Method and apparatus for determining maintenance information of production equipment, and production equipment

Abstract

The application discloses a production equipment maintenance information determination method and device and a production equipment, and belongs to the technical field of production equipment. The production equipment maintenance information determination method comprises the following steps: obtaining a target maintenance instruction; the target maintenance instruction is generated under the condition that a target production equipment fault is detected, and comprises information of a fault category of the target production equipment; based on the target maintenance instruction, maintenance knowledge corresponding to the target maintenance instruction is obtained from a pre-constructed hierarchical knowledge graph; and based on the target maintenance instruction and the maintenance knowledge, maintenance information corresponding to the target maintenance instruction is obtained. The production equipment maintenance information determination method can improve the accuracy and availability of the obtained maintenance information, and is effectively applied to an industrial production environment with high precision requirements.

Description

Technical Field

[0001] This application belongs to the field of production equipment technology, and in particular relates to a method, apparatus and production equipment for determining maintenance information of production equipment. Background Technology

[0002] In practical applications, due to the large variety and quantity of industrial equipment, repairs rely heavily on the experience of maintenance personnel or consulting equipment maintenance technical documents, which is time-consuming and labor-intensive. Related technologies obtain user questions from maintenance personnel and input them into a general-purpose model to generate question-and-answer repair information. However, industrial scenarios are highly specialized, and the accuracy and usability of the repair information output by the general-purpose model are relatively low, making it difficult to apply to industrial production environments with high precision requirements. Summary of the Invention

[0003] This application aims to address at least one of the technical problems existing in the related art. To this end, this application proposes a method, apparatus, and production equipment for determining maintenance information of production equipment, which improves the accuracy and usability of the obtained maintenance information and is effectively applied in industrial production environments with high precision requirements.

[0004] Firstly, this application provides a method for determining maintenance information of production equipment, the method comprising:

[0005] Obtain the target maintenance instruction; the target maintenance instruction is generated when a fault is detected in the target production equipment, and includes information about the fault category of the target production equipment;

[0006] Based on the target maintenance instruction, maintenance knowledge corresponding to the target maintenance instruction is obtained from a pre-constructed hierarchical knowledge graph. The hierarchical knowledge graph is obtained by classifying and categorizing the sample maintenance information of various sample production equipment according to the target classification conditions, and constructing triples from the classified information. The target classification conditions include at least two of the following: basic information of the sample production equipment, component information, component failure causes, and component failure handling information.

[0007] Based on the target maintenance instruction and the maintenance knowledge, the maintenance information corresponding to the target maintenance instruction is obtained.

[0008] According to the method for determining maintenance information of production equipment in this application, the actual maintenance needs are determined by obtaining the target maintenance instruction. Based on the target maintenance instruction, maintenance knowledge corresponding to the target maintenance instruction is obtained from a hierarchical knowledge graph applicable to industrial production scenarios, thereby improving the accuracy of the obtained maintenance knowledge. Based on the maintenance knowledge and the target maintenance instruction, the maintenance information corresponding to the target maintenance instruction is determined, thereby improving the accuracy and usability of the obtained maintenance information and effectively applying it to industrial production environments with high precision requirements.

[0009] According to the method for determining maintenance information of production equipment in this application, the step of obtaining maintenance knowledge corresponding to the target maintenance instruction from a pre-constructed hierarchical knowledge graph based on the target maintenance instruction includes:

[0010] Obtain the feature values ​​included in the target maintenance instruction and the level corresponding to the feature values;

[0011] Based on the feature values, multiple first knowledge items are obtained in the layers corresponding to the feature values ​​of the pre-constructed hierarchical knowledge graph;

[0012] Based on the similarity between the feature value and each of the first knowledge, the maintenance knowledge corresponding to the target maintenance instruction is obtained from the plurality of first knowledge.

[0013] According to the method for determining maintenance information of production equipment in this application, the step of obtaining multiple first knowledge items based on the feature values ​​in the hierarchy corresponding to the feature values ​​of the pre-constructed hierarchical knowledge graph includes:

[0014] The feature values ​​are vectorized to obtain the vectorized information corresponding to the feature values;

[0015] Based on the vectorized information, the plurality of first knowledge is obtained in the layers corresponding to the feature values ​​of the pre-constructed hierarchical knowledge graph.

[0016] According to the method for determining maintenance information of production equipment in this application, before obtaining the target maintenance instruction, the method further includes:

[0017] Based on the target classification conditions, the knowledge graph level is determined, and the sample maintenance information of the various sample production equipment is classified and categorized to obtain the classified information.

[0018] Extract the sample entities, sample relationships, and sample attributes corresponding to the hierarchical and categorized information respectively to construct multiple triples;

[0019] Based on the knowledge graph hierarchy, the multiple triples are stored to obtain the hierarchical knowledge graph.

[0020] According to the method for determining maintenance information of production equipment in this application, obtaining the target maintenance instruction includes:

[0021] Receive the user's first input;

[0022] In response to the first input, the target maintenance instruction is obtained.

[0023] According to the method for determining maintenance information of production equipment in this application, before obtaining the target maintenance instruction, the method further includes:

[0024] If a fault is determined in the target production equipment based on the acquired operating data, first information is output based on the fault status of the target production equipment.

[0025] According to the method for determining maintenance information of production equipment in this application, the output of the first information includes:

[0026] Identify target users based on the maintenance personnel schedule;

[0027] The first information is sent to the user terminal corresponding to the target user.

[0028] According to the method for determining maintenance information of production equipment in this application, obtaining the maintenance information corresponding to the target maintenance instruction based on the target maintenance instruction and the maintenance knowledge includes:

[0029] The target maintenance instruction and the maintenance knowledge are input into the target large model to obtain the maintenance information output by the target large model;

[0030] The target large model is trained based on the sample maintenance data corresponding to the various sample production equipment.

[0031] According to the method for determining maintenance information of production equipment in this application, the target large model is trained based on sample maintenance data corresponding to the various sample production equipment, including:

[0032] Obtain sample maintenance data corresponding to the various sample production equipment;

[0033] The maintenance data of each sample are preprocessed to obtain a sample dataset;

[0034] If the amount of data corresponding to the sample dataset is greater than the first quantity, the target large model is trained based on the sample dataset and preset prompt information; the preset prompt information is a prompt word pre-set based on the maintenance scenario.

[0035] If the amount of data corresponding to the sample dataset is not greater than the first quantity, new sample maintenance data is obtained.

[0036] Secondly, this application provides a device for determining maintenance information of production equipment, the device comprising:

[0037] The first processing module is used to acquire a target maintenance instruction; the target maintenance instruction is generated when a fault is detected in the target production equipment, and includes information about the fault category of the target production equipment.

[0038] The second processing module is used to obtain maintenance knowledge corresponding to the target maintenance instruction from a pre-constructed hierarchical knowledge graph based on the target maintenance instruction. The hierarchical knowledge graph is obtained by classifying and categorizing the sample maintenance information of various sample production equipment according to the target classification conditions, and constructing triples from the classified information. The target classification conditions include at least two of the following: basic information of the sample production equipment, component information, component failure causes, and component failure handling information.

[0039] The third processing module is used to obtain the maintenance information corresponding to the target maintenance instruction based on the target maintenance instruction and the maintenance knowledge.

[0040] The equipment maintenance information determination device according to this application determines the actual maintenance needs by acquiring the target maintenance instruction, and obtains the maintenance knowledge corresponding to the target maintenance instruction from a hierarchical knowledge graph applicable to industrial production scenarios based on the target maintenance instruction, thereby improving the accuracy of the acquired maintenance knowledge. Based on the maintenance knowledge and the target maintenance instruction, the device determines the maintenance information corresponding to the target maintenance instruction, thereby improving the accuracy and usability of the obtained maintenance information and effectively applying it to industrial production environments with high precision requirements.

[0041] Thirdly, this application provides a production apparatus, comprising:

[0042] Based on the maintenance information determination device for production equipment as described in the second aspect.

[0043] Fourthly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the maintenance information determination method for production equipment as described in the first aspect above.

[0044] Fifthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for determining maintenance information of production equipment as described in the first aspect above.

