Intelligent inspection method and system of work order all-in-one machine for pressure-bearing special equipment

Abstract

The invention provides an intelligent inspection method and system of a work order all-in-one machine for pressure-bearing special equipment, and relates to the technical field of equipment inspection.The intelligent inspection method comprises the steps that an inspection request sent by terminal equipment is received, and inspection standards and inspection work order information corresponding to the pressure-bearing special equipment to be inspected are searched for in a pre-stored equipment information database; according to the inspection work order information, inspection parameters corresponding to the to-be-inspected pressure-bearing special equipment are determined, and a prediction model used for predicting the remaining service life of the to-be-inspected pressure-bearing special equipment is constructed based on equipment historical inspection data and equipment use environment data; and using the prediction model to predict the remaining service life of the pressure-bearing special equipment to be detected, taking a prediction result as additional detection reference information, generating a dynamic optimized detection strategy, and repeatedly executing the detection process according to the optimized detection strategy until all detection processes required in detection work order information are completed. And comprehensively generating a test report and sending the test report to the terminal equipment.

Description

Technical Field

[0001] The present invention relates to equipment inspection technology, and in particular to an intelligent inspection method and system for pressure-bearing special equipment using a work order integrated machine. Background Art

[0002] With the continuous development of industrial technology, pressure-bearing special equipment has been widely used in various industries. In order to ensure the safe operation of pressure-bearing special equipment, regular inspection is essential. Traditional inspection methods mainly rely on manual operation, which has problems such as low inspection efficiency and unstable inspection quality. In recent years, intelligent detection technology has made great progress, bringing new development opportunities to the field of pressure-bearing special equipment inspection.

[0003] In the existing technology, there have been some attempts to apply information technology to equipment inspection, which has improved the intelligence level of equipment inspection to a certain extent, but there are still the following shortcomings: First, there is a lack of overall optimization of the entire process of equipment inspection, and it is difficult to achieve adaptive dynamic adjustment of the inspection process; second, the historical inspection data and usage environment data of the equipment are not fully utilized, and the accuracy and reliability of equipment status prediction need to be improved; third, the output method of the inspection results is single, which is not convenient for inspectors to intuitively understand and make decisions. Therefore, a new intelligent inspection method and system is urgently needed to further improve the intelligence and refinement level of pressure-bearing special equipment inspection. Summary of the invention

[0004] The embodiments of the present invention provide a method and system for intelligent inspection of pressure-bearing special equipment using a work order integrated machine, which can solve the problems in the prior art.

[0005] According to a first aspect of the embodiments of the present invention, Provided is a smart inspection method for pressure-bearing special equipment using a work order integrated machine, including: Receive an inspection request sent by a terminal device, which includes equipment information of the pressure-bearing special equipment to be inspected, equipment historical inspection data, and equipment use environment data; search for inspection standards and inspection work order information corresponding to the pressure-bearing special equipment to be inspected from a pre-stored equipment information database based on the equipment information; if the corresponding inspection standards and inspection work order information are not found in the equipment information database, automatically generate corresponding inspection standards and inspection work order information based on the equipment information of the pressure-bearing special equipment to be inspected and the equipment historical inspection data; According to the inspection work order information, the inspected parameters corresponding to the pressure-bearing special equipment to be inspected are determined, and the standard parameter range corresponding to the inspected parameters is obtained from the inspection standard, and a mapping relationship between the inspected parameters and the standard parameter range is established. At the same time, the real-time parameter information of the pressure-bearing special equipment to be inspected is collected, and the collected real-time parameter information is pre-processed and converted into real-time inspection parameters corresponding to the inspected parameters. Based on the historical inspection data of the equipment and the equipment use environment data, a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected is constructed; The prediction model is used to predict the remaining service life of the pressure-bearing special equipment to be inspected, and the prediction result is used as additional inspection reference information. The real-time inspection parameters are compared with the standard parameter range, and the corresponding inspection results are generated according to the comparison results and the reference information. According to the inspection process in the inspection work order information, a dynamically optimized inspection strategy is generated, and the inspection process is repeatedly executed according to the optimized inspection strategy until all inspection processes required in the inspection work order information are completed, and an inspection report is comprehensively generated and sent to the terminal device.

[0006] In an optional embodiment, According to the equipment information, searching the inspection standard and inspection work order information corresponding to the pressure-bearing special equipment to be inspected from the pre-stored equipment information database includes: The core concepts and semantic associations in the field of pressure-bearing special equipment are formally defined through ontology modeling methods, and instantiated and filled based on industry standards and expert knowledge to build a domain knowledge graph covering the entire life cycle of pressure-bearing special equipment; A graph-based multimodal representation learning framework is used to embed the unstructured information of pressure-bearing special equipment into the constructed domain knowledge graph representation space. The information of different modal equipment is aligned in a unified semantic space through attribute mapping and graph convolutional neural network to obtain equipment knowledge representation. By using the obtained equipment knowledge representation, a similar case reasoning algorithm is designed for the pressure-bearing special equipment to be inspected through the attention mechanism. Based on the semantic similarity with historical equipment cases, the inspection item information of the most relevant cases is weighted and fused to generate the inspection work order for the pressure-bearing special initial equipment to be inspected. Based on the obtained initial equipment inspection work order, a multi-objective planning model is constructed, and the multi-objective optimization algorithm is used to solve the multi-objective planning model to obtain the optimal inspection standard and inspection work order information.

[0007] In an optional embodiment, Based on the obtained initial equipment inspection work order, a multi-objective planning model is constructed, and the multi-objective optimization algorithm is used to solve the multi-objective planning model to obtain the optimal inspection standard and inspection work order information including: The inspection item selection problem of pressure-bearing special equipment is transformed into a 0-1 decision variable, and a multi-objective programming model is constructed with maximizing fault coverage, minimizing inspection cost and minimizing inspection time as objective functions, and with dependencies, mutual exclusions and overall resource budgets between inspection items as constraints. The improved multi-objective solution algorithm is used to solve the multi-objective programming model. The population is initialized and a certain number of 0-1 coded individuals are randomly generated. Each individual corresponds to a set of test item selection schemes. The first Pareto optimal solution set is obtained through non-dominated sorting and crowding calculation. The initial particle swarm is randomly generated, each particle represents a combination of test item selections, and the first Pareto optimal solution set is used as the initial position. According to the individual optimal solution and the global optimal solution, the speed and position of each particle are updated, and the fitness value of the current position of each particle is calculated. At the same time, an external archive is introduced to dynamically save non-dominated solutions, and an adaptive grid division mechanism is used to dynamically adjust the grid size of the target space according to the distribution of solutions. In the process of updating the particle position, the individual optimal solution, the global optimal solution and the congestion of the grid where the particle is located are comprehensively considered to construct the particle speed update formula, and iterate repeatedly until the maximum number of iterations is met, and the second Pareto optimal solution set is output; The first Pareto optimal solution set and the second Pareto optimal solution set are merged, and the optimal solution is selected as the final inspection standard and inspection work order information according to the decision maker's preference.

[0008] In an optional embodiment, Based on the historical inspection data of the equipment and the equipment use environment data, a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected is constructed, including: Based on the theory of material science and damage mechanics, a physical mechanism model reflecting the degradation process of key pressure-bearing special equipment is constructed, and the degradation parameter evolution curve and theoretical life estimation value are derived to form a degradation physics knowledge base; Extract characteristic parameters closely related to the degradation process of key pressure-bearing special equipment from the equipment life cycle data, collect the equipment's historical operating parameters, environmental factors, regular inspection data and maintenance records, build a degradation feature set, and use feature selection methods to screen out the target feature subset; Based on the obtained target feature subset, the degradation features are divided into input sequences according to time windows using a long short-term memory network, and the known lifespans at each corresponding moment are used as labels. By minimizing the prediction error between the model output and the label lifespan, the nonlinear mapping relationship between the degradation features and the remaining lifespan is learned, and a data-driven degradation model is constructed. The theoretical degradation curve in the degradation physics knowledge base is discretized and aligned with the degradation feature set in time to form a physical-data hybrid feature set. Taking the physical-data hybrid feature set as input, a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected is constructed using a gated cyclic unit.

