Software operation and maintenance method, system and storage medium

Abstract

The application relates to the technical field of data mining, in particular to a software operation method and system. The method comprises the following steps: acquiring engineering simulation heterogeneous hardware data, and performing multi-physical field coupling tensor compression to obtain a multi-physical field topology snapshot; performing finite element grid nonlinear constraint based on the multi-physical field topology snapshot to obtain a nonlinear contact strategy tensor; encoding a structure dynamics intelligent contract according to the nonlinear contact strategy tensor, and performing simulation software dynamics credible verification to obtain a dynamics credible chain; constructing a fluid mechanics vortex model based on the dynamics credible chain, and performing boundary layer separation simulation flow optimization on the fluid mechanics vortex model to obtain a vortex flow field atlas; performing explicit dynamics explicit integration on the vortex flow field atlas to obtain a convergence optimization field; performing phase field damage evolution according to the convergence optimization field to obtain a multi-physical field evolution twin, and deploying the multi-physical field evolution twin. The application can effectively improve the utilization rate of software computing resources.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data mining, and particularly relates to a software operation and maintenance method and system and a storage medium. BACKGROUND

[0002] Engineering simulation software assists engineers in modeling, optimizing and verifying complex systems through numerical calculation, finite element analysis, multi-physical field coupling simulation and other methods. With the improvement of computer hardware performance and the development of high-performance computing (HPC) technology, engineering simulation software has gradually developed from single-machine operation mode to distributed computing, cloud simulation and parallel computing. However, due to high computing load, complex data interaction and cross-platform compatibility, engineering simulation software operation and maintenance management faces many technical challenges. Traditional engineering simulation software operation and maintenance methods mainly rely on manual management and static rule configuration, such as manual monitoring of server status, regular maintenance of computing clusters, manual adjustment of computing resource allocation and offline analysis of system logs. Engineering simulation software usually involves large-scale numerical calculation and complex data processing, and the execution time of the computing task is longer and the demand for computing resources is higher. The traditional operation and maintenance method mainly relies on manual configuration of computing nodes and storage resources, which is difficult to meet the needs of dynamic computing load, resulting in uneven resource allocation and affecting computing efficiency. SUMMARY

[0003] Therefore, it is necessary to provide a software operation and maintenance method, system and storage medium to solve at least one of the above technical problems.

[0004] To achieve the above-mentioned purpose, a software operation and maintenance method comprises the following steps:

[0005] Step S1: Obtain engineering simulation heterogeneous hardware data, and perform multi-physical field coupling tensor compression on the engineering simulation heterogeneous hardware data to obtain a multi-physical field topology snapshot;

[0006] Step S2: Perform finite element grid nonlinear constraint based on the multi-physical field topology snapshot to obtain a nonlinear contact strategy tensor;

[0007] Step S3: Encode a structure dynamics intelligent contract according to the nonlinear contact strategy tensor, and perform simulation software dynamics credible verification on the structure dynamics intelligent contract to obtain a dynamics credible chain;

[0008] Step S4: Construct a fluid mechanics vortex model based on the dynamics credible chain, and perform boundary layer separation simulation flow optimization on the fluid mechanics vortex model to obtain a vortex flow field atlas;

[0009] Step S5: Explicit dynamic explicit integration is performed on the vortex flow field pattern to obtain a convergent optimization field; a phase field damage evolution is performed according to the convergent optimization field to obtain a multi-physical field evolution twin, and the multi-physical field evolution twin is deployed through a pre-set engineering simulation cloud architecture.

[0010] The present application effectively reduces data redundancy and improves calculation efficiency by obtaining engineering simulation heterogeneous hardware data and performing multi-physical field coupling tensor compression, while ensuring that the interaction information between multi-physical fields is retained, providing high-quality input data for subsequent calculations. Based on the topology snapshot of the multi-physical field, the nonlinear constraint of the finite element grid is executed, which can accurately describe the contact behavior of complex structures and improve the stability and accuracy of contact calculation, thereby improving the physical realism of the simulation. Through the generation of a nonlinear contact strategy tensor and intelligent contract coding, the automatic management of structural dynamics constraint conditions is realized, and the credibility and traceability of the simulation calculation are enhanced by combining with the trusted verification technology, ensuring the reliability of the calculation results. Based on the dynamic trusted chain, a fluid dynamics vortex model is constructed, so that the vortex characteristics can be optimized under the constraint of dynamics, and the flow distribution is optimized through boundary layer separation simulation, improving the convergence and accuracy of fluid calculation. The explicit dynamic integration method can efficiently solve the dynamic characteristics of the vortex flow field, making the calculation process more stable, and finally obtaining a convergent optimization field, which provides accurate dynamic input for subsequent phase field damage evolution. The phase field damage evolution can capture the material evolution process in the multi-physical field system, making the simulation results closer to the actual working conditions, and combining with the engineering simulation cloud architecture for deployment, improving the utilization of computing resources, supporting large-scale parallel computing and remote simulation, and realizing efficient operation and intelligent management of engineering simulation software.

