Household appliance fault processing method, device and system and electronic equipment

Abstract

The invention provides a household appliance fault processing method, device and system and electronic equipment, and belongs to the field of data processing, and the method comprises the steps: responding to a fault recovery request of any household appliance, and generating the fault description of the household appliance according to the state data of the household appliance; querying knowledge fragments matched with the fault description from a preset knowledge base; based on the knowledge fragments, assembling target cue words; inputting the target prompt word into a preset generative large language model, and generating a fault repair strategy by using the generative large language model; and feeding back a fault recovery strategy. Therefore, based on the fault description of the household appliance through a large model technology, the fault reason is accurately positioned in a short time, the fault repair strategy is automatically generated, the most appropriate repair scheme is provided for a user or a maintainer, the fault removal time is greatly shortened, the fault treatment of the household appliance is efficiently and conveniently carried out at low cost, and the user experience is improved. The user experience is effectively improved, and the market demand is met.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method, device, system and electronic equipment for handling household appliance faults. Background Art

[0002] With the rapid development of smart home and Internet of Things technologies, kitchen appliances are becoming increasingly intelligent and complex, becoming an indispensable part of modern family life. These appliances integrate various sensors, microprocessors and other intelligent components to achieve diverse functions from cooking control, energy consumption monitoring to remote operation.

[0003] However, this technological advancement has also brought about the complexity of equipment fault diagnosis and maintenance, so that when equipment fails, the only way to troubleshoot is to refer to the user manual or contact professionals for on-site repairs, which prolongs equipment downtime and increases maintenance costs.

[0004] In view of the fact that the current market has put forward higher requirements on the stability and reliability of kitchen appliances, an efficient, convenient and low-cost equipment failure handling method is developed to improve the user experience and meet the market demand. Summary of the invention

[0005] In view of this, an object of the present invention is to provide a method, device, system and electronic device for handling household appliance faults, which can handle household appliance faults efficiently, conveniently and at low cost, effectively improve user experience and meet market demand.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] In a first aspect, the present invention provides a method for handling a household appliance fault, the method comprising:

[0008] In response to a fault repair request for any household appliance, generating a fault description of the household appliance according to the status data of the household appliance;

[0009] Querying a knowledge fragment matching the fault description from a preset knowledge base; wherein the knowledge fragment is data obtained after knowledge-based fault repair documents;

[0010] Based on the knowledge fragments, assembling target prompt words;

[0011] Inputting the target prompt word into a preset generative large language model, and using the generative large language model to generate a fault repair strategy;

[0012] Feedback on the fault recovery strategy.

[0013] Optionally, the step of assembling target prompt words based on the knowledge fragments includes:

[0014] Obtaining the semantic similarity between the fault description and each of the knowledge fragments, and arranging each of the knowledge fragments in descending order of the semantic similarity to obtain a first prompt word;

[0015] In combination with a preset set role and a preset answer guide, a second prompt word associated with the first prompt word is generated; wherein the second prompt word is used to indicate that the generative large language model serves as the set role, takes the first prompt word as background information, and generates an answer according to the answer guide;

[0016] Generate a third prompt word according to the fault description; wherein the third prompt word is used to prompt the generative large language model of the fault problem to be processed;

[0017] The first prompt word, the second prompt word and the third prompt word are assembled to obtain a target prompt word.

[0018] Optionally, the method further comprises:

[0019] In response to the status data reported by each household appliance, writing the status data and the reporting time of the status data into the status record table of the household appliance;

[0020] The step of generating a fault description of the home appliance according to the status data of the home appliance in response to a fault repair request for any home appliance comprises:

[0021] When receiving a fault repair request, determining the home appliance corresponding to the fault repair request;

[0022] At least one set of the status data whose reporting time is before the current time is retrieved from the status record table, and the fault repair request and the status data are combined to generate a fault description.

[0023] Optionally, the knowledge base includes a plurality of knowledge vectors and knowledge fragments indexed by the knowledge vectors;

[0024] The step of searching for a knowledge fragment matching the fault description from the knowledge base includes:

[0025] Vectorizing the fault description to obtain a description vector;

[0026] Calculating the semantic similarity between the description vector and each knowledge vector in a preset knowledge base respectively, and determining a matching vector from each knowledge vector according to the semantic similarity;

[0027] According to the index of the matching vector, a knowledge segment corresponding to the index is extracted from the knowledge base.

[0028] Optionally, before the step of searching the knowledge base for a knowledge segment matching the fault description, the method further includes:

[0029] Pre-process the fault repair data of each type of household appliance to obtain fault repair documents;

[0030] Slicing the fault repair document to obtain multiple knowledge fragments;

[0031] Performing vector conversion on the knowledge fragments to obtain knowledge vectors of the knowledge fragments, and creating a unique index for each knowledge fragment;

[0032] The knowledge fragment and the knowledge vector are associated with each other through the index of the knowledge fragment and then stored in the knowledge base.

[0033] Optionally, the step of preprocessing the fault repair data of each type of household electrical appliance to obtain a fault repair document includes:

[0034] Cleaning the fault repair data of each type of household electrical appliance to obtain first data;

[0035] Standardizing and normalizing the values ​​in the first data to obtain second data;

[0036] Perform text extraction on the second data to obtain a fault repair document.

[0037] Optionally, the step of feeding back the fault repair strategy includes:

[0038] Determine the identity information of the user of the household appliance, and retrieve the user terminal information of the user based on the identity information;

[0039] The fault repair strategy is sent to the user's terminal device according to the user terminal information.