[0045] Sixthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the method for determining maintenance information of production equipment as described in the first aspect above.

[0046] The above-described one or more technical solutions in the embodiments of this application have at least one of the following technical effects:

[0047] By acquiring the target maintenance instruction, the actual maintenance needs are determined. Based on the target maintenance instruction, the maintenance knowledge corresponding to the target maintenance instruction is obtained from a hierarchical knowledge graph applicable to industrial production scenarios, thereby improving the accuracy of the acquired maintenance knowledge. Based on the maintenance knowledge and the target maintenance instruction, the maintenance information corresponding to the target maintenance instruction is determined, improving the accuracy and usability of the obtained maintenance information, and effectively applying it to industrial production environments with high precision requirements.

[0048] Furthermore, by using target hierarchical conditions, the hierarchical knowledge graph can be effectively determined to include each level. Based on these conditions, sample maintenance information is classified and categorized to obtain hierarchically categorized information. This improves the readability of sample maintenance information, reduces the complexity of sample maintenance data, and allows for more efficient extraction of sample entities, sample relationships, and sample attributes from the hierarchically categorized information. This increases the efficiency of constructing multiple triples, stores multiple triples, and yields a hierarchical knowledge graph with clear relational context and rich knowledge. This improves the usability of the hierarchical knowledge graph and the efficiency of subsequent searches.

[0049] Furthermore, by acquiring sample maintenance data corresponding to various sample production equipment and preprocessing the sample maintenance data to obtain a sample dataset, and when the amount of data corresponding to the sample dataset is greater than the first quantity, a target large model is trained based on the sample dataset and preset prompt information. This improves the understanding ability and output accuracy of the trained target large model in the professional field, enabling the target large model to be effectively applied to industrial scenarios and ensuring that the output maintenance information conforms to the actual application scenario. The maintenance information is output in the required format, thereby improving the user experience.

[0050] Furthermore, by identifying the fault in the target production equipment, and based on the fault status, the system outputs initial information to promptly alert relevant maintenance personnel to pay attention to the target production equipment, thereby reducing losses caused by the fault.

[0051] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0052] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0053] Figure 1 This is one of the flowcharts illustrating the method for determining maintenance information of production equipment provided in the embodiments of this application;

[0054] Figure 2 This is a second flowchart illustrating the method for determining maintenance information of production equipment provided in this application embodiment;

[0055] Figure 3 This is the third flowchart illustrating the method for determining maintenance information of production equipment provided in the embodiments of this application;

[0056] Figure 4 This is the fourth flowchart illustrating the method for determining maintenance information of production equipment provided in the embodiments of this application;

[0057] Figure 5 This is the fifth flowchart illustrating the method for determining maintenance information of production equipment provided in the embodiments of this application;

[0058] Figure 6 This is a schematic diagram of the interface of the method for determining maintenance information of production equipment provided in the embodiments of this application;

[0059] Figure 7 This is the sixth flowchart illustrating the method for determining maintenance information of production equipment provided in this application embodiment;

[0060] Figure 8 This is a schematic diagram of the structure of the maintenance information device for production equipment provided in the embodiments of this application;

[0061] Figure 9 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0062] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0063] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0064] The following description, in conjunction with the accompanying drawings, details the method for determining maintenance information of production equipment, the device for determining maintenance information of production equipment, the production equipment, the electronic equipment, and the readable storage medium provided in this application, through specific embodiments and application scenarios.

[0065] The method for determining maintenance information of production equipment can be applied to the terminal, and can be executed by the hardware or software in the terminal.

[0066] The terminal includes, but is not limited to, portable communication devices such as mobile phones or tablets. It should also be understood that, in some embodiments, the terminal may not be a portable communication device, but rather a desktop computer.

[0067] The following embodiments describe a terminal including a display and a touch-sensitive surface. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, mouse, and joystick.

[0068] The maintenance information determination method for production equipment provided in this application embodiment can be executed by the production equipment, or by a maintenance information determination device for production equipment installed on the production equipment, or by a server electrically connected (wired or wirelessly connected) to the production equipment, or by a user terminal communicatively connected to the production equipment, including but not limited to mobile terminals and non-mobile terminals.

[0069] like Figure 1 As shown, the method for determining the maintenance information of the production equipment includes steps 110, 120 and 130.

[0070] It should be noted that the method for determining maintenance information of production equipment in this application can be applied to production equipment in various industrial scenarios.

[0071] Production equipment may include: CNC machine tools, handling robots, and loading trolleys, etc.

[0072] Production equipment can also be used for household appliances in practical application scenarios.

[0073] Home appliances can include: refrigeration equipment, rice cookers, and televisions, etc.

[0074] Refrigeration equipment can be understood as a broad category of refrigeration and storage equipment, including but not limited to refrigerators, freezers, display cases, beverage cabinets, wine cabinets, refrigerated display cases, and refrigerated vending machines. Refrigeration equipment has diverse structural forms and a wide range of applications.

[0075] Step 110: Obtain the target repair instruction;

[0076] In this step, the target maintenance instruction is generated when a fault is detected in the target production equipment, and includes information about the fault category of the target production equipment.

[0077] The target production equipment is any one of one or more production equipment set up in a real application scenario or a real industrial scenario.

[0078] The fault category refers to the type of fault that occurred in the target production equipment.

[0079] Fault categories can include: target production equipment not working, fan failure, compressor failure, and abnormal temperature, etc.

[0080] The specific category of the fault can be determined based on the actual situation of the production equipment, and this application does not impose any restrictions.

[0081] In actual implementation, equipment sensors can be installed in the target production equipment to monitor the operating status of the target production equipment in real time and determine whether the target production equipment can operate normally, that is, whether a fault has occurred.

[0082] In real-world applications involving multiple production devices, the device sensors can identify the target production device from among these devices by using the device name, device code, channel address, and node identifier of each device.

[0083] The equipment sensors can monitor the hydraulic press temperature, clutch air pressure, pneumatic pressure, temperature, and voltage of the target production equipment in real time.

[0084] Based on data obtained from real-time monitoring by equipment sensors, when a fault in the target production equipment is determined, the equipment sensors can also record information such as the fault time, fault code, and alarm name of the target production equipment.

[0085] In actual execution, once the operating status data collected by the equipment sensors is obtained, the operating status data can be processed through the target prediction model to determine whether the target production equipment has malfunctioned.

[0086] In the event of a malfunction in the target production equipment, a target maintenance instruction will be automatically generated.

[0087] Continue to monitor the operating status of the target production equipment in real time, provided that no malfunction has occurred.

[0088] The target prediction model can be a machine learning model or an artificial intelligence model.

[0089] For example, the target prediction model can be the extreme gradient boosting algorithm (XGBoost), the lightweight gradient boosting machine (LightGBM), and CatBoost.

[0090] The specific category of the target prediction model can be determined based on the actual situation, and this application does not impose any restrictions.

[0091] In actual execution, multiple sample operation status data can be collected in advance, and the sample operation status can be labeled to determine the fault type corresponding to each sample operation status data. The collected data is processed into the format required for the target prediction model input, and the processed data is divided into training set and test set to train the target prediction model. If the training result of the target prediction model meets the recall rate, the trained target prediction model is used to predict whether the target production equipment has failed.

[0092] The specific recall rate can be based on user-defined values ​​or determined based on actual circumstances; this application does not impose any restrictions.

[0093] like Figure 2 As shown, the data processing module can process multiple pre-collected sample running status data, the model training module can train the target prediction model, and the model evaluation module can evaluate the prediction accuracy of the target prediction model.

[0094] Step 120: Based on the target maintenance instruction, obtain the maintenance knowledge corresponding to the target maintenance instruction from the pre-constructed hierarchical knowledge graph;

[0095] In this step, the maintenance knowledge refers to maintenance knowledge related to the fault categories included in the target maintenance instruction.

[0096] For example, if the target maintenance instruction is to troubleshoot an abnormal noise from the compressor, maintenance knowledge can include one or more causes of the abnormal noise and related operations to reduce or eliminate the abnormal noise.

