to the operation and inspection question information is searched from the preset operation and inspection knowledge graph.

[0018] Optionally, the process of constructing the preset operation and inspection knowledge graph includes:

[0019] Acquire a plurality of initial operation and inspection data of the wind turbine generator set equipment;

[0020] Extracting data from the multiple initial operation and inspection data to obtain multiple unstructured operation and inspection data;

[0021] Using a preset model, extracting a plurality of triplet data from the unstructured plurality of inspection and operation data;

[0022] The multiple triple data are stored in the initial graph database to obtain a preset operation and inspection knowledge graph.

[0023] Optionally, the preset model includes a coding layer, a table feature generation layer, a global feature mining layer and a triple generation layer, and the use of the preset model to extract multiple triple data from the multiple unstructured operation data includes:

[0024] Encoding the plurality of operation and inspection data through the encoding layer to obtain a subject feature vector and an object representation vector corresponding to each operation and inspection data;

[0025] Iteratively calculating the subject feature vector and the object representation vector through the table feature generation layer and the global feature mining layer to obtain the iterated subject feature vector and object representation vector;

[0026] The iterated subject feature vector and object representation vector are decoded through the preset label table in the triple generation layer to obtain multiple triple data.

[0027] Optionally, determining, according to the operation and inspection problem information, an operation and inspection problem sequence corresponding to the operation and inspection problem information includes:

[0028] According to a preset language processing strategy, information is extracted from the operation and inspection problem information to obtain the standby entity content and standby relationship type in the operation and inspection problem information;

[0029] The standby entity content and the standby relationship type are used as an operation inspection problem sequence corresponding to the operation inspection problem information.

[0030] A second aspect of the present invention provides an operation and inspection system for wind turbine equipment, the system comprising:

[0031] An acquisition module is configured to acquire the operation and inspection problem information of the equipment to be inspected input by the user into the wind turbine operation and inspection model;

[0032] A first determining module is configured to determine, according to the operation and inspection problem information, an operation and inspection problem sequence corresponding to the operation and inspection problem information;

[0033] The second determination module is configured to determine the operation and inspection answers corresponding to the operation and inspection question information according to the operation and inspection question sequence.

[0034] Optionally, the second determining module is configured to:

[0035] According to the operation and inspection problem sequence, searching for target triple data corresponding to the operation and inspection problem sequence from a preset query statement database;

[0036] Arrange the entity labels and relationship types in the target triple data according to preset query statement rules to obtain a target query statement corresponding to the operation and inspection problem sequence;

[0037] According to the target query statement, an operation and inspection answer corresponding to the operation and inspection question information is obtained.

[0038] Optionally, the second determining module is configured to:

[0039] According to the operation and inspection problem sequence, searching for a plurality of triplet data to be used from a preset query statement database;

[0040] Calculating the similarity between each standby triplet data and the operation and inspection problem sequence to obtain a plurality of standby similarities;

[0041] The standby triple data corresponding to the maximum similarity among the multiple standby similarities is used as the target triple data corresponding to the operation and inspection question sequence.

[0042] Optionally, the second determining module is configured to:

[0043] According to the target query statement, the operation and inspection answer corresponding to the operation and inspection question information is searched from the preset operation and inspection knowledge graph.

[0044] Optionally, the process of constructing the preset operation and inspection knowledge graph includes:

[0045] Acquire a plurality of initial operation and inspection data of the wind turbine generator set equipment;

[0046] Extracting data from the multiple initial operation and inspection data to obtain multiple unstructured operation and inspection data;

[0047] Using a preset model, extracting a plurality of triplet data from the unstructured plurality of inspection and operation data;

[0048] The multiple triple data are stored in the initial graph database to obtain a preset operation and inspection knowledge graph.

