Television intelligent maintenance decision-making method, system and computer program

Abstract

The invention discloses a television intelligent maintenance decision-making method, and the method comprises the following steps: obtaining historical work order information which at least comprises a television type, a fault description, a fault reason, a maintenance scheme, a maintenance category and used spare parts; constructing an intelligent fault library containing standard faults according to the historical work order information; establishing a knowledge graph according to the historical work order information and the intelligent fault database; acquiring new work order information, performing fault matching with the intelligent fault library according to fault information in the new work order information, and performing maintenance scheme recommendation according to the matched fault based on the knowledge graph; the invention also discloses a television intelligent maintenance decision-making system and a computer program. The intelligent, automatic and personalized television maintenance process can be realized, the quality and efficiency of the maintenance service can be greatly improved, the operation cost can be reduced, and the high-quality and reliable maintenance experience can be provided for the user.

Description

Technical Field

[0001] The present invention relates to the technical field of television maintenance, and in particular to a television intelligent maintenance decision method, system and computer program. Background Art

[0002] In the traditional TV repair industry, repair personnel usually rely on their personal experience and technical knowledge to diagnose and repair faults. Although this approach is effective, it has some obvious limitations. First, the repair efficiency is limited by the professional level and experience accumulation of the repair personnel, and novice repair personnel may take longer to identify and solve the problem. Second, due to the lack of a standardized fault database, repair personnel may encounter difficulties when dealing with complex or rare faults, resulting in longer repair cycles and reduced customer satisfaction.

[0003] In addition, existing maintenance management systems often lack intelligent data analysis and decision support functions, and cannot effectively use historical work order data to predict and recommend maintenance plans. This not only limits the improvement of maintenance efficiency, but also increases maintenance costs and wastes resources.

[0004] To sum up, although the existing TV repair technology has met the market demand to a certain extent, there is still much room for improvement in efficiency, accuracy and intelligence. Summary of the invention

[0005] In view of the current technical problem that historical work order data cannot be effectively used to predict and recommend maintenance solutions, the present invention provides a television intelligent maintenance decision-making method, system and computer program, which can solve the above problems and improve the overall service level of the television maintenance industry.

[0006] To achieve the above object, the embodiments of the present invention adopt the following technical solutions:

[0007] A television intelligent maintenance decision-making method, the method comprising the following steps:

[0008] Obtaining historical work order information, wherein the historical work order information at least includes the TV model, fault description, fault cause, repair plan, repair category, and spare parts used;

[0009] Building an intelligent fault library containing standard faults based on historical work order information includes the following steps:

[0010] Fault sorting: Through fine-tuning of prompt word engineering, the DeepSeek model is used to sort out the repair faults in historical work orders, remove redundant information and mark the core faults;

[0011] Fault hierarchical clustering: Use the BGE model to vectorize the fault description, calculate the cosine similarity between fault vectors, and classify similar faults into one category based on the similarity threshold;

[0012] Fault summarization: Summarize the clustered faults to form standardized fault descriptions and generate standard fault entries to build an intelligent fault library. The summarization is based on the similarities and differences of the faults, and similar faults that are not identified by hierarchical clustering are grouped together to obtain duplicate faults. The summary uses prompt word engineering to fine-tune the DeepSeek model and refine the repair faults in the work order into standard descriptions.

[0013] A knowledge graph is established based on historical work order information and the intelligent fault database. A structured knowledge graph network is constructed by sorting and associating TV models, standard faults, maintenance categories, spare parts information and their relationships in historical work orders. The specific steps include:

[0014] Schema definition: defines the basic structure of the graph, including entity nodes such as TV model, fault type, repair category, spare parts, and the relationship types between entities;

[0015] Entity relationship generation and weight setting: Generate weights for the relationships between entity nodes based on historical data. The weights are set based on the following factors:

[0016] a) Historical frequency: Count the frequency of the relationship in historical work orders. The higher the frequency, the greater the weight of the relationship.