[0040] In a second aspect, the present invention provides a household appliance fault processing device, comprising a description generation module, a query matching module, an assembly module, a strategy generation module and a feedback module;

[0041] The description generating module is used to generate a fault description of any household appliance according to the status data of the household appliance in response to a fault repair request of the household appliance;

[0042] The query matching module is used to query the knowledge fragments matching the fault description from the preset knowledge base; wherein the knowledge fragments are data obtained after the fault repair document is converted into knowledge;

[0043] The assembling module is used to assemble the target prompt words based on the knowledge fragments;

[0044] The strategy generation module is used to input the target prompt word into a preset generative large language model, and generate a fault repair strategy using the generative large language model;

[0045] The feedback module is used to feed back the fault repair strategy.

[0046] In a third aspect, the present invention provides a household appliance fault handling system, comprising a central control service device, a household appliance and a terminal device, wherein the central control service device is respectively connected to the household appliance and the terminal device for communication;

[0047] The terminal device is used to send a fault repair request to the central control service device when the home appliance fails;

[0048] The home appliance is used to send a fault repair request to the central control service device when detecting a transmission fault of the home appliance itself;

[0049] The central control service device is used to implement the household appliance fault handling method as described in the first aspect.

[0050] In a fourth aspect, the present invention provides an electronic device, comprising a processor and a memory, wherein the memory stores a computer program executable by the processor, and the processor can execute the computer program to implement the household appliance fault handling method as described in the first aspect.

[0051] In a fifth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for handling household appliance faults as described in the first aspect.

[0052] The household appliance fault handling method, device, system and electronic device provided by the embodiment of the present invention include: in response to a fault repair request for any household appliance, generating a fault description of the household appliance according to the status data of the household appliance; querying a knowledge fragment matching the fault description from a preset knowledge base, wherein the knowledge fragment is data obtained after the fault repair document is knowledge-based; assembling target prompt words based on the knowledge fragment; inputting the target prompt words into a preset generative large language model, and generating a fault repair strategy using the generative large language model; and feeding back the fault repair strategy. In this way, based on the fault description of the household appliance through the large model technology, the cause of the fault can be accurately located in a short time, and a fault repair strategy can be automatically generated, providing the most suitable repair solution for the user or maintenance personnel, greatly shortening the troubleshooting time, realizing efficient, convenient and low-cost fault handling of household appliances, effectively improving the user experience and meeting market demand.

[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0055] Figure 1 A schematic diagram of the system architecture of a household appliance fault handling system provided by an embodiment of the present invention is shown.

[0056] Figure 2 A schematic diagram of a module architecture of an electronic device provided by an embodiment of the present invention is shown.

[0057] Figure 3 One of the flow charts of the method for handling household appliance faults provided in an embodiment of the present invention is shown.

[0058] Figure 4 Shows Figure 3 Schematic diagram of the flow chart of the sub-steps of step 11 in FIG.

[0059] Figure 5 The second flowchart of the method for handling household appliance failure provided by the embodiment of the present invention is shown.

[0060] Figure 6 The third flowchart of the method for handling household appliance failure provided by the embodiment of the present invention is shown.

[0061] Figure 7 Shows Figure 5 and Figure 6 Schematic diagram of the flow chart of the sub-steps of step 22.

[0062] Figure 8 Shows Figure 5 and Figure 6 Schematic diagram of the flow chart of the sub-steps of step 23.

[0063] Fig. 9 Shows Figure 5 and Figure 6 Schematic diagram of the process flow of the sub-steps of step 24.

[0064] Fig.10 Shows Figure 5 and Figure 6 Schematic diagram of the process flow of the sub-steps of step 26.

[0065] Fig.11 A schematic diagram of the module architecture of a device for handling household appliance faults provided by an embodiment of the present invention is shown.

[0066] Explanation of the accompanying drawings: 10 - household appliance fault handling system; 110 - central control service equipment; 120 - household appliance; 130 - terminal equipment; 20 - electronic equipment; 210 - memory; 220 - processor; 230 - communication module; 30 - household appliance fault handling device; 310 - description generation module; 320 - query matching module; 330 - assembly module; 340 - strategy generation module; 350 - feedback module. DETAILED DESCRIPTION

[0067] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0068] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0069] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0070] The household appliance fault processing method provided by the embodiment of the present invention can be applied to Figure 1 The household appliance fault processing system 10 shown in the figure includes a central control service device 110, a household appliance 120 and a terminal device 130. The central control service device 110 can be communicated with the household appliance 120 and the terminal device 130 respectively through the network, WiFi, etc.

[0071] A pre-trained large language model is deployed on the central control service device 110 here.

[0072] The terminal device 130 is used to send a fault repair request to the central control service device 110 when the home appliance 120 fails.

[0073] The home appliance 120 is used to send a fault repair request to the central control service device 110 when detecting that a fault occurs in the home appliance 120 .

[0074] The central control service device 110 is used to implement the household appliance fault handling method provided by the present invention.

[0075] The terminal device 130 may be, but is not limited to, a personal computer, a laptop computer, a tablet computer, a mobile phone, a smart mobile terminal, a wearable portable device, etc. The home appliance 120 may be, but is not limited to, any home appliance such as a range hood, a gas stove, an integrated stove, a dishwasher, an oven, a steamer, a disinfection cabinet, a microwave oven, a water purifier, etc. The central control service device 110 may be, but is not limited to, an independent server, a cloud server, or a server cluster.

[0076] The household appliance fault handling system 10 may further include a development device, which is connected to the central control service device 110 through a network, and is used to construct a large language model and knowledge base used in the household appliance fault handling method.

[0077] It should be noted that the large language model and knowledge base used in constructing the household appliance fault handling method can also be completed by the central control service device 110, and in this case, there is no development equipment.