[0009] In an optional embodiment, The steps after building the prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected also include: According to the physical degradation mechanism of the pressure-bearing special equipment to be inspected, appropriate physical degradation indicators are selected as prior knowledge, and a loss function is constructed based on the data fitting loss and the difference between the physical degradation indicators and the model prediction value. The prediction model is trained by minimizing the loss function, and the model parameters are updated using the gradient descent algorithm. The iteration is repeated until the maximum number of iterations is reached, and finally a trained prediction model is obtained. The loss function calculation formula is as follows: ; Among them, L represents the loss function, T represents the total number of time steps, and β t represents the weight of time step t, represents the predicted value of the remaining life of the pressure-bearing special equipment to be inspected at time step t, y t represents the true value of the remaining life of the pressure-bearing special equipment to be inspected at time step t, λ1 represents the weight coefficient of the regularization term of the physical degradation index, N represents the number of physical degradation indicators, w i represents the weight of the ith physical degradation index in the regularization term, g i (·) represents the theoretical estimation function corresponding to the ith physical degradation index, p t，i represents the i-th physical degradation index at time step t.

[0010] In an optional embodiment, The prediction model is used to predict the remaining service life of the pressure-bearing special equipment to be inspected, and the prediction result is used as additional inspection reference information. The real-time inspection parameters are compared with the standard parameter range, and the corresponding inspection results are generated according to the comparison results and the reference information. According to the inspection process in the inspection work order information, a dynamically optimized inspection strategy is generated, including: According to the type of pressure-bearing special equipment to be inspected, similar cases are selected from the historical database, the corresponding degradation physical parameters and monitoring feature sequences are extracted, and input into the pre-trained prediction model, the prediction model is used to fit the degradation state of the pressure-bearing special equipment to be inspected, and the remaining service life prediction result is output; Deploy an online monitoring system to collect real-time inspection parameters of the pressure-bearing special equipment to be inspected, compare the collected real-time inspection parameters with the preset standard parameter range, and identify abnormal indicators that exceed the normal range; Combined with the remaining service life prediction results and the analysis results of the online monitoring system, the degradation state and failure risk of the pressure-bearing special equipment to be inspected are comprehensively evaluated. When the real-time inspection parameters exceed the preset warning threshold and the remaining service life is lower than the preset safety threshold, it is determined that the pressure-bearing special equipment to be inspected has a failure risk and a risk measurement result is generated; A risk assessment model is constructed using a fuzzy comprehensive evaluation method. The remaining service life prediction results and risk measurement results are combined to form a comprehensive equipment risk assessment matrix. Experts determine the risk level classification threshold and output the risk assessment results. Based on the risk assessment results, a structured rule base is formed according to the inspection standards and industry regulations. A rule-based reasoning engine is used to perform forward reasoning matching on the input degradation physical parameters and monitoring feature sequences, and consistency verification is performed in combination with ontology constraints to generate a preliminary inspection strategy. Based on the preliminary inspection strategy, the inspection strategy optimization model is modeled as a Markov decision process. The Markov decision process is solved by an adaptive optimization algorithm. The weighted combination of risk control effect and inspection cost is used as the reward function. The inspection project combination with the optimal long-term utility under different risk states is learned to generate a dynamically optimized inspection strategy.

[0011] In an optional embodiment, The risk assessment model is constructed using the fuzzy comprehensive evaluation method. The remaining service life prediction results and risk measurement results are combined to form a comprehensive equipment risk assessment matrix. The risk level classification threshold is determined by experts. The output risk assessment results include: The risk assessment model is constructed by using the fuzzy comprehensive evaluation method. The evaluation index membership function in the risk assessment model is used as different evidence sources. Each evidence source provides support for different risk levels. The support is calculated by the evaluation index membership function, indicating the degree of trust in the current risk level and reflecting the influence of each evaluation index on the risk level. By using the evidence combination rules in the evidence reasoning theory, the support provided by different evidence sources is integrated to obtain the comprehensive trust of each risk level; Calculating the confidence and likelihood ratio of each risk level according to the comprehensive confidence, wherein the confidence represents the absolute support for the current risk level, and the likelihood ratio reflects the advantage of the current risk level over other risk levels; The risk levels are ranked and optimized based on the confidence and likelihood comparison, and the ranking results are compared with the risk level division thresholds determined by experts to obtain the risk assessment results.

[0012] According to a second aspect of the embodiments of the present invention, Provide a work order integrated machine for intelligent inspection system of pressure-bearing special equipment, including: The first unit is used to receive an inspection request sent by a terminal device, which includes equipment information of the pressure-bearing special equipment to be inspected, equipment historical inspection data and equipment use environment data, and search for inspection standards and inspection work order information corresponding to the pressure-bearing special equipment to be inspected from a pre-stored equipment information database according to the equipment information. If the corresponding inspection standard and inspection work order information are not found in the equipment information database, the corresponding inspection standard and inspection work order information are automatically generated according to the equipment information of the pressure-bearing special equipment to be inspected and the equipment historical inspection data; The second unit is used to determine the inspected parameters corresponding to the pressure-bearing special equipment to be inspected according to the inspection work order information, and obtain the standard parameter range corresponding to the inspected parameters from the inspection standard, establish a mapping relationship between the inspected parameters and the standard parameter range, and at the same time collect real-time parameter information of the pressure-bearing special equipment to be inspected, convert the collected real-time parameter information into real-time inspection parameters corresponding to the inspected parameters after preprocessing, and construct a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected based on the historical inspection data of the equipment and the equipment use environment data; The third unit is used to use the prediction model to predict the remaining service life of the pressure-bearing special equipment to be inspected, use the prediction result as additional inspection reference information, compare the real-time inspection parameter with the standard parameter range, generate corresponding inspection results according to the comparison result and the reference information, and generate a dynamically optimized inspection strategy according to the inspection process in the inspection work order information, and repeat the inspection process according to the optimized inspection strategy until all inspection processes required in the inspection work order information are completed, and comprehensively generate an inspection report and send it to the terminal device.

[0013] According to a third aspect of the embodiments of the present invention, An electronic device is provided, comprising: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0014] A fourth aspect of the embodiments of the present invention is: A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0015] In this embodiment, by receiving the inspection request sent by the terminal device, the key information of the equipment to be inspected is automatically obtained, and the inspection standard and inspection work order are intelligently generated in combination with the pre-stored equipment information database, thereby reducing manual operation links and improving the efficiency of inspection preparation. The equipment remaining service life prediction model is introduced, and the historical inspection data and the use environment data of the equipment are comprehensively utilized to comprehensively evaluate the health status of the equipment, provide an important reference for the inspection process, and improve the scientificity and accuracy of the inspection results. By constructing a mapping relationship between the inspected parameters and the standard parameter range, the automatic comparison of the real-time inspection parameters and the standard parameters is realized, and intuitive inspection results are generated, which is convenient for inspectors to quickly understand the operating status of the equipment and discover potential problems in time. According to the inspection work order information and real-time inspection data, the inspection strategy is dynamically optimized to realize the adaptive adjustment of the inspection process, improve the inspection efficiency and pertinence, reduce unnecessary repeated inspections, and save inspection resources. Automatically generate a comprehensive inspection report and send it to the terminal device to realize the online sharing and traceability of the inspection results, provide data support for subsequent maintenance and decision-making of the equipment, and improve the level of equipment life cycle management. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of a method for intelligent inspection of pressure-bearing special equipment using a single-piece machine according to an embodiment of the present invention; Figure 2 This is a structural schematic diagram of an intelligent inspection system for pressure-bearing special equipment using an all-in-one workpiece machine according to an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0019] Figure 1 Schematic diagram of the process of the embodiment of the present invention, such as Figure 1 As shown, the intelligent inspection method of the work order integrated machine for pressure-bearing special equipment includes: S101. Receive an inspection request sent by a terminal device, which includes equipment information of the pressure-bearing special equipment to be inspected, equipment historical inspection data and equipment usage environment data; based on the equipment information, search for inspection standards and inspection work order information corresponding to the pressure-bearing special equipment to be inspected from a pre-stored equipment information database; if the corresponding inspection standards and inspection work order information are not found in the equipment information database, automatically generate corresponding inspection standards and inspection work order information based on the equipment information of the pressure-bearing special equipment to be inspected and the equipment historical inspection data.