[0011] Optionally, step S1 is specifically:

[0012] Step S11: Obtain engineering simulation heterogeneous hardware multi-physical field data, and perform time-sensitive synchronization processing on the multi-physical field data to generate an initial multi-physical field data set;

[0013] Step S12: Perform non-Euclidean topology mapping on the initial multi-physical field data set, and perform data standardization and time sequence consistency correction to generate standardized multi-physical field data;

[0014] Step S13: Perform tensor principal component analysis based on the standardized multi-physical field data to extract material nonlinear characteristics including strain-stress coupling parameters, non-uniform thermal expansion coefficients, and contact stiffness evolution parameters, and generate a nonlinear material characteristic matrix;

[0015] Step S14: Construct a multi-physical field topology structure across space-time scales based on the nonlinear material characteristic matrix to generate a multi-physical field topology initial model;

[0016] Step S15: Nonlinear tensor low-rank approximation dimension reduction is performed on the multi-physical field topology initial model, and compression is performed to retain the physical interaction relationship, and a multi-physical field topology snapshot is generated.

[0017] The application performs time-sensitive synchronization processing on multi-physical field data of engineering simulation heterogeneous hardware, ensures the consistency of different physical field data on the time axis, eliminates the timing deviation in the data acquisition process, and improves the accuracy of simulation data. The non-Euclidean topology mapping can adapt to the complex multi-physical field interaction relationship, so that the data structure is more in line with the actual physical characteristics, and the data input quality is improved by combining data standardization and timing consistency correction, and the error accumulation caused by the mismatch of data scales is reduced. The tensor principal component analysis is used to extract the nonlinear characteristics of the key materials, so that important parameters such as strain-stress coupling, non-uniform thermal expansion and contact stiffness can be accurately modeled, and high-precision material behavior description is provided for subsequent simulation calculation. Based on the nonlinear material characteristic matrix, a multi-physical field topology structure across space-time scales is constructed, so that the simulation model can take into account both local physical characteristics and global evolution relationship, thereby enhancing the expression ability of the model. Through nonlinear tensor low-rank approximation dimension reduction, the redundant calculation amount is effectively reduced, and the physical interaction relationship is compressed and retained, so that the multi-physical field topology snapshot generated finally can reduce the calculation complexity while still maintaining high fidelity to the key characteristics of the multi-physical field, providing an efficient and accurate data basis for subsequent engineering simulation calculation.

[0018] Optionally, step S12 is specifically:

[0019] Step S121: Aligning the discrete time point data of the initial multi-physical field data set, interpolating the missing data within a time step Δt=10 -3 s, and setting the maximum offset correction threshold to ±5×10 -3 s, adjusting the time offset of adjacent measuring points, and generating a time-aligned multi-physical field data set;

[0020] Step S122: Non-Euclidean graph topology mapping is performed on the time-aligned multi-physical field data set, and a local physical interaction graph is constructed, generating a non-Euclidean multi-physical field initial graph;

[0021] Step S123: Hierarchical spectral clustering analysis is performed on the non-Euclidean multi-physical field initial graph, the first N=10 eigenvectors are selected for Laplace eigenvalue decomposition, and feature truncation is performed with a threshold value ε=10 -4 , generating a spectral smoothing multi-physical field graph;

[0022] Step S124: Normalizing the spectral smoothing multi-physical field graph to perform data standardization, normalizing the data to the interval [-3σ, 3σ], and performing time series trend correction based on a sliding window length L=50, generating a standardized multi-physical field data;

[0023] Step S125: Bayesian optimal interpolation prediction is performed according to the standardized multi-physical field data, an abnormal threshold δ=3σ is set to remove abnormal points, and interpolation completion is performed under a confidence interval α=95%, to generate the standardized multi-physical field data.

[0024] The application ensures that different physical field data remain synchronized within a time step range by aligning discrete time point data of an initial multi-physical field data set, effectively reduces measurement errors by setting a maximum offset correction threshold, and improves the consistency and reliability of time series data. Non-Euclidean graph topology mapping is adopted to make the data structure more suitable for the complex interaction of multi-physical fields, and the influence of local features on the overall topology structure is enhanced by constructing a local physical interaction graph, thereby improving the simulation accuracy. Combined with hierarchical spectral clustering analysis, it is helpful to extract global features of multi-physical field data, and effectively reduce the amount of redundant calculation and improve the calculation efficiency by Laplace eigenvalue decomposition. At the same time, the feature selection of the data is optimized by threshold truncation, ensuring the integrity of the key features. Normalization data standardization can eliminate the scale difference between physical quantities, making the data distribution more uniform, and based on the sliding window length, the time series trend is corrected, effectively eliminating the non-stationary factors and improving the consistency of the data. Finally, through Bayesian optimal interpolation prediction, the problem of missing abnormal data is effectively solved, the abnormal threshold is set to remove abnormal points, the data quality is improved, and the interpolation completion is performed under the confidence interval to ensure the continuity and accuracy of the data, thereby providing high-quality and reliable standardized multi-physical field data for subsequent simulation calculation.

