Explosion-proof performance evaluation method and system for explosion-proof equipment

Abstract

The invention relates to the technical field of explosion-proof performance detection, and discloses an explosion-proof performance evaluation method and system for explosion-proof equipment. The method comprises the steps that a geometric model of the explosion-proof equipment is established, stress distribution of the surface of a shell in the explosion-proof equipment test process is obtained through a finite element analysis method, a finite element reduced-order model and a stress distribution agent model are constructed based on finite element analysis results, and the two models are fused to construct an optimization model. And inputting a data-level fusion result of the multi-source heterogeneous data into the optimization model, and predicting the stress distribution condition of the explosion-proof equipment shell. And inputting the measured data and the calculated data into the multi-modal fusion evaluation model to obtain the safety level of the explosion-proof equipment. And finally, realizing visual display of the explosion-proof equipment performance detection test in the digital twin system by adopting a three-dimensional visual modeling method. According to the invention, the comprehensiveness and reliability of the performance evaluation of the explosion-proof equipment are improved.

Description

Technical Field

[0001] The present application relates to the technical field of explosion-proof performance detection, and in particular to an explosion-proof performance evaluation method and system for explosion-proof equipment. Background Art

[0002] The explosion-proof performance test is a test to evaluate the safety performance of products in potentially explosive environments. In many industrial fields, such as petroleum, chemical, coal, etc., equipment or systems may generate explosive gas or dust during operation. In order to ensure that these equipment or systems can operate safely in potentially explosive environments, explosion-proof performance tests are particularly important.

[0003] At present, the explosion-proof performance test only verifies the explosion-proof performance and flameproof capability of the equipment through a single pressure data, and cannot accurately evaluate the impact of various physical parameters during the explosion process on the performance of explosion-proof equipment. Traditional pressure detection tests need to be carried out in a closed explosion-proof tank, and the test process is not visible. The construction of the test system and the subsequent repeated testing of the test samples require a lot of funds and manpower and material resources, the preparation cycle is long, and there are certain safety hazards in the test process. Therefore, a new measurement method is needed to evaluate the reliability of explosion-proof equipment and provide a basis for its design, processing and manufacturing. Summary of the invention

[0004] This application provides an explosion-proof performance evaluation method and system for explosion-proof equipment. The multi-source heterogeneous data fusion method can collect different characteristic parameters of explosion-proof equipment through multiple sensors, characterize multiple physical quantities, and construct a multi-source heterogeneous data fusion model to solve the problem of the singleness of data in traditional detection methods. Combined with digital twin technology, the test process is visualized, and the deficiencies of current explosion-proof performance tests are made up through multi-modal fusion evaluation methods.

[0005] Resolve the deficiencies of traditional explosion-proof performance tests and improve the comprehensiveness and reliability of explosion-proof equipment performance evaluation.

[0006] In a first aspect, the present application provides a method for evaluating explosion-proof performance of explosion-proof equipment, the method comprising:

[0007] Provide a data acquisition module to collect multi-source heterogeneous data, and pre-process and fuse the multi-source heterogeneous data to obtain fused data, wherein the multi-source heterogeneous data includes pressure data, vibration data, visual images, and gas concentration change data;

[0008] Provide a model building module, based on the geometric characteristics of explosion-proof equipment, use three-dimensional modeling software to build a geometric model, simplify the geometric model to establish a simplified model of the explosion-proof equipment, mesh the simplified model, and record the coordinates and numbers of the mesh nodes, then perform finite element analysis, build a finite element reduced-order model and a stress distribution proxy model based on the finite element analysis results, and fuse the finite element reduced-order model and the stress distribution proxy model to build a fusion optimization model;

[0009] A performance evaluation module is provided to extract the first pressure data and the first vibration data from the fusion data, establish a feature layer information fusion evaluation model based on the RBF neural network based on the data characteristics of the first pressure data and the first vibration data, and then input the first pressure data and the first vibration data into the feature layer information fusion evaluation model to obtain the safety level of the explosion-proof equipment;

[0010] Provide a visual display module, use digital twin software to develop a visualization interface, build a digital twin system for explosion-proof equipment, and display the performance and status of the explosion-proof equipment during the test in the digital twin system based on fusion data and fusion optimization models.

[0011] In combination with the first aspect, in a first implementation of the first aspect of the present application, the preprocessing and fusion include:

[0012] Convert multi-source heterogeneous data into a unified format and store it in the database;

[0013] The interpolation function of "data point-time" is established by linear fitting to make interpolation predictions for missing data;

[0014] The non-local means method is used to denoise the multi-source heterogeneous data after interpolation.

