An information retrieval method and related equipment for high-end equipment failure maintenance scenarios

Abstract

The present application relates to an information retrieval method and related equipment for high-end equipment fault maintenance scenarios. The method includes: in response to a query request for high-end equipment fault maintenance input by a user, vectorizing the fault query field in the query request, and extracting fault description features based on the vectorization processing result; determining global retrieval weights and local retrieval weights based on the fault description features, the current operating status of the high-end equipment, and historical operation and maintenance data; obtaining a first retrieval result based on a global knowledge base retrieval based on the global retrieval weights and the fault description features; obtaining a second retrieval result based on a local knowledge base retrieval based on the local retrieval weights and the first retrieval result; generating a coherent information retrieval result based on the first retrieval result and the second retrieval result. The method achieves the effect of improving retrieval efficiency in high-end equipment fault maintenance scenarios, and improving retrieval accuracy and comprehensiveness.

Description

Technical Field

[0001] The present application relates to the technical field of high-end equipment fault maintenance, and in particular to an information retrieval method and related equipment for high-end equipment fault maintenance scenarios. Background Art

[0002] With the rapid development of science and technology, high-end equipment such as high-end CNC machine tools, large wind turbines, advanced aerospace vehicles, etc. have been widely used in various key industries, and their complexity and precision continue to rise. As a result, how to accurately and timely maintain the failures of high-end equipment to ensure its long-term stable operation is an urgent problem that needs to be solved.

[0003] In the initial stage, operation and maintenance personnel mainly relied on paper technical manuals to deal with equipment failures. However, in actual operations, when faced with problems with high-end equipment, manually flipping through massive amounts of data was extremely inefficient. Subsequently, the advancement of digital technology gave rise to document retrieval systems, which to a certain extent accelerated the speed of knowledge search. However, high-end equipment failure maintenance scenarios are complex and diverse, involving fault diagnosis at multiple modules and levels. Relying solely on existing document retrieval systems that rely on simple keyword matching for information retrieval in high-end equipment failure maintenance scenarios, there is a problem that the retrieval results are not accurate and comprehensive enough.

[0004] With respect to the above-mentioned problems in the related technologies, no effective solution has been proposed yet. Summary of the invention

[0005] An information retrieval method and related equipment for high-end equipment fault maintenance scenarios provided by an embodiment of the present invention at least solve the problem that the information retrieval method for high-end equipment fault maintenance scenarios in the related art has low retrieval efficiency and poor retrieval accuracy and comprehensiveness.

[0006] In order to solve the above problem, one aspect of an embodiment of the present invention provides an information retrieval method for high-end equipment fault maintenance scenarios, including:

[0007] In response to a query request for high-end equipment fault maintenance input by a user, vectorization is performed on the fault query field in the query request, and fault description features are extracted based on the vectorization processing result;

[0008] Determine the global search weight and local search weight based on the fault description characteristics, the current operating status of high-end equipment, and historical operation and maintenance data;

[0009] According to the global search weight and the fault description feature, the first search result is obtained by searching the global knowledge base; according to the local search weight and the first search result, the second search result is obtained by searching the local knowledge base; wherein the global knowledge base and the local knowledge base are respectively constructed based on the equipment text data corresponding to the high-end equipment;

[0010] A coherent information retrieval result is generated based on the first retrieval result and the second retrieval result as information retrieval feedback to the query request.

[0011] In some of the embodiments, the step of determining the global search weight and the local search weight according to the fault description characteristics, the current operating status of the high-end equipment and the historical operation and maintenance data includes:

[0012] Analyze the correlation between fault description features and equipment submodules; determine suspected fault modules based on the current operating status of high-end equipment; query fault operation and maintenance cases corresponding to fault description features in historical operation and maintenance data, and determine the proportional relationship between local faults and global faults based on fault operation and maintenance cases;

[0013] Based on the degree of association, suspected fault modules and proportional relationships, the local retrieval weight and global retrieval weight are dynamically adjusted.

[0014] In some embodiments, the steps of vectorizing the fault query field in the query request and extracting the fault description feature based on the vectorization result include:

[0015] Parse the fault query field based on the pre-trained language model to capture the semantic information and contextual relationship between words and generate multiple word vectors;

[0016] Feature extraction technology is used to extract key semantic units or keywords from word vectors to generate fault description features.

[0017] In some embodiments, the global knowledge base is constructed based on equipment overall operation parameter data, equipment technical manual data, and equipment operation and maintenance record data in the equipment text data; and the step of obtaining a first search result based on the global knowledge base search according to the global search weight and the fault description feature includes:

[0018] Based on the equipment overall operation situation model, the current operation data of the high-end equipment collected in real time and the global search weight, the preliminary search results corresponding to the fault description characteristics are determined; wherein the preliminary search results include at least one fault type; the equipment overall operation situation model is constructed based on the equipment overall operation parameter data;

[0019] According to the preliminary search results, historical fault cases and the corresponding operation and maintenance data of similar equipment, the candidate equipment submodules involved in different fault types are ranked by probability;

[0020] Determine relevant fault information corresponding to the candidate equipment submodule, and generate a first search result according to the probability sorting result and the relevant fault information.

[0021] In some of the embodiments, the local knowledge base is constructed based on the module design data, module operation parameter data and module failure mode data corresponding to the equipment submodule in the equipment text data; and the step of obtaining the second search result based on the local knowledge base search according to the locality search weight and the first search result includes:

[0022] Based on the first search result, determine the target equipment submodule that needs to be partially searched:

[0023] According to the local knowledge base and the local retrieval weight, maintenance information corresponding to the relevant fault information of the target equipment submodule is queried, and a second retrieval result is generated based on the maintenance information.

[0024] In some embodiments, the step of generating a coherent information retrieval result based on the first retrieval result and the second retrieval result includes:

[0025] The first search result and the second search result are integrated based on the large language model, and a coherent information retrieval result is generated.