[0097] It should be noted that the maintenance knowledge acquired will differ depending on the target maintenance instruction.

[0098] When the target maintenance instruction includes detailed information, the maintenance knowledge acquired will be more specific.

[0099] The hierarchical knowledge graph is constructed by classifying and categorizing sample maintenance information of various sample production equipment according to target classification conditions, and then constructing triples from the classified information.

[0100] It should be noted that the hierarchical knowledge graph can be a graph database such as SPARQL or neo4j-5.13 or later.

[0101] The target classification criteria include at least two of the following: basic information of the sample production equipment, component information, cause of component failure, and component failure handling information.

[0102] The various sample production equipment are of different types.

[0103] Multiple sample production equipment can include: production equipment from different manufacturers and different types of production equipment from the same manufacturer.

[0104] Taking refrigeration equipment as an example of production equipment, various sample production equipment may include: refrigerators produced by manufacturer A, freezers produced by manufacturer A, refrigerators produced by manufacturer B, and vending machines produced by manufacturer B, etc.

[0105] The sample maintenance information includes maintenance information related to various malfunctions that may occur in the sample production equipment.

[0106] The sample repair information may include: repair knowledge, product knowledge, and industry knowledge.

[0107] Sample maintenance information can be obtained from documents such as the maintenance manual, maintenance plan, and maintenance records of the sample production equipment.

[0108] Basic information includes the basic parameters, model, and manufacturer information of the production equipment.

[0109] The component information pertains to components related to problems with the production equipment.

[0110] The cause of component failure refers to the relevant reasons that cause the component to fail.

[0111] Component failure handling information includes solutions, tools, and spare parts for dealing with failures.

[0112] Triples are constructed by extracting entities, relations, and attributes from information that has been hierarchically categorized.

[0113] In actual implementation, based on the target classification conditions, the sample maintenance information of various sample production equipment collected in advance can be divided. Then, the subject, relationship and attribute of the divided information are extracted to construct triples. The triples are stored in the knowledge graphs of each layer determined based on the target classification conditions to obtain the hierarchical knowledge graph.

[0114] After the knowledge graph is pre-built, when a fault is detected in the target production equipment, a maintenance instruction can be automatically generated, and maintenance knowledge can be retrieved from the hierarchical knowledge graph based on the maintenance instruction.

[0115] Step 130: Based on the target maintenance instruction and maintenance knowledge, obtain the maintenance information corresponding to the target maintenance instruction.

[0116] In this step, after obtaining the target maintenance instructions and maintenance knowledge, the maintenance knowledge can be further filtered based on the target maintenance instructions to remove irrelevant or duplicate knowledge.

[0117] Continuing with the example of the target repair instruction being "How to troubleshoot an abnormal noise from the compressor," the retrieved repair knowledge might include repair knowledge related to abnormal noise from the compressors of refrigerators, freezers, and vending machines.

[0118] The target maintenance instruction requires compressor-related maintenance knowledge. Considering that the maintenance methods for compressors of refrigeration equipment may be the same, the maintenance knowledge related to various types of abnormal compressor noises can be integrated to further streamline the acquired maintenance knowledge, and the streamlined maintenance knowledge can be used as maintenance information.

[0119] In other embodiments, after obtaining maintenance knowledge, the maintenance knowledge can be further organized, for example, by developing a detailed maintenance plan based on the maintenance knowledge.

[0120] A repair plan may include: repair steps, the time required for the repair process, and the necessary repair tools.

[0121] The established maintenance plan defines maintenance information so that maintenance personnel can use their maintenance knowledge to repair the target production equipment.

[0122] During the research and development process, the inventors discovered that in related technologies, user questions are input into a general large model by obtaining questions from maintenance personnel to obtain question-and-answer maintenance information. However, industrial scenarios are highly specialized, and the accuracy and usability of the maintenance information output by the general large model are low, making it difficult to apply to industrial production environments with high precision requirements.

[0123] This application obtains the target maintenance instruction to accurately identify the actual maintenance needs. Based on the target maintenance instruction, it retrieves the corresponding maintenance knowledge from a pre-constructed hierarchical knowledge graph. Since the hierarchical knowledge graph is constructed based on sample maintenance information from sample production equipment, it is suitable for industrial scenarios. Furthermore, the knowledge stored in the hierarchical knowledge graph has clear dependencies and stronger semantic understanding and expression capabilities, which can improve the accuracy of the acquired maintenance knowledge. Combining the target maintenance instruction with accurate maintenance knowledge, more precise maintenance information can be obtained, improving the usability of maintenance information and effectively applying it to industrial production environments with high precision requirements.

[0124] According to the method for determining maintenance information of production equipment provided in the embodiments of this application, the actual maintenance needs are determined by obtaining the target maintenance instruction. Based on the target maintenance instruction, maintenance knowledge corresponding to the target maintenance instruction is obtained from a hierarchical knowledge graph applicable to industrial production scenarios, thereby improving the accuracy of the obtained maintenance knowledge. Based on the maintenance knowledge and the target maintenance instruction, the maintenance information corresponding to the target maintenance instruction is determined, thereby improving the accuracy and usability of the obtained maintenance information and effectively applying it to industrial production environments with high precision requirements.

[0125] In some embodiments, step 120 may further include:

[0126] Obtain the feature values ​​included in the target maintenance instruction and the corresponding levels of the feature values;

[0127] Based on feature values, multiple first knowledge points are obtained in the layers corresponding to the feature values ​​of a pre-constructed hierarchical knowledge graph;

[0128] Based on the similarity between the feature value and each piece of first knowledge, the maintenance knowledge corresponding to the target maintenance instruction is obtained from multiple pieces of first knowledge.

[0129] In this embodiment, the feature values ​​are the keywords, entity names, and attributes included in the target maintenance instruction.

[0130] The level corresponding to the feature value is the level in the hierarchical knowledge graph that stores knowledge related to the feature value.

[0131] In actual execution, the target maintenance instruction can be parsed using a natural language processing model to identify keywords, entity names, or attributes in the target maintenance instruction. Then, based on the feature values, each level in the hierarchical knowledge graph can be matched to obtain the level corresponding to the feature values.

[0132] In determining the level corresponding to the feature value, the query scope can be expanded based on the hierarchical relationship of the hierarchical knowledge graph to determine the query path, that is, to obtain the level corresponding to the feature value.

[0133] For example, when querying "compressor repair methods", one can query the layer above storing repair knowledge for the cause of compressor failure.

[0134] The first knowledge is the knowledge related to the target maintenance instruction that is initially retrieved from the hierarchy corresponding to the feature value.

[0135] In actual execution, after determining the feature value and the corresponding level, the relevant knowledge can be retrieved at each level through query statements to obtain multiple primary knowledge.

[0136] It should be noted that there may be a large number of primary knowledge items. In this case, the primary knowledge items can be filtered by the similarity between the feature values ​​and each primary knowledge item, and the primary knowledge items with higher similarity can be retained to obtain the maintenance knowledge.

[0137] In actual implementation, if the similarity between the feature value and the first knowledge is greater than the similarity threshold, the first knowledge can be identified as maintenance knowledge.

[0138] If the similarity between the feature value and the first knowledge is not greater than the similarity threshold, the first knowledge can be removed and not used as maintenance knowledge.

[0139] The specific value of the similarity threshold can be based on user-defined criteria or determined based on actual circumstances; this application does not impose any restrictions.

[0140] According to the method for determining maintenance information of production equipment provided in the embodiments of this application, by obtaining the feature value of the target maintenance instruction and the level corresponding to the feature value, the keywords in the target maintenance instruction are effectively determined. Based on the feature value, multiple first knowledge are retrieved in the level corresponding to the feature value. Based on the similarity between the feature value and the first knowledge, the first knowledge with higher similarity is selected from the multiple first knowledge as maintenance knowledge, thereby improving the accuracy of the obtained maintenance knowledge.

[0141] In some embodiments, obtaining multiple first knowledge points from the levels corresponding to the feature values ​​of a pre-constructed hierarchical knowledge graph based on feature values ​​may further include:

[0142] The eigenvalues ​​are vectorized to obtain the vectorized information corresponding to the eigenvalues;

[0143] Based on vectorized information, multiple first knowledge points are obtained from the layers corresponding to the feature values ​​of a pre-constructed hierarchical knowledge graph.