[0020] First, it is necessary to receive the inspection request from the terminal device. The request can be sent through various communication methods, such as network communication, SMS, phone calls, etc. The request should include the equipment information of the pressure-bearing special equipment to be inspected, the equipment's historical inspection data, and the equipment's use environment data. This information is critical for the generation and determination of subsequent inspection standards and inspection work orders.

[0021] The equipment information of the pressure-bearing special equipment to be inspected is extracted from the received inspection request, including but not limited to the equipment type, equipment model, equipment number, manufacturer, manufacturing date, commissioning date, etc. This information is used to uniquely identify a pressure-bearing special equipment and is the basis for finding the corresponding inspection standard and generating an inspection work order. The equipment historical inspection data reflects the inspection status of the pressure-bearing special equipment in the past period of time, including the time, inspection items, inspection results, etc. of the previous inspections. By analyzing the historical inspection data, the inspection cycle, inspection frequency, past failures, etc. of the equipment can be mastered, providing a reference for determining the standards for this inspection and generating inspection work orders. The equipment use environment will affect the safety performance of the equipment, so it is necessary to obtain the use environment data of the equipment to be inspected, including temperature, humidity, pressure, vibration, corrosive media, etc. On the one hand, environmental data can be used as a condition for screening and determining inspection standards, and on the other hand, it can be used to analyze the degradation trend of the equipment and predict the remaining service life of the equipment.

[0022] According to the equipment information of the equipment to be inspected, check whether there is inspection standard and inspection work order information corresponding to the equipment in the pre-stored equipment information database. The equipment information database stores standard inspection items, inspection cycles, inspection methods, etc. for various types of pressure-bearing special equipment, as well as standard inspection work order templates corresponding to various types of equipment. If there are standards and work orders matching the equipment to be inspected in the database, they can be called directly; if not, they need to be generated according to the characteristics of the equipment.

[0023] When there are no ready-made inspection standards and work orders in the equipment information database, they need to be automatically generated. The information required for generation includes the equipment type and equipment parameters of the equipment to be inspected, which are used to determine the applicable inspection items and technical requirements, and the historical inspection data of the equipment to be inspected, which are used to determine the inspection cycle and parameter settings for specific inspection items. Based on the above input information, and referring to the relevant regulations and standards for pressure-bearing special equipment, the inspection standards and specific inspection work orders that need to be implemented for this inspection can be compiled. The generated inspection standards should clearly specify the technical requirements and qualification criteria for each inspection item, and the generated inspection work orders should clearly specify the specific inspection items, the inspection methods used, the time nodes and other elements.

[0024] In an optional embodiment, According to the equipment information, searching the inspection standard and inspection work order information corresponding to the pressure-bearing special equipment to be inspected from the pre-stored equipment information database includes: The core concepts and semantic associations in the field of pressure-bearing special equipment are formally defined through ontology modeling methods, and instantiated and filled based on industry standards and expert knowledge to build a domain knowledge graph covering the entire life cycle of pressure-bearing special equipment; A graph-based multimodal representation learning framework is used to embed the unstructured information of pressure-bearing special equipment into the constructed domain knowledge graph representation space. The information of different modal equipment is aligned in a unified semantic space through attribute mapping and graph convolutional neural network to obtain equipment knowledge representation. By using the obtained equipment knowledge representation, a similar case reasoning algorithm is designed for the pressure-bearing special equipment to be inspected through the attention mechanism. Based on the semantic similarity with historical equipment cases, the inspection item information of the most relevant cases is weighted and fused to generate the inspection work order for the pressure-bearing special initial equipment to be inspected. Based on the obtained initial equipment inspection work order, a multi-objective planning model is constructed, and the multi-objective optimization algorithm is used to solve the multi-objective planning model to obtain the optimal inspection standard and inspection work order information.

[0025] For example, firstly, through ontology modeling, the core concepts and their semantic associations in the field of pressure-bearing special equipment are formally defined, and then instantiated and filled based on industry standards and expert knowledge to build a domain knowledge graph covering the entire life cycle of pressure-bearing special equipment. Ontology modeling is described using Web Ontology Language (OWL), defining concept classes such as equipment type, component composition, material properties, failure mode, inspection standard, inspection method, and hierarchical and non-hierarchical relationships between concepts. In the process of instantiating and filling concept classes, natural language processing technology is used to extract structured knowledge from industry standard texts, and implicit knowledge is obtained by combining expert interviews and other methods, and finally a large-scale equipment knowledge graph is formed.

[0026] On the basis of constructing the knowledge graph, a graph-based multimodal representation learning framework is further adopted to embed the unstructured information of pressure-bearing special equipment into the constructed knowledge graph representation space. Unstructured information includes text descriptions of equipment, drawings and images, etc. For text information, named entity recognition and relationship extraction are used to map it to the attributes of the corresponding equipment node in the knowledge graph. For image information, a pre-trained convolutional neural network is used to extract image features and associate them with the equipment structure knowledge in the graph. After the heterogeneous information is embedded in the graph, a graph convolutional neural network is used to perform representation learning on the attribute-enhanced heterogeneous graph, so that the equipment information of different modes can be aligned in a unified semantic space. The graph convolutional neural network updates the hidden layer representation of the node by aggregating the node's own features and the features of the neighboring nodes. After multiple layers of iterative updates, a graph embedding representation that combines the topological structure and node semantics is obtained. The aggregation function of the graph convolution can be selected from the mean, max, attention and other mechanisms.

[0027] Using the equipment knowledge representation obtained above, a similar case reasoning algorithm is designed for the pressure-bearing special equipment to be inspected. The most relevant cases to the equipment to be inspected are retrieved from the historical equipment case library, and the initial inspection work order is generated by referring to its inspection item information. The attention mechanism is introduced in the similarity calculation to distinguish the contribution of different attributes of historical cases to the similarity. First, the local similarity between the equipment to be inspected and the historical cases in different attribute dimensions is calculated. Metric functions such as cosine similarity and Euclidean distance can be used. Then, weights are dynamically assigned to each attribute through the attention network. The importance of the attribute is related to its influence on the equipment inspection. Finally, the local similarity and the attention weight are weighted and summed to obtain the comprehensive similarity. The TopK most similar historical cases are selected, and their inspection item sets are extracted for weighted fusion as the initial inspection work order for the equipment to be inspected. The fusion weight is proportional to the comprehensive similarity between the case and the equipment to be inspected.

[0028] On the basis of obtaining the initial inspection work order, further optimize the inspection item combination and balance various influencing factors. Construct a multi-objective planning model, use a multi-objective optimization algorithm to solve the multi-objective planning model, and obtain the optimal inspection standard and inspection work order information.

[0029] In this embodiment, the systematic, structured and standardized representation of knowledge in the field of pressure-bearing special equipment is achieved through ontology modeling and knowledge graph construction. It is helpful for the accumulation, sharing and reuse of knowledge, and improves the efficiency and quality of domain knowledge management. The graph-based multimodal representation learning framework is adopted to achieve the unified representation and semantic alignment of structured and unstructured information, which can describe and understand equipment characteristics more comprehensively and accurately, and improve the efficiency of information utilization. Through the similar case reasoning algorithm, it is possible to automatically generate highly targeted initial inspection work orders, reduce the workload of manual inspection plan formulation, improve the scientificity and rationality of the inspection plan, and use multi-objective planning models and optimization algorithms to achieve the optimization of inspection standards and work orders, which can maximize resource utilization efficiency and reduce inspection costs while ensuring inspection quality.

