Question and answer platform system in electric welding field

Abstract

The invention belongs to the technical field of natural language processing, and discloses an electric welding field question and answer platform system which comprises a knowledge question and answer module, a personalized module and a multi-terminal intercommunication module. The knowledge question and answer module is used for providing efficient and accurate electric welding field knowledge retrieval and answer services; the personalized recommendation module is used for providing highly customized data support in a welding scene so as to optimize user experience and improve working efficiency; the multi-terminal intercommunication module is used for realizing seamless data connection and synchronization among the mobile terminal, the PC terminal and the centralized controller terminal and ensuring that welding parameters of a welding machine can be acquired and updated in real time no matter which terminal device is used by a user to operate or inquire in an electric welding operation scene, and the multi-terminal intercommunication module further comprises a total database for centrally storing and managing historical archived data; the technical threshold of welding workers is lowered, the working efficiency is improved, and powerful support is provided for intelligent development of the welding and cutting industry.

Description

Technical Field

[0001] The present invention belongs to the technical field of natural language processing, and more specifically, relates to a question-and-answer platform system in the field of electric welding. Background Art

[0002] As the welding industry becomes increasingly important in modern industrial production, its application areas continue to expand, covering key industries such as construction, manufacturing, shipbuilding, automobiles, and aerospace. In recent years, the electric welding industry has developed rapidly, and the introduction of external automation and robotics technology has greatly improved welding efficiency and quality. However, the high cost of robots limits their widespread application, and traditional welding equipment still has significant deficiencies in terms of ease of operation and intelligence.

[0003] At present, the welding and cutting industry is facing multiple challenges. On the one hand, the continuous emergence of new welding materials and advanced processes, especially the widespread application of high-strength, lightweight materials such as aluminum alloys and titanium alloys, has put forward higher requirements for welding processes. On the other hand, changes in market demand, such as the lightweight trend in the automotive manufacturing industry, large-scale infrastructure construction, and the rapid development of the new energy field, have all posed new challenges to welding technology.

[0004] However, the fragmentation and inconsistency of the existing industry knowledge system has led to uneven skill levels among welding technicians. Most of them rely on personal experience rather than standardized operations. In addition, the existing welding and cutting equipment is relatively bulky, and technicians need to go back and forth between the workpiece and the equipment to use the knobs on the machine to adjust the appropriate parameters, which is very inconvenient to use. This not only affects the welding quality and reduces the efficiency of welding operations, but also breeds industry chaos, such as false labeling, insufficient safety and other problems.

[0005] To this end, the present invention provides a question-and-answer platform system in the field of electric welding. Summary of the invention

[0006] In view of the above-mentioned problems existing in the existing technology, the purpose of the present invention is to provide a question-and-answer platform system in the field of electric welding, which reduces the technical threshold of welding workers, improves work efficiency, and provides strong support for the intelligent development of the welding and cutting industry.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A question-and-answer platform system in the field of electric welding, comprising a knowledge question-and-answer module, a personalized module and a multi-terminal intercommunication module;

[0009] The knowledge question and answer module is used to provide efficient and accurate knowledge retrieval and answering services in the field of welding;

[0010] The personalized recommendation module is used to provide highly customized data support in welding scenarios to optimize user experience and improve work efficiency;

[0011] The multi-terminal intercommunication module is used to realize seamless data connection and synchronization between the mobile terminal, PC terminal and centralized controller terminal, ensuring that in the welding operation scene, no matter which terminal device the user uses to operate or query, the welding parameters of the welding machine can be obtained and updated in real time;

[0012] It also includes a general database for centralized storage and management of historical archive data.

[0013] As a further preferred technical solution of the present invention, the operation process of the knowledge question and answer module includes the following steps:

[0014] A1: Knowledge base encoding: Establish a welding knowledge base, divide the text in the welding knowledge base into multiple documents, encode the documents into semantic vectors through the retriever, and store them as the semantic vector library of the welding knowledge base;

[0015] A2: User input query encoding: The query text entered by the user is encoded into a semantic vector by the same retriever;

[0016] A3: Vector similarity calculation: The semantic vector of the query is calculated to be similar to all document vectors in the knowledge base. The similarity calculation uses cosine similarity to measure the similarity between semantic vectors, and selects the top k documents that are most similar to the query based on the calculated similarity.