[0049] Optionally, the preset model includes a coding layer, a table feature generation layer, a global feature mining layer and a triple generation layer, and the use of the preset model to extract multiple triple data from the multiple unstructured operation data includes:

[0050] Encoding the plurality of operation and inspection data through the encoding layer to obtain a subject feature vector and an object representation vector corresponding to each operation and inspection data;

[0051] Iteratively calculating the subject feature vector and the object representation vector through the table feature generation layer and the global feature mining layer to obtain the iterated subject feature vector and object representation vector;

[0052] The iterated subject feature vector and object representation vector are decoded through the preset label table in the triple generation layer to obtain multiple triple data.

[0053] Optionally, the first determining module is configured to:

[0054] According to a preset language processing strategy, information is extracted from the operation and inspection problem information to obtain the standby entity content and standby relationship type in the operation and inspection problem information;

[0055] The standby entity content and the standby relationship type are used as an operation inspection problem sequence corresponding to the operation inspection problem information.

[0056] A third aspect of the present invention provides a computer device, comprising: one or more processors;

[0057] The processor is used to store one or more programs;

[0058] When the one or more programs are executed by the one or more processors, the wind turbine equipment operation and inspection method described in the first aspect of the present invention is implemented.

[0059] In a fourth aspect of the present invention, there is provided a computer-readable storage medium having stored thereon a computer program, which when executed, implements the operation and inspection method of a wind turbine unit device described in the first aspect of the present invention above.

[0060] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0061] The present invention provides an operation and inspection method, system, device and medium for a wind turbine unit device. The operation and inspection method of the wind turbine unit device determines an operation and inspection problem sequence corresponding to the operation and inspection problem information by obtaining the operation and inspection problem information of the device to be inspected input by the user to the wind turbine operation and inspection model. According to the operation and inspection problem sequence, the operation and inspection answer corresponding to the operation and inspection problem information can be accurately and effectively determined, improving the standardization and safety of the wind turbine operation and inspection operation, enhancing the intelligent level of fault handling, quickly assisting the maintenance personnel in knowledge query, and thus improving the decision-making efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 is a flowchart of an operation and inspection method for a wind turbine unit device provided by the present invention;

[0063] Figure 2 is a flowchart of an operation and inspection method for a wind turbine unit device provided by the present invention;

[0064] Figure 3 is a flowchart of a construction process of a preset operation and inspection knowledge graph provided by the present invention;

[0065] Figure 4 is a schematic diagram of an operation and inspection system for a wind turbine unit device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] Example 1:

[0067] Figure 1 is a flowchart of an operation and inspection method for a wind turbine unit device provided by the present invention. As Figure 1 shown, the operation and inspection method for the wind turbine unit device may include the following steps:

[0068] In step 101, obtain the operation and inspection problem information of the device to be inspected input by the user to the wind turbine operation and inspection model.

[0069] Exemplarily, the operation and inspection problem information of the device to be inspected input by the user to the wind turbine operation and inspection model may be: "The abnormal noise of the blade is caused by defects (pits, holes) on the blade surface or blade edge. The problematic areas at the blade edge or at the blade tip should be repaired or removed with fiberglass."

[0070] Furthermore, before obtaining the maintenance problem information of the equipment to be inspected input by the user into the wind turbine maintenance model, all wind turbine maintenance triple data in the preset maintenance knowledge graph are queried, and a preset query statement database is generated based on the triple data in the preset maintenance knowledge graph. The preset query statement database contains all wind turbine maintenance entity category labels and relationship types, and stores text data in the form of triples '(:entity label 1)-[:relationship type]->(:entity label 2)', so that the large model can better understand the professional field content and provide a query paradigm for subsequent query knowledge graphs.

[0071] For example, the preset operation and maintenance knowledge graph can be a Neo4j graph database, and the preset query statement database can be a Schema query specification library. All wind turbine operation and maintenance triple data in the Neo4j graph database are queried, and a large model Schema query specification library is generated according to the triples in the graph.

[0072] In step 102, an operation and inspection problem sequence corresponding to the operation and inspection problem information is determined according to the operation and inspection problem information.