[0144] In this embodiment, vectorization processing involves converting textual information into a numerical form that a computer can process.

[0145] Vectorized information is expressed as the numerical form of feature values.

[0146] In actual implementation, the feature values ​​can be vectorized using a vector embedding model to obtain the vectorized information corresponding to the feature values.

[0147] Vector embedding models can include M3E (Moka Massive Mixed Embedding) Chinese vector model, BGE (BAAI General Embedding) series models, and Text Embedding models, among others.

[0148] It should be noted that fine-tuning the vector embedding model using sample maintenance information from industrial scenarios can improve the model's understanding capabilities in specialized fields.

[0149] Before deploying the vector embedding model, various sample maintenance data can be collected in advance. Based on the sample maintenance data, sample maintenance instructions can be generated to obtain a sample training set. The vector embedding model can be fine-tuned using the sample training set and the flagembedding library. After the fine-tuned vector embedding model is evaluated on the test set, the vector embedding model can be deployed.

[0150] like Figure 3 As shown, in the actual execution process, the pre-collected sample maintenance data can be processed through the data processing module, such as data cleaning and format standardization of the sample maintenance data. The sample maintenance instructions can be generated through the data generation module to obtain the sample training set. The vector embedding model can be fine-tuned through the model fine-tuning module. The test set can be tested through the model evaluation module to test the effect of the fine-tuned vector embedding model.

[0151] It should be noted that fine-tuning the vector embedding model using various sample maintenance data can improve the accuracy of subsequent retrieval of maintenance knowledge in professional fields based on target maintenance instructions.

[0152] In actual execution, after obtaining the feature values, the feature values ​​can be vectorized using a vector embedding model to convert them into vector representations and obtain vectorized information. Then, based on the vectorized information, multiple first knowledge points can be obtained from the layers corresponding to the feature values ​​of the pre-constructed hierarchical knowledge graph.

[0153] According to the method for determining maintenance information of production equipment provided in the embodiments of this application, vectorized information is obtained by vectorizing feature values, and multiple first knowledge items are retrieved in the hierarchical knowledge graph based on the vectorized information, which effectively improves the accuracy and efficiency of knowledge retrieval.

[0154] In some embodiments, the method may further include the following prior to step 110:

[0155] Based on the target classification conditions, the knowledge graph hierarchy is determined, and the sample maintenance information of various sample production equipment is classified and categorized to obtain the classified information.

[0156] Extract the sample entities, sample relationships, and sample attributes corresponding to the hierarchical and categorized information respectively to construct multiple triples;

[0157] Based on the knowledge graph hierarchy, multiple triples are stored to obtain a hierarchical knowledge graph.

[0158] In this embodiment, the information after hierarchical classification is the information obtained after dividing the sample maintenance information.

[0159] For example, if the sample maintenance information is the instruction manual of the sample production equipment, the target classification conditions can be used as the classification criteria to divide the information in the manual into multiple parts. For example, the content of the manual that records the basic parameters, model and manufacturer information of the sample production equipment can be divided into basic information, and the specific categories of the multiple components actually included in the sample production equipment can be divided into component information, etc.

[0160] The sample entities are entities in the information after hierarchical classification.

[0161] For example, entities can include production equipment, such as refrigeration equipment, inverters, and loading trolleys.

[0162] Sample relationships refer to the relationships between entities in the hierarchical and categorized information.

[0163] For example, sample relationships can include: production equipment - including - components and components - related causes - failure causes, etc.

[0164] Sample attributes are the attributes included in the entities within the hierarchical and categorized information.

[0165] For example, the attributes of a component may include: name, location, and function.

[0166] In actual implementation, the hierarchical types included in the target hierarchical conditions can be determined as multiple levels that can be set in the hierarchical knowledge graph, thereby determining the knowledge graph hierarchy and constructing the schema of the hierarchical knowledge graph to ensure the consistency and comparability of data in the hierarchical knowledge graph.

[0167] Furthermore, based on target hierarchical conditions, the sample maintenance information of various sample production equipment can be classified and categorized to obtain hierarchical information. Knowledge extraction can then be performed on the hierarchical information to obtain sample entities, sample relationships, and sample attributes. The extracted sample entities, sample relationships, and sample attributes can be combined to construct triples, which are then stored in the corresponding levels to obtain a hierarchical knowledge graph.

[0168] In some embodiments, the extracted sample entities, sample relationships, and sample attributes can be vectorized, and the vectorized information can be stored in the corresponding level to obtain a hierarchical knowledge graph.

[0169] like Figure 4 As shown, in actual execution, the entity, attribute, relationship and hierarchy in the hierarchical knowledge graph can be defined through the graph schema construction module, and the ontology architecture of the hierarchical knowledge graph can be determined.

[0170] The data processing module extracts information from the content of different layers.

[0171] It should be noted that the data processing module can encapsulate large models to extract information from different layers, so that the extracted information is no longer simply extracted production equipment or component entities.

[0172] The vectorization module vectorizes the sample entities, sample relationships, and sample attributes corresponding to the extracted hierarchical and categorized information to obtain sample vectorized information.

[0173] Sample vectorization information is the information obtained after vectorizing the extracted knowledge.

[0174] The graph storage module stores the vectorized information to the corresponding level according to the knowledge graph hierarchy.

[0175] It should be noted that new sample maintenance information can be collected and added to the hierarchical knowledge graph to update it, so that the maintenance knowledge retrieved based on the target maintenance instruction is the most up-to-date knowledge.

[0176] According to the method for determining maintenance information of production equipment provided in the embodiments of this application, the hierarchical knowledge graph can be effectively determined by target hierarchical conditions, and the sample maintenance information is classified and categorized based on the target hierarchical conditions to obtain hierarchically categorized information. This improves the readability of sample maintenance information, reduces the complexity of sample maintenance data, and thus more efficiently extracts sample entities, sample relationships, and sample attributes from the hierarchically categorized information. This improves the efficiency of constructing multiple triples, stores multiple triples, and obtains a hierarchical knowledge graph with clear relationship context and rich knowledge, thereby improving the usability of the hierarchical knowledge graph and subsequent search efficiency.

[0177] In some embodiments, prior to step 110, the method may further include:

[0178] Based on the acquired operating data of the target production equipment, if a fault is determined in the target production equipment, the first information is output based on the fault condition of the target production equipment.

[0179] In this embodiment, the operational data refers to data related to the detection of the working status of the production equipment.

[0180] Operating data may include: voltage, current, temperature, pneumatic pressure, and clutch air pressure, etc.

[0181] It should be noted that the operating data required to detect whether a target production device has malfunctioned may be the same or different for different target production equipment.

[0182] The first message is used to alert relevant users that the target production equipment has malfunctioned.

[0183] The first piece of information may include: information describing the type of fault and a smart chat link for querying the type of fault.

[0184] The information describing the fault type is related to the actual fault condition of the target production equipment.

[0185] like Figure 6 As shown, the information describing the fault type may include: fault alarm time, number of consecutive alarms, alarm name, fault scene phenomenon, and intelligent dialogue link, etc.

[0186] In actual execution, the operating data of the target production equipment can be collected in real time through equipment sensors. The equipment sensors can send the collected operating data to the detection model, and the detection model analyzes the operating data to determine whether the target production equipment has malfunctioned.

[0187] The detection model can be a neural network model, a machine learning model, or a large industrial model, etc.

[0188] When the detection model determines that the target production equipment is faulty based on the operating data, it can output the first message to promptly notify the relevant users that the target production equipment has failed.

[0189] For example, if the target production equipment is a refrigeration unit and the fault is an abnormal noise from the compressor, the system can output a fault in the compressor of the refrigeration unit numbered 10 in area A of the cold storage warehouse.

[0190] like Figure 5 As shown, taking home appliance manufacturing equipment as an example, operational data can be collected through equipment sensor devices, and the fault prediction module can predict whether the home appliance manufacturing equipment will malfunction.

[0191] The fault prediction module can encapsulate fault prediction models.