[0017] A4: Recall document re-ranking: The top k documents most similar to the query are further processed by the re-ranker, which re-ranks the documents based on the deep relationship between the query and the documents to ensure that the most relevant documents are recalled first;

[0018] A5: Generate model response: The re-ranked documents and the input query are concatenated according to the template and used as the input of the generative model. The generative model calls on this input information to generate the final response content.

[0019] As a further preferred technical solution of the present invention, the knowledge question and answer module adopts the retrieval enhancement generation technology in natural language processing, and generates answers or responses by retrieving relevant information from a large-scale electric welding knowledge base;

[0020] The knowledge question answering module needs to use a retriever when performing document retrieval. The retriever is a bidirectional encoder model BERT. The bidirectional encoder model BERT encodes all documents in the welding knowledge base into semantic vectors one by one. The semantic vector captures the semantic information of the welding knowledge base documents and stores them as a knowledge base in vector format rather than natural language text format.

[0021] During retrieval, the text input by the user is encoded into a semantic vector through the bidirectional encoder model BERT. Then, by calculating the cosine similarity between the input query vector and the document vectors in the knowledge base, the document set most relevant to the query text is recalled to complete the preliminary retrieval of the knowledge question answering module.

[0022] The reranker of the knowledge question answering module is used to further optimize the ranking of documents based on the preliminary retrieval. The reranker re-evaluates the relevance of these documents by introducing more context and fine-grained information, ensures that the most relevant documents are ranked first, and re-ranks the top-k documents. The reranker uses the BGE-reranker model obtained by fine-tuning a larger-scale BERT variant on expert data;

[0023] The generative model of the knowledge question answering module generates the final output based on the top-k documents after retrieval and reranking. The generative model adopts the unidirectional attention model LLaMA3 with a decoder-only architecture. LLaMA3 takes the query text and the related documents recalled by the reranker as input, and combines the information of these documents to generate the final text output.

[0024] As a further preferred technical solution of the present invention, the operation process of the personalized recommendation module includes the following steps:

[0025] B1: User data label generation and storage: When users operate on the Q&A platform system, the personalized recommendation module of the Q&A platform system customizes and saves data labels to personal accounts according to the user's own needs. The Q&A platform system gradually builds a personalized preference model for users through the data labels accumulated by the personalized recommendation module.

[0026] B2: Query the general database: When a user faces a new welding scenario for the first time, he or she will first query the general database to see if there is data related to the solution. If no relevant data is found in the general database, the Q&A platform system will automatically recommend new data based on the user's personalized preference model.

[0027] B3: Recommended data adjustment and optimization: The accuracy of new data recommended by the question-answering platform system is optimized and adjusted through the accumulation of data labels of the user's personalized preference model;

[0028] B4: User feedback and data fine-tuning: When users use the recommended new data, they can adjust and fine-tune the recommended new data. If the recommended new data does not meet user needs, users can feedback to the Q&A platform system and submit the adjusted data;

[0029] B5: Personalized database iteration: After receiving feedback data, the personalized preference model of the question-and-answer platform system will continue to learn these adjustment data, and continuously iterate and update the personalized database of the personalized preference model.

[0030] As a further preferred technical solution of the present invention, the data tag not only records the user's preference for specific data, but also covers the user's unique habits in actual operations.

[0031] As a further preferred technical solution of the present invention, the operation process of the multi-terminal intercommunication module includes the following steps:

[0032] C1: Multi-terminal device data intercommunication: The multi-terminal intercommunication module is equipped with a mobile terminal, a PC terminal and a centralized controller. The mobile terminal, the PC terminal and the centralized controller are interconnected. When a user makes a query on any of the mobile and PC terminals, the Q&A platform system will synchronize the data to the centralized controller in real time.

[0033] C2: Wireless remote operation: Users can perform wireless remote operation through the multi-terminal intercommunication module of the question-and-answer platform system and set parameters for multiple welding machines at the same time;

[0034] C3: Real-time parameter monitoring and visualization: The multi-terminal intercommunication module of the question-answering platform system uses graphic visualization technology to show users key data such as voltage and current of the welding machine;

[0035] C4: Parameter recording and interface display: The multi-terminal intercommunication module is equipped with a drawing control. The drawing control set in the multi-terminal intercommunication module is combined with the centralized controller to comprehensively record and display the parameters during the welding process;

[0036] As a further preferred technical solution of the present invention, the mobile terminal, PC terminal and centralized controller multi-terminal interconnection adopts a layered reinforcement design, which mainly includes an upper layer, a middle layer and a hardware layer;