[0073] Possible implementation methods of this step may include: extracting information from the operation and inspection problem information according to a preset language processing strategy to obtain standby entity content and standby relationship types in the operation and inspection problem information; and using the standby entity content and the standby relationship types as the operation and inspection problem sequence corresponding to the operation and inspection problem information.

[0074] It should be noted that the questions are refined according to the context history of the operation and inspection question information and the preset natural language processing capabilities, and the conjunctions, punctuation marks and invalid repeated information in the operation and inspection question information are deleted, and the effective information such as the entity content and relationship type in the question sequence is highlighted to generate the operation and inspection question sequence. By rewriting the operation and inspection question information input by the user, the key information in the user's question content is extracted to obtain the operation and inspection question sequence, and the key information is detailed to make the question more precise.

[0075] In step 103, the operation and inspection answers corresponding to the operation and inspection question information are determined according to the operation and inspection question sequence.

[0076] It should be noted that the operation and maintenance question sequence is generated into a query statement according to a specific statement template, relevant content is searched in the knowledge graph module, and prompt words for the wind turbine operation and maintenance large model are generated according to the query results. After the relevant content in the knowledge graph module is embedded in the prompt words, questions are asked to the large model again, and the final output results of the large model are polished and replied to the user.

[0077] The present invention proposes a method for constructing a large model and system for wind turbine equipment operation and maintenance based on knowledge graph enhancement, which can realize knowledge questions and answers in specific fields of wind turbine equipment operation and maintenance. It can realize the sedimentation of private domain data such as wind turbine maintenance operation records, manufacturer information and equipment ledgers, and form a professional knowledge base in the vertical field of wind turbines. At the same time, retrieval enhancement is performed based on the professional knowledge base to improve the capability boundary and scope of application of the large model, generate real-time wind turbine operation and maintenance strategies, and improve the efficiency of wind turbine operation and maintenance.

[0078] Figure 2 A flow chart of a wind turbine equipment operation and inspection method provided by the present invention, such as Figure 2 As shown above Figure 1 Possible implementations of step 103 may include the following steps:

[0079] In step 1031, according to the operation and inspection problem sequence, the target triple data corresponding to the operation and inspection problem sequence is searched from a preset query statement database.

[0080] The specific implementation method of this step can be: according to the operation and inspection problem sequence, multiple standby triple data are searched from the preset query statement database; similarity is calculated between each standby triple data and the operation and inspection problem sequence to obtain multiple standby similarities; and the standby triple data corresponding to the maximum similarity among the multiple standby similarities is used as the target triple data corresponding to the operation and inspection problem sequence.

[0081] For example, similarity calculation is performed based on the inspection question sequence '{query}' and the triple data in the Schema query specification library to find the triple data with the closest semantics to the question sequence '{query}'.

[0082] In step 1032, the entity labels and relationship types in the target triple data are arranged according to preset query statement rules to obtain a target query statement corresponding to the operation and inspection problem sequence.

[0083] For example, according to the Cypher query statement rules of the Neo4j graph database, the entity label and relationship type of the obtained triple data are filled into the Cypher query statement to obtain the target query statement corresponding to the operation and inspection problem sequence.

[0084] In step 1033, the operation and inspection answer corresponding to the operation and inspection question information is obtained according to the target query statement.

[0085] Possible implementation manners of this step may be: According to the target query statement, search for the operation and maintenance answer corresponding to the operation and maintenance problem information from a preset operation and maintenance knowledge graph. For example, use the filled Cypher query statement (i.e., the target query statement) to query entities and relationships related to the operation and maintenance problem sequence '{query}' in the knowledge graph module (i.e.), and obtain the output result '{kg_results}' of the knowledge graph.

[0086] Optionally, the specific implementation manners of the above Figure 1 and Figure 2 shown method may be:

[0087] After the user inputs the problem sequence, the large model module will use the large model's natural language processing capabilities for context history and itself to refine the problem. Delete the conjunctions, punctuation marks, and invalid repeated information in the original problem, and highlight the effective information such as entity content and relationship types in the problem sequence to generate a new problem sequence '{query}'.