[0025] Optionally, step S2 is specifically:

[0026] Step S21: performing finite element mesh partitioning on the multi-physical field topology snapshot, adjusting the unit size, and generating an initial finite element mesh data set;

[0027] Step S22: identifying a nonlinear contact area based on the initial finite element mesh data set, constructing a contact stiffness distribution matrix, and generating a nonlinear contact area feature matrix;

[0028] Step S23: performing contact constraint optimization on the nonlinear contact area feature matrix, calculating the contact mechanical response, and obtaining a nonlinear contact pair stiffness matrix; dynamically adjusting the stiffness distribution based on the nonlinear contact pair stiffness matrix, and generating a contact optimized finite element mesh;

[0029] Step S24: training a multi-agent reinforcement learning strategy based on the contact optimized finite element mesh, and optimizing the strategy convergence, to generate a nonlinear contact strategy model;

[0030] Step S25: performing strategy tensor encoding on the nonlinear contact strategy model, to generate a nonlinear contact strategy tensor.

[0031] The application improves the grid quality and calculation precision by finite element mesh partitioning and adjusting the unit size, so that the subsequent simulation calculation can be carried out on the basis of better discretization. The nonlinear contact area identification can accurately extract complex contact relationship and construct the contact stiffness distribution matrix, so that the contact characteristics can be accurately quantified, and the reliability of simulation calculation is improved. For the calculation of the contact mechanics response of the nonlinear contact area, dynamic optimization adjustment is carried out in combination with the contact stiffness matrix to ensure the rationality of the stiffness distribution, enhance the physical authenticity of the model, and at the same time, the optimized finite element grid can better adapt to the stress-strain transmission of complex structures, improve the simulation precision. The multi-agent reinforcement learning strategy is used for training, so that the nonlinear contact optimization has stronger adaptability and generalization ability, and through the optimization strategy convergence, it is ensured that the model can still converge quickly and stably in a complex contact environment, thereby improving the simulation calculation efficiency. Finally, the nonlinear contact optimization strategy is quantitatively expressed in the form of a tensor through tensor coding, so that the contact optimization result can be directly applied to subsequent calculation in the form of a tensor, improving the data transmission efficiency and compatibility, and providing an efficient and intelligent contact optimization method for complex engineering simulation.

[0032] Optionally, the coded structure dynamics intelligent contract in step S3 is specifically:

[0033] According to the nonlinear contact strategy tensor, tensor low-rank decomposition is performed, a tensor rank threshold r is set in [5, 15], and a structure dynamics characteristic tensor is obtained;

[0034] Based on the structure dynamics characteristic tensor, a Lagrangian constraint optimization model is constructed, an optimization iteration step size η is set to 0.01, and a convergence error threshold ε is set to ≤10 -5 , and a generalized coordinate transformation matrix is calculated to obtain a set of structure mechanics optimization parameters;

[0035] According to the set of structure dynamics optimization parameters, a dynamics state transition equation is established, and a state-control coupled model is constructed in combination with a preset multi-body system constraint equation, a constraint weight matrix W c is set in [0.1, 1.0], and an intelligent contract basic framework is generated;

[0036] The intelligent contract basic framework is subjected to verifiable calculation conversion, a state consistency verification threshold δ s is set to ≤0.05, a state consistency verification sub is constructed, and a dynamics trusted intelligent contract is generated;

[0037] The dynamics trusted intelligent contract is subjected to consensus mechanism testing, a consensus node number N c is set to 10, a consensus verification round number R is set to ∈[50, 100], and intelligent contract execution time optimization is performed based on the consensus mechanism test result, to obtain a structure dynamics intelligent contract.

[0038] The application extracts key contact feature information by low-rank decomposition of a nonlinear contact strategy tensor, sets a tensor rank threshold, effectively reduces data redundancy, ensures computational efficiency and data compactness, and retains important structural dynamics information. Based on the extracted structural dynamics feature tensor, a Lagrangian constraint optimization model is constructed, and the optimization iteration step and convergence error threshold are set to improve the convergence speed and stability of the optimization calculation. The calculation of the generalized coordinate transformation matrix ensures the consistency of the dynamics parameters in different reference systems, improves the calculation accuracy and the rationality of the system modeling. On this basis, the dynamics state transition equation is established, and combined with the pre-set multibody system constraint equation, the state-control coupled model is constructed to ensure that the dynamics system can achieve optimal performance under controlled conditions, and through the setting of the constraint weight matrix, the system constraint conditions are more flexible to adapt to different working conditions. The construction of the intelligent contract basic framework makes the dynamics optimization process verifiable, and through the state consistency verification sub, the reliability and data security of the intelligent contract execution are improved. The consensus mechanism test further enhances the stability and anti-interference ability of the dynamics trusted intelligent contract, ensures the consistency of the execution of the intelligent contract, and through the optimization of the consensus verification rounds and the number of nodes, the calculation efficiency is improved and the calculation resource consumption is reduced, finally realizing efficient, trusted and stable structural dynamics intelligent contract, providing reliable guarantee for the automatic operation and maintenance of engineering simulation software.