[0026] In some of the embodiments, before the step of constructing the global knowledge base and the local knowledge base based on the equipment text data corresponding to the high-end equipment, the method further includes:

[0027] The equipment text data corresponding to high-end equipment is preprocessed, and the preprocessing includes: text cleaning, word segmentation, and vectorized representation.

[0028] In order to solve the above problem, one aspect of an embodiment of the present invention provides an information retrieval system for high-end equipment fault maintenance scenarios, including:

[0029] A feature extraction unit, for responding to a query request for high-end equipment fault maintenance input by a user, performing vectorization processing on the fault query field in the query request, and extracting fault description features based on the vectorization processing result;

[0030] A search weight determination unit is used to determine the global search weight and the local search weight according to the fault description characteristics, the current operating status of the high-end equipment and the historical operation and maintenance data;

[0031] A retrieval unit is used to obtain a first retrieval result based on a global knowledge base search according to a global retrieval weight and a fault description feature; and obtain a second retrieval result based on a local knowledge base search according to a local retrieval weight and the first retrieval result; wherein the global knowledge base and the local knowledge base are respectively constructed based on equipment text data corresponding to high-end equipment;

[0032] The feedback unit is used to generate a coherent information retrieval result based on the first retrieval result and the second retrieval result as information retrieval feedback for the query request.

[0033] In order to solve the above problems, one aspect of an embodiment of the present invention provides an electronic device, including: a processor, and a memory for storing a program, the program including instructions, and when the instructions are executed by the processor, the processor executes any one of the above-mentioned information retrieval methods for high-end equipment fault maintenance scenarios.

[0034] In order to solve the above problems, one aspect of an embodiment of the present invention provides a non-transitory machine-readable medium storing computer instructions, where the computer instructions are used to enable a computer to execute any of the above information retrieval methods for high-end equipment fault maintenance scenarios.

[0035] The beneficial effects of the embodiments of the present invention are as follows: by adopting a query request for high-end equipment fault maintenance in response to a user input, vectorizing the fault query field in the query request, and extracting fault description features based on the vectorization processing result; determining a global retrieval weight and a local retrieval weight according to the fault description features, the current operating status of the high-end equipment, and historical operation and maintenance data; obtaining a first retrieval result based on a global knowledge base retrieval according to the global retrieval weight and the fault description features; obtaining a second retrieval result based on a local knowledge base retrieval according to the local retrieval weight and the first retrieval result; wherein the global knowledge base and the local knowledge base are respectively constructed based on the equipment text data corresponding to the high-end equipment; generating a coherent information retrieval based on the first retrieval result and the second retrieval result The search results are used as a technical means for information retrieval feedback to query requests. This overcomes the information retrieval method for high-end equipment fault maintenance scenarios in related technologies. Since only simple keyword matching is used, it is difficult to cope with the complexity and diversity of high-end equipment fault maintenance scenarios, resulting in poor accuracy and comprehensiveness of the retrieval results. This method dynamically determines the global retrieval weight and local retrieval weight based on fault description features, the current operating status of high-end equipment, and historical operation and maintenance data. Then, a global search is performed to lock the approximate range, and then a local search is performed to determine the maintenance details. This achieves the technical effect of improving the information retrieval efficiency for high-end equipment fault maintenance scenarios, improving retrieval accuracy and comprehensiveness, and effectively ensuring the long-term stable operation of high-end equipment.

[0036] The details of one or more embodiments of the invention are set forth in the following drawings and description so that other features, objects, and advantages of the invention are more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other embodiments can be obtained based on these drawings without creative work.

[0038] Figure 1 It is a main flow chart of an information retrieval method for high-end equipment fault maintenance scenarios according to an embodiment of the present invention.

[0039] Figure 2 It is a flowchart of an information retrieval method for high-end equipment failure maintenance scenarios according to another embodiment of the present invention.

[0040] Figure 3 It is a main framework diagram of an information retrieval system for high-end equipment failure maintenance scenarios according to an embodiment of the present invention.

[0041] Figure 4 It is a structural schematic diagram of the electronic device of the present invention. DETAILED DESCRIPTION

[0042] Embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not intended to limit the scope of protection of the present invention.

[0043] With the development of retrieval technology, there are three main stages of information retrieval methods for high-end equipment fault maintenance scenarios: In the first stage, paper technical manuals are mainly used to deal with equipment failures. However, in actual operation, when faced with equipment problems, the efficiency of manually reading massive amounts of data is extremely low. In the second stage, the advancement of digital technology has given rise to document retrieval systems, which have accelerated the search speed of fault maintenance information to a certain extent. However, such systems have obvious limitations. They only rely on simple keyword matching, and it is difficult to accurately locate effective information for the complex structure and working conditions of high-end equipment. In the third stage, the big model era has been entered. The RAG (Retrieval-Augmented Generation, a natural language processing technology that combines retrieval and generation) question-and-answer system based on big models has brought new changes to knowledge services. However, the current related technologies still expose significant shortcomings in the process of development. Most RAG systems adopt a single global retrieval strategy in the field of high-end equipment. Although they can grasp the overall situation of the equipment from a macro level, it is still difficult to obtain accurate and comprehensive retrieval results when dealing with specific component failures and fine process problems.

[0044] In order to solve the above problems, the embodiment of the present invention provides an information retrieval method for high-end equipment fault maintenance scenarios, such as Figure 1 As shown, the information retrieval method for high-end equipment fault maintenance scenarios mainly includes:

[0045] Step S101 , in response to a query request for high-end equipment fault maintenance input by a user, vectorization is performed on the fault query field in the query request, and fault description features are extracted based on the vectorization processing result.

[0046] Among them, the high-end equipment provided by the embodiments of the present invention includes but is not limited to high-end CNC machine tools, large wind turbines, advanced aerospace vehicles, etc.