[0088] Subsequently, the large model module calculates the similarity between the new problem sequence '{query}' and the triple data in the Schema query specification library to find the triple data with the closest semantics to the problem sequence '{query}'. Then, according to the Cypher query statement rules of the Neo4j graph database, fill the entity labels and relationship types of the obtained triple data into the Cypher query statement. The system uses the filled Cypher query statement to query entities and relationships related to the problem sequence '{query}' in the knowledge graph module, and obtains the output result '{kg_results}' of the knowledge graph.

[0089] Finally, return the obtained result '{kg_results}' to the large model module, fill it into the prompt template of the large model, and then output the final answer. The prompt template stipulates that if the content of '{kg_results}' is empty, answer the question according to the large model itself without considering the content of the knowledge graph; if the content of '{kg_results}' is not empty, based on the content of the graph, the large model appropriately expands it and then outputs the result, and it is required to indicate whether the output result comes from the knowledge graph information or itself when answering.

[0090] Exemplarily, assume there is the following operation and maintenance text data: "The abnormal noise of the blade is caused by defects (pits, holes) on the blade surface or at the blade edge. The problematic areas at the blade edge or at the blade tip should be repaired or removed with fiberglass." The following triples can be extracted: (blade, fault content, blade abnormal noise) (blade abnormal noise, cause of generation, blade surface or edge defect).

[0091] The Schema entity specification library generated based on this triple is as follows:

[0092] ?? <'labels': 'blade', 'properties': [<'property': 'name', 'type': 'STRING'>]>,

[0093] <'labels': 'fault content', 'properties': [<'property': 'content', 'type': 'STRING'>]>,

[0094] <'labels': 'fault cause', 'properties': [<'property': 'content', 'type': 'STRING'>]>,

[0095] The Schema relationship specification library generated by the large model based on this triple is as follows:

[0096] '(:blade)-[:fault content]->(:fault content)',

[0097] '(:fault content)-[:cause of generation]->(:fault cause)',

[0098] After obtaining this specification library, when the user queries "What is the reason for the abnormal noise of the blade?", the large model module rewrites the user input and generates a Cypher statement based on the specification library, such as:

[0099] cyphers: ["\nMATCH(t:blade)-[:fault content]->(b:fault content)-[:cause of generation]->(p:fault cause)WHERE t.name = 'blade' RETURN p.fault cause"]

[0100] The system retrieves in the knowledge graph through this query statement, obtains the result "surface or edge defect", and then sends the user's question content and "surface or edge defect" into the large model for query at the same time, and gets the final result after polishing and outputs it.

[0101] Figure 3 It is a flowchart of the construction process of a preset operation and maintenance knowledge graph provided by the present invention. As Figure 3 shown, it may include the following steps:

[0102] In step S1, obtain multiple initial operation and maintenance operation data of the wind turbine equipment.

[0103] Among them, the initial operation and maintenance operation data may include technical manuals, operation guides, maintenance manuals, etc. for the operation and maintenance of wind turbines. The formats of the initial operation and maintenance operation data include but are not limited to pictures, texts, and video materials.

[0104] Furthermore, preprocess the initial operation and maintenance operation data.

[0105] Exemplarily, screen out redundant information in the text, delete text directories, serial numbers, and conjunctions, extract data and other contents in the text of wind turbine operation and maintenance, and store them item by item with a full stop as the end point. For image and table data, extract text information to label the images and tables and store them separately.

[0106] In step S2, perform data extraction on the multiple initial operation and maintenance operation data to obtain multiple unstructured operation and maintenance operation data.

[0107] It should be noted that the initial data is preprocessed according to a standardized data structure, unstructured data is extracted, and the unstructured data is processed and data-labeled.

[0108] In step S3, use a preset model to extract multiple triple data from the multiple unstructured operation and maintenance operation data.

[0109] Among them, the preset model includes an encoding layer, a table feature generation layer, a global feature mining layer, and a triple generation layer.