[0039] Optionally, the simulation software dynamics trusted verification in step S3 is specifically:

[0040] The structural dynamics intelligent contract is deployed in the intelligent contract environment, the calculation precision threshold ∈ is set to 10 -5 and the maximum number of iterations I max is set to 500, and the initial environment for simulation software dynamics verification is generated;

[0041] Based on the initial environment for simulation software dynamics verification, the dynamics state variables of contract execution are extracted to construct a state observation matrix;

[0042] The numerical stability of the state observation matrix is analyzed to generate a state stability evaluation report;

[0043] According to the state stability evaluation report, the time step of the dynamics state variable is adaptively adjusted, the initial time step Δt0 is set to 0.001s, the maximum step Δt max is set to 0.1s, and the step dynamic update is performed in combination with the pre-set error control factor γ∈[0.8,1.2] to generate time step optimization dynamics data

[0044] The credibility of the time step optimization dynamics data is calculated, and the lower limit of the credibility P mim= 95%, calculate the consistency error of the smart contract execution, and set the consistency error control criterion E cons ≤ 10 -4 Optimize the reliability of the consistency error calculation result, and generate a dynamic reliability verification report;

[0045] Verify the consensus mechanism of the dynamic reliability verification report, and set the number of consensus nodes N c = 10, the number of consensus verification rounds R = 100, and generate a dynamic reliability chain.

[0046] The present application is deployed in a smart contract environment, sets the calculation precision threshold and the maximum number of iterations, ensures the stability and numerical precision of the simulation calculation, improves the calculation efficiency, and reduces the waste of calculation resources. Based on the initial environment of the simulation software dynamics verification, the dynamics state variables of the contract execution are extracted, and the state observation matrix is constructed, which provides accurate data basis for the numerical analysis of the subsequent dynamic system. Through numerical stability analysis, the stability of the state variable is evaluated, and potential numerical divergence problems are identified, which improves the reliability and data controllability of the dynamics simulation. According to the stability evaluation report, the time step adaptive adjustment strategy is adopted, combined with the error control factor, to realize the dynamic update of the step size, improve the calculation efficiency while ensuring the calculation precision, and avoid the decrease of precision caused by too large step size or too long calculation time caused by too small step size. The reliability calculation ensures the reliability of the smart contract execution process, calculates the consistency error, and optimizes the reliability according to the set error control criterion, which improves the reliability and stability of the smart contract calculation result. Finally, the dynamic reliability verification report is verified by the consensus mechanism, which ensures the consistency of the dynamic data in the distributed computing environment, enhances the reliability and traceability of the simulation calculation, and generates a dynamic reliability chain, which provides a solid guarantee for the efficient, stable and reliable operation of the engineering simulation software.

[0047] Optionally, step S4 is specifically:

[0048] Step S41: Based on the dynamic reliability chain, extract the structural dynamics state variables, and combine the contact stress distribution data in the nonlinear contact strategy tensor to construct an initial dynamics-contact feature model;

[0049] Step S42: Perform energy conservation analysis on the initial dynamics-contact feature model, and calculate the stress gradient and deformation response of the contact area, thereby establishing a contact stiffness dynamic adjustment function;

[0050] Step S43: Perform contact dynamics parameter optimization according to the contact stiffness dynamic adjustment function to obtain an optimized contact dynamics parameter set;

[0051] Step S44: Obtain historical fluid simulation data, and extract fluid-structure coupling relationship from the historical fluid simulation data to obtain a fluid coupling parameter set;

[0052] Step S45: constructing a fluid mechanics vortex model according to the optimized contact kinetics parameter set and the fluid coupling parameter set;

[0053] Step S46: performing simulation credibility verification on the fluid mechanics vortex model, and optimizing the kinetics-fluid interaction characteristic parameters based on the credibility verification result to finally generate a vortex flow field atlas.