[0047] Based on the above settings, through vectorization processing and feature extraction, the accuracy and comprehensiveness of fault diagnosis are significantly improved, while the retrieval efficiency is improved, which can better adapt to the complexity and diversity of high-end equipment fault maintenance scenarios. These technical effects provide a solid foundation for subsequent global and local retrieval, ensuring the effectiveness and reliability of the entire information retrieval method.

[0048] In some of the embodiments, the above-mentioned steps of vectorizing the fault query field in the query request and extracting fault description features based on the vectorization processing results include: parsing the fault query field based on a pre-trained language model to capture semantic information and contextual relationships between words to generate multiple word vectors; using feature extraction technology to extract key semantic units or keywords from the word vectors to generate fault description features.

[0049] Specifically, using pre-trained language models (such as BERT (Bidirectional Encoder Representations from Transformers, a pre-trained language model that can capture the semantic information and contextual relationship between words in user questions, thereby generating high-dimensional word vector representations, commonly used in the field of Natural Language Processing (NLP), GPT (Generative Pre-trained Transformer, Generative Pre-trained Transformer) series models) to vectorize the fault query field can generate high-dimensional word vectors, further improving the accuracy and richness of fault description features. Using feature extraction technology (such as attention mechanism) to extract key semantic units or keywords from word vectors can more comprehensively cover fault information and avoid missing important details. At the same time, the vectorized fault description features can be directly used for subsequent global and local retrieval, reducing information redundancy in the retrieval process and improving retrieval efficiency.

[0050] Given the complexity and diversity of high-end equipment fault maintenance scenarios, the above-mentioned vector processing and feature extraction can better adapt to such complex scenarios and provide more accurate input for subsequent fault diagnosis.

[0051] In some of the embodiments, before the step of constructing a global knowledge base and a local knowledge base based on the equipment text data corresponding to the high-end equipment, the method further includes: preprocessing the equipment text data corresponding to the high-end equipment, and the preprocessing includes: text cleaning, word segmentation processing, and vectorized representation.

[0052] Among them, the text cleaning provided by the embodiment of the present invention helps to remove useless information in the equipment text data, such as HTML tags, punctuation marks, numbers, etc., to make the data more concise and consistent, and reduce the impact of noise data on subsequent processing; word segmentation processing divides the text into independent words, which is particularly important for Chinese, because Chinese has no clear word boundaries, and word segmentation processing can better capture the semantic information in the text; vectorization representation converts the text into a numerical vector, which is convenient for computer processing and analysis and improves the operability of the data.

[0053] Through the above preprocessing steps, redundancy and noise in the data can be reduced, so that the model can better learn the effective information in the data, thereby improving the generalization ability of the model. The preprocessed data is more standardized and consistent, reducing the complexity of the data and improving the data processing efficiency in the retrieval process.

[0054] Step S102, determining the global search weight and the local search weight according to the fault description characteristics, the current operating status of the high-end equipment and the historical operation and maintenance data.

[0055] Based on the above settings, by comprehensively considering the fault description characteristics, the current operating status of high-end equipment and the historical operation and maintenance data, the global and local retrieval weights are dynamically determined, invalid retrieval steps are avoided, and the proportion of global retrieval and local retrieval is dynamically balanced. While improving the retrieval efficiency, the accuracy of fault diagnosis is significantly improved, complex fault scenarios are supported, and the real-time performance of fault diagnosis is improved.

[0056] According to a specific implementation of an embodiment of the present invention, a BERT model can be used to analyze the degree of correlation between the keywords of the user's question (the question indicated in the query request) and the various modules of the equipment, and the weights of the local retrieval and the global retrieval can be adjusted according to the direction of the question; if the question asked by the user clearly points to a local failure of a specific module, the local retrieval weight is increased and the global retrieval weight is reduced; conversely, if the question is more general and involves the overall performance of the equipment or the coordination of multiple modules, the global retrieval weight is increased.

[0057] According to a specific implementation of an embodiment of the present invention, the operating parameters of each module of the equipment can be monitored in real time. If the operating parameters of a module fluctuate abnormally and are related to user problems, the local retrieval weight is increased; if the overall performance indicators decline, the global retrieval weight is increased.

[0058] According to another specific implementation of the embodiment of the present invention, the historical operation and maintenance data can be queried for fault cases similar to the current user problem (the problem indicated in the query request), and the ratio of local faults to global problems in these cases can be counted. If the historical data shows that similar problems are mostly local faults, and the maintenance records of the local faults are detailed and accurate, and have a high reference value for the diagnosis of the current problem, the local search weight is increased; if there are many global problems in the historical cases, and the solutions to the global problems are of reference significance to the current problem, the global search weight is increased.

[0059] Based on the characteristics of the above-mentioned user questions, the current operating status of the equipment, and historical operation and maintenance data, weights are dynamically assigned to local and global searches, ensuring that the retrieved information can not only deeply analyze local problems, but also provide systematic solutions based on the overall machine situation, thereby improving the accuracy and comprehensiveness of the search results.

[0060] In some of the embodiments, the steps of determining the global retrieval weight and the local retrieval weight based on the fault description characteristics, the current operating status of the high-end equipment and the historical operation and maintenance data include: analyzing the degree of correlation between the fault description characteristics and the equipment sub-modules; determining the suspected fault module based on the current operating status of the high-end equipment; querying the fault operation and maintenance cases corresponding to the fault description characteristics in the historical operation and maintenance data, and determining the proportional relationship between the local fault and the global fault based on the fault operation and maintenance cases; dynamically adjusting the local retrieval weight and the global retrieval weight based on the degree of correlation, the suspected fault module and the proportional relationship.