[0017] b) Relationship reliability: according to the repair success rate and spare parts usage rate, the weight of frequently successfully matched repair categories and spare parts is increased;

[0018] c) Node characteristic weight: Based on the common faults of different TV models, the fault nodes with higher frequency are recommended first;

[0019] Relationship Type:

[0020] a) Relationship weight between TV model and standard fault: reflects the common fault types of a certain TV model, and gives priority to recommending standard faults with high weights;

[0021] b) Relationship weight between standard faults and major maintenance categories: shows the optimal maintenance category under specific fault conditions;

[0022] c) The weight of the relationship between maintenance categories and spare parts: Prioritize the recommendation of more suitable spare parts to ensure the accuracy and cost optimization of maintenance;

[0023] Data storage and dynamic update: Store the knowledge graph data in the graph database and support dynamic update so that the weight can be updated according to the newly added work order information to maintain the real-time nature of the knowledge graph;

[0024] Acquire new work order information, where the new work order information includes at least a TV model and a fault description, match the fault with the intelligent fault database according to the fault information in the new work order information, and recommend a maintenance plan based on the matched fault based on the knowledge graph, specifically including:

[0025] Vectorize the fault description of the new work order, generate a fault vector using the BGE model, calculate the cosine similarity between the new work order fault vector and the standard fault vector in the intelligent fault library, and select the standard fault with the highest similarity;

[0026] In the knowledge graph, based on the relationship weight of the matching standard fault node, the associated maintenance categories and spare parts are queried to recommend the optimal maintenance plan.

[0027] According to one aspect of the present invention, the node definition includes:

[0028] TV model node: represents a specific model of TV device;

[0029] Standard fault nodes: represent standardized fault types that TV equipment may encounter;

[0030] Maintenance category node: represents the maintenance category required to repair a specific fault;

[0031] Spare Part Node: represents the spare part or component required during the maintenance process.

[0032] According to one aspect of the present invention, the relationship types include:

[0033] Relationship weight calculation:

[0034] Weight of the relationship between TV model and standard fault: Determine the most common fault type for this model based on the frequency in historical data;

[0035] The weight of the relationship between standard faults and maintenance categories: Based on the maintenance success rate and maintenance frequency, the maintenance category with the highest weight is recommended;

[0036] Weight of the relationship between maintenance categories and spare parts: Based on the spare parts usage rate, the spare parts combination that meets the maintenance needs is recommended first;

[0037] Dynamic adjustment mechanism: Relationship weights are regularly updated based on data such as the frequency of successful matching of new work orders and spare parts usage to ensure the accuracy of recommendation results.

[0038] According to one aspect of the present invention, the method further includes: regularly updating the intelligent fault library; and a dynamic update mechanism of the knowledge graph.

[0039] According to one aspect of the present invention, the periodic updating of the intelligent fault library includes: according to the newly added historical work order data, following the steps of fault cleaning and fault clustering, the standard faults in the intelligent fault library are periodically updated.

[0040] According to one aspect of the present invention, the dynamic update mechanism of the knowledge graph includes: adjusting the relationship weights according to the faults and maintenance records in the new work order data, so that the knowledge graph can reflect the latest maintenance data in real time, thereby optimizing the accuracy of maintenance recommendations.

[0041] According to one aspect of the present invention, the new work order information is obtained, and the new work order information includes at least a TV model and a fault description. The fault matching is performed with the intelligent fault library according to the fault information in the new work order information, including: weight-based priority matching: if the TV model of the new work order has been recorded in the knowledge graph, the most relevant maintenance plan is recommended by the TV model node and its relationship weight with the standard fault; default matching mechanism: if the TV model of the new work order is unknown or not recorded, maintenance suggestions are provided by matching the maintenance category and spare parts with the highest weight of the standard fault.

[0042] A television intelligent maintenance decision system, the television intelligent maintenance decision system comprising:

[0043] A work order acquisition module, used to acquire historical work order information and new work order information, wherein the historical work order information at least includes the TV model, fault description, fault cause, maintenance plan, maintenance category and spare parts used, and the new work order information at least includes fault information;

[0044] Intelligent fault library module, used to build an intelligent fault library containing standard faults based on historical work order information;

[0045] Graph database module, used to build knowledge graph based on historical work order information and intelligent fault database;

[0046] The intelligent recommendation module is used to match faults with the intelligent fault library according to the fault information in the new work order information, and recommend maintenance plans based on the matched faults based on the knowledge graph.

[0047] A computer program, when executed by a processor, implements the steps of the above-mentioned television intelligent maintenance decision-making method.

[0048] Advantages of the present invention:

[0049] Improve maintenance efficiency. By constructing an intelligent fault library and knowledge graph based on historical work order data, the present invention aims to reduce the time spent by maintenance personnel on fault diagnosis and spare parts selection, thereby shortening the maintenance cycle and improving overall maintenance efficiency.

[0050] Improve maintenance accuracy. By using large models and similarity calculation methods, the present invention can provide accurate fault matching and spare parts recommendations for new work orders, reduce maintenance errors caused by human errors, and improve the accuracy and reliability of maintenance services.

[0051] Reduce maintenance costs. Through intelligent data analysis and decision support, the present invention helps to optimize spare parts inventory management, reduce unnecessary spare parts procurement and inventory backlog, thereby reducing maintenance costs and resource waste.

