Material detection method, device, equipment, medium and product

Abstract

The invention discloses a material detection method and device, equipment, a medium and a product, and relates to the technical field of artificial intelligence, and the method comprises the steps: generating a target query condition according to a target material type and a target task parameter, carrying out the detection scheme parameter query in a material detection knowledge graph according to the target query condition, and obtaining a target detection scheme parameter; according to the target detection scheme parameters, a target detection scheme is generated by using the target large language model, so that the target user detects the to-be-detected material based on the target detection scheme; obtaining target detection data, and querying a compliance judgment rule in the material detection knowledge graph to obtain a target judgment rule; and performing compliance judgment on the target detection data by adopting the target judgment rule to obtain a detection result. According to the method, the effects of automatically completing detection scheme making and detection result research and judgment are achieved, the automation and intelligence degree of material detection is improved, the influence of human factors on the detection result is weakened, and the efficiency and accuracy of material detection are improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method, apparatus, equipment, medium, and product for material detection. Background Technology

[0002] Material inspection is a crucial link in the entire lifecycle management of power grid equipment, and equipment quality is closely related to the safe and reliable operation of the power grid. In recent years, supply chain planning and reform plans have proposed to accelerate the construction of a distinctive modern digital supply chain system, build a comprehensive quality management system with digital quality control, and comprehensively improve the quality of equipment entering the grid.

[0003] Existing methods of material inspection typically rely on manual inspection by technicians based on experience. However, this method has a low degree of automation and intelligence, and is easily affected by human factors, resulting in insufficient efficiency and accuracy in material inspection. Summary of the Invention

[0004] This invention provides a method, apparatus, equipment, medium, and product for material testing, in order to solve the problem that existing material testing methods rely on manual testing by technicians based on experience, resulting in insufficient efficiency and accuracy in material testing.

[0005] According to one aspect of the present invention, a method for detecting materials is provided, the method comprising:

[0006] The type of task to be tested is determined based on the acquired task information, and the type of target material corresponding to the material to be tested is determined. The task parameters are extracted from the task information based on the task type to obtain the target task parameters.

[0007] Based on the target material type and the target task parameters, target query conditions are generated, and the detection scheme parameters are queried in the material detection knowledge graph based on the target query conditions to obtain the target detection scheme parameters;

[0008] Based on the target detection scheme parameters, a target detection scheme is generated for the material to be detected using a target large language model, so that the target user can detect the material to be detected based on the target detection scheme;

[0009] Obtain target detection data generated from the detection of the material to be tested, and query the compliance judgment rules in the material detection knowledge graph according to the type of the target material and the type of the task to be tested to obtain the target judgment rules;

[0010] The target detection data is evaluated for compliance using the aforementioned target determination rules to obtain the detection results for the material to be tested.

[0011] According to another aspect of the present invention, a material inspection device is provided, the device comprising:

[0012] The task parameter acquisition module is used to determine the type of the task to be detected and the target material type corresponding to the material to be detected based on the acquired task information to be detected, and to extract task parameters from the task information to be detected based on the task type to obtain the target task parameters.

[0013] The detection scheme parameter acquisition module is used to generate target query conditions based on the target material type and the target task parameters, and to query the detection scheme parameters in the material detection knowledge graph based on the target query conditions to obtain the target detection scheme parameters.

[0014] The detection scheme generation module is used to generate a target detection scheme for the material to be detected based on the target detection scheme parameters using a target large language model, so that the target user can detect the material to be detected based on the target detection scheme;

[0015] The judgment rule acquisition module is used to acquire target detection data generated by detecting the material to be detected, and to query compliance judgment rules in the material detection knowledge graph according to the type of the target material and the type of the task to be detected to obtain the target judgment rule;

[0016] The test result acquisition module is used to perform compliance judgment on the target test data using the target judgment rules, and obtain the test results for the material to be tested.

[0017] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0018] At least one processor; and

[0019] A memory communicatively connected to the at least one processor; wherein,

[0020] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the method described in any one of the present invention.

[0021] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the method described in any one of the present invention.

[0022] According to another aspect of the present invention, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method described in any one of the present invention.

[0023] The beneficial effects of this invention are: it achieves the effect of automatically completing the formulation of testing plans and the evaluation of testing results, improves the automation and intelligence of material testing, reduces the impact of human factors on testing results, and improves the efficiency and accuracy of material testing.

[0024] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart of a material testing method provided in Embodiment 1 of the present invention;

[0027] Figure 2 This is a flowchart of a material testing method provided in Embodiment 2 of the present invention;

[0028] Figure 3 This is a flowchart of a material testing method provided in Embodiment 3 of the present invention;

[0029] Figure 4 This is a flowchart of a material testing method provided in Embodiment 4 of the present invention;

[0030] Figure 5 This is a flowchart of a method for generating a material inspection report provided in Embodiment 5 of the present invention;

[0031] Figure 6 This is a flowchart of a method for generating a knowledge graph for material detection provided in Embodiment Six of the present invention;

[0032] Figure 7 This is a schematic diagram of the structure of a material testing device provided in Embodiment 7 of the present invention;

[0033] Figure 8 This is a schematic diagram of the structure of an electronic device that implements the material detection method of this invention. Detailed Implementation

[0034] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0035] It should be noted that the terms "candidate," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0036] With increasing demands for faster, higher-quality, and more information-based material testing, the management of material testing operations is becoming increasingly challenging. Furthermore, the market competition for material testing is intensifying. To fully leverage the advantages of our in-house testing organizations, there is an urgent need to develop a big data intelligent analysis system specifically for material testing. This system would optimize laboratory testing management, output comprehensive, end-to-end digital information on incoming goods sampling, serve the digital supply chain system, and enhance the core competitiveness of our material testing operations.

[0037] Example 1

[0038] Figure 1 This is a flowchart of a material inspection method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where, during the material inspection process, knowledge graphs and large language models are used to replace manual methods for formulating inspection plans and evaluating inspection results. This method can be executed by a material inspection device, which can be implemented in hardware and / or software. Figure 1 As shown, the method includes:

[0039] S101. Determine the type of task to be tested and the type of target material corresponding to the material to be tested based on the obtained task information, and extract the task parameters from the task information to be tested based on the task type to obtain the target task parameters.

[0040] The "task to be inspected" information refers to the set of relevant data for the task to be inspected input into the system, such as material type, quantity, tender number, submitting unit, and task priority. The "task to be inspected type" refers to the classification of the task to be inspected, defined based on its nature, purpose, or industry standards. For example, the task type could be routine incoming inspection or special quality verification.

[0041] The materials to be tested refer to the actual items or materials that need to be tested, such as transformers, lithium batteries, chemical solvents, or medical instruments. This embodiment does not limit the specific type of materials to be tested. The materials to be tested are the core objects of the test and are identified based on the information of the test task. The target material type refers to the specific classification or category identified from the "materials to be tested," which is the standard type that maps the materials to be tested to the material testing knowledge graph.

[0042] Task parameter extraction refers to the process of extracting core parameters from "task information to be detected". This is an automated step, which usually uses natural language processing or rule engines to parse information. Its importance lies in transforming unstructured information into structured data, which facilitates the system's subsequent query of the material detection knowledge graph.