[0054] The application extracts structural dynamics state variables based on a dynamics credible chain, and combines contact stress distribution data in a nonlinear contact strategy tensor to construct an initial dynamics-contact characteristic model, thereby enhancing the modeling accuracy of the dynamics behavior of the contact area and improving the adaptability of the dynamics simulation to actual working conditions. Energy conservation analysis is performed on the initial dynamics-contact characteristic model to calculate the stress gradient and deformation response of the contact area, ensuring that the contact process conforms to the physical law, and a contact stiffness dynamic adjustment function is established to adaptively adjust the contact stiffness with the change of external load and deformation state, thereby improving the accuracy of simulation calculation. According to the contact stiffness dynamic adjustment function, the contact dynamics parameters are optimized to make the contact mechanics calculation more consistent with the actual physical properties, thereby improving the stability and calculation efficiency of the simulation calculation. By obtaining historical fluid simulation data and extracting fluid-structure coupling relationships, a fluid coupling parameter set is established to enhance the simulation system's modeling capability of the structural response under the action of fluid and improve the accuracy of multi-physical field coupling modeling. A fluid mechanics vortex model is constructed using the optimized contact dynamics parameter set and the fluid coupling parameter set to enhance the physical consistency and calculation accuracy of the vortex modeling, making the fluid dynamics simulation more close to the real engineering application. Finally, the simulation credibility of the fluid mechanics vortex model is verified, and the dynamics-fluid interaction characteristic parameters are optimized based on the verification result to ensure the credibility and accuracy of the vortex flow field simulation result, thereby improving the simulation capability of the engineering simulation software in complex flow environments and providing reliable numerical support for multi-physical field coupling analysis.

[0055] Optionally, the phase field damage evolution in step S5 is specifically:

[0056] Based on the convergence optimization field, an initial damage field is extracted, and time evolution analysis of the damage field is performed to generate a preliminary damage evolution model;

[0057] The preliminary damage evolution model is discretized by finite elements to obtain a local damage evolution characteristic map;

[0058] Based on the local damage evolution characteristic map, time evolution analysis is performed to obtain a damage propagation path;

[0059] According to the damage propagation path, the damage field parameter of the preliminary damage evolution model is optimized, the damage evolution process is updated, and an updated damage evolution model is obtained;

[0060] According to the multi-physics field topology snapshot and the updated damage evolution model, multi-physics field interaction integration is carried out, and a multi-physics field evolution twin is obtained.

[0061] The application can accurately capture the initial damage state inside the material or structure by extracting the initial damage field based on the convergent optimization field and performing time evolution analysis of the damage field, and can reveal the dynamic trend of damage evolution by combining time series analysis, improve the timeliness and accuracy of damage modeling. The preliminary damage evolution model is discretized by finite elements, so that the damage evolution process can be numerically solved at a high resolution spatial scale to obtain a local damage evolution feature map, thereby refining the distribution characteristics of the damage area and improving the precision and local response simulation capability of the simulation. Based on the local damage evolution feature map, time evolution analysis can be performed to effectively predict the damage propagation path and provide reliable damage propagation information for subsequent optimization, thereby improving the accuracy of damage prediction. According to the damage propagation path, the damage field parameters of the preliminary damage evolution model are optimized, the key parameters can be adjusted according to different damage mechanisms, the physical rationality and simulation credibility of the damage modeling are improved, and the model can better adapt to complex working conditions by updating the damage evolution process, thereby improving the application range of engineering simulation. Finally, by combining the multi-physics field topology snapshot and the updated damage evolution model, multi-physics field interaction integration is realized, and a multi-physics field evolution twin is obtained, so that the simulation can not only reflect the damage evolution inside the structure, but also couple the mutual influence between multiple physical fields, providing high-precision, multi-scale and multi-field coupled simulation support for engineering simulation, and improving the reliability and precision of engineering decision-making.

[0062] Optionally, the software operation and maintenance system is used to execute the software operation and maintenance method as described above, and the software operation and maintenance system comprises:

[0063] The coupling tensor compression module is configured to acquire the engineering simulation heterogeneous hardware data, perform multi-physics field coupling tensor compression on the engineering simulation heterogeneous hardware data, and obtain the multi-physics field topology snapshot.

[0064] The nonlinear constraint module is configured to perform finite element grid nonlinear constraint based on the multi-physics field topology snapshot, and obtain the nonlinear contact strategy tensor.

[0065] The dynamics credible verification module is configured to encode the structure dynamics smart contract according to the nonlinear contact strategy tensor, perform simulation software dynamics credible verification on the structure dynamics smart contract, and obtain the dynamics credible chain.

[0066] The vortex flow field construction module is configured to construct a fluid mechanics vortex model based on the dynamics credible chain, perform boundary layer separation simulation flow optimization on the fluid mechanics vortex model, and obtain the vortex flow field atlas.

[0067] A multi-physics interaction evolution module is configured to implement explicit dynamic explicit integration on the vortex flow field atlas to obtain a convergent optimization field; a phase field damage evolution is performed according to the convergent optimization field to obtain a multi-physics evolution twin, and the multi-physics evolution twin is deployed through a preset engineering simulation cloud architecture.

[0068] The software operation system of the application can realize any software operation method of the application, and is used as a medium for joint operation and signal transmission between various modules to complete the software operation method.