[0061] Specifically, by analyzing the degree of correlation between the fault description features and the equipment submodules, the module where the fault is located can be more accurately located. The suspected fault module is determined in combination with the current operating status to further narrow the scope of the fault and improve the accuracy of diagnosis. The fault operation and maintenance cases in the historical operation and maintenance data are used to determine the proportional relationship between local faults and global faults, providing a basis for weight adjustment, thereby improving the accuracy of fault diagnosis. Dynamic adjustment of retrieval weights can enable the model to better adapt to different fault scenarios and improve the generalization ability of the model; at the same time, through the learning of historical data, the model can better handle unknown or rare fault conditions and further enhance the generalization ability.

[0062] Due to the complexity and diversity of high-end equipment fault maintenance scenarios, dynamically adjusting retrieval weights can reduce unnecessary retrieval operations, improve retrieval efficiency, and enable the model to respond to current fault conditions more quickly, thereby improving the real-time nature of fault diagnosis and better adapting to such complex scenarios.

[0063] In the intelligent dynamic weight adjustment method provided by the present invention, weights are dynamically assigned to local retrieval and global retrieval based on the characteristics of user questions, the current operating status of the equipment, and historical operation and maintenance data, ensuring that the retrieved information can not only deeply analyze local problems, but also provide systematic solutions in combination with the overall situation of the machine, thereby achieving a perfect fusion of local and global information. According to a specific implementation of an embodiment of the present invention, the empirical knowledge document in the field of high-end equipment can also be converted into a python code logic to construct a rule engine, and a decision tree machine learning algorithm is used to carry this rule engine. This decision tree algorithm model contains the empirical rules for judging the field, and can automatically infer weights based on the input data characteristics and empirical rules.

[0064] Step S103, according to the global retrieval weight and the fault description characteristics, obtain the first retrieval result based on the global knowledge base retrieval; according to the local retrieval weight and the first retrieval result, obtain the second retrieval result based on the local knowledge base retrieval; wherein the global knowledge base and the local knowledge base are respectively constructed based on the equipment text data corresponding to the high-end equipment.

[0065] Among them, the above-mentioned equipment text data includes product technical documents (such as the vectors obtained by mapping the overall product documents of the crane are global) and module technical documents (such as all vectors obtained by mapping the technical documents of a sub-module are local). The equipment text data can be classified based on this, and then the global knowledge base and the local knowledge base can be constructed respectively.

[0066] The embodiment of the present invention innovatively designs a dual-track parallel search architecture, opening up two independent but interrelated paths: local search and global search. Global search focuses on building an overall operating status model of the equipment, and comprehensively controls the macro performance of the equipment by real-time collection and analysis of collaborative data between modules; local search specifically targets the internal modules of high-end equipment, accurately locks the knowledge information unique to specific components, and ensures that the root cause can be quickly focused on when facing local failures. When the search is started, the system prioritizes running global search, and with its extensive analysis of the overall operating data of the equipment and the collaborative relationship between modules, it can quickly and preliminarily locate the problem. Once the general direction is locked, the system immediately switches to targeted local search, focusing on the key local modules that have been preliminarily judged, digging into the details, accurately locating the cause of the fault, reducing the interference of other irrelevant module information, greatly improving the search efficiency, and ensuring that both the global vision and local accurate diagnosis are taken into account.

[0067] Through the above settings, an embodiment of the present invention provides an adaptive retrieval strategy that combines global retrieval and local retrieval, and combines the retrieval weights corresponding to the global and local retrieval. Since the reasonably allocated global and local retrieval weights optimize the retrieval strategy, by first performing a global retrieval, the approximate scope of the fault can be quickly located from a macro level, providing an overall fault view; then, on the basis of the global retrieval, an in-depth retrieval is performed using the local knowledge base, which can accurately locate the specific modules and details of the fault, improve the accuracy of diagnosis, reduce unnecessary retrieval operations, and improve retrieval efficiency.

[0068] In some of the embodiments, the global knowledge base is constructed based on the overall equipment operating parameter data, equipment technical manual data and equipment operation and maintenance record data in the equipment text data; the step of obtaining the first search result based on the global knowledge base search according to the global retrieval weight and the fault description feature includes: determining the preliminary search results corresponding to the fault description features based on the overall equipment operation status model, the current operation data of high-end equipment collected in real time and the global retrieval weight; wherein the preliminary search results include at least one fault type; the overall equipment operation status model is constructed based on the overall equipment operation parameter data; according to the preliminary retrieval results, historical fault cases and the operation and maintenance data corresponding to the same type of equipment, the candidate equipment sub-modules involved in different fault types are probability ranked; the relevant fault information corresponding to the candidate equipment sub-modules is determined, and the first search result is generated according to the probability ranking result and the relevant fault information.

[0069] Specifically, through the overall operation status model of the equipment, the approximate scope of the fault can be quickly located from a macro level, providing an overall fault view; combined with the current operation data of high-end equipment collected in real time, it can more accurately reflect the current fault status and improve the real-time and accuracy of diagnosis; through global retrieval weights, the retrieval strategy can be dynamically adjusted to further improve the accuracy of fault diagnosis.

[0070] According to a specific implementation of an embodiment of the present invention, a specific execution step of a global retrieval is also provided: keyword extraction and matching: extracting keywords in the fault description features, and searching the global knowledge base for documents or document fragments containing these keywords; using measurement indicators such as cosine similarity or Euclidean distance to calculate the similarity between the candidate documents and the fault description features; sorting the candidate documents according to the similarity calculation results, and selecting the top N documents as the first retrieval results.

[0071] In some of the embodiments, the local knowledge base is constructed based on module design data, module operating parameter data and module fault mode data corresponding to the equipment sub-module in the equipment text data; the step of obtaining the second search result based on the local knowledge base search according to the local retrieval weight and the first search result includes: based on the first search result, determining the target equipment sub-module that needs to be locally searched: based on the local knowledge base and the local retrieval weight, querying the maintenance information corresponding to the relevant fault information of the target equipment sub-module, and generating the second search result based on the maintenance information.