[0043] Target task parameters refer to the final set of parameters obtained through the "task parameter extraction" process. These parameters guide the generation of subsequent target detection solutions. They are structured outputs, including core parameters of the task to be detected, to retrieve matching solutions within the material detection knowledge graph. Their value lies in ensuring the personalization and efficiency of the target detection solution. For example, target task parameters could be "Tender requirements: Additional noise detection," or "Completion deadline: 7 working days," and so on.

[0044] In one implementation, the system obtains the task information to be tested from an external system via an external interface, such as the task information to be tested sent by the power grid experimental management system. Further, an intent recognition model is used to identify the task intent in the task information to be tested, and the type of task to be tested is determined based on the identification result. For example, the task type could be routine incoming inspection or special quality review; alternatively, natural language processing technology can be used to parse the task information to identify keywords, contextual topics, and predefined rules to determine the type of task to be tested.

[0045] Simultaneously, the materials to be tested are extracted from the information of the tasks to be tested, and the target material type corresponding to the materials to be tested is determined by using the material classification system; or, the materials to be tested are mapped to material entity nodes in the material testing knowledge graph by means of entity linking technology, thereby determining the target material type corresponding to the materials to be tested.

[0046] Furthermore, based on the parameter templates predefined for the types of tasks to be detected, the system uses natural language processing technology to automatically scan the information of the tasks to be detected; by matching the rule base corresponding to the types of tasks to be detected, the system locates the key parameter entities in the information of the tasks to be detected, and performs normalization processing on the parameters, and finally outputs structured target task parameters.

[0047] S102. Generate target query conditions based on the target material type and target task parameters, and query the detection scheme parameters in the material detection knowledge graph based on the target query conditions to obtain the target detection scheme parameters.

[0048] Among them, the target query conditions refer to the query statements or conditional expressions generated based on the "target material type" and "target task parameters" in the testing process. These are used to search in the material testing knowledge graph and serve as the input basis for subsequent queries. The purpose is to transform the "target material type" and "target task parameters" into a machine-processable query language to ensure that the query results accurately match the testing requirements.

[0049] The material inspection knowledge graph is a structured semantic knowledge representation system built on graph theory models. It stores professional knowledge in the field of material inspection through nodes and edges, supports complex queries and reasoning, realizes the structured organization, efficient association and machine understanding of knowledge, and provides key technical support for the evolution of artificial intelligence from data-driven to knowledge-driven.

[0050] The query for testing scheme parameters refers to the process of performing a search operation in the "Material Testing Knowledge Graph" using "target query conditions" to obtain target testing scheme parameters related to the target testing scheme. Target testing scheme parameters refer to the set of parameters obtained from the knowledge graph through the "Query for Testing Scheme Parameters," and these parameters are the specific instructions or configuration items for generating the final testing scheme.

[0051] Target inspection plan parameters refer to the set of parameters obtained from the material inspection knowledge graph through the "inspection plan parameter query". These parameters are the specific instructions or configuration items for generating the final target inspection plan. For example, target inspection plan parameters may include mandatory inspection indicators, corresponding technical standards, recommended testing methods, and instrument requirements.

[0052] In one implementation, the target material type and target task parameters are first integrated into a structured target query condition through a preset mapping rule. This condition is usually expressed as a semantic query statement to ensure that the query condition can accurately reflect the detection requirements. Subsequently, the system uses the target query condition to perform graph traversal or semantic matching query in the material detection knowledge graph, and extracts the detection scheme parameters by retrieving relevant nodes and edges. Finally, the system filters and normalizes the target detection scheme parameters from the query results and outputs them.

[0053] S103. Based on the target detection scheme parameters, generate a target detection scheme for the material to be detected using the target large language model, so that the target user can detect the material to be detected based on the target detection scheme.

[0054] The target large language model refers to a domain-customized large language model specifically used in the material inspection process to generate inspection plans. It receives target inspection plan parameters and outputs executable inspection plan text or instructions. The target inspection plan refers to a specific inspection operation guide generated for the target user for the material to be inspected, including but not limited to a list of inspection items, inspection procedures for each item, instrument calibration requirements, environmental control standards, data recording templates, etc.

[0055] The target user refers to the operator or system that actually implements the target testing plan; they are the ultimate executors of the testing task. Testing the materials to be tested refers to the process by which the target user performs specific testing operations on the materials according to the target testing plan.

[0056] In one implementation, a structured prompt adapted to the material testing field is constructed based on the target detection scheme parameters through prompt word engineering. The structured prompt is then input into a target large language model that has been fine-tuned with data from the material testing field to generate a target detection scheme that includes, but is not limited to, a list of detection items, detection procedures for each item, instrument calibration requirements, environmental control standards, and data recording templates.

[0057] Furthermore, the target testing plan is broken down into task sheets according to the testing indicators, for example: "Task 1: Insulation resistance testing. Testing material: 110kV oil-immersed transformer. Instrument: megohmmeter. Testing steps: 1. Power off and discharge for 10 minutes; 2. Connect the high and low voltage terminals; 3. Apply 1000V DC voltage; 4. Read the stable value after 1 minute. Data requirements: Record the measured value, test temperature, humidity, and instrument number."

[0058] Furthermore, the target detection plan is pushed to the detection terminal held by the target user in real time through a preset communication protocol, such as WebSocket. The terminal interface of the detection terminal adopts a visual guidance design to display the target detection plan. The target user then performs the detection on the material to be tested according to the target detection plan.

[0059] S104. Obtain the target testing data generated from testing the materials to be tested, and query the compliance judgment rules in the material testing knowledge graph according to the target material type and the type of the task to be tested to obtain the target judgment rules.

[0060] Among them, target detection data refers to the dataset of detection results generated by the target user after performing actual operations on the materials to be tested based on the target detection solution. It serves as the input basis for compliance judgment, reflects the actual characteristics of the materials to be tested, and aims to provide objective evidence to support subsequent compliance assessments and ensure that the judgment results are based on real operations rather than speculation.

[0061] The query for compliance determination rules refers to the process of performing a search operation in the material testing knowledge graph to obtain compliance determination rules based on the determined target material type and the type of task to be tested. This process utilizes the node and edge relationships of the graph to retrieve matching compliance determination rules. Its purpose is to dynamically extract an authoritative and executable set of rules from the material testing knowledge graph, avoiding the errors and delays of manual searching.

[0062] The target judgment rule refers to the specific set of rules obtained from the material inspection knowledge graph through compliance judgment rule queries. As the foundation of the judgment engine, its value lies in providing a standardized and automated evaluation framework to ensure the consistency and fairness of the judgment results.

[0063] In one implementation, the testing terminal displays a step-by-step pop-up window prompting the user to enter the target testing data generated during the testing of the materials to be tested, and uses data input box format verification to avoid input errors. Furthermore, the system acquires the target testing data uploaded by the testing terminal via a preset protocol, such as MQTT. The data transmission process employs encryption algorithms to ensure security, significantly reducing the proportion of manually entered data and effectively controlling the low input error rate.