[0037] The upper layer is the client, with the mobile terminal and the PC terminal as the core. The mobile terminal and the PC terminal communicate with the server through the network. The server is connected to the knowledge question and answer module and the personalized recommendation module. The mobile terminal and the PC terminal communicate through the network or Bluetooth communication middle layer;

[0038] The middle layer is fieldbus integrated control, with the central controller as the core. The central controller communicates with the hardware layer through network communication or Bluetooth communication upper layer. The central controller is connected to the hardware layer through fieldbus: CAN, RS485, RS232 and other serial communication protocols;

[0039] The hardware layer is the bottom layer of industrial equipment, including various industrial controllers and execution terminals. The industrial controllers are connected to the middle layer through field buses: CAN, RS485, RS232 and other serial communication protocols. The execution terminals are directly connected to industrial equipment. The execution terminals receive hardware drive instructions from the industrial controllers and drive the corresponding industrial equipment to perform corresponding actions.

[0040] As a further preferred technical solution of the present invention, the centralized controller is designed with a personalized onboard design based on the RK3568 high-performance CPU, and software is developed based on the hardware platform. The industrial control software of the centralized controller is designed and developed based on Linux's QT5 to achieve data interaction with mobile phones and computer clients, and complete data collection, processing, storage, upload, status monitoring and control of multiple devices according to the field bus;

[0041] The central controller establishes a network Socket connection with the mobile terminal and the PC terminal through the TCP / IP protocol. The central controller receives instructions and data from the mobile terminal and the PC terminal, and after completing the instruction matching and parsing, sends the data to the corresponding welding machine through the field bus to realize non-contact wireless control of the welding machine.

[0042] At the same time, the centralized controller receives data from each welding machine through the field bus, analyzes it and uploads it to the mobile terminal and PC terminal synchronously.

[0043] As a further preferred technical solution of the present invention, the mobile terminal adopts an APP developed based on the Android operating system, which is compatible with mobile phone or tablet applications. The APP is provided with a voice recognition API, and the APP converts the user's voice input into text input through the API. The text input submits data to the server through a post request, and the parameters returned by the server after querying the total database are parsed and displayed by the APP;

[0044] The APP also exchanges data with the centralized controller through a socket connection, sends the user-set parameters to the centralized controller, and receives the welding machine equipment data returned by the centralized controller.

[0045] As described above, the welding field question and answer platform system provided by the present invention has the following beneficial effects:

[0046] 1. Lowering the technical threshold and promoting the popularization of skills: The present invention innovatively sets up a knowledge question and answer module and a personalized recommendation module. For newcomers in the industry who need to quickly get started with the knowledge in the field of welding, the question and answer platform system can quickly query professional knowledge and reply to the questions asked by newcomers in the industry based on the knowledge question and answer module and the personalized recommendation module. Therefore, newcomers in the industry can quickly get started with actual welding scenarios, use them immediately after checking, and learn welding knowledge in a targeted manner during the use of the product, without having to spend a lot of time and energy on all-round theoretical training for newcomers in the industry as in the past.

[0047] 2. Improve the efficiency of data and knowledge query: When querying knowledge in the field of welding, workers get answers by asking questions directly. RAG uses advanced search engines to retrieve professional knowledge from the massive welding knowledge base. Workers do not need to manually consult books and materials. The system uses efficient and intelligent computing power to quickly generate retrieval based on questions, and organizes them into clear and easy-to-understand natural language to respond to users.

[0048] 3. Through the multi-terminal intercommunication module of the question-and-answer platform system, parameters of existing bulky welding and cutting equipment can be set through wireless remote operation, avoiding the need for technicians to go back and forth between the workpiece and the equipment to use the knobs on the machine to adjust the appropriate parameters, eliminating the trouble of manual adjustment of each machine, greatly saving time and improving welding operation efficiency.

[0049] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0051] Figure 1 This is an operation flow chart of a knowledge question and answer module of a question and answer platform system in the field of electric welding applied for by the present invention;

[0052] Figure 2 This is an example of RAG technology application of a knowledge question and answer module of a question and answer platform system in the field of electric welding applied for by the present invention;

[0053] Figure 3 The RAG technology implementation scheme of the knowledge question and answer module of the welding field question and answer platform system applied for by the present invention;

[0054] Figure 4A multi-terminal intercommunication module of a welding field question and answer platform system applied for by the present invention

[0055] Figure 5 This is a schematic diagram of the interface of a centralized controller of a welding field question-and-answer platform system applied for by the present invention;

[0056] Figure 6 A network communication connection control diagram of a centralized controller of a welding field question-and-answer platform system applied for by the present invention;

[0057] Figure 7 This is a schematic diagram of the technical solution of a mobile terminal of a question-and-answer platform system in the field of welding applied for by the present invention. DETAILED DESCRIPTION

[0058] The following is a description of the implementation of the present invention by means of specific embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification.