[0078] Please refer to Figure 2 , is a block diagram of an electronic device 20, which can Figure 1 The central control service device 110 in the household appliance fault handling system 10 is shown. The electronic device 20 includes a memory 210, a processor 220 and a communication module 230. The memory 210, the processor 220 and the communication module 230 are electrically connected to each other directly or indirectly to achieve data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines.

[0079] The memory 210 is used to store programs or data. The memory 210 may be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc.

[0080] The processor 220 is used to read / write data or programs stored in the memory 210 and execute corresponding functions. For example, Figure 1 In the household appliance fault processing system 10 shown, the processor 220 of the central control service device 110 reads the computer program stored in the memory 210 to implement the household appliance fault processing method provided by the embodiment of the present invention.

[0081] The communication module 230 is used to establish a communication connection between the electronic device 20 and other communication terminals through the network, and to send and receive data through the network. Figure 1 In the household appliance fault handling system 10 shown, the central control service device 110 transmits and receives data with the terminal device 130 and the household appliance 120 respectively through the network.

[0082] It should be understood that Figure 2 The structure shown is only a schematic diagram of the structure of the electronic device 20. The electronic device 20 may also include Figure 2 More or fewer components as shown, or with Figure 2 Different configurations are shown. Figure 2 Each component shown in the figure can be implemented by hardware, software or a combination thereof.

[0083] In order to improve the high cost, low efficiency and inconvenience of fault diagnosis and maintenance of the target home appliance 120, an embodiment of the present invention provides a method for handling home appliance faults, which generates and feeds back corresponding fault repair strategies in real time online according to the fault repair request of the terminal device 130 or the home appliance 120. The implementation of the method includes two parts: preparation before application launch and online application.

[0084] The preparations before the application goes online include the preparation of a large language model and a knowledge base. The prepared large language model and knowledge base will be deployed to Figure 1 The central control server device of the household appliance fault handling system 10 is shown. Figure 1 The central control service device 110 in the household appliance fault handling system 10 shown generates and feeds back corresponding fault repair strategies online in real time according to the fault repair requests of the terminal device 130 or the household appliance 120 based on the pre-deployed large language model and knowledge base.

[0085] In preparation before the application goes online, an unlabeled data set can be used to iteratively train the constructed basic model, and the trained large language model can be deployed to the central control service device 110. The training and construction method of the large language model can be any known model construction training method, which will not be repeated in the embodiment of the present invention.

[0086] In addition, in the preparation before the application is launched, refer to Figure 3 The method for handling household appliance faults further includes steps 11 to 17 to obtain a dedicated knowledge base for handling household appliance faults 120. Thus: Figure 1 In the household appliance fault processing system 10 shown, the central control service device 110 can construct a knowledge base using steps 11 to 17.

[0087] Step 11, pre-processing the fault repair data of each type of household electrical appliance to obtain a fault repair document.

[0088] It should be noted that fault repair data includes information such as equipment operation data related to the description of household appliance faults or repairs, equipment historical data, historical cases, solutions, equipment manuals and technical documents.

[0089] Step 13, slicing the fault repair document to obtain multiple knowledge fragments.

[0090] Step 15, perform vector conversion on the knowledge fragment to obtain the knowledge vector of the knowledge fragment, and create a unique index for each knowledge fragment.

[0091] Step 17, the knowledge fragments and knowledge vectors are associated through the index of the knowledge fragments and stored in the knowledge base.

[0092] In this way, the knowledge base stores knowledge fragments obtained by converting the fault repair document into knowledge, and knowledge vectors representing the knowledge fragments.

[0093] For example, in combination Figure 1 In the household appliance fault handling system 10 shown, the development device collects the fault repair data of each type of household appliance 120, and pre-processes the fault repair data to obtain a fault repair document. Next, the development device slices the fault repair document to obtain multiple knowledge fragments, performs vector conversion on the knowledge fragments obtained by slicing, obtains the knowledge vector of the knowledge fragment, and creates a unique index for each knowledge fragment. Then, the knowledge fragments and knowledge vectors are associated with the index of the knowledge fragment and stored in the knowledge base. Finally, the obtained knowledge base is deployed to the central control service device 110.

[0094] It should be noted that after this, the latest fault repair data of the home appliance 120 can be periodically obtained, and the above steps 11 to 17 can be used to process to obtain new knowledge fragments and knowledge vectors, and the knowledge fragments and knowledge vectors are associated with indexes and added to the knowledge base in the central control service device 110 to update the knowledge base.

[0095] Through the above method, a knowledge base of equipment operation data, equipment history data, historical cases, solutions, equipment manuals and technical documents related to the fault description or repair of each household appliance is constructed. This knowledge base can be used as an external knowledge base of the generative large language model to help the generative large language model quickly and accurately locate the cause of the failure of the household appliance in a short time, and automatically generate accurate fault repair strategies for the fault based on the repair cases and repair manuals of the same or similar faults provided in the knowledge base as reference experience. Therefore, it helps to provide users or maintenance personnel with the most appropriate repair plan for the faulty household appliance, and greatly shortens the troubleshooting time.

[0096] The method of preprocessing the fault repair data in step 11 can be flexibly selected, for example, data denoising, text extraction of the fault repair data, or data expansion. The above methods are all examples, and their implementation methods are not limited.

[0097] In order to improve the quality and comparability of the knowledge base and help improve the accuracy of the subsequent fault repair strategy, data denoising, standardization, normalization and text extraction are introduced in the data preprocessing of step 11. Figure 4 The process of preprocessing the fault repair data in step 11 includes steps 111 to 115.

[0098] Step 111 , clean the fault repair data of each type of household electrical appliance to obtain first data.