[0064] Furthermore, based on the identified target material type and the type of task to be inspected, a structured query statement is constructed through a preset ontology mapping logic. Subsequently, a graph traversal or semantic matching algorithm is executed in the material inspection knowledge graph to search for compliance rule nodes and their relational edges that are directly related to the query conditions. The query process extracts rule content through dynamic filtering and priority sorting, and normalizes the rules. Finally, standardized target determination rules are output.

[0065] S105. Use target determination rules to determine the compliance of target detection data and obtain the detection results for the material to be tested.

[0066] Compliance assessment refers to the process of evaluating and judging target testing data according to target assessment rules. Its purpose is to determine whether the target testing data meets specific requirements, such as industry standards, safety regulations, or internal policies. The testing results for the materials to be tested refer to the final conclusive report obtained after a complete test of the materials. This result is the comprehensive output after the compliance assessment of the target testing data and represents the end point of the process.

[0067] In one implementation, target judgment rules obtained from the material inspection knowledge graph are parsed and converted into operable computational logic. Then, the target inspection data of the material to be inspected is mapped and formatted according to the rule requirements to ensure data alignment with rule conditions. Next, the judgment logic is executed line by line through the rule engine, such as comparing whether the inspection value exceeds the threshold limit, verifying whether the combination of multiple parameters meets the composite conditions, or checking whether the timestamp complies with the regulatory validity period. If any target inspection data triggers an abnormal condition in the target judgment rule, then that target inspection data is determined to be non-compliant. Finally, the judgment conclusions of all target inspection data are integrated. If all target inspection data are compliant, then the inspection result for the material to be inspected is determined to be qualified; if any target inspection data is non-compliant, then the inspection result for the material to be inspected is determined to be unqualified.

[0068] This invention, through its embodiments, determines the type of task to be tested and the target material type corresponding to the material to be tested based on the acquired task information. It then extracts task parameters from the task information based on the task type to obtain target task parameters. Based on the target material type and target task parameters, it generates target query conditions and queries the material testing knowledge graph for testing scheme parameters to obtain target testing scheme parameters. Based on the target testing scheme parameters, it uses a target large language model to generate a target testing scheme for the material to be tested, enabling target users to test the material based on the target testing scheme. It acquires target testing data generated from testing the material to be tested and queries the material testing knowledge graph for compliance judgment rules based on the target material type and task type to obtain target judgment rules. Finally, it uses the target judgment rules to perform compliance judgment on the target testing data to obtain the testing results for the material to be tested. The beneficial effects are:

[0069] It has achieved the effect of automatically completing the formulation of testing plans and the analysis of testing results, improving the automation and intelligence of material testing, reducing the impact of human factors on testing results, and improving the efficiency and accuracy of material testing.

[0070] Example 2

[0071] Figure 2 This is a flowchart of a material testing method provided in Embodiment 2 of the present invention. This embodiment further optimizes and expands the above embodiments and can be combined with the various optional implementation methods described above. For example... Figure 2 As shown, the method includes:

[0072] S201. Determine the type of task to be tested and the type of target material corresponding to the material to be tested based on the obtained task information.

[0073] S202. When the task to be inspected is a routine incoming inspection, extract the general task parameters from the task information as the target task parameters; when the task to be inspected is a special quality review, extract the special task parameters from the task information as the target task parameters.

[0074] Routine arrival sampling inspection refers to random sampling inspection of daily arriving materials, which is a standardized and periodic quality monitoring task. Specialized quality review refers to in-depth inspection tasks initiated in response to specific quality problems or risks, which is a non-routine investigation-type operation. General task parameters refer to basic, universal parameters extracted from routine arrival sampling inspections, applicable to a wide range of testing scenarios for similar materials. Specialized task parameters refer to targeted, refined parameters extracted from specialized quality review, designed for specific problem scenarios.

[0075] General task parameters include at least one of the following: material type, material quantity, and general standard number; special task parameters include at least one of the following: special testing item, special standard number, and historical problem-related information.

[0076] In one implementation, if the task to be inspected is determined to be a routine arrival inspection, then structured general task parameters are extracted from the task information based on a preset general parameter template, and used as target task parameters. If the task to be inspected is determined to be a special quality review, then structured special task parameters are extracted from the task information through a dynamic parsing engine, and used as target task parameters.

[0077] Optionally, the retrieval path of the material testing knowledge graph can be limited according to the type of task to be tested. Routine arrival sampling tasks focus on "materials → mandatory test indicators → general technical standards → routine testing methods", while special quality review tasks focus on "materials → special test indicators → special standards → in-depth testing methods". Valid parameters are bound to construct accurate retrieval conditions. The availability of corresponding resources such as laboratory general / precision instruments, personnel scheduling, and testing environment are checked simultaneously and resource coordination warnings are handled. Then, the priority of rules is adjusted according to the task type to provide accurate and feasible input conditions for subsequent knowledge graph subgraph matching and testing scheme generation.

[0078] By extracting general task parameters from the task information as target task parameters when the task type to be inspected is routine incoming inspection, and extracting specific task parameters from the task information as target task parameters when the task type to be inspected is special quality review, the beneficial effects are: providing "differentiated adaptation basis" for subsequent stages, avoiding "one-size-fits-all" inspection process, and ensuring that the inspection plan, resource scheduling, and judgment standards are accurately matched with the essential needs of the task.

[0079] S203. Generate target query conditions based on the target material type and target task parameters, match material entity nodes in the material detection knowledge graph based on the target material type, and determine the target material entity node from the candidate material entity nodes included in the material detection knowledge graph based on the matching results.

[0080] Among them, material entity nodes refer to standardized entities in the material detection knowledge graph that represent the basic classification of materials. Each node defines the static characteristics of a type of material through a unique identifier and attribute set. Its role is to construct the basic semantic unit of the material detection knowledge graph and provide a classification basis for the detection logic. Target material entity nodes are the set of related nodes selected from the material detection knowledge graph according to the type of the target material during the matching process.

[0081] For example, assuming the target material type is "industrial lithium battery", the material entity nodes are matched in the material detection knowledge graph based on "industrial lithium battery", and the nodes related to "industrial lithium battery" in the candidate material entity nodes are taken as the target material entity nodes.

[0082] S204. Determine the candidate detection item entity nodes in the material detection knowledge graph that have edge relationships with the target material entity node, and determine the target detection item entity node from the candidate detection item entity nodes according to the target task parameters.

[0083] In this context, edge relationships refer to semantic associations connecting two types of entity nodes in the material inspection knowledge graph, describing the logical dependencies between entity nodes through predefined relationship types. Candidate inspection item entity nodes are dynamically selected from the target material entity nodes by traversing direct edge relationships, forming a set of alternative inspection items. Each node represents an executable inspection unit. Target inspection item entity nodes are inspection item entities locked from the candidate inspection item entity nodes based on target task parameters. A unique node is determined through parameter filtering, serving as the decision anchor for generating target inspection scheme parameters.

[0084] In one implementation, based on the identified target material entity nodes, a graph traversal algorithm is used to retrieve all directly connected detection item entity nodes along a predefined edge relationship type to form candidate detection item entity nodes; then, according to the target task parameters, the candidate detection item entity nodes are dynamically filtered, and finally, the accurately matched target detection item entity nodes are output.