[0059] It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the limiting conditions that the present invention can be implemented, so they have no substantial technical significance. Any modification of the structure, change in the proportional relationship, or adjustment of the size should still fall within the scope of the technical content disclosed by the present invention without affecting the effects that the present invention can produce and the purposes that can be achieved. At the same time, the terms such as "upper", "lower", "left", "right", "middle" and "one" quoted in this specification are only for the convenience of description, and are not used to limit the scope of the implementation of the present invention. The change or adjustment of the relative relationship should also be regarded as the scope of the implementation of the present invention without substantial changes in the technical content. The specific structure can be described with reference to the drawings of the patent application.

[0060] The present invention provides a question-answering platform system in the field of welding. Figures 1 to 7 As shown, it includes a knowledge question and answer module, a personalization module, and a multi-terminal intercommunication module;

[0061] The knowledge question and answer module is used to provide efficient and accurate knowledge retrieval and answering services in the field of welding;

[0062] The personalized recommendation module is used to provide highly customized data support in welding scenarios to optimize user experience and improve work efficiency;

[0063] The multi-terminal intercommunication module is used to realize seamless data connection and synchronization between the mobile terminal, PC terminal and centralized controller terminal, ensuring that in the welding operation scene, no matter which terminal device the user uses to operate or query, the welding parameters of the welding machine can be obtained and updated in real time;

[0064] It also includes a general database for centralized storage and management of historical archive data.

[0065] The operation process of the knowledge question answering module includes the following steps: Figure 1 As shown:

[0066] A1: Knowledge base encoding: Establish a welding knowledge base. The texts in the welding knowledge base are derived from various welding knowledge books, industry specification documents, and welding parameter descriptions. The texts in the welding knowledge base are segmented into multiple documents. The documents are encoded into semantic vectors through a retriever and stored as the semantic vector library of the welding knowledge base for subsequent similarity calculation;

[0067] A2: User input query encoding: The query text entered by the user is encoded into a semantic vector by the same retriever;

[0068] A3: Vector similarity calculation: The semantic vector of the query is calculated to be similar to all document vectors in the knowledge base. The similarity calculation uses cosine similarity to measure the similarity between semantic vectors, and selects the top k documents that are most similar to the query based on the calculated similarity.

[0069] A4: Recall document re-ranking: The top k documents most similar to the query are further processed by the re-ranker, which re-ranks the documents based on the deep relationship between the query and the documents to ensure that the most relevant documents are recalled first;

[0070] A5: Generate model response: The re-ranked documents and the input query are concatenated according to the template and used as the input of the generative model. The generative model calls on this input information to generate the final response content.

[0071] The knowledge question-answering module adopts the retrieval-augmented generation (RAG) technology in natural language processing (NLP). RAG is a technology that combines retrieval and generation. It is mainly used in tasks such as question-answering systems and dialogue generation. It retrieves relevant information from a large-scale welding knowledge base and then generates answers or responses, such as Figure 2 As shown;

[0072] The RAG technology implementation scheme of the knowledge question and answer module shown is as follows, combined with Figure 3As shown, the knowledge question answering module needs to use a retriever when performing document retrieval. The retriever is a bidirectional encoder model BERT. The function of the retriever is to retrieve relevant welding knowledge documents from the welding knowledge base according to the input query. The bidirectional encoder model BERT encodes all documents in the welding knowledge base into semantic vectors one by one. The semantic vector captures the semantic information of the welding knowledge base documents and stores them as a knowledge base in vector format rather than natural language text format.

[0073] During retrieval, the text input by the user is encoded into a semantic vector through the bidirectional encoder model BERT. Then, by calculating the cosine similarity between the input query vector and the document vectors in the welding knowledge base, the document set most relevant to the query text is recalled to complete the preliminary retrieval of the knowledge question answering module. This vector similarity calculation can effectively identify documents semantically related to the query text and form a preliminary candidate set.