[0110] A possible implementation of this step can be: encode the multiple operation and maintenance operation data through the encoding layer to obtain a subject feature vector and an object representation vector corresponding to each operation and maintenance operation data; perform iterative calculations on the subject feature vector and the object representation vector through the table feature generation layer and the global feature mining layer to obtain the iterated subject feature vector and object representation vector; decode the iterated subject feature vector and object representation vector through a preset label table in the triple generation layer to obtain multiple triple data.

[0111] It should be noted that the triple extraction task is equivalently transformed into a table filling task: given a wind turbine operation and maintenance text sequence of length n, maintain a table of size n×n for each relationship r∈R in the text sequence, and use (w i , w j ) to represent the token pair in the i-th row and the j-th column, and w i represents the subject token, and w j represents the object token. Then the set of labels filled in the cell corresponding to (w i , w j ) is:

[0112] {'N / A', 'MMH', 'MMT', 'MSH', 'MST', 'SMH', 'SMT', 'SS'}

[0113] Among them, 'N / A' indicates that there is no any association relationship between two tokens, and other tags indicate (w i , w j ) are associated with the same (subject token, object token) entity pair. The first character of the tag indicates whether the subject is a multi-token entity (corresponding to 'M') or a single-token entity (corresponding to 'S'), the second character of the tag indicates whether the object is a multi-token entity (corresponding to 'M') or a single-token entity (corresponding to 'S'), and the third character of the tag indicates whether the token pair (w i , w j ) is the start (corresponding to 'H') or the end (corresponding to 'T') of the two entities.

[0114] After obtaining the table filling task, the encoding layer, table feature generation layer, global feature mining layer and triple generation layer of the GRTE model process the text in sequence.

[0115] Since the table feature generation layer and the global mining layer require multiple rounds of iterative calculation, let in the t-th round, the subject and object features be and The table feature of the relationship r is Each item of the feature among them is (w i , w j ) corresponding feature value, that is

[0116]

[0117] The table feature generation layer learns the local association features between token pairs and generates the table features of each relationship. The global mining layer mines and models two types of global features on this basis, and generates new subject and object features, which are passed into the table feature generation layer for the next round of calculation.

[0118] First, the table features of all the previous relationships are concatenated to obtain a unified table feature TF (t) , and the subject-oriented table vector and the object-oriented table vector are obtained through max pooling and fully connected layers respectively, that is:

[0119]

[0120] Subsequently, a Transformer-based model is used to mine the global associations between relationships and between token pairs, respectively at Perform multi-head self-attention calculation on the token sequence H, and then use the fully connected layer to generate new subject features and object features:

[0121]

[0122] Finally, use residual connection to generate the final object and subject feature representations:

[0123]

[0124] The triple generation layer performs decoding on the padding table TF obtained after N rounds of calculation (N) and infers the existing triples, that is, for each relationship r, predicts the label of each item in its relationship table, and finally performs triple judgment based on the label table. After decoding, the final triples are generated, which are:

[0125]

[0126] The Neo4j database is a graph database that uses a graph data structure to store, process, and query data. Different from traditional relational databases, the Neo4j graph database is particularly suitable for processing complex wind turbine operation and maintenance triple data with a large number of relationships. After obtaining the wind turbine operation and maintenance triples through the GRTE model, the knowledge graph module uploads the triple data to the Neo4j database for knowledge representation, providing data support for the subsequent large model module to query the wind turbine operation and maintenance content.

[0127] In step S4, store the multiple triple data in the initial graph database to obtain a preset operation and maintenance knowledge graph.

[0128] It should be noted that the obtained triple data is stored in the preset graph database for knowledge representation to obtain the wind turbine operation and maintenance knowledge graph;

[0129] Embodiment 2:

[0130] Figure 4 As shown in the schematic diagram of an operation and maintenance system for wind turbine equipment provided by the present invention, Figure 4 as shown, the operation and maintenance system for wind turbine equipment may include:

[0131] An acquisition module 401, configured to acquire the operation and maintenance problem information of the equipment to be inspected input by the user to the wind turbine operation and maintenance model;

[0132] A first determination module 402, configured to determine an operation and maintenance problem sequence corresponding to the operation and maintenance problem information according to the operation and maintenance problem information;

[0133] The second determination module 403 is configured to determine the operation and inspection answer corresponding to the operation and inspection question information according to the operation and inspection question sequence.