[0085] S205. Identify the target entity nodes in the material testing knowledge graph that have target edge relationships with the target testing item entity nodes, and determine the target testing scheme parameters based on the target entity nodes.

[0086] Among them, target edge relationship refers to the specific semantic association connecting the target detection project entity node with other entity nodes. It carries industry rules and technical specifications through predefined relationship types. Target edge relationship includes at least one of the following: mandatory inspection indicators, corresponding technical standards, recommended testing methods, and instrument requirements.

[0087] Target entity nodes are terminal parameter carriers that are directly connected to the target test item entity nodes through target edge relationships. Each node represents a specific technical element. For example, when the target edge relationship is a mandatory test indicator, the target entity node can correspond to "insulation resistance, no-load loss, and short-circuit impedance," etc.; when the target edge relationship is a recommended test method, the target entity node can correspond to "megohmmeter method and wattmeter method," etc.; when the target edge relationship is an instrument requirement, the target entity node can correspond to "megohmmeter range 0-5000MΩ and wattmeter accuracy class 0.5," etc.

[0088] By matching material entity nodes in the material detection knowledge graph according to the target material type, and determining the target material entity node from the candidate material entity nodes included in the material detection knowledge graph based on the matching results; determining candidate detection project entity nodes in the material detection knowledge graph that have edge relationships with the target material entity node, and determining the target detection project entity node from the candidate detection project entity nodes based on the target task parameters; determining the target entity node in the material detection knowledge graph that has target edge relationships with the target detection project entity node, and determining the target detection scheme parameters based on the target entity node, the beneficial effects are: compared with the traditional method of determining detection scheme parameters by relying on experience or manual data query, this scheme automatically searches for target detection scheme parameters through nodes and edge relationships in the material detection knowledge graph, improving the efficiency of determining target detection scheme parameters, and further improving the efficiency of subsequent target detection scheme determination.

[0089] S206. Based on the target detection scheme parameters, use the target large language model to generate a target detection scheme for the material to be detected, so that the target user can detect the material to be detected based on the target detection scheme.

[0090] S207. Obtain the target testing data generated from testing the materials to be tested, and query the compliance judgment rules in the material testing knowledge graph according to the target material type and the type of the task to be tested to obtain the target judgment rules; use the target judgment rules to make compliance judgments on the target testing data to obtain the testing results for the materials to be tested.

[0091] Example 3

[0092] Figure 3This is a flowchart of a material testing method provided in Embodiment 3 of the present invention. This embodiment further optimizes and expands the above embodiments and can be combined with the various optional implementation methods described above. For example... Figure 3 As shown, the method includes:

[0093] S301. Determine the type of task to be tested and the type of target material corresponding to the material to be tested based on the obtained task information to be tested, and extract the task parameters from the task information to be tested based on the task type to obtain the target task parameters.

[0094] S302. Generate target query conditions based on the target material type and target task parameters, and query the detection scheme parameters in the material detection knowledge graph based on the target query conditions to obtain the target detection scheme parameters.

[0095] S303. Construct initial large model prompt words based on the target detection scheme parameters, and obtain the fine-tuned large model prompt words corresponding to the initial large model prompt words based on the manual fine-tuning operation of the initial large model prompt words.

[0096] The initial large-scale model prompts refer to a preliminary instruction framework automatically generated based on the parameters of the object detection scheme. Structured templates are used to transform these parameters into machine-understandable semantic instructions. Manual fine-tuning involves domain experts providing targeted optimization interventions to the initial large-scale model prompts, improving their quality through additions, deletions, and modifications. The fine-tuned large-scale model prompts are the final version of the instructions after manual fine-tuning, combining machine parsingability with the accuracy of human intent. Their core value lies in bridging the gap between the parameters of the object detection scheme and the understanding of the large language model.

[0097] For example, the prompt for fine-tuning the large model could be: "Based on GB / T, generate a routine inspection plan for 5 110kV oil-immersed transformers upon arrival. The plan must include mandatory inspection indicators and a specific noise detection plan, clearly defining the testing steps, instrument parameters, environmental requirements (temperature 20±5℃, humidity ≤70%), and data recording format for each indicator. The plan must comply with the specifications of China Southern Power Grid."

[0098] S304. Input the fine-tuned large model prompt words into the target large language model, so that the target large language model generates a detection plan based on the fine-tuned large model prompt words and outputs the target detection plan for the material to be detected, so that the target user can detect the material to be detected based on the target detection plan.

[0099] By constructing initial large-scale model prompts based on target detection scheme parameters, and obtaining fine-tuned large-scale model prompts corresponding to the initial large-scale model prompts through manual fine-tuning, the fine-tuned large-scale model prompts are input into the target large-scale language model. This allows the target large-scale language model to generate a detection scheme based on the fine-tuned prompts, outputting a target detection scheme for the material to be detected. The beneficial effects are:

[0100] Traditional manual development of detection schemes is time-consuming and struggles to guarantee compliance. In contrast, this solution automatically builds a basic framework using initial prompts and generates initial large-scale model prompts. These prompts are then fine-tuned manually, incorporating industry experience. The resulting fine-tuned large-scale model prompts enable the target language model to quickly respond and generate the target detection scheme, reducing the time required for scheme generation. Furthermore, the self-checking process of the initial large-scale model prompts, manual fine-tuning, and the target language model ensures the compliance of the target detection scheme.

[0101] S305. Obtain the target testing data generated from testing the materials to be tested, and query the compliance judgment rules in the material testing knowledge graph according to the type of target materials and the type of the task to be tested to obtain the target judgment rules.

[0102] S306. Use target determination rules to determine the compliance of target detection data and obtain the detection results for the materials to be tested.

[0103] Example 4

[0104] Figure 4 This is a flowchart of a material testing method provided in Embodiment 4 of the present invention. This embodiment further optimizes and expands the above embodiments and can be combined with the various optional implementation methods described above. For example... Figure 4 As shown, the method includes:

[0105] S401. Determine the type of task to be tested and the type of target material corresponding to the material to be tested based on the obtained task information, and extract the task parameters from the task information to be tested according to the task type to obtain the target task parameters.

[0106] S402. Generate target query conditions based on the target material type and target task parameters, and query the detection scheme parameters in the material detection knowledge graph based on the target query conditions to obtain the target detection scheme parameters; generate a target detection scheme for the material to be detected using the target large language model based on the target detection scheme parameters, so that the target user can detect the material to be detected based on the target detection scheme.

[0107] S403. Obtain the target detection data generated by the detection of the materials to be tested, and match the material entity nodes in the material detection knowledge graph according to the type of the target materials. Based on the matching results, determine the target material entity node from the candidate material entity nodes included in the material detection knowledge graph.

[0108] S404. Determine the target detection indicators included in the target detection scheme parameters, and determine the target detection indicator entity nodes from the candidate detection indicator entity nodes that have an edge relationship with the target material entity nodes based on the target detection indicators.

[0109] Among them, the target detection index refers to the core performance parameters that the detection task needs to verify. These are derived from the parameter analysis of the target detection scheme and serve as the quantitative anchor point driving the design of the detection scheme. The candidate detection index entity node refers to the set of potential detection indices retrieved from the target material entity node through edge relationships. The target detection index entity node refers to the unique detection index entity locked from the candidate detection index entity nodes based on the target detection index.