[0069] Optionally, the present specification also provides a computer readable storage medium, wherein a computer program is stored, and the computer program is executed to realize the software operation method as described above. BRIEF DESCRIPTION OF DRAWINGS

[0070] Other features, objects and advantages of the application will become more apparent from the following detailed description of non-restrictive embodiments made with reference to the accompanying drawings:

[0071] Fig. 1 The figure is a schematic diagram of the step flow of the software operation method of the application;

[0072] Fig. 2 The figure is a schematic diagram of the detailed step flow of step S1 in the application;

[0073] Fig. 3 The figure is a schematic diagram of the detailed step flow of step S2 in the application;

[0074] The implementation, functional features and advantages of the application will be further described with reference to the accompanying drawings. DETAILED DESCRIPTION

[0075] The technical method of the application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0076] In addition, the accompanying drawings are only schematic diagrams of the application, and are not necessarily drawn to scale. The same reference signs in the drawings represent the same or similar parts, and thus repeated descriptions thereof will be omitted. Some block diagrams shown in the drawings are functional entities, and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0077] It should be understood that, although the terms "first", "second" and the like can be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of the example embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0078] To achieve the above object, please refer to Figs. 1 to 3 The application provides a software operation and maintenance method, which comprises the following steps:

[0079] Step S1: acquiring engineering simulation heterogeneous hardware data, and performing multi-physical field coupling tensor compression on the engineering simulation heterogeneous hardware data to obtain a multi-physical field topology snapshot;

[0080] In this embodiment, an engineering simulation software containing a GPU cluster, embedded sensors and FPGA accelerators is used to realize real-time collection and preprocessing of engineering simulation data. Through IEEE1588 clock synchronization technology, a 1kHz sampling rate is achieved, and multi-physical field data such as temperature, stress, vibration and electromagnetic field are collected, with a data resolution of 16 bits. Then, the original data is processed by tensor compression using a high-order singular value decomposition (HOSVD) algorithm, and the tensor rank threshold is set to 10, so as to effectively reduce data redundancy while retaining the coupling characteristics between physical fields, and finally form a multi-physical field topology snapshot with high precision and low redundancy, providing a solid data foundation for subsequent simulation analysis.

[0081] Step S2: performing finite element grid nonlinear constraint based on the multi-physical field topology snapshot to obtain a nonlinear contact strategy tensor;

[0082] In this embodiment, ABAQUS or ANSYS and other finite element software are used for adaptive meshing and nonlinear constraint modeling. In specific implementation, the initial mesh element size is set to 2mm, the nonlinear contact constraint algorithm based on the variational principle is used to identify the key contact areas and calculate the stress field distribution of each area, a contact stiffness distribution matrix is constructed, and local grid density fine adjustment is adopted to realize accurate description of the contact behavior, so as to finally obtain a nonlinear contact strategy tensor which can truly reflect the complex contact characteristics and ensure good numerical stability and physical authenticity of the model under high load conditions.

[0083] Step S3: encoding a structure dynamics smart contract according to the nonlinear contact strategy tensor, and performing simulation software dynamics credible verification on the structure dynamics smart contract to obtain a dynamics credible chain;

[0084] In this embodiment, the nonlinear contact strategy tensor is used as the core input, the Solidity language is used to develop the smart contract, and the Ethereum private chain platform is deployed. In the specific process, the Lagrangian constraint optimization method is used, and the iteration step is set to 0.01 seconds and the convergence error threshold is 1x10 -5 At the same time, the generalized coordinate transformation matrix is calculated to ensure the consistency of the parameters in different reference systems; through the simulation environment, the real-time verification and safety test of the smart contract execution are carried out, and the dynamic credible chain with traceability, automatic feedback and fault monitoring capability is constructed, which provides high reliability data support for complex engineering simulation.

[0085] Step S4: based on the dynamic credible chain, a fluid mechanics vortex model is constructed, and the fluid mechanics vortex model is subjected to boundary layer separation simulation flow optimization to obtain a vortex flow field atlas;

[0086] In this embodiment, the fluid mechanics vortex model is constructed in ANSYS Fluent, and the large eddy simulation (LES) method and boundary layer separation technology are used to solve the flow field. In the specific implementation, the flow field sampling step is set to 0.005 seconds, and the experience correction coefficient 0.85 is introduced to correct the boundary layer separation point position; through the iterative solution of the pressure and velocity distribution in the flow field and the boundary condition correction, a high-precision vortex flow field atlas is obtained, which fully reflects the coupling effect between the fluid and the structure, and provides key fluid dynamics parameters for subsequent dynamic integration and damage evolution.

[0087] Step S5: explicit dynamics explicit integration is performed on the vortex flow field atlas to obtain a convergence optimization field; according to the convergence optimization field, phase field damage evolution is carried out to obtain a multi-physical field evolution twin, and the multi-physical field evolution twin is deployed through a pre-set engineering simulation cloud architecture.