[0099] Here, a reasonable range of various types of fault repair data can be preset, so that the development equipment can clean each fault repair data according to the reasonable range, remove outliers and noise, and then deduplicate the duplicate data. In this way, removing interference data and irrelevant data in the original fault repair data will help improve the semantic richness of the knowledge fragments obtained after subsequent segmentation, improve the quality of the knowledge base, and make the fault repair strategy generated by the generative large language model based on the knowledge base more accurate.

[0100] Step 113: standardize and normalize the values ​​in the first data to obtain second data.

[0101] Among them, the standardization and normalization methods can be flexibly selected. For example, it can be minimum-maximum normalization, that is, all data are scaled to the [0,1] interval. The formula is: It can also be Z-score normalization, which converts the data into a distribution with a mean of 0 and a standard deviation of 1. The formula is: It can also be any other standardized normalization method, and its implementation method is not limited.

[0102] After standardization and normalization, each value has the same dimension, format and scale, eliminating noise, data redundancy and differences in format, scale and other aspects between data, thereby increasing the consistency and comparability of the data, making the data easier to understand and read, and helping to improve the efficiency and accuracy of knowledge fragment matching in the subsequent home appliance fault handling process.

[0103] Step 115: extract text from the second data to obtain a fault repair document.

[0104] For the second data in different formats (such as PDF, pictures, etc.), electronic document parsing or OCR technology is used to perform parsing and recognition to extract texts from the data in various formats to obtain fault repair documents.

[0105] Through the above steps 111 to 113, noise, outliers and duplicate data in the data are filtered out, and at the same time, through standardization and normalization, it is ensured that the data in each knowledge fragment in the knowledge base has the same dimension and format, and is unified and comparable in different dimensions, thereby greatly improving the quality of fault repair documents and data comparability, which helps to improve the quality of the subsequently constructed knowledge base, as well as the accuracy and fault resolution speed of the fault repair strategy generated for home appliances using the knowledge base.

[0106] After obtaining the fault repair document, in step 13, any slicing method can be used to slice the fault repair document. For example, the RecursiveCharacterTextSplitter method provided by langchain can be used to slice the fault repair document according to a preset block size (i.e., chunkSize, which can be 50 or 60, and the value is not limited). The slicing function of the NLP natural language text processing library (such as spaCy of Python and CPython) can also be used to slice the fault repair document. The Natural Language Processing Toolkit (NLTK) provided by the Python library is also used to slice the fault repair document. The above method is only an example, and its implementation method is not limited.

[0107] After slicing to obtain all knowledge fragments, in step 15, you can call the HuggingFaceEmbeddings method provided by langchain to convert each knowledge fragment into a knowledge vector. You can also use the bag of words model, TF-IDF, word embedding or any other document vectorization method to convert each knowledge fragment into a knowledge vector.

[0108] In addition, Faiss is used in step 15 to create an IndexIVFPQ index for each knowledge fragment. Then, in step 17, the knowledge fragment and the knowledge vector of the knowledge fragment are associated through the index and written into the knowledge base. And the knowledge base can be a faiss database. By building a knowledge base through faiss, even if the scale of the knowledge base is huge (such as millions or even billions), it is possible to rely on the advanced indexing technology and algorithm optimization strategy adopted by Faiss to match the search in a very short time, which helps to improve the speed and efficiency of home appliance fault handling.

[0109] It should be noted that the faiss database storing knowledge fragments and knowledge vectors can also be replaced by any other search tool such as Milvus, and the corresponding index creation can also be replaced synchronously.

[0110] Through the above method, the fault repair document is sliced ​​and converted into a vector representation and stored in the knowledge base, so as to support fast vector similarity calculation and retrieval, so as to facilitate subsequent efficient calculation and retrieval.

[0111] After preparing the large language model and knowledge base, refer to Figure 5 The household appliance fault handling method provided in the embodiment of the present invention may further include steps 22 to 26, so that Figure 1 The central control service device 110 in the household appliance fault handling system 10 is Figure 2The structure shown, when the processor 220 executes the computer program corresponding to steps 22 to 26 stored in the memory 210, generates and feeds back the corresponding fault repair strategy in real time according to the fault repair request of the terminal device 130 or the household appliance 120.

[0112] Step 22: In response to a fault repair request for any household electrical appliance, generate a fault description of the household electrical appliance according to the status data of the household electrical appliance.

[0113] Step 23: query the knowledge fragments matching the fault description from the preset knowledge base.

[0114] The knowledge fragments are data obtained by converting the fault repair document into knowledge. The knowledge fragments obtained in step 23 are equipment operation data, equipment history data, repair cases, repair experience, manuals and other information of the same type or kind of fault as the home appliance.

[0115] Step 24: assemble target prompt words based on the knowledge fragments. Step 25: input the target prompt words into a preset generative large language model, and use the generative large language model to generate a fault repair strategy.

[0116] The assembled target prompt words contain information such as equipment operation data, equipment historical data, repair cases, repair experience, and instructions for the same type of faults or the same type of faults of the faulty home appliances. This will help the generative large language model to quickly locate the cause of the fault and quickly generate a fault repair strategy after determining the cause of the fault. For example, using the equipment operation data and equipment historical data of the faulty equipment in the knowledge fragment as a reference, it is possible to determine whether it is the type of fault in the historical repair case based on the current operation data and historical data of the home appliance, so as to determine the cause of the fault. After determining the cause of the fault, the generative large language model can understand the information in the fault repair cases, repair data, and technical documents caused by the same fault cause in the knowledge fragment to generate a fault repair strategy for the faulty home appliance.

[0117] Step 26: Feedback the fault repair strategy.