[0110] S405. Based on the type of task to be detected, determine the target judgment rule entity node from the candidate judgment rule entity nodes that have an edge relationship with the target detection index entity node, and determine the target judgment rule based on the target judgment rule entity node.

[0111] Among them, the candidate decision rule entity node refers to the rule base set connected to the target detection indicator node through edge relationships, storing compliance thresholds, environmental correction formulas, and logical decision conditions. The target decision rule entity node is the final decision rule entity node determined from the candidate decision rule entity nodes based on the type of the task to be detected, and its attributes generate the final target decision rule.

[0112] In one implementation, firstly, candidate detection indicator entity nodes that are connected to the target material entity node by edge relationships are traversed based on the target detection indicator, and the target detection indicator entity node is locked by semantic similarity calculation; then, according to the type of the task to be detected, candidate judgment rule entity nodes are retrieved from the target detection indicator entity node along specific edge relationships, and the target judgment rule entity node is selected by combining the type of the task to be detected. Finally, the structured attributes of the node are parsed to generate an executable target judgment rule.

[0113] By matching material entity nodes in the material detection knowledge graph according to the target material type, and determining the target material entity node from the candidate material entity nodes included in the material detection knowledge graph based on the matching results; determining the target detection indicators included in the target detection scheme parameters, and determining the target detection indicator entity node from the candidate detection indicator entity nodes that have edge relationships with the target material entity node based on the target detection indicator; determining the target judgment rule entity node from the candidate judgment rule entity nodes that have edge relationships with the target detection indicator entity node based on the type of the task to be detected, and determining the target judgment rule based on the target judgment rule entity node. The beneficial effects are: traditional solutions rely on technical personnel to formulate judgment rules based on experience, which consumes a lot of manpower and cannot guarantee the adaptability of the judgment rules. In contrast, this solution uses the material detection knowledge graph to automatically retrieve the target judgment rules, reducing the time spent on generating judgment rules and ensuring the adaptability of the generated judgment rules to the material type and detection indicators.

[0114] S406. Determine the target detection indicators included in the target detection scheme parameters, and determine the indicator detection data corresponding to each target detection indicator from the target detection data.

[0115] Among them, the index detection data refers to the raw quantitative results of the target detection index collected by experimental instruments.

[0116] S407. Determine the indicator judgment rules corresponding to each target detection indicator from the target judgment rules, and use each indicator judgment rule to judge the compliance of each indicator detection data; if any indicator detection data is non-compliant, determine that the detection result of the material to be tested is unqualified.

[0117] In one implementation, the 3σ principle is used to remove outliers from the indicator detection data, and the indicator detection data is then corrected. Further, the corrected indicator detection data is matched with the judgment thresholds or logical rules in the indicator judgment rules to obtain the compliance judgment results for each indicator detection data. The compliance judgment results for each indicator detection data are statistically analyzed. If any indicator detection data is non-compliant, the test result for the material to be tested is determined to be unqualified; if all indicator detection data are compliant, the test result for the material to be tested is determined to be qualified.

[0118] By determining the target detection indicators included in the target detection scheme parameters, and identifying the corresponding indicator detection data for each target detection indicator from the target detection data; by determining the indicator judgment rules corresponding to each target detection indicator from the target judgment rules, and using each indicator judgment rule to judge the compliance of each indicator detection data; and by determining that the test result of the material to be tested is unqualified if any indicator detection data is non-compliant, the beneficial effects are:

[0119] Firstly, by using indicator judgment rules to automatically determine the compliance of indicator detection data, the risk of human error can be avoided, and the efficiency of compliance judgment can also be improved.

[0120] Secondly, the global veto mechanism for "inspection failure" triggered by non-compliance of a single indicator further enhances the testing efforts for materials.

[0121] Example 5

[0122] Figure 5 This is a flowchart of a method for generating a material inspection report according to Embodiment 5 of the present invention. This embodiment further optimizes and expands the above embodiments and can be combined with the various optional implementation methods described above. Figure 5 As shown, the method includes:

[0123] S501. Match material entity nodes in the material detection knowledge graph according to the target material type, and determine the target material entity node from the candidate material entity nodes included in the material detection knowledge graph based on the matching results.

[0124] S502. Determine the target detection indicators included in the target detection scheme parameters, and determine the target detection indicator entity nodes from the candidate detection indicator entity nodes that have an edge relationship with the target material entity nodes based on the target detection indicators.

[0125] S503. Determine the target detection standard entity nodes that have edge relationships with the target detection index entity nodes, and determine the standard basis information based on the target detection standard entity nodes.

[0126] Among them, the target testing standard entity node refers to the testing standard entity directly associated with the target testing indicator entity node through predefined edge relationships in the material testing knowledge graph. It represents the technical regulations or industry standards that testing must comply with, serving as the legal basis for testing compliance. The standard basis information is the standard reference content extracted from the target testing standard entity node that can be directly embedded into the testing report. Its core value lies in transforming the testing standard entity nodes in the material testing knowledge graph into human-readable legal text, ensuring that every testing result has clear legal support and avoiding disputes caused by ambiguous descriptions of testing standards.

[0127] In one implementation, firstly, based on the graph structure of the material inspection knowledge graph, all edge relationships directly connected to the target inspection indicator entity nodes are traversed to filter out the associated candidate inspection standard entity nodes; then, the most relevant target inspection standard entity nodes are matched; then, predefined attribute data is extracted from these target inspection standard entity nodes and integrated into structured information; finally, standard basis information is generated by parsing the attribute data.

[0128] S504. Determine the basic information of the task based on the information of the task to be tested, and determine the summary of the testing plan based on the target testing plan. Based on the basic information of the task, the summary of the testing plan, the target testing data, the testing results and the standard basis information, generate a testing report for the material to be tested.

[0129] The task-based information refers to the core metadata extracted and structured from the task information to be tested, used to uniquely identify and manage the basic elements of the testing task. The testing plan summary is a concise outline generated based on the target testing plan. It condenses the core content of the target testing plan, omitting lengthy details to form an easily understandable summary of key points. This summary is obtained through the refinement and abstraction of the target testing plan. The testing report is a comprehensive document generated for the materials to be tested. It integrates the task-based information, testing plan summary, target testing data, testing results, and standard reference information to form a structured and operational final conclusion document. This report is the final output of the testing process, presenting the overall status and assessment details of the materials in a standardized format.

[0130] In one implementation, based on the standard documents stored in the material testing knowledge graph, a template rendering engine integrates basic task information, testing plan summary, target testing data, testing results, and standard reference information, while automatically inserting data visualization charts to generate a testing report. After the testing report is generated, a laboratory digital signature is added using electronic signature technology to ensure the report is tamper-proof, and finally, a PDF format report is generated, supporting browsing on testing terminals and access by the experimental management system.