[0088] In this embodiment, the explicit solver in LS-DYNA is used to perform explicit dynamics integration on the vortex flow field atlas, and the integration step is accurately set to 0.001 seconds. The system monitors the dynamic response of each time step in real time during the iterative solution process, and the residual error between the two consecutive iterations is reduced to 1x10 -4The following is judged as convergence, thereby generating a stable convergent optimization field. The convergent optimization field not only fully reflects the coupling effect between the fluid and the structure, but also provides accurate initial conditions for subsequent phase field damage evolution. In the phase field damage evolution process, an improved phase field model is used in combination with a finite element discrete method to model the damage propagation and crack evolution inside the structure. In specific operations, adaptive mesh refinement technology is used in key damage areas to control the mesh element size to within 0.5 millimeters, and the simulation time step is set to 0.01 seconds to ensure that subtle crack propagation changes can be captured. After multiple iterations of simulation, an updated damage evolution model is generated, which can accurately predict the crack propagation trend of the structure under stress. Finally, the updated damage evolution model is integrated with the multi-physical field topology snapshot data to construct a multi-physical field evolution twin, which is deployed in real time online through a pre-set engineering simulation cloud architecture (such as AWS EC2 container cluster deployment), with data refreshed every second to ensure the real-time nature of online monitoring and data feedback, thereby providing efficient and accurate technical support for intelligent operation and decision support of engineering simulation software.

[0089] Optionally, step S1 is specifically:

[0090] Step S11: Obtain engineering simulation heterogeneous hardware multi-physical field data, and perform time-sensitive synchronization processing on the multi-physical field data to generate an initial multi-physical field data set;

[0091] In this embodiment, through deployment in the engineering simulation software, a heterogeneous hardware platform composed of a high-performance GPU cluster, embedded sensors, and FPGA accelerators, real-time collection of multi-physical field data such as temperature, stress, vibration, and electromagnetic field during the engineering simulation process, time calibration of the data using a distributed time synchronization protocol (IEEE 1588) with a synchronization accuracy of 1 microsecond to ensure the consistency of the multi-physical field data. Subsequently, the data is filtered for noise, and a fifth-order Savitzky-Golay filter is used for smoothing processing of the measurement signal to eliminate high-frequency noise and preserve key physical characteristics, ultimately generating an initial multi-physical field data set for simulation calculation.

[0092] Step S12: Perform non-Euclidean topology mapping on the initial multi-physical field data set, and perform data standardization and timing consistency correction to generate standardized multi-physical field data;

[0093] In this embodiment, a non-Euclidean topological mapping algorithm based on graph theory is used to embed data points into a high-dimensional topological space, thereby depicting the complex spatial relationship between various physical fields; at the same time, the Z-score normalization method is used to standardize the data, convert all data values to the standard normal distribution range (mean value is 0, standard deviation is 1), and set the error threshold ± 0.001 to eliminate outliers. At the same time, in order to ensure the accuracy of the time series data, the sliding window autoregressive method (window length 50 steps) is used to correct the time series consistency of the data, so that the time series signal remains smooth in the local trend of change, and finally generates standardized multi-physical field data.

[0094] Step S13: Perform tensor principal component analysis based on the standardized multi-physical field data to extract material nonlinear characteristics including strain-stress coupling parameters, non-uniform thermal expansion coefficients, and contact stiffness evolution parameters, and generate a nonlinear material characteristic matrix.

[0095] In this embodiment, the tensor principal component analysis (TPCA) technique is used to extract material nonlinear characteristics, and the principal component relationship between physical variables is analyzed by constructing a third-order tensor (space-time-physical variable). The retention criterion is set to be greater than 95% of the cumulative contribution rate of the principal component, so as to extract key characteristics, including strain-stress coupling parameters (range 0.2-0.9), non-uniform thermal expansion coefficients (1.2×10 -5 –5.8×10 -5 K -1 ) and contact stiffness evolution parameters (500-5000 MPa). At the same time, the QR decomposition method is used to reduce the dimension of the characteristic matrix to reduce computational redundancy, and finally obtain the nonlinear material characteristic matrix.

[0096] Step S14: Construct a multi-physical field topological structure across space-time scales according to the nonlinear material characteristic matrix, and generate a multi-physical field topological initial model.

[0097] In this embodiment, a multi-layer graph neural network technology is further used to construct a multi-physical field topological structure across space-time scales. This method uses the coupling relationship between physical fields to construct a multi-layer connected graph, sets the edge weight threshold in the graph to 0.05 to filter out weak interactions, and ensures that key coupling relationships are retained. Then, the graph attention network (GAT) is used to calculate the interaction weight between each physical field, and the dynamic time warping (DTW) method is used to analyze the dynamic characteristics across space-time scales, so that the topological structure can adapt to the multi-physical field interaction relationship under different scales, and finally generate an initial multi-physical field topological model, laying a solid foundation for subsequent simulation analysis and optimization.

[0098] Step S15: Perform nonlinear tensor low-rank approximation dimension reduction on the multi-physical field topological initial model, and compress the retained physical interaction relationship to generate a multi-physical field topological snapshot.