[0072] According to a specific implementation of an embodiment of the present invention, a specially designed feature extraction module can be used to carry out a detailed analysis of the complex structures inside each module of high-end equipment, going deep into the physical characteristics of each sensitive device and detailed information in its production process; and then accurately locating the root causes of possible local faults, ensuring that when facing local faults, the root causes can be quickly focused on to achieve local retrieval.

[0073] The construction of the local knowledge base is based on multiple data sources, so that the model can better adapt to different fault scenarios. Through the construction of the local knowledge base, the specific modules and details of the fault can be accurately located, and the comprehensiveness and accuracy of the retrieval results can be improved; the module design data, operating parameter data and fault mode data are used to provide rich detailed information for fault diagnosis.

[0074] Step S104: generating a coherent information retrieval result based on the first retrieval result and the second retrieval result as information retrieval feedback for the query request.

[0075] By integrating the first search result and the second search result, a coherent information retrieval result is generated, which significantly improves the coherence and accuracy of information retrieval, enhances the user interaction experience, improves the adaptability of the retrieval system, and optimizes the relevance of the retrieval results.

[0076] In some of the embodiments, the step of generating a coherent information retrieval result based on the first retrieval result and the second retrieval result includes: integrating the first retrieval result and the second retrieval result based on a large language model, and generating a coherent information retrieval result.

[0077] Specifically, the Large Language Model (LLM) has a strong context understanding and generation capability, and can effectively integrate the first search result and the second search result to generate coherent and accurate information retrieval results. By integrating global and local search results, the LLM can better understand the context of the user's query, generate responses that better meet user needs, and improve the user experience. At the same time, the LLM can dynamically adapt to different fault scenarios, and better handle complex fault scenarios through multi-round interaction and feedback mechanisms. Among them, multi-round interaction also enhances the adaptability of RAG in complex dialogue scenarios, enabling it to handle knowledge integration and complex reasoning across multiple rounds. Specifically, in the process of content integration, the priority of the search results can also be optimized through the rearrangement model to ensure that the most relevant documents are placed in the position closest to the query in the prompt, thereby improving the performance of RAG. For example, using a relevant feedback mechanism (such as the Rocchio algorithm), the query is expanded and better query results (i.e., the above-mentioned information retrieval results) are returned based on the user's relevance annotation of the preliminary results.

[0078] According to a specific implementation of the embodiment of the present invention, Figure 2 As shown, a flowchart of an information retrieval method based on an information retrieval system for high-end equipment fault maintenance scenarios is provided. The specific steps include:

[0079] 1. First, use the unstructured loader to load local documents (the above-mentioned equipment text data, which can include files in .docx, .xlsx, .pdf and other formats), and convert these files into processable text data.

[0080] 2. Extract text data Text from the loaded document and prepare for the next step of processing.

[0081] 3. Divide the extracted text data into smaller text blocks (Chunks) for subsequent processing and retrieval.

[0082] 4. Represents the text data after slicing. These text blocks will be further converted into vectors.

[0083] 5. Vectorization (Embedding): The text blocks are vectorized (Embedding) and converted into vectors. These vectors can capture the semantic information of the text and are used for subsequent similarity calculation and retrieval.

[0084] 6. Store the vector representations of all text blocks and establish a vector database (it should be noted that since this solution implements two parallel and related retrieval strategies of global retrieval and local retrieval, the established vector database includes two types of knowledge bases, a global knowledge base and a local knowledge base, which are used in global retrieval and local retrieval respectively) for fast retrieval of text blocks related to the query.

[0085] 7. The question raised by the user is received (i.e., the query request input by the user is received), which is a query request regarding the fault maintenance of high-end equipment.

[0086] 8. Vectorize the user's question (i.e. the query request mentioned above) and convert it into a vector that can be compared with the vectors in the database.

[0087] 9. The vector representing the user's problem after vectorization (i.e., the fault description feature mentioned above) is used to retrieve relevant information in the vector database.

[0088] 10. Search the vector database for the text block that is most similar to the question vector and find the relevant paragraph (Context) that may contain the answer. It is understandable that the first search result needs to be obtained based on the global knowledge base query, and then the second search result needs to be obtained based on the local knowledge base query.

[0089] 11. Combine the prompt module to preliminarily sort out the first search results and the second search results in the relevant paragraphs.

[0090] 12. Combine the user's question (query request) and the retrieved related paragraphs (the first search result and the second search result) to generate prompt words or prompt templates to guide the large language model to generate answers.

[0091] 13. Use large language models (such as the GPT series models) to process prompt words or prompt templates to generate coherent and accurate answers.

[0092] 14. The final answer generated by the large language model (i.e. the above-mentioned coherent information retrieval result) will be fed back to the user as information retrieval feedback for his query request.

[0093] The above-mentioned information retrieval method for high-end equipment fault maintenance scenarios provided by the embodiment of the present invention adopts a method of responding to a query request for high-end equipment fault maintenance input by a user, performing vectorization processing on the fault query field in the query request, and extracting fault description features based on the vectorization processing result; determining a global retrieval weight and a local retrieval weight according to the fault description features, the current operating status of the high-end equipment, and historical operation and maintenance data; obtaining a first retrieval result based on a global knowledge base retrieval according to the global retrieval weight and the fault description features; obtaining a second retrieval result based on a local knowledge base retrieval according to the local retrieval weight and the first retrieval result; wherein the global knowledge base and the local knowledge base are respectively constructed based on the equipment text data corresponding to the high-end equipment; and obtaining a second retrieval result based on the first retrieval result and the second retrieval result. The retrieval results generate coherent information retrieval results as a technical means for information retrieval feedback to query requests, overcoming the information retrieval method for high-end equipment fault maintenance scenarios in related technologies. Since only simple keyword matching is used, it is difficult to cope with the complexity and diversity of high-end equipment fault maintenance scenarios, resulting in poor accuracy and comprehensiveness of the retrieval results. It realizes the dynamic determination of global retrieval weights and local retrieval weights based on fault description features, the current operating status of high-end equipment, and historical operation and maintenance data, and then first performs a global search to lock the approximate range, and then performs a local search to determine the maintenance details information, thereby achieving the technical effect of improving the information retrieval efficiency for high-end equipment fault maintenance scenarios, improving retrieval accuracy and comprehensiveness, and effectively ensuring the long-term stable operation of high-end equipment.