[0030] In an optional embodiment, Based on the obtained initial equipment inspection work order, a multi-objective planning model is constructed, and the multi-objective optimization algorithm is used to solve the multi-objective planning model to obtain the optimal inspection standard and inspection work order information including: The inspection item selection problem of pressure-bearing special equipment is transformed into a 0-1 decision variable, and a multi-objective programming model is constructed with maximizing fault coverage, minimizing inspection cost and minimizing inspection time as objective functions, and with dependencies, mutual exclusions and overall resource budgets between inspection items as constraints. The improved multi-objective solution algorithm is used to solve the multi-objective programming model. The population is initialized and a certain number of 0-1 coded individuals are randomly generated. Each individual corresponds to a set of test item selection schemes. The first Pareto optimal solution set is obtained through non-dominated sorting and crowding calculation. The initial particle swarm is randomly generated, each particle represents a combination of test item selections, and the first Pareto optimal solution set is used as the initial position. According to the individual optimal solution and the global optimal solution, the speed and position of each particle are updated, and the fitness value of the current position of each particle is calculated. At the same time, an external archive is introduced to dynamically save non-dominated solutions, and an adaptive grid division mechanism is used to dynamically adjust the grid size of the target space according to the distribution of solutions. In the process of updating the particle position, the individual optimal solution, the global optimal solution and the congestion of the grid where the particle is located are comprehensively considered to construct the particle speed update formula, and iterate repeatedly until the maximum number of iterations is met, and the second Pareto optimal solution set is output; The first Pareto optimal solution set and the second Pareto optimal solution set are merged, and the optimal solution is selected as the final inspection standard and inspection work order information according to the decision maker's preference.

[0031] Exemplarily, first, the inspection item selection problem is abstracted into a 0-1 decision problem, in which each inspection item corresponds to a 0-1 decision variable, a value of 1 indicates that the inspection item is selected, and a value of 0 indicates that it is not selected. On this basis, a multi-objective planning model is constructed, considering three optimization objectives: maximizing fault coverage, minimizing inspection costs, and minimizing inspection time. Fault coverage indicates the degree of coverage of the potential failure modes of the equipment by the selected inspection items. The inspection cost includes labor costs, material costs, and equipment depreciation, and the inspection time reflects the time efficiency of the inspection plan. At the same time, multiple constraints are introduced into the model, including dependencies, mutual exclusions, and overall resource budget constraints between inspection items, to ensure that the generated inspection plan is feasible and reasonable.

[0032] For the above multi-objective programming model, this paper adopts an improved multi-objective solution algorithm, which combines the advantages of genetic algorithm and particle swarm optimization algorithm. In the solution process, the population is first initialized, and a certain number of 0-1 coded individuals are randomly generated, each of which represents a set of test item selection schemes. Then, the first Pareto optimal solution set is obtained by non-dominated sorting and crowding calculation as the starting point of optimization. On this basis, the initial particle swarm is randomly generated, and the first Pareto optimal solution set is used as the initial position of the particle. The speed and position of each particle are updated according to the individual optimal solution and the global optimal solution. In order to improve the convergence and diversity of the algorithm, an external archiving mechanism is introduced to dynamically save the non-dominated solutions in the optimization process. At the same time, an adaptive grid division strategy is adopted to dynamically adjust the grid size of the target space according to the distribution of the solution. When the particle position is updated, the individual optimal solution, the global optimal solution and the crowding degree of the particle are comprehensively considered to construct the particle speed update formula. The algorithm repeats the iteration until the maximum number of iterations is met, and the second Pareto optimal solution set is output.

[0033] Finally, the first Pareto optimal solution set and the second Pareto optimal solution set are merged to form the final non-dominated solution set. According to the decision maker's preference information, the optimal solution is selected as the result of the test item selection, and the corresponding test standard and test work order are generated.

[0034] To better understand the above method, a specific example is given below. Assume that a pressure-bearing equipment has 10 potential failure modes and there are 15 candidate inspection items. Each inspection item has different coverage capabilities for different failure modes and has different cost and duration attributes. The decision-making goal is to select a set of inspection items to maximize the fault coverage and minimize the inspection cost and duration while meeting the resource budget.

[0035] First, each test item is represented as a 0-1 decision variable, and a multi-objective programming model is constructed. Then, 50 0-1 coded individuals are randomly generated, each of which corresponds to a set of test item selection schemes. Through non-dominated sorting and crowding calculation, the first Pareto optimal solution set is obtained, assuming that it contains 10 non-dominated solutions. Next, 20 particles are randomly generated, and the 10 non-dominated solutions obtained previously are used as the initial positions of the particles, and the particle swarm optimization solution is started. In each iteration, the speed and position of the particles are updated according to the individual optimal solution of the particles, the global optimal solution, and the crowding of the grid where they are located, and the newly generated non-dominated solutions are added to the external archive. Assuming that the algorithm iterates 100 times, the second Pareto optimal solution set is finally obtained, which contains 15 non-dominated solutions.

[0036] The first Pareto optimal solution set and the second Pareto optimal solution set are combined to form a final solution set containing 25 non-dominated solutions. Assuming that the decision maker prefers the fault coverage and inspection cost indicators, an optimal solution that balances the two indicators is selected from the solution set, and the corresponding inspection item combination is {2, 4, 7, 9, 12}. Based on this combination, an inspection standard file and a specific inspection work order are generated to guide the actual inspection work.

[0037] In this embodiment, multiple objectives such as fault coverage, inspection cost and inspection time are comprehensively considered, and the constraint relationship between inspection items is introduced to generate a more reasonable and optimized inspection item selection scheme. This method integrates the ideas of multi-objective planning modeling and heuristic optimization solution, has strong adaptability and robustness, and provides a new solution for the optimization of inspection items of pressure-bearing special equipment.

[0038] S102. Determine the inspected parameters corresponding to the pressure-bearing special equipment to be inspected based on the inspection work order information, obtain the standard parameter range corresponding to the inspected parameters from the inspection standards, establish a mapping relationship between the inspected parameters and the standard parameter range, and collect real-time parameter information of the pressure-bearing special equipment to be inspected at the same time. Pre-process the collected real-time parameter information and convert it into real-time inspection parameters corresponding to the inspected parameters. Based on the historical inspection data of the equipment and the equipment usage environment data, construct a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected.

[0039] Among them, the inspection standard stipulates the qualified range of each parameter to be inspected, that is, the standard parameter range. In order to facilitate subsequent rapid judgment, it is necessary to establish a one-to-one mapping relationship between the inspected parameters and the standard parameter range. Specifically, the technical parameters related to the equipment to be inspected in the inspection standard are extracted to form a set of inspected parameters; then the qualified value range of each inspected parameter specified in the standard is searched one by one to form a standard parameter range set. The two sets correspond one to one, forming a mapping relationship. Based on this mapping relationship, the collected real-time inspection parameters are compared with the standard range to quickly determine whether they are qualified.

[0040] When inspecting the equipment to be inspected, it is necessary to collect real-time parameter information during the operation of the equipment. The relevant parameters are obtained in real time through online monitoring or portable detection equipment. The types of parameters collected must be consistent with the parameter set to be inspected, and the time and frequency of collection must meet the requirements of the inspection work order.

[0041] The parameter data collected in real time usually has some quality problems, such as uneven sampling, missing data, outliers, etc. In order to ensure the reliability of the test results, the real-time data needs to be preprocessed. Common preprocessing methods include using smoothing filtering and other methods to remove random noise; interpolation repair of missing data; and using statistical methods or machine learning algorithms to identify and correct outliers. The preprocessed real-time parameter data can be used as real-time test parameters for subsequent qualification determination.

[0042] A data-driven modeling method is used to build a prediction model based on the historical inspection data and usage environment data of the equipment. The basic idea of ​​modeling is to extract characteristic parameters related to equipment degradation and failure from historical data, such as the degradation rate of key parameters, the distribution law of historical failures, the correlation between environmental factors and failures, etc.; then use machine learning algorithms, such as support vector machines and neural networks, to establish a mapping relationship between characteristic parameters and the remaining service life of the equipment to form a prediction model. The input of the model is the real-time inspection parameters and the current environmental parameters, and the output is the remaining service life of the equipment.