[0131] By matching material entity nodes in the material detection knowledge graph according to the target material type, the target material entity node is determined from the candidate material entity nodes included in the material detection knowledge graph based on the matching results. The target detection indicators included in the target detection scheme parameters are determined, and the target detection indicator entity node is determined from the candidate detection indicator entity nodes that have edge relationships with the target material entity node. The target detection standard entity nodes that have edge relationships with the target detection indicator entity nodes are determined, and the standard basis information is determined based on the target detection standard entity nodes. The basic task information is determined based on the task information to be detected, and the detection scheme summary is determined based on the target detection scheme. Based on the task's basic information, the testing plan summary, the target testing data, the testing results, and the standard reference information, a testing report is generated for the materials to be tested. The benefits are as follows: the material testing knowledge graph is used to intelligently match the target material entity nodes and the target testing indicator entity nodes, avoiding the time-consuming and biased manual search. At the same time, the standard reference information is automatically determined based on the target testing standard entity nodes, ensuring that the report complies with industry standards. In addition, by integrating the task's basic information, the testing plan summary, the target testing data, the testing results, and the standard reference information, a comprehensive and structured report is generated, which facilitates users' rapid decision-making, auditing, and quality control, ultimately improving the professionalism and credibility of the overall material testing process.

[0132] Example 6

[0133] Figure 6 This is a flowchart of a method for generating a knowledge graph for material detection according to Embodiment Six of the present invention. This embodiment further optimizes and expands upon the above embodiments and can be combined with the various optional implementation methods described above. For example... Figure 6 As shown, the method includes:

[0134] S601. Obtain heterogeneous resource information in the field of material testing.

[0135] Heterogeneous resource information refers to a collection of raw data in the field of materials testing that is diverse in origin, format, and structure. Heterogeneous resource information includes at least one of the following: normative documents for materials testing, business-related documents for materials testing, and business data for materials testing. Normative documents for materials testing refer to mandatory technical standards and regulations issued by industry associations or international organizations, such as industry technical standards and local quality control management specifications. Business-related documents for materials testing are process documents generated internally by testing institutions, such as bidding documents for materials and testing contracts, specifying special testing items, testing methods promised by the bidder, testing scope, and time limits. Business data for materials testing are structured records generated during the testing process, including historical and real-time business data, such as historical laboratory testing reports (including structured tabular data and unstructured descriptions of the testing process), testing instrument information (testing instrument model, accuracy level, and calibration records), and information on tasks to be tested (material type, quantity, and manufacturer).

[0136] In one implementation, heterogeneous resource information in the field of material inspection is acquired and preprocessed, including: first, text cleaning is performed, using regular expressions to remove redundant characters such as page breaks and garbled characters, and correcting recognition errors using a terminology dictionary for the field of material inspection; second, format standardization is performed, converting heterogeneous formats into UTF-8 encoded text streams and structured JSON data; finally, terminology normalization is completed, eliminating synonyms and ensuring semantic consistency of heterogeneous resource information through ontology mapping.

[0137] S602. Entity recognition is performed based on heterogeneous resource information to obtain candidate recognition entities, and entity relationship extraction is performed based on heterogeneous resource information to obtain candidate entity relationships between each candidate recognition entity.

[0138] Entity recognition is a technology that automatically identifies and labels key objects from heterogeneous resource information. Candidate entities are objects in the material testing field initially extracted in the entity recognition process, including but not limited to material entities (such as material model, manufacturer, etc.), testing entities (such as insulation resistance, short-circuit impedance, noise value, including units, test environment requirements, etc.), testing standard entities (such as release year, revision version, etc.), judgment rule entities (such as numerical range, applicable environment, etc., for example: ≥1000MΩ at 20℃), testing method entities, and historical case entities.

[0139] Entity relation extraction refers to mining the semantic relationships between candidate entities to obtain candidate entity relation triples. Candidate entity relations refer to the semantic connections obtained from entity relation extraction.

[0140] In one implementation, a hybrid architecture combining pre-trained language models and conditional random fields, such as BERT-Based-CRF, is employed to identify core entity types in the field of material detection for heterogeneous resource information, yielding candidate entities. Further, a BiLSTM-Attention model combined with a remote supervision strategy is used to identify semantic relationships between candidate entities, resulting in candidate entity relations and triples. Finally, through entity linking, conflict detection, and fusion, a precise and high-quality set of triples is formed. Conflict detection and fusion follow the principle of prioritizing the latest released specifications and supplementing with industry conventions.

[0141] For example, candidate entity relationship triples can be: <110kV transformer, mandatory inspection index, insulation resistance>, <insulation resistance, judgment basis, GB / T>, <insulation resistance, test method, megohmmeter method>, and <insulation resistance, judgment threshold, ≥1000MΩ>, etc.

[0142] S603. Construct a knowledge graph for material detection based on each candidate identification entity and the candidate entity relationships between them.

[0143] In one implementation, the ontology model is first defined, and a top-level ontology framework is constructed based on the business logic of material testing. It includes five core ontology classes: material category, testing indicators, technical standards, testing methods, and historical cases. Each class implements semantic constraints through attribute definitions: for example, the material category class contains data attributes such as model, manufacturer, manufacturing date, and calibration time; the testing indicators class contains data attributes such as unit, test environment temperature range, and test environment humidity range. Classes are associated with each other through object attributes to form a unified semantic framework.

[0144] Secondly, we carried out graph storage and index optimization, and selected Neo4j graph database as storage carrier to adapt to efficient query of complex network relationships. The entity-relationship triples were imported in the form of node-edge-attribute: entities are mapped to nodes (such as 110kV transformer nodes), relations are mapped to edges (such as mandatory inspection index edges), and attributes are key-value pairs of nodes or edges (such as the unit attribute value of insulation resistance nodes is MΩ).

[0145] To improve retrieval efficiency, a tag propagation algorithm is used to cluster entities. For example, materials such as transformers, switch cabinets, and cables can be clustered into a unified category of materials. A two-level index in the form of material type-inspection index is constructed, so that the retrieval response time of a single material inspection knowledge is controlled within 100ms, which meets the real-time business requirements.

[0146] Finally, a dynamic update mechanism was established, employing a strategy that combines incremental and full updates. Incremental updates target single detection cases, writing newly generated triples into the material detection knowledge graph in real time via API interfaces to ensure timely replenishment of case knowledge. Full updates are performed monthly, addressing major knowledge changes such as revisions to industry standards and the introduction of new detection methods. This involves updating old triples in the graph using an alignment algorithm based on entity similarity while deleting outdated knowledge to ensure the timeliness of the material detection knowledge graph.

[0147] By acquiring heterogeneous resource information in the field of material testing, including at least one of material testing normative documents, material testing business-related documents, and material testing business data, entity identification is performed based on the heterogeneous resource information to obtain candidate identification entities. Furthermore, entity relationship extraction is performed based on the heterogeneous resource information to obtain candidate entity relationships between each candidate identification entity. Based on each candidate identification entity and the candidate entity relationships between them, a material testing knowledge graph is constructed. The beneficial effects are: by integrating scattered normative documents, business-related documents, and business data through the material testing knowledge graph, structured association and intelligent retrieval of multi-source knowledge are achieved, solving the problems of knowledge dispersion and poor reusability in traditional testing, and further improving the accuracy and reliability of material testing.