[0094] Based on the above-mentioned information retrieval method for high-end equipment fault maintenance scenarios provided by the embodiment of the present invention, the embodiment of the present invention also provides an information retrieval system for high-end equipment fault maintenance scenarios, such as Figure 3 As shown, the information retrieval system 300 for high-end equipment fault maintenance scenarios includes:

[0095] The feature extraction unit 301 is used to respond to a query request for high-end equipment fault maintenance input by a user, perform vectorization processing on the fault query field in the query request, and extract fault description features based on the vectorization processing result.

[0096] Based on the above settings, through vectorization processing and feature extraction, the accuracy and comprehensiveness of fault diagnosis are significantly improved, while the retrieval efficiency is improved, which can better adapt to the complexity and diversity of high-end equipment fault maintenance scenarios. These technical effects provide a solid foundation for subsequent global and local retrieval, ensuring the effectiveness and reliability of the entire information retrieval method.

[0097] In some of the embodiments, the feature extraction unit 301 is further used to: parse the fault query field based on a pre-trained language model to capture semantic information and contextual relationships between words to generate multiple word vectors; and extract key semantic units or keywords from the word vectors using feature extraction technology to generate fault description features.

[0098] By using a pre-trained language model to vectorize the fault query field, a high-dimensional word vector can be generated, further improving the accuracy and richness of the fault description features. By using feature extraction technology (such as the attention mechanism) to extract key semantic units or keywords from the word vector, fault information can be more comprehensively covered to avoid missing important details. At the same time, the vectorized fault description features can be directly used for subsequent global and local retrieval, reducing information redundancy in the retrieval process and improving retrieval efficiency.

[0099] In some of the embodiments, the information retrieval system 300 for high-end equipment fault maintenance scenarios also includes a preprocessing unit. Before the step of constructing a global knowledge base and a local knowledge base based on the equipment text data corresponding to the high-end equipment, the preprocessing unit is used to: preprocess the equipment text data corresponding to the high-end equipment, and the preprocessing includes: text cleaning, word segmentation processing, and vectorized representation.

[0100] Through the above preprocessing steps, redundancy and noise in the data can be reduced, so that the model can better learn the effective information in the data, thereby improving the generalization ability of the model. The preprocessed data is more standardized and consistent, reducing the complexity of the data and improving the data processing efficiency in the retrieval process.

[0101] The search weight determination unit 302 is used to determine the global search weight and the local search weight according to the fault description characteristics, the current operating status of the high-end equipment and the historical operation and maintenance data.

[0102] Based on the above settings, by comprehensively considering the fault description characteristics, the current operating status of high-end equipment and the historical operation and maintenance data, the global and local retrieval weights are dynamically determined, invalid retrieval steps are avoided, and the proportion of global retrieval and local retrieval is dynamically balanced. While improving the retrieval efficiency, the accuracy of fault diagnosis is significantly improved, complex fault scenarios are supported, and the real-time performance of fault diagnosis is improved.

[0103] In some of the embodiments, the retrieval weight determination unit 302 is also used to: analyze the degree of correlation between fault description features and equipment sub-modules; determine suspected fault modules based on the current operating status of high-end equipment; query historical operation and maintenance data for fault operation and maintenance cases corresponding to the fault description features, and determine the proportional relationship between local faults and global faults based on the fault operation and maintenance cases; dynamically adjust the local retrieval weight and the global retrieval weight based on the degree of correlation, suspected fault modules and the proportional relationship.

[0104] By analyzing the degree of correlation between the fault description features and the equipment submodules, the module where the fault is located can be more accurately located. The suspected fault module can be determined in combination with the current operating status to further narrow the fault scope and improve the accuracy of diagnosis. The fault operation and maintenance cases in the historical operation and maintenance data are used to determine the proportional relationship between local faults and global faults, providing a basis for weight adjustment, thereby improving the accuracy of fault diagnosis. Dynamic adjustment of retrieval weights can make the model better adapt to different fault scenarios.

[0105] The retrieval unit 303 is used to obtain a first retrieval result based on a global knowledge base search according to a global retrieval weight and a fault description feature; and to obtain a second retrieval result based on a local knowledge base search according to a local retrieval weight and the first retrieval result; wherein the global knowledge base and the local knowledge base are respectively constructed based on equipment text data corresponding to high-end equipment.

[0106] Through the above settings, an embodiment of the present invention provides an adaptive retrieval strategy that combines global retrieval and local retrieval, and combines the retrieval weights corresponding to the global and local retrieval. Since the reasonably allocated global and local retrieval weights optimize the retrieval strategy, by first performing a global retrieval, the approximate scope of the fault can be quickly located from a macro level, providing an overall fault view; then, on the basis of the global retrieval, an in-depth retrieval is performed using the local knowledge base, which can accurately locate the specific modules and details of the fault, improve the accuracy of diagnosis, reduce unnecessary retrieval operations, and improve retrieval efficiency.