[0118] For example, in combination Figure 1In the household appliance fault handling system 10 shown, when a household appliance 120 fails, the household appliance 120 with a self-checking and reporting function sends a fault repair request to the central control service device 110, or the user of the household appliance 120 can use his own terminal device 130 to interact with the central control service device 110 by inputting text or voice on the interactive interface to initiate a fault repair request. After receiving the fault repair request, the central control service device 110 generates a fault description of the household appliance 120 according to the status data of the household appliance 120, and queries one or more knowledge fragments matching the fault description from a preset knowledge base.

[0119] Then, the central control service device 110 assembles the target prompt words based on the matched one or more knowledge fragments, and inputs the assembled target prompt words into the preset generative large language model, and generates the fault repair strategy using the generative large language model. After obtaining the fault repair strategy output by the generative large language model, the central control service device 110 feeds back the fault repair strategy to the user terminal of the user of the home appliance 120 or the terminal device 130 of the maintenance personnel. Thus, the user can repair the home appliance 120 with a fault according to the fault repair strategy.

[0120] In steps 22 to 26 of the above-mentioned household appliance fault handling method, the knowledge fragment that best matches the fault description of the household appliance is first obtained from the external knowledge base, so as to obtain the equipment operation data, equipment history data, historical repair cases, repair data and repair instructions for the fault, etc. of the device with the same fault, and the target prompt words of the generative large language model are assembled with the queried matching fragments, so that the generative large language model can accurately locate the cause of the fault in a short time based on the knowledge fragment matching the fault (i.e., the equipment operation data, equipment history data, historical repair cases, repair data and repair instructions for the fault corresponding to the fault), and automatically generate an accurate fault repair strategy for the fault based on the repair cases and repair instructions of the same or same faults in the knowledge fragment as reference experience, so as to provide users or maintenance personnel with the most suitable repair plan for the faulty household appliance, greatly shorten the troubleshooting time, realize efficient, convenient and low-cost fault handling of household appliances, effectively improve user experience and meet market demand.

[0121] Among them, the method of generating the fault description in step 22 can be flexibly selected. For example, the user can manually input the status data of the home appliance 120, and the central control service device 110 uses the status data as the fault description. The fault description can also be generated according to preset rules, and its implementation method is not limited.

[0122] In order to make the generated fault description more accurately reflect the fault condition of the home appliance 120, the home appliance 120 periodically and automatically reports status data to the central control service device 110, and the central control service device 110 selects the status data that best matches the fault to generate the fault description. Figure 6 The household appliance fault handling method further includes step 21, referring to Figure 7 The process of generating a fault description in step 22 includes steps 221 to 222.

[0123] Step 21 , in response to the status data reported by each household appliance, the status data and the reporting time of the status data are written into a status record table of the household appliance.

[0124] During operation, the home appliance 120 can obtain its current status data at regular intervals and report the status data to the central control service device 110. After receiving the status data reported by any home appliance 120, the central control service device 110 associates the status data with the reporting time and writes it into the status record table of the home appliance 120.

[0125] The status data may be in the form of "device model: CQ926; error code: E001; device status: stopped running; operation log: motor overheating, automatic shutdown protection triggered".

[0126] The status data of each home appliance 120 is recorded in a dedicated status record table, and the index of the status record table can be a unique identifier such as the device number of the home appliance 120. Alternatively, the status data of all home appliances 120 can be recorded in one status record table. The setting method of the status record table is not limited.

[0127] Step 221: When a fault repair request is received, determine the home appliance corresponding to the fault repair request.

[0128] Step 222: retrieve at least one set of status data with a reporting time before the current time from the status record table, and generate a fault description by combining the fault repair request and the status data.

[0129] The fault repair request may include a description of the fault problem, for example, a description of any problem of the home appliance 120 such as “What should I do if the 'water shortage' sign of my all-in-one machine keeps flashing?” or “What should I do if my range hood does not work after being powered on?”.

[0130] Through the above-mentioned step 21, and steps 221 to 222, one or more groups of status data whose reporting time is before the current moment when the fault repair request is received and closest to the current moment are used as the fault description together with the fault repair request, so that the fault description can more accurately describe the operating conditions of the household appliance 120 before the fault, which helps to improve the accuracy of subsequent fault cause location and repair strategy location.

[0131] After obtaining the fault description, refer to Figure 8 The process after step 23 obtains the knowledge fragment matching the fault description from the knowledge base includes steps 231 to 233.

[0132] Step 231, vectorize the fault description to obtain a description vector.

[0133] Here, the method of vectorizing the fault description to obtain the description vector is the same as the method of vectorizing the knowledge fragment to obtain the knowledge vector in the above text. Please refer to the above content and explanation, and will not be repeated here.

[0134] Step 232, respectively calculate the semantic similarity between the description vector and each knowledge vector in the preset knowledge base, and determine the matching vector from each knowledge vector according to the semantic similarity.

[0135] It should be noted that the semantic similarity can be the cosine similarity between the description vector and the knowledge vector. In this case, the calculation formula of the semantic similarity is: cosθ represents semantic similarity, A and B represent description vector and knowledge vector respectively. Semantic similarity can also be the Euclidean distance between description vector and knowledge vector. In this case, the calculation formula of semantic similarity is: d(A,B)=‖AB‖ 2 .

[0136] In addition, the semantic similarity can also be calculated by any vector similarity calculation method such as Harman distance, Jaccard similarity coefficient, Hamming distance, etc. The implementation method is not limited.

[0137] After obtaining the semantic similarity between the description vector and each knowledge vector, the first n knowledge vectors with the highest semantic similarity are selected as matching vectors in descending order of semantic similarity. The value of n can be flexibly set, for example, it can be 3, 1, or 5, and its value is not limited.

[0138] Alternatively, the knowledge vectors whose semantic similarity is higher than a preset similarity threshold value may be used as matching vectors. In this embodiment, there is no limitation on the method of determining the matching vectors.