[0134] Optionally, the second determining module 403 is configured to:

[0135] According to the operation and inspection problem sequence, searching for target triple data corresponding to the operation and inspection problem sequence from a preset query statement database;

[0136] Arrange the entity labels and relationship types in the target triple data according to preset query statement rules to obtain a target query statement corresponding to the operation and inspection problem sequence;

[0137] According to the target query statement, an operation and inspection answer corresponding to the operation and inspection question information is obtained.

[0138] Optionally, the second determining module 403 is configured to:

[0139] According to the operation and inspection problem sequence, searching for a plurality of triplet data to be used from a preset query statement database;

[0140] Calculating the similarity between each standby triplet data and the operation and inspection problem sequence to obtain a plurality of standby similarities;

[0141] The standby triple data corresponding to the maximum similarity among the multiple standby similarities is used as the target triple data corresponding to the operation and inspection question sequence.

[0142] Optionally, the second determining module 403 is configured to:

[0143] According to the target query statement, the operation and inspection answer corresponding to the operation and inspection question information is searched from the preset operation and inspection knowledge graph.

[0144] Optionally, the process of constructing the preset operation and inspection knowledge graph includes:

[0145] Acquire a plurality of initial operation and inspection data of the wind turbine generator set equipment;

[0146] Extracting data from the multiple initial operation and inspection data to obtain multiple unstructured operation and inspection data;

[0147] Using a preset model, extracting a plurality of triplet data from the unstructured plurality of inspection and operation data;

[0148] The multiple triple data are stored in the initial graph database to obtain a preset operation and inspection knowledge graph.

[0149] Optionally, the preset model includes a coding layer, a table feature generation layer, a global feature mining layer and a triple generation layer, and the use of the preset model to extract multiple triple data from the multiple unstructured operation data includes:

[0150] Encoding the plurality of operation and inspection data through the encoding layer to obtain a subject feature vector and an object representation vector corresponding to each operation and inspection data;

[0151] Iteratively calculating the subject feature vector and the object representation vector through the table feature generation layer and the global feature mining layer to obtain the iterated subject feature vector and object representation vector;

[0152] The iterated subject feature vector and object representation vector are decoded through the preset label table in the triple generation layer to obtain multiple triple data.

[0153] Optionally, the first determining module 401 is configured to:

[0154] According to a preset language processing strategy, information is extracted from the operation and inspection problem information to obtain the standby entity content and standby relationship type in the operation and inspection problem information;

[0155] The standby entity content and the standby relationship type are used as an operation inspection problem sequence corresponding to the operation inspection problem information.

[0156] Embodiment 3:

[0157] Based on the same inventive concept, the present invention also provides a computer device, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function, so as to implement the steps of the operation and inspection method of a wind turbine equipment in the above embodiment.

[0158] Embodiment 4:

[0159] Based on the same inventive concept, the present invention further provides a storage medium, specifically a computer-readable storage medium (Memory). The computer-readable storage medium is a memory device in a computer device and is used to store programs and data. It can be understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and, of course, the extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space, and this storage space stores the operating system of the terminal. Moreover, in this storage space, there is also stored one or more instructions suitable for being loaded and executed by the processor. These instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. One or more instructions stored in the computer-readable storage medium can be loaded and executed by the processor to implement the steps of the operation and maintenance method of a wind turbine device in the above embodiment.

[0160] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0161] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0162] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction means which implements the functions specified in one or more of the processes Figure 1 one or more of the processes and / or blocks Figure 1 specified in the block or blocks.

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one or more of the processes Figure 1 one or more of the processes and / or blocks Figure 1 specified in the block or blocks.

[0164] The above are only embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included within the scope of the claims of the present invention awaiting approval.