[0192] In actual implementation, historical operating data of various sample production equipment can be collected in advance. The historical operating data can be used as samples, and the corresponding operating results can be used as labels to train the fault prediction model.

[0193] According to the maintenance information determination method for production equipment provided in the embodiments of this application, by determining the fault of the target production equipment, the first information is automatically output based on the fault status of the target production equipment, and relevant maintenance personnel are promptly reminded to pay attention to the target production equipment, thereby reducing the losses caused by the fault of the target production equipment.

[0194] In some embodiments, outputting the first information may further include:

[0195] Identify target users based on the maintenance personnel schedule;

[0196] Send the first message to the client corresponding to the target user.

[0197] In this embodiment, the maintenance personnel schedule is a pre-set schedule based on actual work conditions.

[0198] The maintenance personnel schedule can be user-defined or determined based on actual circumstances; this application does not impose any restrictions.

[0199] The maintenance personnel schedule may include information such as name, phone number, and length of service.

[0200] It should be noted that the maintenance personnel schedule can be updated regularly based on actual circumstances.

[0201] The target users are those who need to repair the target production equipment on the day the malfunction is confirmed.

[0202] In actual operation, after the target production equipment malfunction is identified, the maintenance personnel schedule can be automatically queried to determine the maintenance worker on duty that day, i.e. the target user, and then the first information can be sent to the user terminal corresponding to the target user.

[0203] According to the method for determining maintenance information of production equipment provided in the embodiments of this application, the target user is effectively identified through the maintenance personnel schedule, thereby automatically sending the first information to the user terminal corresponding to the target user, so as to promptly remind relevant personnel to maintain the target production equipment, reducing the losses caused by incorrect notification to the user or notification to multiple users.

[0204] In some embodiments, outputting the first information may include:

[0205] The first information is sent to the user terminal through at least one of voice broadcasting, text output, and image output.

[0206] In this embodiment, voice broadcasting outputs information via voice.

[0207] Text output outputs information in text format.

[0208] Image output is the output of information in the form of an image.

[0209] For example, image output can output an image of the location of the fault in the target production equipment.

[0210] The user terminal is the terminal held by maintenance personnel who need to receive the first information.

[0211] Continue to refer to Figure 5 In actual execution, the fault warning module can send the first information to the user terminal.

[0212] The user terminal can connect to the enterprise's office system, such as the enterprise's email and work groups, and send fault alarm information, fault images, and intelligent dialogue links to the user terminal's work groups.

[0213] In addition, the API for making phone calls can be called to make calls to the user's end and output the first information through voice broadcast.

[0214] According to the method for determining maintenance information of production equipment provided in the embodiments of this application, the first information is output through one or more of the following methods: voice broadcast, text output, and image output. This ensures that the user corresponding to the user terminal can receive the first information in a timely manner, reduces the risk that the user corresponding to the user terminal may not receive the first information in a timely manner, and provides multiple safeguards.

[0215] In some embodiments, step 110 may further include:

[0216] Receive the user's first input;

[0217] In response to the first input, obtain the target maintenance instruction.

[0218] In this embodiment, the first input user inputs the target repair command.

[0219] The first input can be in at least one of the following ways:

[0220] Firstly, the first input can be a touch operation, including but not limited to click, swipe, and press operations.

[0221] In this embodiment, receiving the user's first input can be receiving the user's touch operation on the display area of ​​the terminal screen.

[0222] To reduce the rate of user error, the effective area of ​​the first input can be limited to a specific area, such as the upper middle area of ​​the maintenance query interface; or, while the maintenance query interface is displayed, a target control can be displayed on the current interface, and touching the target control will enable the first input; or the first input can be set to a series of taps on the display area within a target time interval.

[0223] Secondly, the first input can be a physical button input.

[0224] In this embodiment, the terminal is equipped with physical buttons on its body corresponding to querying maintenance information. The first input received by the user can be the operation of the user pressing the corresponding physical button; the first input can also be a combination operation of pressing multiple physical buttons simultaneously.

[0225] Thirdly, the first input can be voice input.

[0226] In this implementation, the terminal can trigger the display of a repair information query interface when it receives a voice message such as "How to repair and maintain a compressor malfunction".

[0227] Of course, in other embodiments, the first input may also be in other forms, including but not limited to character input, etc., which can be determined according to actual needs, and this application embodiment does not limit it.

[0228] Continue to refer to Figure 5 When the target user receives the first information through the user terminal, and the first information includes a smart dialogue link, the target user can click the smart dialogue link to enter the question and answer interface and directly consult about the problem of the faulty equipment. When the maintenance worker inputs the question (i.e. the target maintenance instruction), the input question will be converted into a vector representation through the vector embedding module, and then the corresponding equipment maintenance knowledge will be retrieved from the equipment maintenance knowledge graph (i.e. the hierarchical knowledge graph).

[0229] Target users can interact with the device through the intelligent maintenance module to ask and answer questions and input their desired maintenance commands.

[0230] In actual execution, when the target user determines that the target production equipment is faulty, the target user can input the information they want to query into the user terminal. The user terminal receives the user's first input and responds to the first input to obtain the target maintenance instruction.

[0231] According to the method for determining maintenance information of production equipment provided in the embodiments of this application, by receiving the user's first input, the target maintenance instruction can be effectively obtained, and the user's query needs can be processed in real time when the user needs to query maintenance information, thereby improving the user experience.

[0232] In some embodiments, step 130 may further include:

[0233] Input the target maintenance instructions and maintenance knowledge into the target large model, and obtain the maintenance information output by the target large model;

[0234] The target large model is trained based on sample maintenance data corresponding to various sample production equipment.

[0235] In this embodiment, the target large model is a model used to integrate and analyze target maintenance instructions and maintenance knowledge in order to output more accurate maintenance information.

[0236] like Figure 5 As shown, in actual execution, when a fault is detected in the target production equipment, the operating data of the target production equipment and the prompt words that limit the output operating data format can be automatically sent to the target big model. The target big model then sends the first information to the target user. After receiving a phone notification or seeing a message in the work group, the target user can view the specific content of the first information in the work group. Based on the first information, the user can determine the problem to be queried (i.e., the target maintenance instruction), input the problem to be queried into the intelligent dialogue link, and conduct maintenance interaction.

[0237] The target large model has the ability to call tools, including API interfaces for making phone calls and sending messages.

[0238] Intelligent dialogue links can encapsulate hierarchical knowledge graphs and target large models.

[0239] After receiving a target maintenance instruction, the intelligent dialogue connection first queries maintenance knowledge in the hierarchical knowledge graph. Then, it inputs the maintenance knowledge and the target maintenance instruction into the target big model. The target big model can use the maintenance knowledge as a reference answer to generate maintenance information, thereby improving the accuracy of the generated maintenance information.

[0240] The target large model can comprehensively process various types of information.

[0241] The target large model can output maintenance knowledge that matches the actual application scenario, such as outputting maintenance documents.

[0242] Repair documentation may include information such as the tools required, estimated repair time, and repair methods.

[0243] In actual implementation, from equipment fault detection to the ability to call the target big model tool to realize alarm notification of target production equipment faults, and then to intelligently recommend equipment maintenance solutions in an interactive manner, the target big model intelligently generates maintenance reports for the maintenance process, forming a closed-loop control.

[0244] According to the method for determining maintenance information of production equipment provided in the embodiments of this application, the target maintenance instructions and maintenance knowledge are processed by the target large model. Based on the processing capabilities of the target large model itself, the maintenance knowledge is further combined to effectively improve the accuracy, reliability, interpretability and efficiency of the output maintenance information and the processing efficiency of maintenance knowledge.

[0245] In some embodiments, the target large model is trained based on sample maintenance data corresponding to multiple sample production equipment, and may further include:

[0246] Obtain sample maintenance data corresponding to various sample production equipment;

[0247] The maintenance data of each sample are preprocessed to obtain the sample dataset;

[0248] When the amount of data corresponding to the sample dataset is greater than the first quantity, train the target large model based on the sample dataset and the preset prompt information;

[0249] If the amount of data in the sample dataset is no greater than the first quantity, new sample maintenance data is obtained.