[0074] The reranker of the knowledge question-answering module is used to further optimize the ranking of documents based on the preliminary retrieval. During the preliminary retrieval, the retriever will return multiple relevant documents, but their relevance may not be completely accurate. The reranker re-evaluates the relevance of these documents by introducing more context and fine-grained information, ensuring that the most relevant documents are ranked first, and reranks the top-k documents. The reranker uses a larger-scale BERT variant to fine-tune the model BGE-reranker on expert data, so as to have a deeper understanding of the relationship between the input query and the candidate documents. Through reranking, the question-answering platform system can more accurately rank candidate documents, because the reranker can recall more accurate and relevant enhanced information for the generation model, so the quality of the final generated content can be further improved;

[0075] The generative model of the knowledge question-answering module generates the final output based on the top-k documents after retrieval and reranking. The generative model adopts the unidirectional attention model LLaMA3 with a decoder-only architecture, which is dedicated to natural language text generation tasks. LLaMA3 takes the query text and the related documents recalled by the reranker as input, and combines the information of these documents to generate the final text output. LLaMA3 uses the content of documents in the welding field as context information and integrates it into each step of the generation process, so that the generated content can not only closely fit the input query text, but also accurately reflect the retrieved information. This method can significantly enhance the answering ability of the generative model in the welding professional field.

[0076] It should be noted that: on the one hand, RAG technology does not require a lot of computing resources to train a large language generation model, and on the other hand, it can update the welding knowledge base in real time and flexibly. For example, when the welding industry standard changes, it only needs to replace the knowledge base to control the intelligent question-answering system to generate answers to the new standard;

[0077] For example, in practical applications of the knowledge question-and-answer module, when a user asks a question such as "What is the pulse base current generally equal to the pulse peak current?", the retriever first recalls relevant documents from the knowledge base - several relevant documents in the "Welding Parameters Guide", and then the re-ranker puts the documents that can best help answer the question in the front according to their relevance to the question, and finally inputs the question and the top-ranked documents into the generative large model, which outputs the answer: "The pulse base current is generally 10% to 20% of the pulse peak current".

[0078] The operation process of the personalized recommendation module includes the following steps:

[0079] B1: User data label generation and storage: When users operate on the Q&A platform system, the personalized recommendation module of the Q&A platform system customizes and saves data labels to personal accounts according to the user's own needs. The Q&A platform system gradually builds a personalized preference model for users through the data labels accumulated by the personalized recommendation module.

[0080] B2: Query the total database: When the user faces a new welding scenario for the first time, he / she first queries the total database to see whether there is data related to the solution. If there is relevant data in the total database, the Q&A platform system will automatically match and recommend the most suitable data to help the user quickly get started with the new solution. At the same time, the Q&A platform system combines the user's personalized preference model to continuously adjust the accuracy of the recommendation to ensure that the recommended data not only meets the requirements of the new solution, but also meets the user's actual operation needs. If no relevant data is found in the total database, the Q&A platform system automatically recommends new data based on the user's personalized preference model, generates and recommends new data that is closest to the user's needs. Therefore, even if the user faces a completely new welding scenario, he / she can quickly find suitable reference data and quickly get started with the new solution through the system's intelligent recommendation.

[0081] B3: Recommended data adjustment and optimization: The accuracy of the new data recommended by the question-answering platform system is optimized and adjusted through the accumulation of data labels of the user's personalized preference model to ensure that the recommended new data not only meets the requirements of the new solution, but also fits the user's operating habits;

[0082] B4: User feedback and data fine-tuning: When users use the recommended new data, they can adjust and fine-tune the recommended new data. If the recommended new data does not meet user needs, the user will feedback to the question-and-answer platform system and submit the adjusted data. The technical team will audit the user's feedback to ensure the accuracy of the feedback and update the user's personalized preference model accordingly.

[0083] B5: Personalized database iteration: After receiving feedback data, the personalized preference model of the question-and-answer platform system will continue to learn these adjustment data, and continuously iterate and update the personalized database of the personalized preference model. Through this continuous iteration and optimization process of the personalized database, the question-and-answer platform system will continue to enhance its adaptability and intelligence level, so that each user can obtain highly personalized and accurate welding data support during use.

[0084] The data labels not only record the user's preference for specific data, but also cover the user's unique habits in actual operations.

[0085] It should be noted that the personalized preference model of the personalized recommendation module of the question-and-answer platform system, a deep learning method based on personalized data labels, enables the system to understand user needs more accurately and recommend data solutions that best suit their habits and needs in future use.