[0015] The Kalman filter algorithm is used to process the redundancy of multi-source heterogeneous data after denoising;

[0016] The redundantly processed multi-source heterogeneous data are synchronously processed through time frames to ensure the alignment of data sources at different time scales.

[0017] In combination with the first aspect, in the second implementation method of the first aspect of the present application, the simplification process includes: analyzing the geometric model, deleting the screw holes and welds in the geometric model according to the analysis results, and homogenizing the elastic modulus and density of the material.

[0018] In combination with the first aspect, in a third implementation method of the first aspect of the present application, performing finite element analysis includes: using a dynamic analysis method to construct a shell structure analysis model of the explosion process of the explosion-proof equipment, and setting the gas type and concentration in the explosion-proof equipment at the initial moment, as well as the initial pressure, temperature, ignition position and flame shape of the explosion, and outputting finite element analysis results, which include stress magnitude and deformation of the grid division.

[0019] In combination with the first aspect, in a fourth implementation of the first aspect of the present application, a reduced-order model is trained on the shell structure analysis model using a POD algorithm to construct a finite element reduced-order model, including:

[0020] Conduct multiple finite element analyses to obtain simulation data under different working conditions, where the simulation data includes the distribution of stress in space and time;

[0021] Extract any simulation data, generate a simulation vector based on any simulation data, and combine all simulation vectors to generate a data matrix;

[0022] Calculate the covariance matrix of the data matrix, perform eigendecomposition on the covariance matrix, and obtain eigenvalues ​​and eigenvectors;

[0023] The eigenvectors are sorted based on the size of the eigenvalues, and the first k eigenvectors with high contribution are selected and defined as the main eigenvectors. The finite element reduced-order model is constructed based on the main eigenvectors.

[0024] In combination with the first aspect, in the fifth implementation method of the first aspect of the present application, the method for constructing a stress distribution proxy model is: obtaining simulation data, dividing the simulation data into a training set and a test set according to a preset ratio, using the training set to train a random forest network to obtain a proxy model, and then using the test set to verify and optimize the proxy model to generate a stress distribution proxy model.

[0025] In combination with the first aspect, in the sixth implementation method of the first aspect of the present application, the method for constructing the fusion optimization model is: inputting the simulation data into the stress distribution proxy model to obtain the predicted value, using the determination coefficient to calculate the similarity between the label value and the predicted value, evaluating the prediction result of the stress distribution proxy model, and correcting the output of the finite element reduced-order model based on the predicted result to obtain the fusion optimization model.

[0026] In combination with the first aspect, in a seventh implementation of the first aspect of the present application, before obtaining the security level, the first diagnostic result of the RBF neural network is optimized using the improved DS evidence theory, including:

[0027] Use CNN neural network to diagnose the fused data and obtain the second diagnosis result;

[0028] Normalizing the first diagnosis result and the second diagnosis result, and using them as a new basic probability allocation value BPA;

[0029] The evidence of the two neural networks is re-obtained through the weighted idea, and the decision-level fusion is achieved through the combination rules of evidence theory to obtain the security level.

[0030] In combination with the first aspect, in an eighth implementation of the first aspect of the present application, the visual display module is further used to:

[0031] According to the coordinates and numbers of the nodes after meshing the simplified model, the geometric model is reconstructed in the digital twin system using 3D visualization technology;

[0032] Input the fused data into the fusion optimization model to obtain the stress magnitude and deformation degree of the nodes of the simplified mesh model;

[0033] According to the stress magnitude and deformation degree, different color gradients are used for mapping, and corresponding colors are assigned to each node of the geometric model to show the performance and status of the explosion-proof equipment during the test.

[0034] In a second aspect, the present application provides an explosion-proof performance evaluation system for explosion-proof equipment, which implements the explosion-proof performance evaluation method for explosion-proof equipment described in the first aspect, and the system includes: a data acquisition module, a model building module, a performance evaluation module and a visual display module;

[0035] A data acquisition module is used to collect multi-source heterogeneous data, and pre-process and fuse the multi-source heterogeneous data to obtain fused data, wherein the multi-source heterogeneous data includes pressure data, vibration data, visual images, and gas concentration change data;