[0043] In an optional embodiment, Based on the historical inspection data of the equipment and the equipment use environment data, a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected is constructed, including: Based on the theory of material science and damage mechanics, a physical mechanism model reflecting the degradation process of key pressure-bearing special equipment is constructed, and the degradation parameter evolution curve and theoretical life estimation value are derived to form a degradation physics knowledge base; Extract characteristic parameters closely related to the degradation process of key pressure-bearing special equipment from the equipment life cycle data, collect the equipment's historical operating parameters, environmental factors, regular inspection data and maintenance records, build a degradation feature set, and use feature selection methods to screen out the target feature subset; Based on the obtained target feature subset, the degradation features are divided into input sequences according to time windows using a long short-term memory network, and the known lifespans at each corresponding moment are used as labels. By minimizing the prediction error between the model output and the label lifespan, the nonlinear mapping relationship between the degradation features and the remaining lifespan is learned, and a data-driven degradation model is constructed. The theoretical degradation curve in the degradation physics knowledge base is discretized and aligned with the degradation feature set in time to form a physical-data hybrid feature set. Taking the physical-data hybrid feature set as input, a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected is constructed using a gated cyclic unit.

[0044] For example, it is first necessary to analyze the degradation process of the equipment from the perspective of physical mechanism. By studying the evolution law of the mechanical properties of the material and the damage accumulation mechanism, a physical model that can quantitatively describe the key degradation modes of the equipment is established. Taking fatigue failure as an example, the fatigue crack propagation theory in fracture mechanics is used to derive the quantitative relationship between crack size and the number of load cycles, and obtain the theoretical remaining fatigue life of the equipment in the current state. Similarly, for other degradation modes such as corrosion and creep, corresponding physical failure models are constructed respectively. Through numerical analysis techniques such as finite element simulation, the evolution law of the degradation parameters of the model under actual operating conditions is solved to form a complete degradation physics knowledge base, which provides theoretical support for subsequent model fusion.

[0045] While acquiring knowledge of physical degradation, it is also necessary to explore the degradation laws of equipment from a data-driven perspective. To this end, it is first necessary to extract characteristic indicators that are highly correlated with the degradation process from the data of the entire life cycle of the equipment. Collect multi-source heterogeneous information such as environmental parameters, operating parameters, inspection and maintenance records, and regular inspection data since the equipment was put into operation, and organize them into a structured degradation feature set according to time series. Statistical indicators such as the Pearson correlation coefficient and mutual information are used to measure the correlation between each candidate feature and the known life, and automatically screen out the target feature subset that contributes most to the prediction performance, while reducing the data dimension while retaining the degradation trend information to the greatest extent.

[0046] Based on the selected target feature subset, deep learning technology is used to build a data-driven degradation model for remaining life prediction. Model structures such as long short-term memory networks that can mine long-term dependencies in time series are selected to learn the complex nonlinear mapping from degradation feature history to remaining life. The target feature subset is divided into a series of input sequences according to a fixed time window, and the known life corresponding to each time segment is extracted as a training label. The model adaptively fits the degradation trajectory of each sample through end-to-end supervised learning. The loss function is defined using evaluation indicators such as mean square error to measure the deviation between the model's predicted life and the actual life, and the back-propagation algorithm is used to minimize the loss function and continuously optimize the model parameters. Through training on a large-scale degradation dataset, the data-driven model can capture the intrinsic correlation between the evolution law of degradation features and the remaining life in a statistical sense.

[0047] In order to integrate physical degradation knowledge and data-driven laws, a hybrid degradation modeling method is proposed. First, the theoretical degradation curve in the physical knowledge base is discretized at a certain time interval to align it with the input sequence of the data-driven model in the time dimension. Then, the discretized physical degradation index is used as an additional feature and cascaded with the original target feature subset to form a physical-data hybrid feature set. With this hybrid feature set as input, a hybrid degradation mapping model is constructed using a gated cyclic unit to predict the remaining service life of the pressure-bearing special equipment to be inspected. During the model training process, the physical degradation index is added as a priori knowledge to constrain the data features so that the learning direction of the model parameters conforms to the physical degradation mechanism. At the same time, the remaining life prediction value output by the model is superimposed with the physical theoretical estimate according to a certain weight to obtain a hybrid prediction result that comprehensively considers the two types of knowledge. The hybrid degradation model based on the gated cyclic unit can flexibly adapt to the degradation data distribution under actual working conditions while following the physical laws, balance the mechanism explanation and data fitting, and significantly improve the accuracy and reliability of the remaining life prediction.

[0048] In an optional embodiment, The steps after building the prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected also include: According to the physical degradation mechanism of the pressure-bearing special equipment to be inspected, appropriate physical degradation indicators are selected as prior knowledge, and a loss function is constructed based on the data fitting loss and the difference between the physical degradation indicators and the model prediction value. The prediction model is trained by minimizing the loss function, and the model parameters are updated using the gradient descent algorithm. The iteration is repeated until the maximum number of iterations is reached, and finally a trained prediction model is obtained. The loss function calculation formula is as follows: ; Among them, L represents the loss function, T represents the total number of time steps, and β t represents the weight of time step t, represents the predicted value of the remaining life of the pressure-bearing special equipment to be inspected at time step t, y t represents the true value of the remaining life of the pressure-bearing special equipment to be inspected at time step t, λ1 represents the weight coefficient of the regularization term of the physical degradation index, N represents the number of physical degradation indicators, w i represents the weight of the ith physical degradation index in the regularization term, g i (·) represents the theoretical estimation function corresponding to the ith physical degradation index, p t，i represents the i-th physical degradation index at time step t.

[0049] In this embodiment, by combining the material science and damage mechanics theory with the data-driven method, an organic integration of theory and practice is achieved, and the accuracy and interpretability of the remaining life prediction are improved. Features are extracted from the data of the entire life cycle of the equipment to ensure a comprehensive characterization of the degradation process and improve the model's ability to express complex degradation processes. Through long short-term memory networks and gated recurrent units, the timing characteristics can be effectively captured and the prediction accuracy can be improved. The combination of physical models and data-driven models further improves the accuracy and stability of the prediction. The use of physical-data hybrid feature sets enhances the model's generalization ability for different types of equipment and working conditions, reduces dependence on large amounts of historical data, and improves the performance of the model in data-sparse conditions. It greatly improves the accuracy and reliability of the remaining life prediction of pressure-bearing special equipment, and also provides strong support for the intelligent management and decision-making of equipment.

[0050] S103. Use the prediction model to predict the remaining service life of the pressure-bearing special equipment to be inspected, use the prediction result as additional inspection reference information, compare the real-time inspection parameters with the standard parameter range, generate corresponding inspection results based on the comparison results and reference information, and generate a dynamically optimized inspection strategy based on the inspection process in the inspection work order information, and repeat the inspection process according to the optimized inspection strategy until all inspection processes required in the inspection work order information are completed, and comprehensively generate an inspection report and send it to the terminal device.

[0051] In an optional embodiment, The prediction model is used to predict the remaining service life of the pressure-bearing special equipment to be inspected, and the prediction result is used as additional inspection reference information. The real-time inspection parameters are compared with the standard parameter range, and the corresponding inspection results are generated according to the comparison results and the reference information. According to the inspection process in the inspection work order information, a dynamically optimized inspection strategy is generated, including: According to the type of pressure-bearing special equipment to be inspected, similar cases are selected from the historical database, the corresponding degradation physical parameters and monitoring feature sequences are extracted, and input into the pre-trained prediction model, the prediction model is used to fit the degradation state of the pressure-bearing special equipment to be inspected, and the remaining service life prediction result is output; Deploy an online monitoring system to collect real-time inspection parameters of the pressure-bearing special equipment to be inspected, compare the collected real-time inspection parameters with the preset standard parameter range, and identify abnormal indicators that exceed the normal range; Combined with the remaining service life prediction results and the analysis results of the online monitoring system, the degradation state and failure risk of the pressure-bearing special equipment to be inspected are comprehensively evaluated. When the real-time inspection parameters exceed the preset warning threshold and the remaining service life is lower than the preset safety threshold, it is determined that the pressure-bearing special equipment to be inspected has a failure risk and a risk measurement result is generated; A risk assessment model is constructed using a fuzzy comprehensive evaluation method. The remaining service life prediction results and risk measurement results are combined to form a comprehensive equipment risk assessment matrix. Experts determine the risk level classification threshold and output the risk assessment results. Based on the risk assessment results, a structured rule base is formed according to the inspection standards and industry regulations. A rule-based reasoning engine is used to perform forward reasoning matching on the input degradation physical parameters and monitoring feature sequences, and consistency verification is performed in combination with ontology constraints to generate a preliminary inspection strategy. Based on the preliminary inspection strategy, the inspection strategy optimization model is modeled as a Markov decision process. The Markov decision process is solved by an adaptive optimization algorithm. The weighted combination of risk control effect and inspection cost is used as the reward function. The inspection project combination with the optimal long-term utility under different risk states is learned to generate a dynamically optimized inspection strategy.