[0107] In some of the embodiments, the global knowledge base is constructed based on the equipment overall operating parameter data, equipment technical manual data and equipment operation and maintenance record data in the equipment text data; the retrieval unit 303 is also used to: determine the preliminary retrieval results corresponding to the fault description characteristics based on the equipment overall operation status model, the current operation data of high-end equipment collected in real time and the global retrieval weight; wherein the preliminary retrieval results include at least one fault type; the equipment overall operation status model is constructed based on the equipment overall operation parameter data; according to the preliminary retrieval results, historical fault cases and the operation and maintenance data corresponding to the same type of equipment, the candidate equipment sub-modules involved in different fault types are probability ranked; the relevant fault information corresponding to the candidate equipment sub-module is determined, and the first retrieval result is generated according to the probability ranking result and the relevant fault information.

[0108] Through the overall operation status model of the equipment, the approximate scope of the fault can be quickly located from a macro level, providing an overall fault view; combined with the current operation data of high-end equipment collected in real time, it can more accurately reflect the current fault status and improve the real-time and accuracy of diagnosis; through global retrieval weights, the retrieval strategy can be dynamically adjusted to further improve the accuracy of fault diagnosis.

[0109] In some of the embodiments, the above-mentioned local knowledge base is constructed based on the module design data, module operating parameter data and module failure mode data corresponding to the equipment sub-module in the equipment text data; the above-mentioned retrieval unit 303 is also used to: determine the target equipment sub-module that needs to be locally retrieved based on the first retrieval result: query the maintenance information corresponding to the relevant fault information of the target equipment sub-module according to the local knowledge base and the local retrieval weight, and generate a second retrieval result based on the maintenance information.

[0110] The construction of the local knowledge base is based on multiple data sources, so that the model can better adapt to different fault scenarios. Through the construction of the local knowledge base, the specific modules and details of the fault can be accurately located, and the comprehensiveness and accuracy of the diagnostic retrieval results can be improved; the module design data, operating parameter data and fault mode data are used to provide rich detailed information for fault diagnosis.

[0111] The feedback unit 304 is configured to generate a coherent information retrieval result based on the first retrieval result and the second retrieval result as information retrieval feedback for the query request.

[0112] By integrating the first search result and the second search result, a coherent information retrieval result is generated, which significantly improves the coherence and accuracy of information retrieval, enhances the user interaction experience, improves the adaptability of the retrieval system, and optimizes the relevance of the retrieval results.

[0113] In some of the embodiments, the feedback unit 304 is further configured to: integrate the first search result and the second search result based on the large language model, and generate a coherent information search result.

[0114] By integrating global and local search results, the large language model can better understand the context of user queries and generate responses that better meet user needs, improving user experience. At the same time, the large language model can dynamically adapt to different fault scenarios and better handle complex fault scenarios through multiple rounds of interaction and feedback mechanisms.

[0115] The above-mentioned information retrieval system for high-end equipment fault maintenance scenarios provided by an embodiment of the present invention has a feature extraction unit, which is used to respond to a query request for high-end equipment fault maintenance input by a user, vectorize the fault query field in the query request, and extract the fault description feature based on the vectorization processing result; a retrieval weight determination unit, which is used to determine the global retrieval weight and the local retrieval weight according to the fault description feature, the current operating status of the high-end equipment, and the historical operation and maintenance data; a retrieval unit, which is used to obtain a first retrieval result based on the global knowledge base retrieval according to the global retrieval weight and the fault description feature; and obtain a second retrieval result based on the local knowledge base retrieval according to the local retrieval weight and the first retrieval result; wherein the global knowledge base and the local knowledge base are respectively constructed based on the equipment text data corresponding to the high-end equipment; a feedback unit, which is used to generate a coherent information retrieval result based on the first retrieval result and the second retrieval result as information retrieval feedback for the query request. It realizes the dynamic determination of global and local retrieval weights based on fault description characteristics, the current operating status of high-end equipment, and historical operation and maintenance data. Then, a global search is performed first to lock the approximate range, and then a local search is performed to determine the maintenance details. This achieves the technical effect of improving the efficiency of information retrieval for high-end equipment fault maintenance scenarios, enhancing the accuracy and comprehensiveness of retrieval, and effectively ensuring the long-term stable operation of high-end equipment.

[0116] An embodiment of the present invention further provides a non-transitory machine-readable medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to execute the method of the embodiment of the present invention.

[0117] The embodiment of the present invention further provides a computer program product, including a computer program, wherein the computer program is used to make the computer perform the method of the embodiment of the present invention when executed by a processor of the computer. The computer program product should be understood as a software product that mainly implements the above method of the present application through a computer program.

[0118] An embodiment of the present invention further provides an electronic device, comprising: at least one processor; and a memory in communication with the at least one processor. The memory stores a computer program executable by the at least one processor, and when the computer program is executed by the at least one processor, the electronic device executes the method of the embodiment of the present invention.

[0119] refer to Figure 4 , a block diagram of an electronic device that can be used as a server or client of an embodiment of the present invention will now be described, which is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0120] like Figure 4 As shown, the electronic device includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 to a random access memory (RAM) 403. In RAM 403, various programs and data required for the operation of the electronic device can also be stored. The computing unit 401, ROM 402, and RAM 403 are connected to each other via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0121] Multiple components in the electronic device are connected to the I / O interface 405, including: an input unit 406, an output unit 407, a storage unit 408, and a communication unit 409. The input unit 406 can be any type of device that can input information to the electronic device, and the input unit 406 can receive input digital or character information, and generate key signal input related to user settings and / or function control of the electronic device. The output unit 407 can be any type of device that can present information, and can include but is not limited to a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 408 can include but is not limited to a disk, an optical disk. The communication unit 409 allows the electronic device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and can include but is not limited to a modem, a network card, an infrared communication device, and / or a wireless communication transceiver, such as a Bluetooth device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0122] The computing unit 401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a CPU, a graphics processing unit (GPU), various special artificial intelligence (AI) computing units, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 401 performs the various methods and processes described above. For example, in some embodiments, the method embodiments of the present invention may be implemented as a computer program, which is tangibly contained in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on an electronic device via ROM 402 and / or communication unit 409. In some embodiments, the computing unit 401 may be configured to perform the above-described method in any other appropriate manner (e.g., by means of firmware).