[0036] A model building module is used to build a geometric model using three-dimensional modeling software according to the geometric characteristics of the explosion-proof equipment, simplify the geometric model to establish a simplified model of the explosion-proof equipment, mesh the simplified model, and record the coordinates and numbers of the mesh nodes, and then perform finite element analysis. Based on the results of the finite element analysis, a finite element reduced-order model and a stress distribution proxy model are constructed, and the finite element reduced-order model and the stress distribution proxy model are fused to construct a fusion optimization model;

[0037] A performance evaluation module is used to extract the first pressure data and the first vibration data from the fused data, establish a feature layer information fusion evaluation model based on the RBF neural network based on the data characteristics of the first pressure data and the first vibration data, and then input the first pressure data and the first vibration data into the feature layer information fusion evaluation model to obtain the safety level of the explosion-proof equipment;

[0038] The visual display module is used to develop a visualization interface using digital twin software, build a digital twin system for explosion-proof equipment, and display the performance and status of the explosion-proof equipment during the test in the digital twin system based on fused data and fused optimization models.

[0039] Compared with the prior art, the beneficial effects of the technical solution of the present application are at least as follows:

[0040] 1. By collecting multi-source heterogeneous data and performing preprocessing and fusion, the performance of explosion-proof equipment in actual use can be more comprehensively reflected, thereby improving the accuracy of the evaluation.

[0041] 2. Use 3D modeling software to build the geometric model of explosion-proof equipment, simplify it and divide the mesh. Combined with finite element analysis, it can accurately simulate the stress distribution and deformation of explosion-proof equipment under various working conditions, effectively predict the performance of explosion-proof equipment in actual applications, and improve the accuracy and reliability of the evaluation. At the same time, the finite element reduced-order model and the stress distribution proxy model are integrated to build a fusion optimization model, which can further optimize the evaluation results and improve the accuracy and efficiency of the evaluation.

[0042] 3. By establishing a feature layer information fusion evaluation model based on RBF neural network and inputting multi-source data into the model for comprehensive analysis, the safety level of explosion-proof equipment can be evaluated more comprehensively, the introduction of DS evidence theory and CNN neural network can be improved, and more reliable evaluation results can be provided.

[0043] 4. Using digital twin technology to build a visualization interface can display the performance and status of explosion-proof equipment in real time, allowing operators to intuitively understand the operation of the equipment. The stress changes and deformation degree are displayed through color mapping technology, which further enhances the intuitiveness and readability of the visualization, helps to quickly identify and locate problems, and improves the efficiency and accuracy of the assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0045] Figure 1 A schematic diagram of an embodiment of a method for evaluating explosion-proof performance of explosion-proof equipment in an embodiment of the present application;

[0046] Figure 2 This is a schematic diagram of an embodiment of a method for generating fusion data in an embodiment of the present application;

[0047] Figure 3 A schematic diagram of an embodiment of a method for generating a fusion optimization model in an embodiment of the present application;

[0048] Figure 4 This is a schematic diagram of an embodiment of an explosion-proof performance evaluation system for explosion-proof equipment in an embodiment of the present application. DETAILED DESCRIPTION

[0049] The embodiment of the present application provides an explosion-proof performance evaluation method and system for explosion-proof equipment. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0050] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 , an embodiment of the explosion-proof performance evaluation method for explosion-proof equipment in the embodiments of the present application includes:

[0051] A data acquisition module is provided to collect multi-source heterogeneous data, and pre-process and fuse the multi-source heterogeneous data to obtain fused data, wherein the multi-source heterogeneous data includes pressure data, vibration data, visual images, and gas concentration change data.

[0052] Specifically, the pressure data, vibration data and visual images of the explosion-proof equipment during the explosion-proof performance test are collected through pressure sensors, laser Doppler, and high-speed cameras, and the changes in gas concentration (such as methane, oxygen, hydrogen, etc.) during the test are obtained through gas concentration sensors. Pressure data is used to evaluate the pressure bearing capacity of the equipment under the action of the explosion shock wave, and to detect whether there is overpressure or uneven pressure; vibration data is used to evaluate the vibration characteristics of the equipment, detect whether there is resonance or excessive vibration, and evaluate its structural stability; visual images are used for visual monitoring to detect whether there are cracks, deformations or other damage on the surface of the equipment, and evaluate its appearance integrity and structural changes; gas concentration changes are used to evaluate whether the explosion-proof equipment effectively suppresses the leakage of harmful gases during the explosion process, and to detect whether there is gas leakage or concentration exceeds the standard. The accuracy and comprehensiveness of data collection directly affect the reliability of the entire evaluation process.