[0250] In this embodiment, the sample repair data consists of documents related to the production equipment used for repairing the samples.

[0251] Sample repair data may include documents such as repair manuals, repair plans, and repair records.

[0252] Data preprocessing can include operations such as data cleaning and data format standardization.

[0253] In actual implementation, the data format of sample maintenance data can be uniformly processed into txt text format.

[0254] The choice of data format can be based on user-defined criteria or determined based on actual circumstances; this application does not impose any restrictions.

[0255] The sample dataset is the training set used to train the target large model.

[0256] The first quantity is a preset value used to determine the size of the data in the sample dataset.

[0257] The specific value of the first quantity can be determined based on the actual situation, and this application does not impose any restrictions.

[0258] The preset prompts are prompts pre-set based on the maintenance scenario.

[0259] Preset prompts may include: fault time, fault cause, fault scenario, repair method, and repair tools.

[0260] The specific content of the preset prompt message can be based on user customization or determined based on the actual situation; this application does not impose any restrictions.

[0261] In actual execution, sample maintenance data can be continuously collected, and the sample maintenance data can be preprocessed to obtain a sample dataset. The size relationship between the data volume in the sample dataset and the first quantity can be compared to determine whether the currently acquired sample dataset can be used to train the target large model.

[0262] If the amount of data corresponding to the sample dataset is no greater than the first quantity, new sample maintenance data needs to be obtained to increase the amount of data corresponding to the sample dataset.

[0263] If the amount of data corresponding to the sample dataset is greater than the first quantity, the target large model can be trained based on the sample dataset and preset prompt information, and then the target large model can be fine-tuned to make it a vertical industrial large model that meets the needs of production equipment in industrial scenarios.

[0264] In actual implementation, the evaluation results of the target large model can be compared with the first evaluation threshold to determine whether the target large model meets the needs of industrial scenarios.

[0265] The evaluation results can be obtained based on the predicted maintenance information output by the target large model and the sample maintenance information corresponding to the sample maintenance instructions.

[0266] Deploy the target large model if it meets the needs of industrial scenarios.

[0267] like Figure 7 As shown, in actual execution, the data processing module can be used to preprocess the sample maintenance data.

[0268] It should be noted that different document processing modules can be developed based on commonly used document formats of enterprises to preprocess sample maintenance data.

[0269] The fine-tuning module can be activated to determine whether the amount of data in the training set is greater than the first quantity. If the amount of data in the training set is greater than the first quantity, the large model fine-tuning can be activated.

[0270] The large model fine-tuning module can use LoRa technology to fine-tune the target large model.

[0271] Of course, in actual implementation, the target large model can be fine-tuned in any other feasible way, such as vector-based random matrix adaptation (Vera) and prefix tuning, etc., which are not limited in this application.

[0272] The large model evaluation module can evaluate the target large model by comparing the predicted maintenance information output by the target large model with the sample maintenance information corresponding to the sample maintenance instructions, and then scoring the results.

[0273] The deployment module can compare the evaluation results of the target large model with the first evaluation threshold, and deploy the model if the evaluation reaches the first evaluation threshold.

[0274] The model update and deployment module can fine-tune the target large model based on continuously collected new sample datasets and deploy the updated target large model.

[0275] In actual implementation, after the target user completes the maintenance of the target production equipment based on the maintenance information, the target big model can generate a maintenance report based on the interaction records between the target user and the intelligent dialogue link, and use the maintenance report as new sample maintenance data to update the target big model and hierarchical knowledge graph.

[0276] It should be noted that the training process of the target large model can be achieved with a high degree of automation. During the training process, based on the accumulation of collected sample maintenance data and the evaluation of model performance, the need for fine-tuning and deployment can be controlled by starting the fine-tuning module and the deployment module, thus meeting the requirements for low-code automatic updates and continuously strengthening the model's capabilities.

[0277] According to the method for determining maintenance information of production equipment provided in the embodiments of this application, sample maintenance data corresponding to various sample production equipment is obtained, and the sample maintenance data is preprocessed to obtain a sample dataset. When the amount of data corresponding to the sample dataset is greater than a first quantity, a target large model is trained based on the sample dataset and preset prompt information to improve the understanding ability and output accuracy of the trained target large model in the professional field, so that the target large model can be effectively applied to industrial scenarios, and the output maintenance information can conform to the actual application scenario and output maintenance information in the required format to improve the user experience.

[0278] In some embodiments, data preprocessing is performed on each sample maintenance data to obtain a sample dataset, which may further include:

[0279] Based on the sample maintenance data corresponding to the sample production equipment, obtain various sample maintenance instructions and sample maintenance information corresponding to each sample maintenance instruction;

[0280] The sample maintenance instruction corresponding to the sample production equipment, and the sample maintenance information corresponding to the sample target maintenance instruction, are determined as a training sample corresponding to the sample production equipment.

[0281] The sample dataset is obtained based on multiple training samples corresponding to each sample production equipment.

[0282] In this embodiment, the sample repair instruction is an instruction corresponding to the repair information in the sample repair data.

[0283] The sample repair information is the repair information that matches the sample repair instructions.

[0284] Sample repair information can be obtained from sample repair data.

[0285] In actual execution, after receiving the sample maintenance instructions corresponding to the sample production equipment, the sample maintenance data can be segmented, each segment of sample maintenance data can be vectorized, and the vectorized data can be stored in a vector database.

[0286] Each segment of sample maintenance data is sent to the large model to generate various sample maintenance instructions. Then, the generated problems are used one by one to generate sample maintenance information using knowledge retrieval enhancement technology, so as to generate the sample training set required for fine-tuning the target large model and improve the quality of maintenance information output by the target large model.

[0287] The large model can be a general large model or an industrial large model that has been fine-tuned based on sample maintenance data.

[0288] The following section uses the alarm repair solution for the motor inverter of the automatic door frame feeding trolley as a sample repair data example to illustrate the sample repair data processing process.

[0289] Sample repair data may include the following:

[0290] 1. Check the trolley track and wheels: Check for foreign objects in the track / whether the wheels are on the track / whether the wheel axles are broken or misaligned. If there is any damage, the wheels need to be replaced and adjusted and the track needs to be cleaned. Routine maintenance standard: The trolley guide rails / wheels need to be cleaned / inspected / lubricated every week.

[0291] 2. Input Voltage: Check if the input voltage is within the normal range and if there is any voltage difference / phase loss. If there is any damage, the equipment input line and circuit breaker need to be replaced. Routine maintenance standard: Check the energy usage of the equipment weekly. Then check the usage of the frequency converter: Check if the output voltage is within the normal range and if the frequency converter is damaged. If there is any damage, the frequency converter needs to be replaced. Routine maintenance standard: Clean the dust from the distribution cabinet every quarter.

[0292] 3. Check for circuit and cable faults: When the trolley is running, the movement of the tank chain causes wear on the motor cable. Short circuits, phase loss, or leakage during operation will trigger an alarm from the frequency converter. Determine if there is damage by measuring the continuity and grounding of the circuit. If there is damage, the cable needs to be replaced. Routine maintenance standard: Check the trolley cable every six months and replace any damaged cables.

[0293] 4. Check if the drive motor is faulty: Check if the three-phase resistance of the drive motor is balanced and grounded, and check if the motor is stuck and not turning. If there is any abnormality, the damaged motor needs to be replaced. Routine maintenance standard: Lubricate the motor lubrication port every year.

[0294] Sending the above sample repair data to the large model will yield the following sample repair instructions and corresponding sample repair information:

[0295] {“Sample Maintenance Instruction”: “The inverter motor of the automatic door frame feeding trolley is alarming. What should be done first?”, “Sample Maintenance Information”: “First, check the trolley track and wheels. Check for foreign objects in the track / whether the wheels are on the track / whether the wheel axles are broken or misaligned. If there is any damage, the wheels need to be replaced, adjusted, and the track cleaned.”, “Reference Original Content”: “First, check the trolley track and wheels: check for foreign objects in the track / whether the wheels are on the track / whether the wheel axles are broken or misaligned. If there is any damage, the wheels need to be replaced, adjusted, and the track cleaned (Routine maintenance standard: the feeding trolley guide rails / wheels need to be cleaned / inspected / lubricated weekly)”}.