[0086] The operation process of the multi-terminal intercommunication module includes the following steps:

[0087] C1: Multi-terminal device data intercommunication: The multi-terminal intercommunication module is equipped with a mobile terminal, a PC terminal and a centralized controller. The mobile terminal, the PC terminal and the centralized controller are interconnected. When the user queries on any of the mobile terminal and the PC terminal, the question-and-answer platform system will synchronize the data to the centralized controller in real time. This synchronization mechanism ensures the reliability of the welding parameters of the welding machine in a specific welding scenario and avoids invalid welding due to user operation errors and other reasons;

[0088] C2: Wireless remote operation: Users can perform wireless remote operation through the multi-terminal intercommunication module of the question-and-answer platform system and set parameters for multiple welding machines at the same time. This operation mode saves the trouble of manual adjustment of each machine, greatly saves time and labor costs, and the system synchronizes all operation instructions to related equipment to ensure the efficient operation of the production process and reduce human operation errors;

[0089] C3: Real-time parameter monitoring and visualization: The multi-terminal intercommunication module of the question-answering platform system uses graphic visualization technology to show users key data such as voltage and current of the welding machine. The graphic visualization technology uses the OxyPlot open source graphic library. Users can intuitively monitor parameter changes during the welding process through various graphic display forms such as waveform graphs, line graphs, and bar graphs. This real-time monitoring function enables users to grasp the welding status at any time and make adjustments in time to ensure the stability and consistency of welding quality;

[0090] C4: Parameter recording and interface display: The multi-terminal intercommunication module is equipped with a drawing control. The drawing control set in the multi-terminal intercommunication module is combined with the centralized controller to comprehensively record and display the parameters during the welding process;

[0091] The multi-terminal interconnection between the mobile terminal, PC terminal and centralized controller adopts a layered reinforcement design, combined with Figure 4 As shown, it mainly includes the upper layer, the middle layer and the hardware layer;

[0092] The upper layer is the client, with the mobile terminal and the PC terminal as the core. The mobile terminal and the PC terminal communicate with the server through the network. The server is connected to the knowledge question and answer module and the personalized recommendation module. The mobile terminal and the PC terminal communicate through the network or Bluetooth communication middle layer;

[0093] The middle layer is fieldbus integrated control, with the central controller as the core. The central controller communicates with the hardware layer through network communication or Bluetooth communication upper layer. The central controller is connected to the hardware layer through fieldbus: CAN, RS485, RS232 and other serial communication protocols;

[0094] The hardware layer is the bottom layer of industrial equipment, including various industrial controllers and execution terminals. The industrial controllers are connected to the middle layer through field buses: CAN, RS485, RS232 and other serial communication protocols. The execution terminals are directly connected to industrial equipment. The execution terminals receive hardware drive instructions from the industrial controllers and drive the corresponding industrial equipment to perform corresponding actions.

[0095] The centralized controller is designed with personalized onboard customization based on the RK3568 high-performance CPU, and software is developed based on this hardware platform. The industrial control software of the centralized controller is designed and developed based on Linux's QT5 to realize data interaction with mobile phones and computer clients. At the same time, it completes data collection, processing, storage, and upload of multiple devices according to the field bus, and status monitoring and control of multiple devices. The centralized controller has onboard Bluetooth, optical fiber network, WIFI and other interfaces, and supports wireless communication in multiple modes. The interface diagram is as follows: Figure 5 As shown;

[0096] like Figure 6As shown, a network Socket connection is established between the central controller and the mobile terminal and the PC terminal through the TCPIP protocol. The central controller receives instructions and data from the mobile terminal and the PC terminal, and after completing the instruction matching and parsing, sends the data to the corresponding welding machine equipment through the field bus to realize non-contact wireless control of the welding machine. At the same time, the central controller receives data from each welding machine equipment through the field bus, parses it and uploads it synchronously to the mobile terminal and the PC terminal, including basic welding parameter data such as current, voltage, wire feeding speed, etc. during the welding process, as well as the working mode and working status of the welding equipment.

[0097] It should be noted that: with the help of the centralized controller, users can realize dynamic monitoring of the status of on-site equipment, complete the collection and recording of real-time data of the equipment in the form of log files, which is convenient for subsequent review and backtracking, and the network connection method improves the flexibility of equipment connection, and users can realize remote equipment monitoring and control. In actual working scenarios, users can connect to the centralized controller through a mobile terminal or PC, and complete non-contact parameter adjustment and status monitoring of the welding equipment through remote control, without the need for additional space movement to operate the actual equipment, thereby improving the flexibility and convenience of the working space.