[0053] A model building module is provided. Based on the geometric characteristics of explosion-proof equipment, a geometric model is constructed using three-dimensional modeling software. The geometric model is simplified to establish a simplified model of the explosion-proof equipment. The simplified model is meshed and the coordinates and numbers of the mesh nodes are recorded. Finite element analysis is then performed. Based on the finite element analysis results, a finite element reduced-order model and a stress distribution proxy model are constructed. The finite element reduced-order model and the stress distribution proxy model are fused to construct a fusion optimization model.

[0054] Specifically, the geometric characteristics of explosion-proof equipment usually come from design drawings, CAD models or actual measurement data. These geometric characteristics are used to construct the geometric model of the explosion-proof equipment through 3D modeling software (such as SolidWorks, AutoCAD, ANSYS, etc.). Among them, the appropriate 3D modeling software is selected according to the complexity and precision requirements of the explosion-proof equipment. For example, for complex curved surface structures, software with powerful curved surface modeling functions can be selected.

[0055] In order to reduce the amount of calculation and improve the efficiency of analysis, the geometric model is simplified after it is built to obtain a simplified model, and then the simplified model is meshed and discretized into a finite number of units for numerical calculation. Commonly used meshing methods include tetrahedral meshes and hexahedral meshes. Selecting the appropriate mesh type and density can improve the accuracy and efficiency of analysis. Based on the theory of elastic mechanics and plastic mechanics, the stress condition of explosion-proof equipment is decomposed into the mechanical behavior of a finite number of units for solution, in order to analyze the stress distribution and deformation of explosion-proof equipment under the action of explosion shock waves, and evaluate its structural strength and stability.

[0056] In order to reduce the complexity of the model, improve the computational efficiency, and maintain the key analysis accuracy, model reduction techniques such as principal component analysis (PCA) and singular value decomposition (SVD) are used to extract the main features of the model and construct a finite element reduced-order model. At the same time, in order to establish the mathematical relationship between stress distribution and geometric parameters, material properties, etc., so as to quickly predict stress distribution, regression analysis, neural network and other methods are used to establish a stress distribution proxy model. Through data fusion, model combination and other methods, the finite element reduced-order model and the stress distribution proxy model are integrated to form a comprehensive fusion optimization model to further improve the performance and computational accuracy of the model.

[0057] A performance evaluation module is provided to extract the first pressure data and the first vibration data from the fusion data, and a feature layer information fusion evaluation model based on the RBF neural network is established based on the data characteristics of the first pressure data and the first vibration data. Subsequently, the first pressure data and the first vibration data are input into the feature layer information fusion evaluation model to obtain the safety level of the explosion-proof equipment.

[0058] Specifically, according to the data characteristics of pressure and vibration signals, different methods are used for feature extraction. Combined with the nonlinear relationship between pressure and vibration data and the semantic variables of the evaluation method, a feature layer information fusion evaluation model based on RBF neural network is established.

[0059] Using different characteristic information collected by pressure sensor and laser Doppler as input layer and safety level as output layer, the characteristic data is obtained by finite element method analysis. , the data center of the RBF neural network is , the weight between the hidden layer and the output layer is The mathematical model of the established RBF neural network is , output It indicates the degree of membership of explosion-proof equipment to various safety levels. Generally, its safety level is determined according to the principle of maximum membership, and the reliability of explosion-proof equipment is evaluated at the feature level.

[0060] Provide a visual display module, use digital twin software to develop a visualization interface, build a digital twin system for explosion-proof equipment, and display the performance and status of the explosion-proof equipment during the test in the digital twin system based on fusion data and fusion optimization models.

[0061] Specifically, a visualization interface is developed using digital twin software to build a virtual measurement system for explosion-proof equipment. Pressure sensors, laser Doppler, high-speed cameras and gas concentration sensors are used to collect pressure signals, vibration signals, visual images and measured data of various gas concentrations during the explosion-proof performance test of explosion-proof equipment in real time, and the measured data are converted into measured data packets and uploaded to the database server. The digital twin system sends a request instruction for obtaining measured data to the server through wireless communication. After the server verifies that the instruction is consistent with the preset instruction in the server, the server sends the measured data packet to the digital twin system. The obtained measured data packet is parsed by the digital twin system to obtain the measured data of the explosion-proof equipment and classify and store it in the digital twin system. The measured data and calculated data are input into the fusion optimization model to realize virtual measurement and reliability evaluation of the explosion-proof equipment.

[0062] In a specific embodiment, preprocessing and fusion include:

[0063] (1) Convert multi-source heterogeneous data into a unified format and store it in a database.