[0296] {“Sample Maintenance Instruction”: “How to check the input voltage?”, “Sample Maintenance Information”: “Check if the input voltage is within the normal range, and whether there is a voltage difference / phase loss. If there is damage, the equipment input line and circuit breaker need to be replaced.”, “Reference Original Content”: “Check Input Voltage: Check if the input voltage is within the normal range, and whether there is a voltage difference / phase loss. If there is damage, the equipment input line and circuit breaker need to be replaced (Routine Maintenance Standard: Weekly inspection of equipment energy usage)”}.

[0297] {“Sample Maintenance Instruction”: “How to check the condition of the frequency converter?”, “Sample Maintenance Information”: “Check if the output voltage is within the normal range, and whether the frequency converter is damaged. If damaged, the frequency converter needs to be replaced.”, “Reference Original Content”: “Check the condition of the frequency converter: Check if the output voltage is within the normal range, and whether the frequency converter is damaged. If damaged, the frequency converter needs to be replaced (Routine maintenance standard: Clean the dust from the distribution cabinet every quarter)”}.

[0298] {“Sample Repair Instruction”: “How to handle circuit cable faults?”, “Sample Repair Information”: “Determine if there is damage by measuring the continuity / grounding of the line. If there is damage, the cable needs to be replaced.”, “Reference Original Content”: “Check for circuit cable faults: When the trolley is running, the movement of the tank chain causes wear on the motor cable. Short circuits / phase loss / leakage during operation will cause the inverter to alarm. Determine if there is damage by measuring the continuity / grounding of the line. If there is damage, the cable needs to be replaced (routine maintenance standard: inspect the trolley cable every six months and replace the damaged cable)”}

[0299] {“Sample Repair Instructions”: “How to check for drive motor faults?”, “Sample Repair Information”: “Check if the three-phase resistance of the drive motor is balanced and grounded; check if the motor is stuck and not turning; if there are any abnormalities, the damaged motor needs to be replaced”, “Reference Original Content”: “Check if the drive motor is faulty: Check if the three-phase resistance of the drive motor is balanced and grounded; check if the motor is stuck and not turning; if there are any abnormalities, the damaged motor needs to be replaced (Routine maintenance standard: Lubricate the motor lubrication port annually)”}.

[0300] In actual implementation, the sample maintenance instructions corresponding to the sample production equipment obtained through the above method, as well as the sample maintenance information corresponding to the sample target maintenance instructions, can be identified as a training sample corresponding to the sample production equipment, and the training sample can be stored.

[0301] After obtaining the training samples, they can be evaluated to determine the data quality of the training samples.

[0302] If the evaluation result corresponding to the training data is greater than the second evaluation threshold, the training sample is stored. By storing multiple training samples corresponding to each sample production device, a sample dataset is obtained.

[0303] The specific value of the second evaluation threshold can be based on user-defined criteria or determined based on actual circumstances; this application does not impose any restrictions.

[0304] Continue to refer to Figure 7 In actual execution, the data generation module can segment the sample maintenance data after data preprocessing, then vectorize it and store it in the vector database. Based on the vectorized data, sample maintenance instructions and corresponding sample maintenance information can be generated.

[0305] The data quality of the generated sample repair instructions and the corresponding sample repair information is evaluated using the data quality assessment module.

[0306] The sample repair instruction and the corresponding sample repair information can be considered as a question-and-answer pair.

[0307] It should be noted that the data quality assessment module can use RAGAS technology to evaluate the quality of the generated question-and-answer pairs.

[0308] Of course, in actual implementation, question-answer pairs can be evaluated in any other feasible way, and this application does not limit this.

[0309] The training samples are stored through the data storage module to obtain the sample data volume.

[0310] The data storage module can store training samples in the format required for training the target large model and store them in the database.

[0311] According to the method for determining maintenance information of production equipment provided in the application embodiment, multiple sample maintenance instructions and sample maintenance information corresponding to each sample maintenance instruction are obtained through sample maintenance data corresponding to the sample production equipment. Question-answer pairs matching the sample maintenance instructions and sample maintenance information are effectively obtained, thereby storing multiple question-answer pairs corresponding to each sample production equipment to obtain a sample training set, improving the data instructions of the constructed sample training set, and thus improving the training effect of the target large model.

[0312] The method for determining maintenance information of production equipment provided in this application can be executed by a device for determining maintenance information of production equipment. This application uses the example of a device for determining maintenance information of production equipment executing the method for determining maintenance information of production equipment to illustrate the device for determining maintenance information of production equipment provided in this application.

[0313] This application also provides a device for determining maintenance information of production equipment.

[0314] like Figure 8 As shown, the maintenance information determination device for the production equipment includes: a first processing module 810, a second processing module 820, and a third processing module 830.

[0315] The first processing module 810 is used to acquire the target maintenance instruction; the target maintenance instruction is generated when a fault is detected in the target production equipment, and includes information about the fault category of the target production equipment.

[0316] The second processing module 820 is used to obtain maintenance knowledge corresponding to the target maintenance instruction from a pre-constructed hierarchical knowledge graph based on the target maintenance instruction. The hierarchical knowledge graph is obtained by classifying and categorizing the sample maintenance information of various sample production equipment according to the target classification conditions, and constructing triples from the classified information. The target classification conditions include at least two of the following: basic information of the sample production equipment, component information, component failure causes, and component failure handling information.

[0317] The third processing module 830 is used to obtain the maintenance information corresponding to the target maintenance instruction based on the target maintenance instruction and maintenance knowledge.

[0318] The equipment maintenance information determination device for production equipment provided in the embodiments of this application determines the actual maintenance needs by acquiring the target maintenance instruction, and obtains the maintenance knowledge corresponding to the target maintenance instruction from a hierarchical knowledge graph applicable to industrial production scenarios based on the target maintenance instruction, thereby improving the accuracy of the acquired maintenance knowledge. Based on the maintenance knowledge and the target maintenance instruction, the device determines the maintenance information corresponding to the target maintenance instruction, thereby improving the accuracy and usability of the obtained maintenance information and effectively applying it to industrial production environments with high precision requirements.

[0319] In some embodiments, the second processing module 820 may also be used for:

[0320] Obtain the feature values ​​included in the target maintenance instruction and the corresponding levels of the feature values;

[0321] Based on feature values, multiple first knowledge points are obtained in the layers corresponding to the feature values ​​of a pre-constructed hierarchical knowledge graph;

[0322] Based on the similarity between the feature value and each piece of first knowledge, the maintenance knowledge corresponding to the target maintenance instruction is obtained from multiple pieces of first knowledge.

[0323] In some embodiments, the second processing module 820 may also be used for:

[0324] The eigenvalues ​​are vectorized to obtain the vectorized information corresponding to the eigenvalues;

[0325] Based on vectorized information, multiple first knowledge points are obtained from the layers corresponding to the feature values ​​of a pre-constructed hierarchical knowledge graph.

[0326] In some embodiments, the device may further include a fourth processing module for:

[0327] Based on the target classification conditions, the knowledge graph hierarchy is determined, and the sample maintenance information of various sample production equipment is classified and categorized to obtain the classified information.

[0328] Extract the sample entities, sample relationships, and sample attributes corresponding to the hierarchical and categorized information respectively to construct multiple triples;

[0329] Based on the knowledge graph hierarchy, multiple triples are stored to obtain a hierarchical knowledge graph.

[0330] In some embodiments, the first processing module 810 may also be used for:

[0331] Receive the user's first input;

[0332] In response to the first input, obtain the target maintenance instruction.

[0333] In some embodiments, the device may further include a fifth processing module for:

[0334] Based on the acquired operating data of the target production equipment, if a fault is determined in the target production equipment, the first information is output based on the fault condition of the target production equipment.

[0335] In some embodiments, the device may further include a sixth processing module for:

[0336] Identify target users based on the maintenance personnel schedule;

[0337] Send the first message to the client corresponding to the target user.