[0052] Exemplarily, first, according to the type of equipment to be inspected, similar cases are selected from the historical database to extract the corresponding degradation physical parameters and monitoring feature sequences. Degradation physical parameters refer to physical quantities that reflect the intrinsic degradation state of equipment material performance, structural integrity, etc., such as material yield strength, fracture toughness, etc. Monitoring feature sequences refer to time series data collected by sensors that reflect the external working conditions of the equipment, such as temperature, pressure, vibration, etc. The extracted case data is input into a pre-trained prediction model, and the model automatically fits the degradation state of the current equipment and outputs the prediction result of its remaining service life. While obtaining the remaining life prediction result, it is also necessary to monitor the status of the equipment in real time. By deploying an online monitoring system, the key operating parameters of the equipment are collected in real time. According to relevant industry standards and equipment management specifications, the upper and lower limit thresholds of the normal working range of each monitoring parameter are pre-set. Compare the monitoring parameters collected in real time with the standard range to identify abnormal indicators that exceed the normal range.

[0053] Comprehensive analysis of the remaining life prediction results and online monitoring data can comprehensively evaluate the current degradation state and potential failure risks of the equipment. When the key monitoring parameters of the equipment are close to or exceed the warning threshold, and the predicted remaining life is lower than the set safety margin, it can be judged that the equipment has a high failure risk. The system automatically generates corresponding risk measurement results for subsequent risk assessment and inspection strategy optimization.

[0054] In order to quantify the comprehensive risk level of the equipment, a fuzzy comprehensive evaluation method is used to construct a risk assessment model. This method can organically combine qualitative expert experience knowledge with quantitative data analysis results to more comprehensively and accurately judge the risk status of the equipment. The remaining life prediction value and the risk measurement value of the online monitoring parameter are used as the input of the fuzzy comprehensive evaluation, and they are mapped to the corresponding risk level domain through the membership function. In the risk level domain, the division thresholds of each level are pre-determined by experts to form a risk assessment matrix. According to the membership distribution of the equipment parameters, its comprehensive membership at each risk level is calculated, and the level with the largest membership is taken as the comprehensive risk level of the equipment.

[0055] According to the risk assessment results, it is necessary to further optimize the equipment inspection strategy and formulate a targeted inspection project combination. First, a set of structured inspection rule bases are formed according to relevant inspection standards and industry regulations. A rule-based reasoning engine is used to forward match the input equipment degradation physical parameters and monitoring feature sequences to identify potentially applicable inspection items. At the same time, ontology knowledge is used to constrain and verify the consistency of the reasoning results, eliminate redundant or contradictory inspection items, and form a preliminary feasible inspection strategy.

[0056] On this basis, the dynamic optimization of the inspection strategy also needs to be considered. Since the degradation state of the equipment evolves dynamically over time, the same inspection strategy may not be able to continuously meet the needs of risk control. To this end, the inspection process is modeled as a Markov decision process, and the optimal strategy is sought through reinforcement learning. The state transition probability of the equipment under different risk states is used as the environmental model, the various combinations of inspection items are used as the action space, and the weighted combination of risk control effect and inspection cost is used as the reward function. In one of the optional methods, the Q-Learning algorithm is used to find the optimal combination of inspection items under various risk states through continuous trial and error and learning, and form a dynamically optimized inspection strategy. Based on the real-time risk assessment results of the equipment, the system automatically recommends the corresponding optimal inspection plan to guide on-site inspection practices.

[0057] In this embodiment, through similar case analysis and pre-training models, the accuracy of remaining life prediction is improved, and personalized life prediction for different types of equipment is realized. The online monitoring system realizes real-time monitoring of equipment status, can detect abnormal indicators in time, and improves the timeliness of risk identification. Combining life prediction and real-time monitoring data, a more comprehensive risk assessment is achieved, and the fuzzy comprehensive evaluation method improves the scientificity and reliability of risk assessment. The rule-based reasoning engine automatically generates preliminary inspection strategies, reducing manual intervention, and the ontology constraints ensure the consistency and compliance of the generated strategies. The Markov decision process realizes the dynamic optimization of the inspection strategy, and the adaptive optimization algorithm ensures the long-term optimality of the inspection plan. It not only greatly improves the safety management level and operation and maintenance efficiency of pressure-bearing special equipment, but also realizes the transformation of management mode from passive response to active prevention. The system provides strong technical support for the intelligent and refined management of pressure-bearing special equipment.

[0058] In an optional embodiment, The risk assessment model is constructed using the fuzzy comprehensive evaluation method. The remaining service life prediction results and risk measurement results are combined to form a comprehensive equipment risk assessment matrix. The risk level classification threshold is determined by experts. The output risk assessment results include: The risk assessment model is constructed by using the fuzzy comprehensive evaluation method. The evaluation index membership function in the risk assessment model is used as different evidence sources. Each evidence source provides support for different risk levels. The support is calculated by the evaluation index membership function, indicating the degree of trust in the current risk level and reflecting the influence of each evaluation index on the risk level. By using the evidence combination rules in the evidence reasoning theory, the support provided by different evidence sources is integrated to obtain the comprehensive trust of each risk level; Calculating the confidence and likelihood ratio of each risk level according to the comprehensive confidence, wherein the confidence represents the absolute support for the current risk level, and the likelihood ratio reflects the advantage of the current risk level over other risk levels; The risk levels are ranked and optimized based on the confidence and likelihood comparison, and the ranking results are compared with the risk level division thresholds determined by experts to obtain the risk assessment results.

[0059] For example, first, a risk assessment model is constructed, and a fuzzy comprehensive evaluation method is used to deal with the uncertainty and ambiguity of the evaluation indicators. Each evaluation indicator in the risk assessment model is regarded as a different source of evidence, and each source of evidence is calculated through a membership function to support different risk levels. The membership function represents the degree of membership of the evaluation indicator to the risk level, reflecting the mapping relationship between the indicator value and the risk level. The support indicates the degree of trust in the current indicator for a specific risk level. The larger the value, the higher the support of the indicator for the risk level.

[0060] After obtaining the support of each source of evidence, the evidence combination rule in the theory of evidence reasoning is used to fuse information. The evidence combination rule is a reasoning method used to process uncertain information. It combines the information provided by different sources of evidence to obtain the comprehensive trust in each proposition. In risk assessment problems, propositions correspond to different risk levels, and the comprehensive trust represents the overall trust in each risk level after considering all evaluation indicators. The evidence combination rule is based on the independence assumption between evidence sources. It obtains the comprehensive trust by calculating the product of the support of the evidence sources and normalizing it.

[0061] In order to further judge the risk level, two indicators, confidence and likelihood ratio, are introduced. Confidence indicates the absolute support for the current risk level, which is calculated by multiplying the comprehensive confidence and the prior probability of the risk level. The likelihood ratio reflects the advantage of the current risk level over other risk levels, which is calculated by the ratio of the comprehensive confidence of the current risk level to the sum of the comprehensive confidences of other risk levels. The larger the confidence and likelihood ratio, the greater the possibility that the current risk level is supported.

[0062] Finally, the risk levels are sorted and optimized according to the confidence and likelihood ratio. The risk levels are sorted from large to small according to the confidence and likelihood ratio to obtain the priority order of the risk levels. At the same time, experts determine the risk level division threshold based on experience and historical data, and compare the sorting results with the threshold to obtain the final risk assessment results. The risk assessment results include the current risk level of the equipment and the confidence of each risk level.