[0064] (2) Use linear fitting to establish the “data point-time” interpolation function and perform interpolation prediction on missing data.

[0065] (3) The non-local means method is used to denoise the multi-source heterogeneous data after interpolation processing.

[0066] (4) Use the Kalman filter algorithm to perform redundant processing on the multi-source heterogeneous data after denoising.

[0067] (5) Synchronize the redundantly processed multi-source heterogeneous data through time frames to ensure the alignment of data sources at different time scales.

[0068] Specifically, the acquisition formats of different sensors are converted into a unified format (such as JSON, CSV, etc.) and stored in the database; the complete data points in the data set are used to find the rules of the data sequence, and the "data point-time" interpolation function is established by linear fitting. Finally, the true value of the missing data in the real-time sequence is predicted based on the interpolation function; the non-local mean method is used to denoise the multi-source heterogeneous data, effectively remove the dirty data, and retain and strengthen the valid data in the time series as much as possible; the Kalman filter algorithm is used to perform redundant processing on the multi-source heterogeneous data, so that it can obtain a good filtering estimation effect. After the filtering algorithm, the data redundancy can be better reduced, and a better data combination can be achieved with higher accuracy; the collected multi-source data is synchronously processed through the time frame, and the multi-source heterogeneous data is modeled in the database, and a data table is created. Different sensor data are used as different columns, and the unit length of the time frame is set to ensure that the data sources of different time scales are aligned, so as to achieve the fusion of multi-source heterogeneous data at the data level.

[0069] The flowchart of the fusion data generation method is as follows: Figure 2 shown.

[0070] In a specific embodiment, the simplification process includes: analyzing the geometric model, deleting the screw holes and welds in the geometric model according to the analysis results, and homogenizing the elastic modulus and density of the material.

[0071] Specifically, through the analysis of the geometric model, the screw holes and welds are simply processed using the deletion method, uniform material properties are used to replace the non-uniform material distribution, and it is assumed that the elastic modulus and density of the material are uniform in the entire structure to establish a simplified model of explosion-proof equipment.

[0072] In a specific embodiment, performing finite element analysis includes: using a dynamic analysis method to construct a shell structure analysis model of the explosion process of the explosion-proof equipment, and setting the gas type and concentration in the explosion-proof equipment at the initial moment, as well as the initial pressure, temperature, ignition position and flame shape of the explosion, and outputting the finite element analysis results, which include the stress size and deformation of the grid division.

[0073] Specifically, a dynamic analysis method in the finite element analysis method is used to construct a shell structure analysis model of the explosion process of explosion-proof equipment, and the gas type and concentration in the explosion-proof equipment at the initial moment, the initial pressure, temperature, ignition position and flame shape of the explosion are set, and the finite element analysis results of the geometric model of the explosion-proof equipment are output. The finite element analysis results of the geometric model of the explosion-proof equipment include the stress size and deformation of the mesh division of the geometric model of the explosion-proof equipment. Indicates The stress distribution in the simulation results is are spatial coordinates, It's time.

[0074] In a specific embodiment, the POD algorithm is used to perform reduced-order model training on the shell structure analysis model to construct a finite element reduced-order model, including:

[0075] (1) Perform multiple finite element analyses to obtain simulation data under different working conditions, where the simulation data includes the distribution of stress in space and time.

[0076] (2) Extract any simulation data, generate a simulation vector based on any simulation data, and combine all simulation vectors to generate a data matrix.

[0077] (3) Calculate the covariance matrix of the data matrix, perform eigendecomposition on the covariance matrix, and obtain eigenvalues ​​and eigenvectors.

[0078] (4) Sort the eigenvectors based on the size of the eigenvalues, select the top k eigenvectors with the highest contribution, define them as the main eigenvectors, and construct the finite element reduced-order model based on the main eigenvectors.

[0079] Specifically, the POD algorithm is used to reduce the order of the original finite element model and the stress distribution of a simulation sample in space and time is represented as a vector ,have Different simulation results are combined into a The data matrix , for the matrix , calculate its covariance matrix , for the covariance matrix Perform eigenvalue decomposition to obtain eigenvalues ​​and eigenvectors: , select the first few eigenvectors with large eigenvalues ​​and high contribution to construct the reduced-order model. The original stress-strain field can be approximated in the following way: ,in is a time function, indicating The time-varying coefficients of each mode are calculated. The calculation speed of the finite element model is improved by reducing the order.