[0052] Enhance user experience: the implementation of the present invention will make maintenance services faster and more accurate, improve customer satisfaction, and enhance user trust and loyalty to maintenance services.

[0053] Promoting knowledge sharing, by building a standardized fault library and knowledge graph, the present invention helps to accumulate and share maintenance knowledge and improve the technical level and service quality of the entire industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 A flowchart of a TV intelligent maintenance decision-making method according to the present invention;

[0056] Figure 2 This is a schematic diagram of the knowledge graph described in the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] Embodiment 1

[0059] like Figure 1 and Figure 2 As shown, a television intelligent maintenance decision-making method comprises the following steps:

[0060] Step S1: Obtaining historical work order information, wherein the historical work order information at least includes the TV model, fault description, fault cause, maintenance plan, maintenance category and spare parts used;

[0061] In order to obtain historical work order information, we need to understand the specific content of these work orders in detail. These historical work order information should at least include two main parts: fault information and maintenance plan. The fault information part will describe in detail the problem that occurred during the customer's use so that we can accurately identify and analyze the root cause of the problem. The maintenance plan part should contain more detailed information to ensure the smooth progress of the maintenance work.

[0062] Specifically, the maintenance plan should include at least the following key information:

[0063] 1. TV model: Clearly indicate the specific model of the TV that needs to be repaired so that the maintenance personnel can understand the specific structure and potential problems of the model and carry out the repair work in a targeted manner.

[0064] 2. Maintenance categories: Classify faults and maintenance work for easy management and statistics. For example, maintenance work can be divided into circuit problems, screen problems, software problems, etc., so as to quickly locate the problem and take corresponding maintenance measures.

[0065] 3. Maintenance spare parts: List the spare parts needed during the maintenance process in detail, including the name, model, quantity and other information of the spare parts. This will help maintenance personnel prepare the required spare parts in advance and ensure the smooth progress of maintenance work.

[0066] By obtaining this detailed historical work order information, we can better understand customer problems and develop more accurate maintenance plans, thereby improving maintenance efficiency and customer satisfaction.

[0067] Step S2: construct an intelligent fault library containing standard faults based on historical work order information;

[0068] An intelligent fault library containing standard faults is constructed based on historical work order information. The specific process is as follows:

[0069] (1) Fault sorting: Through fine-tuning of prompt word engineering, the DeepSeek model is used to sort out the fault information in historical work orders, including fault identification, removal of redundant information, and extraction of the core content of the fault. This step can separate multiple coexisting faults to ensure that the information is clear and suitable for subsequent analysis.

[0070] (2) Fault hierarchical clustering: The pre-trained BGE model is used to vectorize the sorted fault information to generate a high-dimensional fault vector. By calculating the cosine similarity between every two fault vectors, faults with high similarity are clustered together, making it easier to merge similar faults into the same category. After setting an appropriate similarity threshold, the faults are hierarchically clustered according to the score.

[0071] (3) Fault induction: Summarize the fault categories after clustering, and use the DeepSeek model again to generate standardized fault descriptions. The standard fault descriptions generated in this step will be included in the intelligent fault library to provide a standard basis for subsequent maintenance plan recommendations. Among them, induction is to summarize according to the similarities and differences of the faults, and group similar faults that are not identified by hierarchical clustering together to obtain deduplicated faults; summary is to use prompt word engineering to fine-tune the DeepSeek model and refine the repair faults in the work order into standard descriptions;

[0072] Through the above steps, the intelligent fault library constructed in this embodiment contains the cleaned, clustered and standardized fault information in the historical data, which is convenient for rapid query and analysis in the subsequent fault matching of new work orders.

[0073] The method of constructing an intelligent fault library containing standard faults according to historical work order information includes: performing fault cleaning on the fault information in the historical work order information, summarizing and forming standard fault descriptions, and constructing an intelligent fault library.

[0074] Among them, the fault cleaning of the fault information in the historical work order information is summarized to form a standard fault description, and the construction of an intelligent fault library includes: fault sorting, fault hierarchical clustering and fault summarization; the fault sorting is to sort the reported faults in the historical work orders through fine-tuning of prompt word engineering and using the DeepSeek model, including fault identification, removal of redundant information, fault extraction, and identification and separation of multiple faults; the fault hierarchical clustering is to encode the sorted faults using a pre-trained BGE model to generate a high-dimensional vector, vectorize the fault description, calculate the cosine similarity between each two fault vectors, set a distance threshold, and hierarchically cluster the faults according to the similarity score; the fault summarization is to summarize the faults of each category after clustering through fine-tuning of prompt word engineering and using the DeepSeek model to form a standard fault description, thereby building an intelligent fault library.