[0148] Optionally, the method also includes:

[0149] The complete case from this inspection is transformed into a new entity-relationship triple and written into the material inspection knowledge graph through the incremental update interface, realizing the transformation of inspection cases into knowledge units. Historical inspection data in the material inspection knowledge graph is statistically analyzed monthly, and latent knowledge is mined using association rule mining algorithms. This statistical knowledge is then added to the material inspection knowledge graph in the form of triples. This new knowledge will support the optimization of subsequent inspection schemes, ultimately achieving a continuous improvement loop and driving the iterative upgrade of material inspection capabilities.

[0150] Example 7

[0151] Figure 7 This is a schematic diagram of a material inspection device provided in Embodiment 7 of the present invention. It is applicable to situations where, during the material inspection process, knowledge graphs and large language models can replace manual methods for formulating inspection plans and evaluating inspection results. Figure 7 As shown, the device includes:

[0152] The task parameter acquisition module 71 is used to determine the type of the task to be detected and the target material type corresponding to the material to be detected based on the acquired task information to be detected, and to extract task parameters from the task information to be detected based on the task type to be detected, so as to obtain the target task parameters.

[0153] The detection scheme parameter acquisition module 72 is used to generate target query conditions based on the target material type and the target task parameters, and to query the detection scheme parameters in the material detection knowledge graph based on the target query conditions to obtain the target detection scheme parameters.

[0154] The detection scheme generation module 73 is used to generate a target detection scheme for the material to be detected based on the target detection scheme parameters using a target large language model, so that the target user can detect the material to be detected based on the target detection scheme;

[0155] The judgment rule acquisition module 74 is used to acquire target detection data generated by detecting the material to be detected, and to query the compliance judgment rule in the material detection knowledge graph according to the target material type and the type of the task to be detected, so as to obtain the target judgment rule.

[0156] The test result acquisition module 75 is used to perform compliance judgment on the target test data using the target judgment rules to obtain the test results for the material to be tested.

[0157] Optionally, the task parameter acquisition module 71 is specifically used for:

[0158] When the task to be inspected is a routine incoming inspection, the general task parameters in the task information are extracted as the target task parameters; wherein, the general task parameters include at least one of material type, material quantity and general standard number;

[0159] When the task to be tested is a special quality review, the special task parameters in the task information are extracted as the target task parameters; wherein, the special task parameters include at least one of special testing items, special standard numbers and historical problem association information.

[0160] Optionally, the detection scheme parameter acquisition module 72 is specifically used for:

[0161] According to the target material type, material entity nodes are matched in the material detection knowledge graph, and the target material entity node is determined from the candidate material entity nodes included in the material detection knowledge graph based on the matching results.

[0162] Identify candidate detection item entity nodes in the material detection knowledge graph that have edge relationships with the target material entity node, and determine the target detection item entity node from the candidate detection item entity nodes according to the target task parameters;

[0163] Identify the target entity nodes in the material testing knowledge graph that have target edge relationships with the target testing item entity nodes, and determine the target testing scheme parameters based on the target entity nodes; wherein, the target edge relationship includes at least one of mandatory testing indicators, corresponding technical standards, recommended testing methods, and instrument requirements.

[0164] Optionally, the detection scheme generation module 73 is specifically used for:

[0165] An initial large model prompt word is constructed based on the target detection scheme parameters, and a fine-tuned large model prompt word corresponding to the initial large model prompt word is obtained based on the manual fine-tuning operation of the initial large model prompt word;

[0166] The fine-tuned large model prompts are input into the target large language model, so that the target large language model generates a detection scheme based on the fine-tuned large model prompts and outputs a target detection scheme for the material to be detected.

[0167] Optionally, the determination rule acquisition module 74 is specifically used for:

[0168] According to the target material type, material entity nodes are matched in the material detection knowledge graph, and the target material entity node is determined from the candidate material entity nodes included in the material detection knowledge graph based on the matching results.

[0169] The target detection indexes included in the target detection scheme parameters are determined, and the target detection index entity nodes are determined from the candidate detection index entity nodes that have an edge relationship with the target material entity node based on the target detection indexes.

[0170] Based on the type of task to be detected, a target determination rule entity node is determined from candidate determination rule entity nodes that have an edge relationship with the target detection index entity node, and the target determination rule is determined based on the target determination rule entity node.

[0171] Optionally, the detection result acquisition module 75 is specifically used for:

[0172] The target detection indexes included in the target detection scheme parameters are determined, and the index detection data corresponding to each target detection index is determined from the target detection data.

[0173] The indicator determination rules corresponding to each of the target detection indicators are determined from the target determination rules, and the compliance determination of each indicator detection data is performed by each of the indicator determination rules.

[0174] If any of the aforementioned indicator test data is non-compliant, the test result of the material to be tested shall be determined as unqualified.

[0175] Optionally, the device further includes a test report generation module, specifically used for:

[0176] According to the target material type, material entity nodes are matched in the material detection knowledge graph, and the target material entity node is determined from the candidate material entity nodes included in the material detection knowledge graph based on the matching results.

[0177] The target detection indexes included in the target detection scheme parameters are determined, and the target detection index entity nodes are determined from the candidate detection index entity nodes that have an edge relationship with the target material entity node based on the target detection indexes.

[0178] Identify the target detection standard entity nodes that have edge relationships with the target detection index entity nodes, and determine the standard basis information based on the target detection standard entity nodes;

[0179] Based on the task information to be tested, basic task information is determined, and based on the target testing scheme, a testing scheme summary is determined. Based on the basic task information, the testing scheme summary, the target testing data, the testing results, and the standard reference information, a testing report for the material to be tested is generated.

[0180] Optionally, the material detection knowledge graph is generated in the following way:

[0181] Acquire heterogeneous resource information in the field of materials testing; wherein, the heterogeneous resource information includes at least one of materials testing normative documents, materials testing business-related documents, and materials testing business data;

[0182] Entity identification is performed based on the heterogeneous resource information to obtain candidate identification entities, and entity relationship extraction is performed based on the heterogeneous resource information to obtain candidate entity relationships between each candidate identification entity.

[0183] The material detection knowledge graph is constructed based on each of the candidate identification entities and the candidate entity relationships between them.

[0184] The material testing device provided in the embodiments of the present invention can execute the material testing method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0185] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0186] Example 8

[0187] Figure 8 A schematic diagram of an electronic device 80 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, 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 processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0188] like Figure 8 As shown, the electronic device 80 includes at least one processor 81 and a memory, such as a read-only memory (ROM) 82 and a random access memory (RAM) 83, communicatively connected to the at least one processor 81. The memory stores computer programs executable by the at least one processor. The processor 81 can perform various appropriate actions and processes based on the computer program stored in the ROM 82 or loaded from storage unit 88 into the RAM 83. The RAM 83 can also store various programs and data required for the operation of the electronic device 80. The processor 81, ROM 82, and RAM 83 are interconnected via a bus 84. An input / output (I / O) interface 85 is also connected to the bus 84.

[0189] Multiple components in electronic device 80 are connected to I / O interface 85, including: input unit 86, such as keyboard, mouse, etc.; output unit 87, such as various types of monitors, speakers, etc.; storage unit 88, such as disk, optical disk, etc.; and communication unit 89, such as network card, modem, wireless transceiver, etc. Communication unit 89 allows electronic device 80 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0190] Processor 81 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 81 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 81 performs the various methods and processes described above, such as material inspection methods.