[0338] In some embodiments, the third processing module 830 can also be used for:

[0339] Input the target maintenance instructions and maintenance knowledge into the target large model, and obtain the maintenance information output by the target large model;

[0340] The target large model is trained based on sample maintenance data corresponding to various sample production equipment.

[0341] In some embodiments, the third processing module 830 can also be used for:

[0342] Obtain sample maintenance data corresponding to various sample production equipment;

[0343] The maintenance data of each sample are preprocessed to obtain the sample dataset;

[0344] When the amount of data corresponding to the sample dataset is greater than the first quantity, a target large model is trained based on the sample dataset and preset prompt information; the preset prompt information is a prompt word pre-set based on the maintenance scenario.

[0345] If the amount of data in the sample dataset is no greater than the first quantity, new sample maintenance data is obtained.

[0346] The device for determining maintenance information of production equipment in this application embodiment can be production equipment, or an electronic device communicatively connected to the production equipment, or a component in the production equipment or electronic device, such as an integrated circuit or chip. The electronic device can be a terminal, or other devices besides a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. It can also be a server, network attached storage (NAS), personal computer (PC), television (TV), ATM, or self-service machine, etc. This application embodiment does not specifically limit the device.

[0347] The device for determining maintenance information of production equipment in this embodiment can be a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this embodiment does not specifically limit its use.

[0348] The equipment maintenance information determination device for production equipment provided in this application embodiment can achieve... Figures 1 to 7 The various processes implemented in the method implementation examples will not be described again here to avoid repetition.

[0349] This application also provides a production device.

[0350] In some embodiments, the production equipment includes: a maintenance information determination device for the production equipment as described in any of the above embodiments.

[0351] In this embodiment, the maintenance information determination device for production equipment can monitor the operating status of the production equipment and promptly send a prompt message to the target user in the event of a malfunction, so as to prompt the target user to carry out maintenance on the production equipment.

[0352] Target users can access maintenance information through intelligent Q&A to repair production equipment under the guidance of this information.

[0353] According to the production equipment provided in the embodiments of this application, by obtaining the target maintenance instruction, the actual maintenance needs are determined. Based on the target maintenance instruction, maintenance knowledge corresponding to the target maintenance instruction is obtained from a hierarchical knowledge graph applicable to industrial production scenarios, thereby improving the accuracy of the obtained maintenance knowledge. Based on the maintenance knowledge and the target maintenance instruction, maintenance information corresponding to the target maintenance instruction is determined, thereby improving the accuracy and usability of the obtained maintenance information and effectively applying it to industrial production environments with high precision requirements.

[0354] In some embodiments, such as Figure 9 As shown, this application embodiment also provides an electronic device 900, including a processor 901, a memory 902, and a computer program stored in the memory 902 and executable on the processor 901. When the program is executed by the processor 901, it implements the various processes of the above-described method embodiment for determining maintenance information of production equipment and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0355] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0356] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described method for determining maintenance information of production equipment and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0357] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0358] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for determining maintenance information of production equipment.

[0359] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0360] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above-described method embodiment for determining maintenance information of production equipment, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0361] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0362] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0363] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the related technology, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0364] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0365] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0366] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for determining maintenance information of production equipment, characterized in that, include: Obtain the target maintenance instruction; the target maintenance instruction is generated when a fault is detected in the target production equipment, and includes information about the fault category of the target production equipment; Based on the target maintenance instruction, maintenance knowledge corresponding to the target maintenance instruction is obtained from a pre-constructed hierarchical knowledge graph. The hierarchical knowledge graph is obtained by classifying and categorizing the sample maintenance information of various sample production equipment according to the target classification conditions, and constructing triples from the classified information. The target classification conditions include at least two of the following: basic information of the sample production equipment, component information, component failure causes, and component failure handling information. Based on the target maintenance instruction and the maintenance knowledge, the maintenance information corresponding to the target maintenance instruction is obtained.

2. The method for determining maintenance information of production equipment according to claim 1, characterized in that, The step of obtaining maintenance knowledge corresponding to the target maintenance instruction from a pre-constructed hierarchical knowledge graph based on the target maintenance instruction includes: Obtain the feature values ​​included in the target maintenance instruction and the level corresponding to the feature values; Based on the feature values, multiple first knowledge items are obtained in the layers corresponding to the feature values ​​of the pre-constructed hierarchical knowledge graph; Based on the similarity between the feature value and each of the first knowledge, the maintenance knowledge corresponding to the target maintenance instruction is obtained from the plurality of first knowledge.

3. The method for determining maintenance information of production equipment according to claim 2, characterized in that, Based on the feature values, multiple first knowledge items are obtained from the levels corresponding to the feature values ​​in the pre-constructed hierarchical knowledge graph, including: The feature values ​​are vectorized to obtain the vectorized information corresponding to the feature values; Based on the vectorized information, the plurality of first knowledge is obtained in the layers corresponding to the feature values ​​of the pre-constructed hierarchical knowledge graph.

4. The method for determining maintenance information of production equipment according to any one of claims 1-3, characterized in that, Prior to obtaining the target maintenance instruction, the method further includes: Based on the target classification conditions, the knowledge graph level is determined, and the sample maintenance information of the various sample production equipment is classified and categorized to obtain the classified information. Extract the sample entities, sample relationships, and sample attributes corresponding to the hierarchical and categorized information respectively to construct multiple triples; Based on the knowledge graph hierarchy, the multiple triples are stored to obtain the hierarchical knowledge graph.

5. The method for determining maintenance information of production equipment according to any one of claims 1-3, characterized in that, The acquisition of the target maintenance instruction includes: Receive the user's first input; In response to the first input, the target maintenance instruction is obtained.

6. The method for determining maintenance information of production equipment according to any one of claims 1-3, characterized in that, Before obtaining the target maintenance instruction, the method further includes: If a fault is determined in the target production equipment based on the acquired operating data, first information is output based on the fault status of the target production equipment.

7. The method for determining maintenance information of production equipment according to claim 6, characterized in that, The output of the first information includes: Identify target users based on the maintenance personnel schedule; The first information is sent to the user terminal corresponding to the target user.

8. The method for determining maintenance information of production equipment according to any one of claims 1-3, characterized in that, The step of obtaining the maintenance information corresponding to the target maintenance instruction based on the target maintenance instruction and the maintenance knowledge includes: The target maintenance instruction and the maintenance knowledge are input into the target large model to obtain the maintenance information output by the target large model; The target large model is trained based on the sample maintenance data corresponding to the various sample production equipment.

9. The method for determining maintenance information of production equipment according to claim 8, characterized in that, The target large model is trained based on sample maintenance data corresponding to the various sample production equipment, including: Obtain sample maintenance data corresponding to the various sample production equipment; The maintenance data of each sample are preprocessed to obtain a sample dataset; If the amount of data corresponding to the sample dataset is greater than the first quantity, the target large model is trained based on the sample dataset and preset prompt information; the preset prompt information is a prompt word pre-set based on the maintenance scenario. If the amount of data corresponding to the sample dataset is not greater than the first quantity, new sample maintenance data is obtained.

10. A device for determining maintenance information of production equipment, characterized in that, include: The first processing module is used to acquire a target maintenance instruction; the target maintenance instruction is generated when a fault is detected in the target production equipment, and includes information about the fault category of the target production equipment. The second processing module is used to obtain maintenance knowledge corresponding to the target maintenance instruction from a pre-constructed hierarchical knowledge graph based on the target maintenance instruction. The hierarchical knowledge graph is obtained by classifying and categorizing the sample maintenance information of various sample production equipment according to the target classification conditions, and constructing triples from the classified information. The target classification conditions include at least two of the following: basic information of the sample production equipment, component information, component failure causes, and component failure handling information. The third processing module is used to obtain the maintenance information corresponding to the target maintenance instruction based on the target maintenance instruction and the maintenance knowledge.

11. A production equipment, characterized in that, include: Based on the maintenance information determination device for production equipment as described in claim 10.

12. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the method for determining maintenance information of production equipment as described in any one of claims 1-9.

13. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for determining maintenance information of production equipment as described in any one of claims 1-9.

Images