[0075] In the process of building an intelligent fault library containing standard faults, it is first necessary to perform a series of cleaning and sorting work on the fault information in the historical work order information. This process mainly includes cleaning and summarizing the fault information and finally forming a standard fault description, so as to build an intelligent fault library. Specifically, this process can be divided into the following steps:

[0076] First, perform fault sorting. In this step, the DeepSeek model is used to fine-tune the prompt word engineering and carefully sort out the reported faults in the historical work orders. This includes identifying the faults, removing redundant information, extracting the core fault content, and being able to identify and separate multiple simultaneous faults. Through this series of operations, the accuracy and conciseness of the fault information can be ensured.

[0077] Secondly, perform hierarchical fault clustering. In this step, the pre-trained BGE model is used to encode the sorted faults to generate high-dimensional vectors. These high-dimensional vectors can vectorize the fault description, making it easier to perform mathematical calculations. Next, the cosine similarity between each two fault vectors is calculated, and the faults are hierarchically clustered according to the similarity score by setting a suitable distance threshold. This process can classify similar faults into the same category, making the faults in the fault library more orderly and easier to manage.

[0078] Finally, perform fault summarization. In this step, the DeepSeek model is used to summarize the faults of each category after clustering through fine-tuning of the prompt word engineering. Through this process, faults of the same category can be summarized into a standard fault description, making the fault descriptions in the intelligent fault library more standardized and normalized. Finally, through these steps, an intelligent fault library containing standard faults is constructed to provide strong support for subsequent fault handling and maintenance work.

[0079] In summary, the construction of an intelligent fault library containing standard faults based on historical work order information includes: fault cleaning of the fault information in the historical work order information, summarizing to form a standard fault description, and constructing an intelligent fault library. Among them, the fault cleaning of the fault information in the historical work order information, summarizing to form a standard fault description, and constructing an intelligent fault library include: fault sorting, fault hierarchical clustering, and fault summarization; the fault sorting is to sort the repair faults in the historical work order by fine-tuning the prompt word engineering, using the DeepSeek model, including fault identification, removal of redundant information, fault extraction, and identification and separation of multiple faults; the fault hierarchical clustering is to encode the sorted faults using the pre-trained BGE model, generate a high-dimensional vector, vectorize the fault description, calculate the cosine similarity between each two fault vectors, and set the distance threshold, and hierarchically cluster the faults according to the similarity score; the fault summarization is to summarize the faults of each category after clustering by fine-tuning the prompt word engineering, using the DeepSeek model, to form a standard fault description, thereby constructing an intelligent fault library.

[0080] Step S3: Establish a knowledge graph based on historical work order information and intelligent fault library;

[0081] Based on the constructed intelligent fault database and historical work order information, a structured knowledge graph containing weight relationships is established. The specific steps are as follows:

[0082] (1) Schema definition: Define the basic structure of the knowledge graph, including nodes such as TV models, standard faults, maintenance categories, and spare parts, as well as the relationship types between nodes.

[0083] (2) Node relationship and weight setting: According to the statistical data of historical work orders, the weight is set for each node relationship, including the following:

[0084] Weight of the relationship between TV models and standard faults: Based on the common fault frequencies of different TV models, high-weighted fault types are recommended first.

[0085] Weight of the relationship between standard faults and maintenance categories: Set the weight of the best maintenance category based on the matching frequency and maintenance success rate between the fault and maintenance category.

[0086] Weight of the relationship between major maintenance categories and spare parts: Based on the historical spare parts usage rate and adaptability, high-weighted spare parts combinations are recommended to improve maintenance accuracy.

[0087] (3) Graph database storage and dynamic update: Store the knowledge graph data in the graph database and support dynamic update. Update the relationship weights regularly so that the knowledge graph maintains adaptability and accuracy to the latest fault information.

[0088] The establishment of a knowledge graph based on historical work order information and an intelligent fault library includes: by organizing and associating TV models, fault information, major repair categories and repair spare parts in historical work orders, combined with standard faults in intelligent faults, a network containing these entities and their relationships is formed to establish a structured knowledge graph and store it in a graph database.

[0089] Build a structured knowledge graph, which organizes and associates TV models, reported faults, major repair categories, repair spare parts in historical work orders, and standard faults matched from the intelligent fault library to form a network containing these entities and their relationships. The data of the knowledge graph is stored in the graph database, and the knowledge graph and graph database are dynamically updated.