Claims

1. An operation and maintenance method for a wind turbine unit, characterized in that, The method comprises: Obtaining the operation and inspection problem information of the equipment to be inspected input by the user into the wind turbine unit operation and inspection model; Determine, according to the operation and inspection problem information, an operation and inspection problem sequence corresponding to the operation and inspection problem information; According to the operation inspection question sequence, an operation inspection answer corresponding to the operation inspection question information is determined.

2. The method according to claim 1, wherein Determining the operation and inspection answers corresponding to the operation and inspection question information according to the operation and inspection question sequence includes: According to the operation and inspection problem sequence, searching for target triple data corresponding to the operation and inspection problem sequence from a preset query statement database; Arrange the entity labels and relationship types in the target triple data according to preset query statement rules to obtain a target query statement corresponding to the operation and inspection problem sequence; According to the target query statement, an operation and inspection answer corresponding to the operation and inspection question information is obtained.

3. The method according to claim 2, characterized in that, The step of searching, according to the operation and inspection problem sequence, a preset query statement database for target triple data corresponding to the operation and inspection problem sequence includes: According to the operation and inspection problem sequence, searching for a plurality of triplet data to be used from a preset query statement database; Calculating the similarity between each standby triplet data and the operation and inspection problem sequence to obtain a plurality of standby similarities; The standby triple data corresponding to the maximum similarity among the multiple standby similarities is used as the target triple data corresponding to the operation and inspection question sequence.

4. The method according to claim 2, wherein The step of obtaining an operation and inspection answer corresponding to the operation and inspection question information according to the target query statement includes: According to the target query statement, the operation and inspection answer corresponding to the operation and inspection question information is searched from the preset operation and inspection knowledge graph.

5. The method according to claim 2, wherein The construction process of the preset operation and inspection knowledge graph includes: Acquire a plurality of initial operation and inspection data of the wind turbine generator set equipment; Extracting data from the multiple initial operation and inspection data to obtain multiple unstructured operation and inspection data; Using a preset model, extracting a plurality of triplet data from the unstructured plurality of inspection and operation data; The multiple triple data are stored in the initial graph database to obtain a preset operation and inspection knowledge graph.

6. The method according to claim 5, wherein The preset model includes a coding layer, a table feature generation layer, a global feature mining layer and a triple generation layer. The preset model is used to extract multiple triple data from the multiple unstructured operation data, including: Encoding the plurality of operation and inspection data through the encoding layer to obtain a subject feature vector and an object representation vector corresponding to each operation and inspection data; Iteratively calculating the subject feature vector and the object representation vector through the table feature generation layer and the global feature mining layer to obtain the iterated subject feature vector and object representation vector; The iterated subject feature vector and object representation vector are decoded through the preset label table in the triple generation layer to obtain multiple triple data.

7. The method according to claim 1, wherein The step of determining, according to the operation and inspection problem information, an operation and inspection problem sequence corresponding to the operation and inspection problem information comprises: According to a preset language processing strategy, information is extracted from the operation and inspection problem information to obtain the standby entity content and standby relationship type in the operation and inspection problem information; Use the to-be-processed entity content and the to-be-processed relationship type as the operation and maintenance problem sequence corresponding to the operation and maintenance problem information.

8. An operation and maintenance system for a wind turbine unit, characterized in that, The system includes: An acquisition module, configured to acquire operation and maintenance problem information of equipment to be inspected input by a user to an operation and maintenance model of a wind turbine generator set; A first determination module, configured to determine an operation and maintenance problem sequence corresponding to the operation and maintenance problem information according to the operation and maintenance problem information; A second determination module, configured to determine an operation and maintenance answer corresponding to the operation and maintenance problem information according to the operation and maintenance problem sequence.

9. A computer device, characterized in that, It includes: One or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, an operation and maintenance method for a wind turbine generator set device as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and when the computer program is executed, an operation and maintenance method for a wind turbine generator set device as described in any one of claims 1 to 7 is implemented.

Images