[0063] Assume that a certain equipment has five risk assessment indicators, namely, remaining service life, probability of failure, severity of failure consequences, difficulty of detection, and maintenance cost. Each indicator has a corresponding membership function, which maps the indicator value to the support for three levels of low risk, medium risk, and high risk. For example, the membership function of the remaining service life indicator can be defined as: when the remaining service life is greater than 10 years, the low risk support is 0.8, the medium risk support is 0.2, and the high risk support is 0; when the remaining service life is between 5 and 10 years, the low risk support is 0.4, the medium risk support is 0.5, and the high risk support is 0.1; when the remaining service life is less than 5 years, the low risk support is 0.1, the medium risk support is 0.3, and the high risk support is 0.6.

[0064] According to the membership function, the support of each indicator for the three risk levels is calculated to form the evidence source. Then, the support of the five indicators is integrated using the evidence combination rule to obtain the comprehensive trust for low risk, medium risk and high risk, which are 0.6, 0.3 and 0.1 respectively. Next, the confidence and likelihood ratio of each risk level are calculated. Assuming that the prior probabilities of the three risk levels are 0.2, 0.5 and 0.3 respectively, the confidence of low risk is 0.6×0.2=0.12, and the likelihood ratio is 0.6 / (0.3+0.1)=1.5; the confidence of medium risk is 0.3×0.5=0.15, and the likelihood ratio is 0.3 / (0.6+0.1)≈0.43; the confidence of high risk is 0.1×0.3=0.03, and the likelihood ratio is 0.1 / (0.6+0.3)≈0.11.

[0065] According to the confidence and likelihood ratio, the three risk levels are sorted, and the priority order is: medium risk > low risk > high risk. Assume that experts determine the risk level classification thresholds based on historical experience: low risk confidence threshold is 0.1, medium risk confidence threshold is 0.2, and high risk confidence threshold is 0.3. Comparing the sorting results with the thresholds, it is found that the device is currently at a medium risk level with a confidence of 0.15.

[0066] In this embodiment, a more comprehensive and reliable risk assessment is achieved by integrating fuzzy comprehensive evaluation and evidence reasoning theory. The fuzzy comprehensive evaluation method allows for comprehensive consideration of complex risk factors, while the evidence reasoning theory provides a scientific method to deal with the uncertainty and conflict of different sources of evidence. Taking the evaluation index membership function as the source of evidence fully utilizes the advantages of fuzzy theory in dealing with uncertainty and ambiguity. This method can better reflect the complexity and uncertainty in actual risk assessment and improve the accuracy and reliability of risk assessment. It enhances the scientificity, reliability and practicality of risk assessment. It provides more powerful technical support for the safety management of pressure-bearing special equipment, helps to improve the safety of equipment operation, optimize resource allocation, and provides a reliable basis for formulating preventive maintenance strategies.

[0067] Figure 2 Schematic diagram of the structure of the system of the embodiment of the present invention, as shown in Figure 2 As shown, the work order integrated machine is used for the intelligent inspection system of pressure-bearing special equipment, including: The first unit is used to receive an inspection request sent by a terminal device, which includes equipment information of the pressure-bearing special equipment to be inspected, equipment historical inspection data and equipment use environment data, and search for inspection standards and inspection work order information corresponding to the pressure-bearing special equipment to be inspected from a pre-stored equipment information database according to the equipment information. If the corresponding inspection standard and inspection work order information are not found in the equipment information database, the corresponding inspection standard and inspection work order information are automatically generated according to the equipment information of the pressure-bearing special equipment to be inspected and the equipment historical inspection data; The second unit is used to determine the inspected parameters corresponding to the pressure-bearing special equipment to be inspected according to the inspection work order information, and obtain the standard parameter range corresponding to the inspected parameters from the inspection standard, establish a mapping relationship between the inspected parameters and the standard parameter range, and at the same time collect real-time parameter information of the pressure-bearing special equipment to be inspected, convert the collected real-time parameter information into real-time inspection parameters corresponding to the inspected parameters after preprocessing, and construct a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected based on the historical inspection data of the equipment and the equipment use environment data; The third unit is used to use the prediction model to predict the remaining service life of the pressure-bearing special equipment to be inspected, use the prediction result as additional inspection reference information, compare the real-time inspection parameter with the standard parameter range, generate corresponding inspection results according to the comparison result and the reference information, and generate a dynamically optimized inspection strategy according to the inspection process in the inspection work order information, and repeat the inspection process according to the optimized inspection strategy until all inspection processes required in the inspection work order information are completed, and comprehensively generate an inspection report and send it to the terminal device.

[0068] According to a third aspect of the embodiments of the present invention, An electronic device is provided, comprising: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0069] A fourth aspect of the embodiments of the present invention is: A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0070] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart inspection method for pressure-bearing special equipment using a single-machine machine, characterized in that: include: Receive an inspection request sent by a terminal device, which includes equipment information of the pressure-bearing special equipment to be inspected, equipment historical inspection data, and equipment use environment data; search for inspection standards and inspection work order information corresponding to the pressure-bearing special equipment to be inspected from a pre-stored equipment information database based on the equipment information; if the corresponding inspection standards and inspection work order information are not found in the equipment information database, automatically generate corresponding inspection standards and inspection work order information based on the equipment information of the pressure-bearing special equipment to be inspected and the equipment historical inspection data; According to the inspection work order information, the inspected parameters corresponding to the pressure-bearing special equipment to be inspected are determined, and the standard parameter range corresponding to the inspected parameters is obtained from the inspection standard, and a mapping relationship between the inspected parameters and the standard parameter range is established. At the same time, the real-time parameter information of the pressure-bearing special equipment to be inspected is collected, and the collected real-time parameter information is pre-processed and converted into real-time inspection parameters corresponding to the inspected parameters. Based on the historical inspection data of the equipment and the equipment use environment data, a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected is constructed; The prediction model is used to predict the remaining service life of the pressure-bearing special equipment to be inspected, and the prediction result is used as additional inspection reference information. The real-time inspection parameters are compared with the standard parameter range, and the corresponding inspection results are generated according to the comparison results and the reference information. According to the inspection process in the inspection work order information, a dynamically optimized inspection strategy is generated, and the inspection process is repeatedly executed according to the optimized inspection strategy until all inspection processes required in the inspection work order information are completed, and an inspection report is comprehensively generated and sent to the terminal device.

2. The method according to claim 1, characterized in that According to the equipment information, searching the inspection standard and inspection work order information corresponding to the pressure-bearing special equipment to be inspected from the pre-stored equipment information database includes: The core concepts and semantic associations in the field of pressure-bearing special equipment are formally defined through ontology modeling methods, and instantiated and filled based on industry standards and expert knowledge to build a domain knowledge graph covering the entire life cycle of pressure-bearing special equipment; A graph-based multimodal representation learning framework is used to embed the unstructured information of pressure-bearing special equipment into the constructed domain knowledge graph representation space. The information of different modal equipment is aligned in a unified semantic space through attribute mapping and graph convolutional neural network to obtain equipment knowledge representation. By using the obtained equipment knowledge representation, a similar case reasoning algorithm is designed for the pressure-bearing special equipment to be inspected through the attention mechanism. Based on the semantic similarity with historical equipment cases, the inspection item information of the most relevant cases is weighted and fused to generate the inspection work order for the pressure-bearing special initial equipment to be inspected. Based on the obtained initial equipment inspection work order, a multi-objective planning model is constructed, and the multi-objective optimization algorithm is used to solve the multi-objective planning model to obtain the optimal inspection standard and inspection work order information.