[0139] Step 233: extract the knowledge fragment corresponding to the index from the knowledge base according to the index of the matching vector.

[0140] It should be noted that only knowledge segments indexed by matching vectors may be extracted in step 233. In order to obtain sufficient data to make subsequent generation of fault repair strategies more accurate, knowledge segments indexed by matching vectors and knowledge segments having contextual relationships with knowledge segments indexed by matching vectors may also be extracted.

[0141] Through the above steps 231 to 233, all knowledge fragments matching the fault description are extracted quickly and accurately from the knowledge base.

[0142] After all knowledge fragments matching the fault description are obtained in the above manner, the knowledge fragments and the fault description may be assembled together as target prompt words in step 24, or the target prompt words may be assembled according to a preset rule. Here, the assembly method of the target prompt words is not limited.

[0143] In order to enable the generated large language model to generate a fault repair strategy that accurately corresponds to the fault description request and the fault of the home appliance 120, the concept of guiding the generated large language model to give a targeted answer and prompting the prompt words to answer the question is introduced in the target prompt word assembly of step 24. Fig. 9 The process of assembling the target prompt word in step 24 includes steps 241 to 244.

[0144] Step 241 , obtaining the semantic similarity between the fault description and each knowledge fragment, and arranging each knowledge fragment in descending order of semantic similarity to obtain a first prompt word.

[0145] The higher the semantic similarity, the greater the probability that the fault described in the knowledge fragment (i.e., fault repair case, repair data, and equipment operation data) is the same or similar to the fault of the household appliance to be solved, and the more reference it has for fault repair. Therefore, it helps the generative large language model to prioritize the knowledge fragments with the highest reference for fault repair, which in turn helps to locate the cause of the fault more quickly and accurately and generate a fault repair strategy.

[0146] Step 242: Generate a second prompt word associated with the first prompt word in combination with the preset set role and the preset answer guidance.

[0147] The second prompt word is used to instruct the generative large language model to act as a set role, use the first prompt word as background information, and generate an answer according to the answer guidance.

[0148] By setting the role and answering guidance in the second prompt, the generative large language model always uses the set role to perform tasks such as knowledge segment understanding, fault location, and fault repair strategy generation, thereby preventing the work of the generative large language model from deviating from the requirements and ensuring the accuracy of the fault repair strategy. At the same time, the answering guidance can ensure that the answer of the generative large language model does not deviate from the fault problem to be solved, and when it is impossible to locate the cause of the fault and generate a fault repair strategy, it can also provide a fault repair strategy that conforms to the fault handling response, such as "Hello, I cannot provide a solution for this fault for the time being, please call the customer service number to handle it", so as to improve the user experience.

[0149] Step 243: Generate a third prompt word according to the fault description.

[0150] The third prompt word is used to prompt the generative large language model of the fault problem to be processed. By setting the third prompt word, it is ensured that all processing of the generative large language model is aimed at solving the fault problem, which helps to improve the accuracy of the fault repair strategy.

[0151] Step 244: Assemble the first prompt word, the second prompt word, and the third prompt word to obtain a target prompt word.

[0152] It should be noted that in different application scenarios or answering requirements, the specific content of the role and answering instructions set in the second prompt word may be different, and the specific content is not limited. For example, the role setting may be "As my personal knowledge base, your task is to manage and use document content to answer questions", or "You are a kitchen appliance troubleshooting expert who can provide detailed equipment troubleshooting methods." The answering instructions may be "Use the background information provided to describe in detail a series of steps or methods to solve or answer the user's questions", or "If the information is not enough to answer the question, there is no need to mention the source document, but directly state that it cannot be answered. The standard response in this case is: XXXX".

[0153] For example, there are three knowledge fragments matching the fault description, and the question in the fault repair request is "What should I do if the 'water shortage' sign of my all-in-one machine keeps flashing?" The target prompt words finally generated can be shown in Table 1.

[0154] Table 1

[0155]

[0156]

[0157] In Table 1, the first item in the second prompt is to set the role, and the second, third and fourth items in the second prompt are the answer guidance.

[0158] Through the above steps 241 to 244, relevant knowledge fragments are provided in the target prompt words as reference content for the answer, and the fault problem to be answered and the answer strategy are prompted, so as to help the generative large language model understand the knowledge fragments that match the fault description in the order of high semantic similarity (fault repair reference) and low fault repair reference, and always solve the fault problem with the set role to perform knowledge fragment understanding, fault location, fault repair strategy generation and other tasks, and through answer guidance, the answer of the generative large language model does not deviate from the fault problem to be solved, and when it is impossible to locate the cause of the fault and generate a fault repair strategy, it can also give a fault repair strategy that conforms to the fault processing reply. Thereby, the generative large language model locates the cause of the fault more quickly and accurately, and generates a fault repair strategy to solve the fault problem, and then the generative large language model generates a fault repair strategy that accurately corresponds to the fault description request and the fault of the home appliance 120, ensuring the pertinence and accuracy of the generated fault repair strategy.

[0159] Thus, after receiving the target prompt word, the generative large language model generates the final answer or solution based on the injected knowledge fragment and the user's original question. This answer can be a direct solution, a recommended operation step, or a more detailed problem diagnosis process.

[0160] The above method is used to make the generative large language model generate fault repair strategies, and actual verification is carried out. When the target prompt words are the contents of Table 1 above, the fault repair strategies output by the generative large language model are shown in Table 2 below.

[0161]

[0162]

[0163] After the above steps are used to obtain the fault repair strategy, in step 26, the fault repair strategy can be fed back to the user or maintenance personnel's terminal. Fig.10 , the process includes step 261 to step 262.