[0090] The structured knowledge graph is constructed, which forms a network containing these entities and their relationships by sorting and associating TV models, reported faults, major repair categories, repair spare parts, and standard faults matched from the intelligent fault library in historical work orders, including:

[0091] Using the method in step S2 to sort out the reported faults;

[0092] The pre-trained BGE model is used to vectorize the sorted work order faults and the standard faults in the intelligent fault library, the cosine similarity between the work order faults and the standard faults is calculated, and the standard fault with the largest similarity is selected as the matching standard fault.

[0093] In practical applications, the data of the knowledge graph is stored in the graph database, including:

[0094] According to the TV model and standard fault dimension, the weights of different maintenance categories and maintenance spare parts are counted and recorded as X;

[0095] According to the standard failure dimension, the weights of different maintenance categories are calculated and recorded as Y;

[0096] The nodes of the graph database include TV models, standard faults, maintenance categories, and spare parts. The standard fault nodes represent various standard faults that a TV may encounter, the TV model nodes represent different TV models, the maintenance category nodes represent the categories of maintenance spare parts under different fault conditions, and the spare parts nodes represent the spare parts that may be needed when repairing a TV.

[0097] Build the relationship between the TV model node and the standard fault node to reflect the standard faults that have occurred in the TV model in the historical work orders;

[0098] Build a relationship between the TV model node and the maintenance category node. The relationship attributes include standard fault and weight X. When a standard fault occurs on a specific TV, the corresponding maintenance category is recommended in the order of weight.

[0099] Construct the relationship between the TV model node and the spare part node to indicate the affiliation between the spare part and the TV model;

[0100] Build the relationship between the standard fault node and the maintenance category node. The relationship attribute includes the weight Y. When the aircraft model is unknown or not recorded, the corresponding maintenance category is recommended in the order of weight.

[0101] Build the relationship between the maintenance category node and the spare parts node to indicate the affiliation between the spare parts and the maintenance category.

[0102] In order to build a structured knowledge graph, we need to first organize and analyze the historical work orders in detail. This includes extracting key information such as TV models, reported faults, repair categories, and repair spare parts, and associating and integrating this information. In this way, we can form a network that contains these entities and their relationships. In addition, we also need to match the corresponding standard faults from the intelligent fault library to further enrich and improve our knowledge graph.

[0103] This knowledge graph not only contains entity information, but also covers various relationships between entities, such as the association between TV models and reported faults, the correspondence between major repair categories and repair spare parts, etc. Through this structured representation, we can manage and retrieve relevant information more effectively, thereby improving maintenance efficiency and quality.

[0104] Storing the knowledge graph data in a graph database is a key step. Graph databases are particularly suitable for processing complex relational networks due to their efficient data storage and query capabilities. By storing the knowledge graph in a graph database, we can achieve fast data retrieval and dynamic updates.

[0105] In order to keep the knowledge graph timely and accurate, we need to update it dynamically on a regular basis. This includes adding new work order data, updating existing entity information, and correcting incorrect or outdated relationships. Through this continuous update mechanism, the knowledge graph can continuously adapt to new maintenance needs and technological developments, thereby maintaining its value and effectiveness in practical applications.

[0106] Step S4: Obtain new work order information, match faults with the intelligent fault database based on the fault information in the new work order information, and recommend maintenance plans based on the matched faults based on the knowledge graph.

[0107] To obtain new work order information, including TV model and fault description, perform the following steps:

[0108] (1) Vectorization of new work order faults: The fault description in the new work order is vectorized using the BGE model to generate the fault vector of the new work order.

[0109] (2) Standard fault matching: Calculate the cosine similarity between the new work order fault vector and the standard fault vectors in the intelligent fault library, and select the standard fault with the highest similarity.

[0110] (3) Recommend maintenance solutions based on knowledge graph: In the knowledge graph, query the associated maintenance categories and spare parts according to the matching standard fault node, and recommend the most appropriate maintenance solution based on the weight. If the TV model of the new work order is not recorded in the knowledge graph, the corresponding maintenance category and spare parts are directly recommended based on the standard fault weight.

[0111] The obtaining of new work order information, matching faults with the intelligent fault library according to the fault information in the new work order information, and recommending maintenance plans according to the matched faults based on the knowledge graph include: querying and obtaining new work order information; organizing and vectorizing the work order faults of the new work order; querying the most similar standard faults in the intelligent fault library by a cosine similarity calculation method; querying the maintenance plan corresponding to the standard fault in the knowledge graph, thereby recommending the most appropriate maintenance plan.