3. The method according to claim 2, characterized in that Based on the obtained initial equipment inspection work order, a multi-objective planning model is constructed, and the multi-objective optimization algorithm is used to solve the multi-objective planning model to obtain the optimal inspection standard and inspection work order information including: The inspection item selection problem of pressure-bearing special equipment is transformed into a 0-1 decision variable, and a multi-objective programming model is constructed with maximizing fault coverage, minimizing inspection cost and minimizing inspection time as objective functions, and with dependencies, mutual exclusions and overall resource budgets between inspection items as constraints. The improved multi-objective solution algorithm is used to solve the multi-objective programming model. The population is initialized and a certain number of 0-1 coded individuals are randomly generated. Each individual corresponds to a set of test item selection schemes. The first Pareto optimal solution set is obtained through non-dominated sorting and crowding calculation. The initial particle swarm is randomly generated, each particle represents a combination of test item selections, and the first Pareto optimal solution set is used as the initial position. According to the individual optimal solution and the global optimal solution, the speed and position of each particle are updated, and the fitness value of the current position of each particle is calculated. At the same time, an external archive is introduced to dynamically save non-dominated solutions, and an adaptive grid division mechanism is used to dynamically adjust the grid size of the target space according to the distribution of solutions. In the process of updating the particle position, the individual optimal solution, the global optimal solution and the congestion of the grid where the particle is located are comprehensively considered to construct the particle speed update formula, and iterate repeatedly until the maximum number of iterations is met, and the second Pareto optimal solution set is output; The first Pareto optimal solution set and the second Pareto optimal solution set are merged, and the optimal solution is selected as the final inspection standard and inspection work order information according to the decision maker's preference.

4. The method according to claim 1, characterized in that: Based on the historical inspection data of the equipment and the equipment use environment data, a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected is constructed, including: Based on the theory of material science and damage mechanics, a physical mechanism model reflecting the degradation process of key pressure-bearing special equipment is constructed, and the degradation parameter evolution curve and theoretical life estimation value are derived to form a degradation physics knowledge base; Extract characteristic parameters closely related to the degradation process of key pressure-bearing special equipment from the equipment life cycle data, collect the equipment's historical operating parameters, environmental factors, regular inspection data and maintenance records, build a degradation feature set, and use feature selection methods to screen out the target feature subset; Based on the obtained target feature subset, the degradation features are divided into input sequences according to time windows using a long short-term memory network, and the known lifespans at each corresponding moment are used as labels. By minimizing the prediction error between the model output and the label lifespan, the nonlinear mapping relationship between the degradation features and the remaining lifespan is learned, and a data-driven degradation model is constructed. The theoretical degradation curve in the degradation physics knowledge base is discretized and aligned with the degradation feature set in time to form a physical-data hybrid feature set. Taking the physical-data hybrid feature set as input, a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected is constructed using a gated cyclic unit.

5. The method according to claim 4, characterized in that The steps after building the prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected also include: According to the physical degradation mechanism of the pressure-bearing special equipment to be inspected, appropriate physical degradation indicators are selected as prior knowledge, and a loss function is constructed based on the data fitting loss and the difference between the physical degradation indicators and the model prediction value. The prediction model is trained by minimizing the loss function, and the model parameters are updated using the gradient descent algorithm. The iteration is repeated until the maximum number of iterations is reached, and finally a trained prediction model is obtained. The loss function calculation formula is as follows: ； Among them, L represents the loss function, T represents the total number of time steps, and β t represents the weight of time step t, represents the predicted value of the remaining life of the pressure-bearing special equipment to be inspected at time step t, y t represents the true value of the remaining life of the pressure-bearing special equipment to be inspected at time step t, λ1 represents the weight coefficient of the regularization term of the physical degradation index, N represents the number of physical degradation indicators, w i represents the weight of the ith physical degradation index in the regularization term, g i (·) represents the theoretical estimation function corresponding to the ith physical degradation index, p t，i represents the i-th physical degradation index at time step t.

6. The method according to claim 1, characterized in that The prediction model is used to predict the remaining service life of the pressure-bearing special equipment to be inspected, and the prediction result is used as additional inspection reference information. The real-time inspection parameters are compared with the standard parameter range, and the corresponding inspection results are generated according to the comparison results and the reference information. According to the inspection process in the inspection work order information, a dynamically optimized inspection strategy is generated, including: According to the type of pressure-bearing special equipment to be inspected, similar cases are selected from the historical database, the corresponding degradation physical parameters and monitoring feature sequences are extracted, and input into the pre-trained prediction model, the prediction model is used to fit the degradation state of the pressure-bearing special equipment to be inspected, and the remaining service life prediction result is output; Deploy an online monitoring system to collect real-time inspection parameters of the pressure-bearing special equipment to be inspected, compare the collected real-time inspection parameters with the preset standard parameter range, and identify abnormal indicators that exceed the normal range; Combined with the remaining service life prediction results and the analysis results of the online monitoring system, the degradation state and failure risk of the pressure-bearing special equipment to be inspected are comprehensively evaluated. When the real-time inspection parameters exceed the preset warning threshold and the remaining service life is lower than the preset safety threshold, it is determined that the pressure-bearing special equipment to be inspected has a failure risk and a risk measurement result is generated; A risk assessment model is constructed using a fuzzy comprehensive evaluation method. The remaining service life prediction results and risk measurement results are combined to form a comprehensive equipment risk assessment matrix. Experts determine the risk level classification threshold and output the risk assessment results. Based on the risk assessment results, a structured rule base is formed according to the inspection standards and industry regulations. A rule-based reasoning engine is used to perform forward reasoning matching on the input degradation physical parameters and monitoring feature sequences, and consistency verification is performed in combination with ontology constraints to generate a preliminary inspection strategy. Based on the preliminary inspection strategy, the inspection strategy optimization model is modeled as a Markov decision process. The Markov decision process is solved by an adaptive optimization algorithm. The weighted combination of risk control effect and inspection cost is used as the reward function. The inspection project combination with the optimal long-term utility under different risk states is learned to generate a dynamically optimized inspection strategy.

7. The method according to claim 6, characterized in that The risk assessment model is constructed using the fuzzy comprehensive evaluation method. The remaining service life prediction results and risk measurement results are combined to form a comprehensive equipment risk assessment matrix. The risk level classification threshold is determined by experts. The output risk assessment results include: The risk assessment model is constructed by using the fuzzy comprehensive evaluation method. The evaluation index membership function in the risk assessment model is used as different evidence sources. Each evidence source provides support for different risk levels. The support is calculated by the evaluation index membership function, indicating the degree of trust in the current risk level and reflecting the influence of each evaluation index on the risk level. By using the evidence combination rules in the evidence reasoning theory, the support provided by different evidence sources is integrated to obtain the comprehensive trust of each risk level; Calculating the confidence and likelihood ratio of each risk level according to the comprehensive confidence, wherein the confidence represents the absolute support for the current risk level, and the likelihood ratio reflects the advantage of the current risk level over other risk levels; The risk levels are ranked and optimized based on the confidence and likelihood comparison, and the ranking results are compared with the risk level division thresholds determined by experts to obtain the risk assessment results.

8. A smart inspection system for pressure-bearing special equipment using a single-machine machine, used to implement the method described in any one of claims 1 to 7, characterized in that: include: The first unit is used to receive an inspection request sent by a terminal device, which includes equipment information of the pressure-bearing special equipment to be inspected, equipment historical inspection data and equipment use environment data, and search for inspection standards and inspection work order information corresponding to the pressure-bearing special equipment to be inspected from a pre-stored equipment information database according to the equipment information. If the corresponding inspection standard and inspection work order information are not found in the equipment information database, the corresponding inspection standard and inspection work order information are automatically generated according to the equipment information of the pressure-bearing special equipment to be inspected and the equipment historical inspection data; The second unit is used to determine the inspected parameters corresponding to the pressure-bearing special equipment to be inspected according to the inspection work order information, and obtain the standard parameter range corresponding to the inspected parameters from the inspection standard, establish a mapping relationship between the inspected parameters and the standard parameter range, and at the same time collect real-time parameter information of the pressure-bearing special equipment to be inspected, convert the collected real-time parameter information into real-time inspection parameters corresponding to the inspected parameters after preprocessing, and construct a prediction model for predicting the remaining service life of the pressure-bearing special equipment to be inspected based on the historical inspection data of the equipment and the equipment use environment data; The third unit is used to use the prediction model to predict the remaining service life of the pressure-bearing special equipment to be inspected, use the prediction result as additional inspection reference information, compare the real-time inspection parameter with the standard parameter range, generate corresponding inspection results according to the comparison result and the reference information, and generate a dynamically optimized inspection strategy according to the inspection process in the inspection work order information, and repeat the inspection process according to the optimized inspection strategy until all inspection processes required in the inspection work order information are completed, and comprehensively generate an inspection report and send it to the terminal device.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method described in any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images