[0080] In a specific embodiment, a method for constructing a stress distribution proxy model is as follows: obtaining simulation data, dividing the simulation data into a training set and a test set according to a preset ratio, using the training set to train a random forest network to obtain a proxy model, and then using the test set to verify and optimize the proxy model to generate a stress distribution proxy model.

[0081] In a specific embodiment, the method for constructing the fusion optimization model is: inputting simulation data into the stress distribution proxy model to obtain the predicted value, using the determination coefficient to calculate the similarity between the label value and the predicted value, evaluating the prediction result of the stress distribution proxy model, and correcting the output of the finite element reduced-order model based on the prediction result to obtain the fusion optimization model.

[0082] Specifically, according to the simulation data, the training set and the test set are divided into training set and test set in the ratio of 7:3. The proxy model of the surface stress change of the explosion-proof equipment shell is trained by the random forest network. The determination coefficient is used to calculate the similarity between the label value and the predicted value, and the prediction result of the proxy model is evaluated. The prediction result of the proxy model is used to correct the output of the finite element reduced-order model to improve the calculation accuracy of the model.

[0083] The flowchart of the fusion optimization model generation method is as follows: Figure 3 shown.

[0084] In a specific embodiment, before obtaining the security level, the first diagnosis result of the RBF neural network is optimized using the improved DS evidence theory, including:

[0085] (1) Use CNN neural network to diagnose the fused data and obtain the second diagnosis result.

[0086] (2) The first diagnosis result and the second diagnosis result are normalized and used as the new basic probability allocation value BPA.

[0087] (3) The evidence of the two neural networks is re-obtained through the weighted idea, and the decision-level fusion is achieved through the combination rules of evidence theory to obtain the security level.

[0088] Specifically, in view of the problems existing in RBF neural network diagnosis, the improved DS evidence theory is used for optimization. The RBF and CNN neural networks are combined with the evidence theory. The diagnostic outputs of each neural network are processed in a normalized manner and the results are used as the new basic probability distribution value (BPA). The normalization formula is:

[0089] ;

[0090] in represent Class fault BPA value, Represents the actual diagnostic output value of the network; Represents the network output error, and the error calculation formula is:

[0091] ;

[0092] in , Respectively represent Network The ideal output and actual output of a neuron.

[0093] The evidences of the two neural networks are re-obtained using the weighted idea, and then the decision-level fusion is realized through the combination rules of evidence theory to obtain the final safety level evaluation result.

[0094] Semantic variable set The quantitative analysis is carried out, and the test set explosion-proof shell damage sample data is input into the network for safety level assessment. The BPA value is significantly improved compared with the single neural network through the improved evidence theory fusion method, which verifies the effectiveness of the method and realizes the evaluation of the safety level of explosion-proof equipment.

[0095] In a specific embodiment, the visual display module is also used to:

[0096] (1) Reconstruct the geometric model in the digital twin system using 3D visualization technology based on the coordinates and numbers of the nodes after meshing the simplified model;

[0097] (2) Input the fused data into the fusion optimization model to obtain the stress magnitude and deformation degree of the nodes in the simplified mesh model;

[0098] (3) Different color gradients are used for mapping according to the stress magnitude and deformation degree, and corresponding colors are assigned to each node of the geometric model to display the performance and status of the explosion-proof equipment during the test.

[0099] Specifically, pressure sensors, laser Doppler, high-speed cameras and gas concentration sensors are used to collect the pressure signals, vibration signals, visual images and measured data of various gas concentrations during the explosion-proof performance test of explosion-proof equipment in real time, and the measured data are formed into measured data packets and uploaded to the database server; the digital twin system sends a request instruction for obtaining measured data to the server through wireless communication, and after the server verifies that the instruction is consistent with the preset instruction in the server, the server sends the measured data packet to the digital twin system; the obtained measured data packet is parsed by the digital twin system to obtain the measured data of the explosion-proof equipment and classify and store it in the digital twin system; according to The coordinates and numbers of the mesh nodes of the simplified geometric model of explosion-proof equipment are used to reconstruct the geometric model of explosion-proof equipment in the digital twin system using three-dimensional visualization technology; the measured data are input into the optimized shell surface stress change proxy model to predict the stress change and deformation degree of the nodes of the simplified mesh model of explosion-proof equipment, and the data is stored; the digital twin system uses color mapping technology to assign corresponding colors to the nodes of the geometric model of explosion-proof equipment according to the calculated structural stress and structural deformation data for the reconstructed geometric model of explosion-proof equipment, display the performance and status of the explosion-proof equipment during the test, and realize virtual measurement of the explosion-proof equipment.