[0112] The fault matching is performed based on the fault information in the new work order information and the intelligent fault library, including: obtaining the work order fault in the new work order information; using the pre-trained BGE model to vectorize the obtained work order fault and the standard fault in the intelligent fault library, calculating the cosine similarity between the work order fault and the standard fault, and selecting the standard fault with the largest similarity as the matching standard fault.

[0113] By obtaining new work order information and using this information to match faults with the intelligent fault library, we can recommend corresponding maintenance plans based on the knowledge graph. The specific steps are as follows: First, we need to query and obtain new work order information, which is the basis of the entire process. Next, we will organize the fault information in the new work order and convert it into vector form for subsequent calculations and matching. In order to find the standard fault that is most similar to the work order fault, we will use the cosine similarity calculation method to query in the intelligent fault library to find the most similar fault records. Finally, in the knowledge graph, we will query the maintenance plan corresponding to the selected standard fault and recommend the most appropriate maintenance plan based on these plans.

[0114] In addition, in order to ensure the accuracy of fault matching, we will explain in detail the process of how to match faults with the intelligent fault library based on the fault information in the new work order information. First, we need to obtain the work order faults in the new work order information, which is the first step in the matching process. Next, we will use the pre-trained BGE model to vectorize the obtained work order faults and the standard faults in the intelligent fault library. Through this vectorization process, we can convert text information into numerical form for easy mathematical calculations. Then, we will calculate the cosine similarity between the work order fault and the standard fault, which is a method to measure the similarity of two vectors. Through calculation, we can find the standard fault that is most similar to the work order fault and use it as the matching result. In this way, we can ensure the accuracy and efficiency of fault matching, thereby providing a solid foundation for subsequent maintenance plan recommendations.

[0115] By carefully cleaning the repair faults in the historical work orders, we can summarize and organize a series of standard fault descriptions. These descriptions will be used to build an intelligent fault library to facilitate future fault processing and query. In this process, we focus on the key information in the historical work orders, including the TV model, the specific content of the repair fault, the main repair category, and the repair spare parts used.

[0116] In order to further improve the efficiency and accuracy of fault handling, we built a structured knowledge graph. This graph organizes and associates the data in historical work orders, organically links TV models, reported faults, major repair categories, repair parts, and standard faults matched from the intelligent fault library, forming a network that includes these entities and their relationships. This knowledge graph not only helps us better understand and analyze fault modes, but also stores related data in the graph database for dynamic update and maintenance.

[0117] When processing a new work order, we first organize the reported faults and convert them into vector form. Then, we use the cosine similarity calculation method to query the intelligent fault database for the standard fault that is most similar to the new work order reported fault. In this way, we can quickly locate the most likely cause of the fault. Subsequently, we query the graph database for the maintenance category or spare parts information corresponding to the standard fault, so as to recommend the most appropriate maintenance plan. This process not only improves maintenance efficiency, but also ensures the accuracy and reliability of the maintenance plan.

[0118] Embodiment 2

[0119] A television intelligent maintenance decision system, the television intelligent maintenance decision system comprising:

[0120] A work order acquisition module, used to acquire historical work order information and new work order information, wherein the historical work order information at least includes the TV model, fault description, fault cause, maintenance plan, maintenance category and spare parts used, and the new work order information at least includes fault information;

[0121] The main function of the work order acquisition module is to extract and retrieve historical work order information and the latest work order information. Specifically, historical work order information contains detailed fault information and corresponding maintenance plans, which are of great reference value for analyzing and solving similar problems. New work order information at least covers fault information, which helps to quickly locate and handle new problems. Through this module, users can easily obtain the required historical and new work order data, thereby improving work efficiency and problem solving efficiency.

[0122] Intelligent fault library module, used to build an intelligent fault library containing standard faults based on historical work order information;

[0123] The intelligent fault library module is designed to utilize historical work order information, analyze and organize this data, and build an intelligent fault library containing standard fault cases. This library is designed to provide technicians with a reference resource to help them quickly identify and resolve common equipment or system failures. By accumulating fault information from historical work orders, the intelligent fault library can continuously enrich and improve its content, thereby improving the efficiency and accuracy of fault handling. In addition, the intelligent fault library can also continuously optimize its fault diagnosis and solution recommendations through machine learning algorithms, making it more intelligent and efficient.

[0124] Graph database module, used to build knowledge graph based on historical work order information and intelligent fault database;

[0125] The main function of the graph database module is to use the historical work order information and the data in the intelligent fault library to build a powerful knowledge graph. This graph can effectively integrate and associate various information, thereby providing strong support for fault diagnosis and problem solving. By analyzing the data in the historical work orders, the system can identify common failure modes and solutions and store this information in the graph database. At the same time, the data in the intelligent fault library will also be integrated to further enrich the content of the graph. In this way, when a new fault occurs, the system can quickly find the relevant failure modes and solutions by querying the knowledge graph, thereby improving the efficiency and accuracy of fault handling.