[0191] In some embodiments, the material inspection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 88. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 80 via ROM 82 and / or communication unit 89. When the computer program is loaded into RAM 83 and executed by processor 81, one or more steps of the material inspection method described above may be performed. Alternatively, in other embodiments, processor 81 may be configured to perform the material inspection method by any other suitable means (e.g., by means of firmware).

[0192] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0193] Computer programs used to implement the methods 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 of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0194] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0195] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0196] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0197] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product within the cloud computing service system to address the shortcomings of traditional physical hosts and virtual private servers, such as high management difficulty and weak business scalability.

[0198] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and no limitation is imposed herein.

[0199] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for testing materials, characterized in that, The method includes: The type of task to be tested is determined based on the acquired task information, and the type of target material corresponding to the material to be tested is determined. The task parameters are extracted from the task information based on the task type to obtain the target task parameters. Based on the target material type and the target task parameters, target query conditions are generated, and the detection scheme parameters are queried in the material detection knowledge graph based on the target query conditions to obtain the target detection scheme parameters; Based on the target detection scheme parameters, a target detection scheme is generated for the material to be detected using a target large language model, so that the target user can detect the material to be detected based on the target detection scheme; Obtain target detection data generated from the detection of the material to be tested, and query the compliance judgment rules in the material detection knowledge graph according to the type of the target material and the type of the task to be tested to obtain the target judgment rules; The target detection data is evaluated for compliance using the aforementioned target determination rules to obtain the detection results for the material to be tested.

2. The method according to claim 1, characterized in that, The step of extracting task parameters from the task information to be detected based on the task type to obtain target task parameters includes: When the task to be inspected is a routine arrival inspection, the general task parameters in the task information are extracted as the target task parameters; wherein, the general task parameters include at least one of material type, material quantity and general standard number; When the task to be tested is a special quality review, the special task parameters in the task information are extracted as the target task parameters; wherein, the special task parameters include at least one of special testing items, special standard numbers and historical problem association information.

3. The method according to claim 1, characterized in that, The step of querying the detection scheme parameters in the material detection knowledge graph according to the target query conditions to obtain the target detection scheme parameters includes: According to the target material type, material entity nodes are matched in the material detection knowledge graph, and the target material entity node is determined from the candidate material entity nodes included in the material detection knowledge graph based on the matching results. Identify candidate detection item entity nodes in the material detection knowledge graph that have edge relationships with the target material entity node, and determine the target detection item entity node from the candidate detection item entity nodes according to the target task parameters; Identify the target entity nodes in the material testing knowledge graph that have target edge relationships with the target testing item entity nodes, and determine the target testing scheme parameters based on the target entity nodes; wherein, the target edge relationship includes at least one of mandatory testing indicators, corresponding technical standards, recommended testing methods, and instrument requirements.

4. The method according to claim 1, characterized in that, The step of generating a target detection scheme for the material to be detected using a target large language model based on the target detection scheme parameters includes: An initial large model prompt word is constructed based on the target detection scheme parameters, and a fine-tuned large model prompt word corresponding to the initial large model prompt word is obtained based on the manual fine-tuning operation of the initial large model prompt word; The fine-tuned large model prompts are input into the target large language model, so that the target large language model generates a detection scheme based on the fine-tuned large model prompts and outputs a target detection scheme for the material to be detected.

5. The method according to claim 1, characterized in that, The step involves querying the compliance determination rules in the material detection knowledge graph based on the target material type and the type of the task to be detected, to obtain the target determination rules, including: According to the target material type, material entity nodes are matched in the material detection knowledge graph, and the target material entity node is determined from the candidate material entity nodes included in the material detection knowledge graph based on the matching results. The target detection indexes included in the target detection scheme parameters are determined, and the target detection index entity nodes are determined from the candidate detection index entity nodes that have an edge relationship with the target material entity node based on the target detection indexes. Based on the type of task to be detected, a target determination rule entity node is determined from candidate determination rule entity nodes that have an edge relationship with the target detection index entity node, and the target determination rule is determined based on the target determination rule entity node.

6. The method according to claim 1, characterized in that, The step of using the target determination rule to determine the compliance of the target detection data and obtaining the detection results for the material to be tested includes: The target detection indexes included in the target detection scheme parameters are determined, and the index detection data corresponding to each target detection index is determined from the target detection data. The indicator determination rules corresponding to each of the target detection indicators are determined from the target determination rules, and the compliance determination of each indicator detection data is performed by each of the indicator determination rules. If any of the aforementioned indicator test data is non-compliant, the test result of the material to be tested shall be determined as unqualified.

7. The method according to claim 1, further comprising, after obtaining the test result for the material to be tested: According to the target material type, material entity nodes are matched in the material detection knowledge graph, and the target material entity node is determined from the candidate material entity nodes included in the material detection knowledge graph based on the matching results. The target detection indexes included in the target detection scheme parameters are determined, and the target detection index entity nodes are determined from the candidate detection index entity nodes that have an edge relationship with the target material entity node based on the target detection indexes. Identify the target detection standard entity nodes that have edge relationships with the target detection index entity nodes, and determine the standard basis information based on the target detection standard entity nodes; Based on the task information to be tested, basic task information is determined, and based on the target testing scheme, a testing scheme summary is determined. Based on the basic task information, the testing scheme summary, the target testing data, the testing results, and the standard reference information, a testing report for the material to be tested is generated.

8. The method according to claim 1, characterized in that, The material testing knowledge graph is generated in the following way: Acquire heterogeneous resource information in the field of materials testing; wherein, the heterogeneous resource information includes at least one of materials testing normative documents, materials testing business-related documents, and materials testing business data; Entity identification is performed based on the heterogeneous resource information to obtain candidate identification entities, and entity relationship extraction is performed based on the heterogeneous resource information to obtain candidate entity relationships between each candidate identification entity. The material detection knowledge graph is constructed based on each of the candidate identification entities and the candidate entity relationships between them.

9. A material detection device, characterized in that, The device includes: The task parameter acquisition module is used to determine the type of the task to be detected and the target material type corresponding to the material to be detected based on the acquired task information to be detected, and to extract the task parameters from the task information to be detected based on the task type to obtain the target task parameters. The detection scheme parameter acquisition module is used to generate target query conditions based on the target material type and the target task parameters, and to query the detection scheme parameters in the material detection knowledge graph based on the target query conditions to obtain the target detection scheme parameters. The detection scheme generation module is used to generate a target detection scheme for the material to be detected based on the target detection scheme parameters using a target large language model, so that the target user can detect the material to be detected based on the target detection scheme; The judgment rule acquisition module is used to acquire target detection data generated by detecting the material to be detected, and to query compliance judgment rules in the material detection knowledge graph according to the type of the target material and the type of the task to be detected to obtain the target judgment rule; The test result acquisition module is used to perform compliance judgment on the target test data using the target judgment rules to obtain the test results for the material to be tested.

10. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a processor to perform the method of any one of claims 1-8.

12. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-8.

Images