[0164] Step 261, determining the identity information of the user of the household appliance, and retrieving the user terminal information of the user based on the identity information.

[0165] Here, the central control service device 110 may be pre-set with an information table of home appliances 120, which records the user identity information indexed by the device model (or device number) of each home appliance 120, and the user terminal information indexed by the identity information. Therefore, the information table of home appliances 120 may be queried according to the device model (or device number) of the home appliance 120 corresponding to the fault repair request to obtain the user terminal information of the user. The user terminal information of the user may also be included in the fault repair request. The above methods are all examples, and the method of obtaining user terminal information is not limited.

[0166] Step 262: Send the fault recovery strategy to the user's terminal device according to the user terminal information.

[0167] In the case where the fault repair request is reported by the maintenance personnel using their own terminal device 130 , the fault repair strategy can be directly fed back to the terminal device 130 .

[0168] Based on the same inventive concept as the above household appliance fault handling method, refer to Fig.11 The embodiment of the present invention also provides a household appliance fault processing device 30, which can be applied to Figure 1 The central control service device 110 of the household appliance fault processing system 10 shown includes a description generation module 310 , a query matching module 320 , an assembly module 330 , a strategy generation module 340 and a feedback module 350 .

[0169] The description generating module 310 is used to generate a fault description of the home appliance according to the status data of the home appliance in response to a fault repair request for any home appliance.

[0170] The query matching module 320 is used to query the knowledge fragments matching the fault description from the preset knowledge base, wherein the knowledge fragments are data obtained by converting the fault repair document into knowledge.

[0171] The assembly module 330 is used to assemble the target prompt words based on the knowledge fragments. The obtained knowledge fragments are equipment operation data, equipment history data, repair cases, repair experience, manuals and other information of the same type or kind of fault as the household appliance.

[0172] The strategy generation module 340 is used to input the target prompt word into a preset generative large language model, and use the generative large language model to generate a fault repair strategy.

[0173] The target prompt words assembled by the strategy generation module 340 contain information such as device operation data, device historical data, repair cases, repair experience, and instructions for the same type of faults or the same type of faults of the faulty home appliance. This will help the generative large language model to quickly locate the cause of the fault and quickly generate a fault repair strategy after determining the cause of the fault. For example, using the device operation data and device historical data of the faulty device in the knowledge fragment as a reference, it is possible to determine whether it is the type of fault in the historical repair case based on the current operation data and historical data of the home appliance, so as to determine the cause of the fault. After determining the cause of the fault, the generated large language model can understand the information in the fault repair cases, repair data, and technical documents caused by the same fault cause in the knowledge fragment to generate a fault repair strategy for the faulty home appliance.

[0174] The feedback module 350 is used to provide feedback on the fault repair strategy.

[0175] Furthermore, the household appliance fault processing device 30 may also include a knowledge base construction module and a data recording module.

[0176] The data recording module is used to respond to the status data reported by each household appliance and write the status data and the reporting time of the status data into the status recording table of the household appliance.

[0177] The knowledge base construction module is used to: pre-process the fault repair data of each type of household appliance to obtain fault repair documents; slice the fault repair documents to obtain multiple knowledge fragments; vectorize the knowledge fragments to obtain the knowledge vectors of the knowledge fragments, and create a unique index for each knowledge fragment; associate the knowledge fragments and knowledge vectors through the index of the knowledge fragments and store them in the knowledge base.

[0178] The above-mentioned household appliance fault handling device 30, under the coordinated action of the description generation module 310, the query matching module 320, the assembly module 330, the strategy generation module 340 and the feedback module 350, first obtains the knowledge fragment that best matches the fault description of the household appliance from the external knowledge base, so as to obtain the equipment operation data, equipment history data, historical repair cases, repair data and repair instructions of the fault with the same fault, and assembles the target prompt words of the generative large language model with the matching fragments obtained by the query, so that the generative large language model can accurately locate the cause of the fault in a short time based on the knowledge fragment matching the fault (that is, the equipment operation data, equipment history data, historical repair cases, repair data and repair instructions of the fault corresponding to the fault), and automatically generates a fault repair strategy based on the maintenance cases and maintenance instructions of the same or same type of faults in the knowledge fragment as reference experience, so as to provide users or maintenance personnel with the most suitable repair plan for the faulty household appliance, greatly shorten the troubleshooting time, realize efficient, convenient and low-cost fault handling of household appliances, effectively improve user experience and meet market demand.

[0179] For the specific implementation and effect of the household appliance fault handling device 30, please refer to the description of the implementation of the household appliance fault handling method in the above text, such as, for the specific implementation and effect of the description generation module 310, please refer to the description of the relevant content of step 22 in the above text, for the specific implementation and effect of the query matching module 320, please refer to the description of the relevant content of step 23 in the above text, for the specific implementation and effect of the assembly module 330, please refer to the description of the relevant content of step 24 in the above text, for the specific implementation and effect of the strategy generation module 340, please refer to the description of the relevant content of step 25 in the above text, for the specific implementation and effect of the feedback module 350, please refer to the description of the relevant content of step 27 in the above text, and no further details will be given here.

[0180] In addition, each module of the above-mentioned household appliance fault processing device 30 can be implemented in whole or in part by software, hardware, or a combination thereof. The above-mentioned modules can be embedded in or independent of the processor 220 in the electronic device 20 in the form of hardware, or can be stored in the memory 210 of the electronic device 20 in the form of software, so that the processor 220 can call and execute the operations corresponding to the above-mentioned modules to implement the household appliance fault processing method provided above.