[0098] Combination Figure 7 As shown, the technical solution of the mobile terminal is as follows:

[0099] The mobile terminal adopts an APP developed based on the Android operating system, which is compatible with mobile phone or tablet applications. The APP is provided with a voice recognition API. The APP converts the user's voice input into text input based on iFlytek voice technology through the API. The text input submits data to the server through a post request, and queries the database stored in the server through the HTTP protocol. The parameters returned by the server after querying the total database are parsed and displayed by the APP to guide the user to perform welding operations;

[0100] The APP also exchanges data with the centralized controller through socket connection and TCP / IP protocol, sends the user-set parameters to the centralized controller, and receives the welding machine equipment data returned by the centralized controller.

[0101] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A question-answering platform system in the field of welding, characterized in that: Including knowledge question and answer module, personalization module and multi-terminal intercommunication module; The knowledge question and answer module is used to provide efficient and accurate knowledge retrieval and answering services in the field of welding; The personalized recommendation module is used to provide highly customized data support in welding scenarios to optimize user experience and improve work efficiency; The multi-terminal intercommunication module is used to realize seamless data connection and synchronization between the mobile terminal, PC terminal and centralized controller terminal, ensuring that in the welding operation scene, no matter which terminal device the user uses to operate or query, the welding parameters of the welding machine can be obtained and updated in real time; It also includes a general database for centralized storage and management of historical archive data.

2. A welding field question and answer platform system according to claim 1, characterized in that: The operation process of the knowledge question answering module includes the following steps: A1: Knowledge base encoding: Establish a welding knowledge base, divide the text in the welding knowledge base into multiple documents, encode the documents into semantic vectors through the retriever, and store them as the semantic vector library of the welding knowledge base; A2: User input query encoding: The query text entered by the user is encoded into a semantic vector by the same retriever; A3: Vector similarity calculation: The semantic vector of the query is calculated to be similar to all document vectors in the knowledge base. The similarity calculation uses cosine similarity to measure the similarity between semantic vectors, and selects the top k documents that are most similar to the query based on the calculated similarity. A4: Recall document re-ranking: The top k documents most similar to the query are further processed by the re-ranker, which re-ranks the documents based on the deep relationship between the query and the documents to ensure that the most relevant documents are recalled first; A5: Generate model response: The re-ranked documents and the input query are concatenated according to the template and used as the input of the generative model. The generative model calls on this input information to generate the final response content.

3. A welding field question and answer platform system according to claim 2, characterized in that: The knowledge question-answering module adopts the retrieval enhancement generation technology in natural language processing, and generates answers or responses by retrieving relevant information from a large-scale electric welding knowledge base; The knowledge question answering module needs to use a retriever when performing document retrieval. The retriever is a bidirectional encoder model BERT. The bidirectional encoder model BERT encodes all documents in the welding knowledge base into semantic vectors one by one. The semantic vector captures the semantic information of the welding knowledge base documents and stores them as a knowledge base in vector format rather than natural language text format. During retrieval, the text input by the user is encoded into a semantic vector through the bidirectional encoder model BERT. Then, by calculating the cosine similarity between the input query vector and the document vectors in the knowledge base, the document set most relevant to the query text is recalled to complete the preliminary retrieval of the knowledge question answering module. The reranker of the knowledge question answering module is used to further optimize the ranking of documents based on the preliminary retrieval. The reranker re-evaluates the relevance of these documents by introducing more context and fine-grained information, ensures that the most relevant documents are ranked first, and re-ranks the top-k documents. The reranker uses the BGE-reranker model obtained by fine-tuning a larger-scale BERT variant on expert data; The generative model of the knowledge question answering module generates the final output based on the top-k documents after retrieval and reranking. The generative model adopts the unidirectional attention model LLaMA3 with a decoder-only architecture. LLaMA3 takes the query text and the related documents recalled by the reranker as input, and combines the information of these documents to generate the final text output.