[0126] The intelligent recommendation module is used to match faults with the intelligent fault library according to the fault information in the new work order information, and recommend maintenance plans based on the matched faults based on the knowledge graph.

[0127] The main function of the intelligent recommendation module is to perform in-depth matching and comparative analysis with the intelligent fault library based on the fault information contained in the new work order information. Through this process, the system can identify fault cases that match the faults described in the new work order. Subsequently, based on a powerful knowledge graph, the module can further analyze the matched fault cases and generate and recommend corresponding maintenance plans based on the relevant information of these cases. These maintenance plans not only cover the specific maintenance steps, but also include the tools and spare parts that may be needed, as well as matters that need to be paid attention to during the maintenance process, thereby providing comprehensive guidance and support for maintenance personnel. In this way, the intelligent recommendation module can significantly improve the efficiency and accuracy of fault handling, reduce maintenance time, and improve overall work efficiency.

[0128] Embodiment 3

[0129] A computer program, when executed by a processor, implements the steps of the above-mentioned television intelligent maintenance decision-making method; the steps include:

[0130] Obtaining historical work order information, wherein the historical work order information at least includes the TV model, fault description, fault cause, repair plan, repair category, and spare parts used;

[0131] Build an intelligent fault library containing standard faults based on historical work order information;

[0132] Establish a knowledge graph based on historical work order information and intelligent fault database;

[0133] Obtain new work order information, match faults with the intelligent fault database based on the fault information in the new work order information, and recommend maintenance plans based on the matched faults based on the knowledge graph.

[0134] Prior to this invention, the industry's solutions mainly relied on manual experience and technical knowledge to repair TVs, but there were problems with low repair efficiency, insufficient accuracy and high costs. This invention can significantly improve repair efficiency, increase accuracy and reduce costs through the following practices:

[0135] Build an intelligent fault library. Use big models to deeply analyze and learn fault information in historical work orders, and build a comprehensive and standardized fault library to quickly and accurately identify and classify faults in new work orders.

[0136] Establish a knowledge graph. Integrate information such as machine models, faults, maintenance categories, and maintenance spare parts in historical work orders to build a dynamically updated knowledge graph to enhance the system's ability to understand and apply maintenance knowledge.

[0137] Implement similarity calculation. For new work orders, use advanced similarity calculation methods to match the reported faults with the faults in the fault database, so as to recommend the most appropriate maintenance category and spare parts information.

[0138] Develop an intelligent recommendation system. Based on the similarity calculation results, combined with real-time inventory information and user feedback, develop an intelligent recommendation system to automatically generate the optimal maintenance plan, reduce human intervention, and improve decision-making efficiency and accuracy.

[0139] Through the above-mentioned approach, the present invention can realize the intelligence, automation and personalization of the television repair process, greatly improve the quality and efficiency of the repair service, while reducing the operating cost, and provide users with a better quality and more reliable repair experience.