[0181] An embodiment of the present invention also provides an electronic device 20, including a processor 220 and a memory 210, wherein the memory 210 stores a computer program that can be executed by the processor 220, and the processor 220 can execute the computer program to implement the household appliance fault handling method proposed in the embodiment of the present invention.

[0182] The embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by the processor 220, the method for handling a household appliance fault as proposed in the embodiment of the present invention is implemented.

[0183] In summary, the household appliance fault handling method, device, system and electronic device provided by the embodiments of the present invention have at least the following beneficial effects:

[0184] (1) Accurately locate the cause of the fault in a short time, automatically generate a fault repair strategy, and provide the most appropriate repair solution for users or maintenance personnel, greatly shortening the troubleshooting time, achieving efficient, convenient, and low-cost troubleshooting of home appliance equipment, effectively improving user experience and meeting market demand;

[0185] (2) Through the first prompt word, the second prompt word and the third prompt word in the target prompt word, the large language model is generated to generate a fault repair strategy that accurately corresponds to the fault description request and the fault of the home appliance, greatly improving the accuracy of the fault repair strategy.

[0186] In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0187] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0188] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk.

[0189] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for handling household electrical appliance failure, characterized in that: The method comprises: In response to a fault repair request for any household appliance, generating a fault description of the household appliance according to the status data of the household appliance; Querying a knowledge fragment matching the fault description from a preset knowledge base; wherein the knowledge fragment is data obtained after knowledge-based fault repair documents; Based on the knowledge fragments, assembling target prompt words; Inputting the target prompt word into a preset generative large language model, and using the generative large language model to generate a fault repair strategy; Feedback on the fault repair strategy.

2. The method for handling household appliance failure according to claim 1, characterized in that: The step of assembling target prompt words based on the knowledge fragments includes: Obtaining the semantic similarity between the fault description and each of the knowledge fragments, and arranging each of the knowledge fragments in descending order of the semantic similarity to obtain a first prompt word; In combination with a preset set role and a preset answer guide, a second prompt word associated with the first prompt word is generated; wherein the second prompt word is used to indicate that the generative large language model serves as the set role, takes the first prompt word as background information, and generates an answer according to the answer guide; Generate a third prompt word according to the fault description; wherein the third prompt word is used to prompt the generative large language model of the fault problem to be processed; The first prompt word, the second prompt word and the third prompt word are assembled to obtain a target prompt word.

3. The method for handling household appliance failure according to claim 1 or 2, characterized in that: The method further comprises: In response to the status data reported by each household appliance, writing the status data and the reporting time of the status data into the status record table of the household appliance; The step of generating a fault description of the home appliance according to the status data of the home appliance in response to a fault repair request for any home appliance comprises: When receiving a fault repair request, determining the home appliance corresponding to the fault repair request; At least one set of the status data whose reporting time is before the current time is retrieved from the status record table, and the fault repair request and the status data are combined to generate a fault description.

4. The method for handling household appliance failure according to claim 1 or 2, characterized in that: The knowledge base includes a plurality of knowledge vectors and knowledge fragments indexed by the knowledge vectors; The step of searching for a knowledge fragment matching the fault description from the knowledge base includes: Vectorizing the fault description to obtain a description vector; Calculating the semantic similarity between the description vector and each knowledge vector in a preset knowledge base respectively, and determining a matching vector from each knowledge vector according to the semantic similarity; According to the index of the matching vector, a knowledge segment corresponding to the index is extracted from the knowledge base.

5. The method for handling household appliance failure according to claim 4, characterized in that: Before the step of searching the knowledge base for a knowledge segment matching the fault description, the method further includes: Pre-process the fault repair data of each type of household appliance to obtain fault repair documents; Slicing the fault repair document to obtain multiple knowledge fragments; Performing vector conversion on the knowledge fragments to obtain knowledge vectors of the knowledge fragments, and creating a unique index for each knowledge fragment; The knowledge fragment and the knowledge vector are associated with each other through the index of the knowledge fragment and then stored in the knowledge base.

6. The method for handling household appliance failure according to claim 5, characterized in that: The step of preprocessing the fault repair data of each type of household electrical appliance to obtain a fault repair document includes: Cleaning the fault repair data of each type of household electrical appliance to obtain first data; Standardizing and normalizing the values ​​in the first data to obtain second data; Perform text extraction on the second data to obtain a fault repair document.

7. The method for handling household appliance failure according to claim 1 or 2, characterized in that: The step of feeding back the fault repair strategy includes: Determine the identity information of the user of the household appliance, and retrieve the user terminal information of the user based on the identity information; The fault repair strategy is sent to the user's terminal device according to the user terminal information.

8. A household appliance fault handling device, characterized in that: It includes description generation module, query matching module, assembly module, strategy generation module and feedback module; The description generating module is used to generate a fault description of any household appliance according to the status data of the household appliance in response to a fault repair request of the household appliance; The query matching module is used to query the knowledge fragments matching the fault description from the preset knowledge base; wherein the knowledge fragments are data obtained after the fault repair document is converted into knowledge; The assembling module is used to assemble the target prompt words based on the knowledge fragments; The strategy generation module is used to input the target prompt word into a preset generative large language model, and generate a fault repair strategy using the generative large language model; The feedback module is used to feed back the fault repair strategy.

9. A household appliance fault handling system, characterized in that: It includes a central control service device, a home appliance device and a terminal device, wherein the central control service device is respectively connected to the home appliance device and the terminal device for communication; The terminal device is used to send a fault repair request to the central control service device when the home appliance fails; The home appliance is used to send a fault repair request to the central control service device when detecting a transmission fault of the home appliance itself; The central control service device is used to implement the household appliance fault handling method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program executable by the processor, and the processor can execute the computer program to implement the method for handling household appliance faults as claimed in any one of claims 1 to 7.

Images