[0123] The computer programs for implementing the methods of the embodiments of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer programs are executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer programs may be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0124] In the context of an embodiment of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable signal medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, or infrared systems, devices, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0125] It should be noted that the term "including" and its variations used in the embodiments of the present invention are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments". The modifications of "one" and "multiple" mentioned in the embodiments of the present invention are illustrative rather than restrictive. Those skilled in the art should understand that unless the context clearly indicates otherwise, it should be understood as "one or more".

[0126] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0127] The various steps described in the method implementation methods provided in the embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method implementation methods may include additional steps and / or omit the steps shown. The scope of protection of the present invention is not limited in this respect.

[0128] The term "embodiment" in this specification refers to specific features, structures or characteristics described in conjunction with the embodiment that can be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. The various embodiments in this specification are described in a related manner, and the same or similar parts between the various embodiments refer to each other. In particular, for the device, equipment, and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts refer to the partial description of the method embodiment.

[0129] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the attached claims.

Claims

1. An information retrieval method for high-end equipment fault maintenance scenarios, characterized in that: include: In response to a query request for high-end equipment fault maintenance input by a user, vectorization is performed on the fault query field in the query request, and fault description features are extracted based on the vectorization processing result; Determine the global search weight and the local search weight according to the fault description characteristics, the current operating status of the high-end equipment and the historical operation and maintenance data; According to the global search weight and the fault description feature, a first search result is obtained based on a global knowledge base search; according to the local search weight and the first search result, a second search result is obtained based on a local knowledge base search; wherein the global knowledge base and the local knowledge base are respectively constructed based on equipment text data corresponding to high-end equipment; A coherent information retrieval result is generated based on the first retrieval result and the second retrieval result to serve as information retrieval feedback for the query request.

2. The method according to claim 1, characterized in that The step of determining the global search weight and the local search weight according to the fault description characteristics, the current operating status of the high-end equipment and the historical operation and maintenance data includes: Analyze the degree of correlation between the fault description features and the equipment submodules; determine the suspected fault module based on the current operating status of the high-end equipment; query the fault operation and maintenance cases corresponding to the fault description features in the historical operation and maintenance data, and determine the proportional relationship between local faults and global faults based on the fault operation and maintenance cases; Based on the association degree, the suspected fault module and the proportional relationship, the local search weight and the global search weight are dynamically adjusted.

3. The method according to claim 1, characterized in that The step of vectorizing the fault query field in the query request and extracting fault description features based on the vectorization processing result includes: Parsing the fault query field based on a pre-trained language model to capture semantic information and contextual relationships between words to generate multiple word vectors; Feature extraction technology is used to extract key semantic units or keywords from the word vector to generate fault description features.

4. The method according to claim 1, characterized in that: The global knowledge base is constructed based on the equipment overall operation parameter data, equipment technical manual data and equipment operation and maintenance record data in the equipment text data; The step of obtaining a first search result based on a global knowledge base search according to the global search weight and the fault description feature comprises: Based on the equipment overall operation situation model, the current operation data of the high-end equipment collected in real time and the global search weight, determine the preliminary search results corresponding to the fault description features; wherein the preliminary search results include at least one fault type; the equipment overall operation situation model is constructed based on the equipment overall operation parameter data; According to the preliminary search results, historical fault cases and operation and maintenance data corresponding to similar equipment, the candidate equipment submodules involved in different fault types are probabilistically ranked; Determine relevant fault information corresponding to the candidate equipment submodule, and generate a first search result according to the probability ranking result and the relevant fault information.

5. The method according to claim 1, characterized in that The local knowledge base is constructed based on the module design data, module operation parameter data and module failure mode data corresponding to the equipment submodule in the equipment text data; the step of obtaining the second search result based on the local knowledge base search according to the local search weight and the first search result includes: Based on the first search result, determine the target equipment submodule that needs to be locally searched: Maintenance information corresponding to relevant fault information of the target equipment submodule is queried according to the local knowledge base and the local retrieval weight, and a second retrieval result is generated based on the maintenance information.

6. The method according to claim 1, characterized in that The step of generating a coherent information retrieval result based on the first retrieval result and the second retrieval result comprises: The first search result and the second search result are integrated based on the large language model to generate a coherent information retrieval result.

7. The method according to claim 1, characterized in that Before the step of constructing the global knowledge base and the local knowledge base based on the equipment text data corresponding to the high-end equipment, the method further includes: The equipment text data corresponding to the high-end equipment is preprocessed, and the preprocessing includes: text cleaning, word segmentation processing, and vectorized representation.

8. An information retrieval system for high-end equipment failure maintenance scenarios, characterized in that: include: A feature extraction unit, configured to, in response to a query request for high-end equipment fault maintenance input by a user, perform vectorization processing on the fault query field in the query request, and extract fault description features based on the vectorization processing result; A search weight determination unit, used to determine a global search weight and a local search weight according to the fault description characteristics, the current operating status of the high-end equipment, and historical operation and maintenance data; A retrieval unit, configured to obtain a first retrieval result based on a global knowledge base search according to the global retrieval weight and the fault description feature; and obtain a second retrieval result based on a local knowledge base search according to the local retrieval weight and the first retrieval result; wherein the global knowledge base and the local knowledge base are respectively constructed based on equipment text data corresponding to high-end equipment; A feedback unit is used to generate a coherent information retrieval result based on the first retrieval result and the second retrieval result as information retrieval feedback for the query request.

9. An electronic device, comprising: A processor, and a memory storing a program, wherein the program comprises instructions, which, when executed by the processor, cause the processor to perform the method according to any one of claims 1 to 7.

10. A non-transitory machine-readable medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-7.

Images