[0100] The measured data and calculated data are input into the multimodal fusion evaluation model, and the safety level of the explosion-proof equipment is output to achieve the reliability evaluation of the explosion-proof equipment.

[0101] The above describes the explosion-proof performance evaluation method for explosion-proof equipment in the embodiment of the present application. The following describes the explosion-proof performance evaluation system for explosion-proof equipment in the embodiment of the present application. Figure 4 In one embodiment of the present application, an explosion-proof performance evaluation system for explosion-proof equipment includes: a data acquisition module 10, a model building module 20, a performance evaluation module 30 and a visual display module 40.

[0102] The data acquisition module 10 is used to collect multi-source heterogeneous data, and pre-process and fuse the multi-source heterogeneous data to obtain fused data, wherein the multi-source heterogeneous data includes pressure data, vibration data, visual images, and gas concentration change data.

[0103] The model building module 20 is used to build a geometric model based on the geometric characteristics of the explosion-proof equipment using three-dimensional modeling software, simplify the geometric model to establish a simplified model of the explosion-proof equipment, mesh the simplified model, and record the coordinates and numbers of the mesh nodes, and then perform finite element analysis. Based on the finite element analysis results, a finite element reduced-order model and a stress distribution proxy model are constructed, and the finite element reduced-order model and the stress distribution proxy model are merged to construct a fusion optimization model.

[0104] The performance evaluation module 30 is used to extract the first pressure data and the first vibration data from the fused data, establish a feature layer information fusion evaluation model based on the RBF neural network based on the data characteristics of the first pressure data and the first vibration data, and then input the first pressure data and the first vibration data into the feature layer information fusion evaluation model to obtain the safety level of the explosion-proof equipment.

[0105] The visual display module 40 is used to develop a visualization interface using digital twin software, build a digital twin system for explosion-proof equipment, and display the performance and status of the explosion-proof equipment during the test in the digital twin system based on fused data and fused optimization models.

[0106] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0107] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0108] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application 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 make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for evaluating explosion-proof performance of explosion-proof equipment, characterized in that: The method comprises: Provide a data acquisition module for acquiring multi-source heterogeneous data, and preprocessing and fusing the multi-source heterogeneous data to obtain fused data, wherein the multi-source heterogeneous data includes pressure data, vibration data, visual images, and gas concentration change data; A model building module is provided. Based on the geometric characteristics of the explosion-proof equipment, a geometric model is constructed using three-dimensional modeling software. The geometric model is simplified to establish a simplified model of the explosion-proof equipment. The simplified model is meshed and the coordinates and numbers of the mesh nodes are recorded. Then, a finite element analysis is performed. A finite element reduced-order model and a stress distribution proxy model are constructed based on the finite element analysis results. The finite element reduced-order model and the stress distribution proxy model are fused to construct a fused optimization model. providing a performance evaluation module, extracting first pressure data and first vibration data from the fused data, establishing a feature layer information fusion evaluation model based on RBF neural network based on data characteristics of the first pressure data and the first vibration data, and then inputting the first pressure data and the first vibration data into the feature layer information fusion evaluation model to obtain the safety level of the explosion-proof equipment; A visual display module is provided, a visualization interface is developed using digital twin software, a digital twin system of explosion-proof equipment is constructed, and the performance and status of the explosion-proof equipment during the test are displayed in the digital twin system based on the fusion data and the fusion optimization model.

2. The explosion-proof performance evaluation method for explosion-proof equipment according to claim 1, characterized in that: The preprocessing and fusion include: Convert the multi-source heterogeneous data into a unified format and store them in a database; The interpolation function of "data point-time" is established by linear fitting to make interpolation predictions for missing data; Using a non-local means method to perform denoising on the multi-source heterogeneous data after interpolation processing; Using a Kalman filter algorithm to perform redundancy processing on the multi-source heterogeneous data after denoising; The multi-source heterogeneous data after redundant processing is synchronously processed through a time frame to ensure that data sources at different time scales are aligned.

3. The explosion-proof performance evaluation method for explosion-proof equipment according to claim 1, characterized in that: The simplification process includes: analyzing the geometric model, deleting the screw holes and welding seams in the geometric model according to the analysis results, and homogenizing the elastic modulus and density of the material.