[0140] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the art within the technical scope disclosed in the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A television intelligent maintenance decision-making method, characterized in that: The method comprises the following steps: Obtaining historical work order information, wherein the historical work order information at least includes the TV model, fault description, fault cause, repair plan, repair category, and spare parts used; Building an intelligent fault library containing standard faults based on historical work order information includes the following steps: Fault sorting: Through fine-tuning of prompt word engineering, the DeepSeek model is used to sort out the repair faults in historical work orders, remove redundant information and mark the core faults; Fault hierarchical clustering: Use the BGE model to vectorize the fault description, calculate the cosine similarity between fault vectors, and classify similar faults into one category based on the similarity threshold; Fault summarization: Summarize the clustered faults to form standardized fault descriptions and generate standard fault entries to build an intelligent fault library. The summarization is based on the similarities and differences of the faults, and similar faults that are not identified by hierarchical clustering are grouped together to obtain duplicate faults. The summary uses prompt word engineering to fine-tune the DeepSeek model and refine the repair faults in the work order into standard descriptions. A knowledge graph is established based on historical work order information and the intelligent fault database. A structured knowledge graph network is constructed by sorting and associating TV models, standard faults, maintenance categories, spare parts information and their relationships in historical work orders. The specific steps include: Schema definition: defines the basic structure of the graph, including entity nodes such as TV model, fault type, repair category, spare parts, and the relationship types between entities; Entity relationship generation and weight setting: Generate weights for the relationships between entity nodes based on historical data. The weights are set based on the following factors: a) Historical frequency: Count the frequency of the relationship in historical work orders. The higher the frequency, the greater the weight of the relationship. b) Relationship reliability: according to the repair success rate and spare parts usage rate, the weight of frequently successfully matched repair categories and spare parts is increased; c) Node characteristic weight: Based on the common faults of different TV models, the fault nodes with higher frequency are recommended first; Relationship Type: a) Relationship weight between TV model and standard fault: reflects the common fault types of a certain TV model, and gives priority to recommending standard faults with high weights; b) Relationship weight between standard faults and major maintenance categories: shows the optimal maintenance category under specific fault conditions; c) The weight of the relationship between maintenance categories and spare parts: Prioritize the recommendation of more suitable spare parts to ensure the accuracy and cost optimization of maintenance; Data storage and dynamic update: Store the knowledge graph data in the graph database and support dynamic update so that the weight can be updated according to the newly added work order information to maintain the real-time nature of the knowledge graph; Acquire new work order information, where the new work order information includes at least a TV model and a fault description, match the fault with the intelligent fault database according to the fault information in the new work order information, and recommend a maintenance plan based on the matched fault based on the knowledge graph, specifically including: Vectorize the fault description of the new work order, generate a fault vector using the BGE model, calculate the cosine similarity between the new work order fault vector and the standard fault vector in the intelligent fault library, and select the standard fault with the highest similarity; In the knowledge graph, based on the relationship weight of the matching standard fault node, the associated maintenance categories and spare parts are queried to recommend the optimal maintenance plan.

2. The TV intelligent maintenance decision-making method according to claim 1, characterized in that: The node definition includes: TV model node: represents a specific model of TV device; Standard fault nodes: represent standardized fault types that TV equipment may encounter; Maintenance category node: represents the maintenance category required to repair a specific fault; Spare Part Node: represents the spare part or component required during the maintenance process.

3. The TV intelligent maintenance decision-making method according to claim 2, characterized in that: The relationship types include: Relationship weight calculation: Weight of the relationship between TV model and standard fault: Determine the most common fault type for this model based on the frequency in historical data; The weight of the relationship between standard faults and maintenance categories: Based on the maintenance success rate and maintenance frequency, the maintenance category with the highest weight is recommended; Weight of the relationship between maintenance categories and spare parts: Based on the spare parts usage rate, the spare parts combination that meets the maintenance needs is recommended first; Dynamic adjustment mechanism: Relationship weights are regularly updated based on data such as the frequency of successful matching of new work orders and spare parts usage to ensure the accuracy of recommendation results.

4. The TV intelligent maintenance decision-making method according to claim 1, characterized in that: The method also includes: regularly updating the intelligent fault library; and a dynamic updating mechanism of the knowledge graph.

5. The TV intelligent maintenance decision-making method according to claim 4, characterized in that: The periodic updating of the intelligent fault library includes: according to the newly added historical work order data, following the steps of fault cleaning and fault clustering, the standard faults in the intelligent fault library are periodically updated.

6. The TV intelligent maintenance decision-making method according to claim 4, characterized in that: The dynamic update mechanism of the knowledge graph includes: adjusting the relationship weights according to the faults and maintenance records in the new work order data, so that the knowledge graph can reflect the latest maintenance data in real time, thereby optimizing the accuracy of maintenance recommendations.

7. The television intelligent maintenance decision-making method according to any one of claims 1 to 6, characterized in that: The obtaining of new work order information, wherein the new work order information includes at least a TV model and a fault description, and matching the fault with the intelligent fault library according to the fault information in the new work order information include: weight-based priority matching: if the TV model of the new work order has been recorded in the knowledge graph, the most relevant maintenance plan is recommended by priority through the TV model node and its relationship weight with the standard fault; default matching mechanism: if the TV model of the new work order is unknown or not recorded, maintenance suggestions are provided by matching the maintenance category and spare parts with the highest weight of the standard fault.

8. A television intelligent maintenance decision system, characterized in that: The television intelligent maintenance decision-making system comprises: A work order acquisition module, used to acquire historical work order information and new work order information, wherein the historical work order information at least includes the TV model, fault description, fault cause, maintenance plan, maintenance category and spare parts used, and the new work order information at least includes the TV model and fault description; Intelligent fault library module, used to build an intelligent fault library containing standard faults based on historical work order information; Graph database module, used to build knowledge graph based on historical work order information and intelligent fault database; The intelligent recommendation module is used to match faults with the intelligent fault library according to the fault information in the new work order information, and recommend maintenance plans based on the matched faults based on the knowledge graph.

9. A computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images