4. The welding field question and answer platform system according to claim 1, characterized in that: The operation process of the personalized recommendation module includes the following steps: B1: User data label generation and storage: When users operate on the Q&A platform system, the personalized recommendation module of the Q&A platform system customizes and saves data labels to personal accounts according to the user's own needs. The Q&A platform system gradually builds a personalized preference model for users through the data labels accumulated by the personalized recommendation module. B2: Query the general database: When a user faces a new welding scenario for the first time, he or she will first query the general database to see if there is data related to the solution. If no relevant data is found in the general database, the Q&A platform system will automatically recommend new data based on the user's personalized preference model. B3: Recommended data adjustment and optimization: The accuracy of new data recommended by the question-answering platform system is optimized and adjusted through the accumulation of data labels of the user's personalized preference model; B4: User feedback and data fine-tuning: When users use the recommended new data, they can adjust and fine-tune the recommended new data. If the recommended new data does not meet user needs, users can feedback to the Q&A platform system and submit the adjusted data; B5: Personalized database iteration: After receiving feedback data, the personalized preference model of the question-and-answer platform system will continue to learn these adjustment data, and continuously iterate and update the personalized database of the personalized preference model.

5. The welding field question and answer platform system according to claim 4, characterized in that: The data labels not only record the user's preference for specific data, but also cover the user's unique habits in actual operations.

6. The welding field question and answer platform system according to claim 1, characterized in that: The operation process of the multi-terminal intercommunication module includes the following steps: C1: Multi-terminal device data intercommunication: The multi-terminal intercommunication module is equipped with a mobile terminal, a PC terminal and a centralized controller. The mobile terminal, the PC terminal and the centralized controller are interconnected. When a user makes a query on any of the mobile and PC terminals, the Q&A platform system will synchronize the data to the centralized controller in real time. C2: Wireless remote operation: Users can perform wireless remote operation through the multi-terminal intercommunication module of the question-and-answer platform system and set parameters for multiple welding machines at the same time; C3: Real-time parameter monitoring and visualization: The multi-terminal intercommunication module of the question-answering platform system uses graphic visualization technology to show users key data such as voltage and current of the welding machine; C4: Parameter recording and interface display: The multi-terminal intercommunication module is provided with a drawing control. The drawing control provided in the multi-terminal intercommunication module is combined with the centralized controller to comprehensively record and display the parameters in the welding process.

7. The welding field question and answer platform system according to claim 6, characterized in that: The multi-terminal interconnection between the mobile terminal, PC terminal and centralized controller adopts a layered reinforcement design, which mainly includes an upper layer, a middle layer and a hardware layer; The upper layer is the client, with the mobile terminal and the PC terminal as the core. The mobile terminal and the PC terminal communicate with the server through the network. The server is connected to the knowledge question and answer module and the personalized recommendation module. The mobile terminal and the PC terminal communicate through the network or Bluetooth communication middle layer; The middle layer is fieldbus integrated control, with the central controller as the core. The central controller communicates with the hardware layer through network communication or Bluetooth communication upper layer. The central controller is connected to the hardware layer through fieldbus: CAN, RS485, RS232 and other serial communication protocols; The hardware layer is the bottom layer of industrial equipment, including various industrial controllers and execution terminals. The industrial controllers are connected to the middle layer through field buses: CAN, RS485, RS232 and other serial communication protocols. The execution terminals are directly connected to industrial equipment. The execution terminals receive hardware drive instructions from the industrial controllers and drive the corresponding industrial equipment to perform corresponding actions.

8. The welding field question and answer platform system according to claim 7, characterized in that: The centralized controller is designed with personalized onboard customization based on the RK3568 high-performance CPU, and software is developed based on the hardware platform. The industrial control software of the centralized controller is designed and developed based on Linux's QT5 to achieve data interaction with mobile phones and computer clients, and complete data collection, processing, storage, upload, status monitoring and control of multiple devices based on the field bus; The central controller establishes a network Socket connection with the mobile terminal and the PC terminal through the TCPIP protocol. The central controller receives instructions and data from the mobile terminal and the PC terminal, and after completing the instruction matching and parsing, sends the data to the corresponding welding machine through the field bus to realize non-contact wireless control of the welding machine. At the same time, the centralized controller receives data from each welding machine through the field bus, analyzes it and uploads it to the mobile terminal and PC terminal synchronously.

9. The welding field question and answer platform system according to claim 7, characterized in that: The mobile terminal uses an APP developed based on the Android operating system, which is compatible with mobile phone or tablet applications. The APP is provided with a voice recognition API, through which the APP converts the user's voice input into text input, and the text input submits data to the server through a post request. The parameters returned after the server queries the total database are parsed and displayed by the APP; The APP also exchanges data with the centralized controller through a socket connection, sends the user-set parameters to the centralized controller, and receives the welding machine equipment data returned by the centralized controller.

Images