4. The explosion-proof performance evaluation method for explosion-proof equipment according to claim 1, characterized in that: The finite element analysis includes: using a dynamic analysis method to construct a shell structure analysis model of the explosion process of the explosion-proof equipment, and setting the gas type and concentration in the explosion-proof equipment at the initial moment, as well as the initial pressure, temperature, ignition position and flame shape of the explosion, and outputting the finite element analysis results, which include the stress size and deformation of the grid division.

5. The explosion-proof performance evaluation method for explosion-proof equipment according to claim 4, characterized in that: The POD algorithm is used to perform order reduction model training on the shell structure analysis model to construct a finite element order reduction model, including: Performing the finite element analysis multiple times to obtain simulation data under different working conditions, wherein the simulation data includes the distribution of stress in space and time; Extract any simulation data, generate a simulation vector based on any of the simulation data, and combine all simulation vectors to generate a data matrix; Calculating the covariance matrix of the data matrix, performing eigendecomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors; The eigenvectors are sorted based on the size of the eigenvalues, the first k eigenvectors with high contribution are selected and defined as main eigenvectors, and the finite element reduced-order model is constructed based on the main eigenvectors.

6. The explosion-proof performance evaluation method for explosion-proof equipment according to claim 5, characterized in that: The method for constructing the stress distribution proxy model is as follows: obtaining the simulation data, dividing the simulation data into a training set and a test set according to a preset ratio, using the training set to train a random forest network to obtain a proxy model, and then using the test set to verify and optimize the proxy model to generate the stress distribution proxy model.

7. The explosion-proof performance evaluation method for explosion-proof equipment according to claim 6, characterized in that: The method for constructing the fusion optimization model is: inputting the simulation data into the stress distribution proxy model to obtain a predicted value, using a determination coefficient to calculate the similarity between the label value and the predicted value, evaluating the prediction result of the stress distribution proxy model, and correcting the output of the finite element reduced-order model based on the prediction result to obtain the fusion optimization model.

8. The explosion-proof performance evaluation method for explosion-proof equipment according to claim 1, characterized in that: Before obtaining the security level, the first diagnosis result of the RBF neural network is optimized by using the improved DS evidence theory, including: Using a CNN neural network to diagnose the fused data to obtain a second diagnosis result; Normalizing the first diagnosis result and the second diagnosis result, and using the normalized values ​​as a new basic probability allocation value BPA; The evidence of the two neural networks is re-obtained through the weighted idea, and the decision-level fusion is realized through the combination rules of evidence theory to obtain the security level.

9. The explosion-proof performance evaluation method for explosion-proof equipment according to claim 1, characterized in that: The visual display module is also used to: Reconstructing the geometric model in the digital twin system using three-dimensional visualization technology according to the coordinates and numbers of the nodes after meshing the simplified model; Inputting the fused data into the fusion optimization model to obtain the stress magnitude and deformation degree of the nodes of the simplified mesh model; According to the stress magnitude and the deformation degree, different color gradients are used for mapping, and corresponding colors are assigned to each node of the geometric model to display the performance and status of the explosion-proof equipment during the test.

10. An explosion-proof performance evaluation system for explosion-proof equipment, which implements the explosion-proof performance evaluation method for explosion-proof equipment according to any one of claims 1 to 9, characterized in that: The system includes: a data acquisition module, a model building module, a performance evaluation module and a visual display module; The data acquisition module is used to collect multi-source heterogeneous data, and pre-process and fuse the multi-source heterogeneous data to obtain fused data, wherein the multi-source heterogeneous data includes pressure data, vibration data, visual images, and gas concentration change data; The model building module is used to build a geometric model using three-dimensional modeling software according to the geometric characteristics of the explosion-proof equipment, simplify the geometric model to establish a simplified model of the explosion-proof equipment, mesh the simplified model, and record the coordinates and numbers of the mesh nodes, then perform finite element analysis, build a finite element reduced-order model and a stress distribution proxy model based on the finite element analysis results, and fuse the finite element reduced-order model and the stress distribution proxy model to build a fusion optimization model; The performance evaluation module is used to extract the first pressure data and the first vibration data from the fused data, establish a feature layer information fusion evaluation model based on the RBF neural network based on the data characteristics of the first pressure data and the first vibration data, and then input the first pressure data and the first vibration data into the feature layer information fusion evaluation model to obtain the safety level of the explosion-proof equipment; The visual display module is used to develop a visualization interface using digital twin software, build a digital twin system for explosion-proof equipment, and display the performance and status of the explosion-proof equipment during the test in the digital twin system based on the fusion data and the fusion optimization model.

Images