[0099] In this embodiment, the initial multi-physical field topology model is processed by dimension reduction using a nonlinear tensor low-rank approximation method such as Tucker decomposition, and the decomposition rank threshold is set to 10 to compress redundant data while retaining the core interaction features between the physical fields. The alternating least squares (ALS) method is used to optimize the tensor decomposition process to improve convergence speed and computational stability. The compressed model still accurately represents the complex coupling relationship, thereby generating an efficient, low-redundancy, and high-fidelity multi-physical field topology snapshot, providing accurate data support for efficient operation and subsequent multi-physical field coupling simulation of engineering simulation software.

[0100] Optionally, step S12 is specifically:

[0101] Step S121: Aligning discrete time point data of the initial multi-physical field data set, interpolating missing data within a time step Δt = 10 -3 s, and setting the maximum offset correction threshold to ±5x10 -3 s; adjusting the time offset of adjacent measurement points to generate a time series alignment multi-physical field data set.

[0102] In this embodiment, the initial multi-physical field data set is aligned at discrete time points. Engineering simulation data includes measurement data of multiple physical fields (such as temperature field, stress field, and displacement field) at different time points, and due to the precision limitations of hardware or sensors, some data points may be missing. In this case, a linear interpolation method is used, the time step is set to 10 -3 seconds (s = 10 -3 s), and the missing data is interpolated and supplemented according to the time step. For each pair of adjacent measurement points, the maximum offset correction threshold is set to ±5x10 -3 s seconds to ensure that the supplemented data within the time step does not cause excessive time offset. In this way, the generated time series alignment data set ensures the consistency and synchronization of the multi-physical field data on the same time axis.

[0103] Step S122: Non-Euclidean graph topology mapping of the time series alignment multi-physical field data set, and constructing a local physical interaction graph to generate a non-Euclidean multi-physical field initial graph.

[0104] In this embodiment, the multi-physical field data set after time alignment is mapped to a non-Euclidean graph topology. A nearest neighbor-based graph embedding method is used for topology mapping. Specifically, a K-nearest neighbor (KNN) algorithm is used to construct a graph topology, and K=8 is set, that is, each data point is connected to its nearest 8 neighbor nodes. Through adjacency matrix normalization processing, the weight threshold W th =0.05 is set to ensure that the interaction relationship above the threshold is effectively preserved, and the edges below the threshold are removed, and finally a non-Euclidean multi-physical field initial graph (MP-GraphInit) is generated. This mapping method effectively captures the nonlinear relationship between complex physical fields and provides a structured graphical representation for subsequent analysis.

[0105] Step S123: Hierarchical spectral clustering analysis is performed on the non-Euclidean multi-physical field initial graph, the first N=10 eigenvectors are selected for Laplace eigenvalue decomposition, and a threshold value ε=10 -4 The feature truncation generates a spectral smoothing multi-physical field graph.

[0106] In this embodiment, hierarchical spectral clustering analysis is performed on the non-Euclidean multi-physical field initial graph. In this step, the spectral clustering algorithm is used to cluster the multi-physical field data set, the number of clusters is set to 10 (N=10), and the eigenvectors of each cluster are extracted. Then, the Laplace eigenvalue decomposition method is applied for eigenvalue decomposition, the first N most important eigenvectors are selected, and the threshold value is set to 10 -4 only the components with eigenvalues greater than the threshold value are retained. This process effectively reduces the data dimension, extracts the main physical patterns and structural information in the multi-physical field data, and lays the foundation for subsequent analysis, generating a spectral smoothing multi-physical field graph.

[0107] Step S124: The spectral smoothing multi-physical field graph is normalized and standardized, the data is normalized to the [-3σ, 3σ] interval, and the time series trend is corrected based on a sliding window length L=50 to generate standardized multi-physical field data.

[0108] In this embodiment, the spectral smoothing multi-physical field graph is standardized. To ensure that different physical field data have the same scale and dimension, the minimum-maximum normalization method is applied to normalize the data to the [-3σ, 3σ] interval (σ refers to the standard deviation of the data set), to prevent different dimensional physical fields from affecting the model imbalance. Then, based on a sliding window length of 50, the time series trend is corrected, and the window step is set to 5 to ensure that the trend in each time window is smoothly adjusted. In this process, the data is denoised to ensure the stability and consistency of the multi-physical field data.

[0109] Step S125: Bayesian optimal interpolation prediction is performed according to the standardized multi-physical field data, an abnormal threshold δ=3σ is set to remove abnormal points, and interpolation completion is performed under a confidence interval α=95% to generate standardized multi-physical field data.

[0110] In this embodiment, by analyzing the missing points in the data, a Bayesian regression model is used to predict the missing data. The abnormal threshold is set to 3 times the standard deviation, and the abnormal values beyond this range are removed to ensure the reliability of the data in the interpolation process. In order to optimize the interpolation effect, interpolation completion is performed within a confidence interval α=95%, and the Monte Carlo method is used for multiple sampling to generate multiple interpolation results, thereby ensuring the accuracy and stability of the interpolation. Finally, the generated standardized multi-physical field data set has high precision and strong fault tolerance for missing values and abnormal points, providing a reliable data basis for further simulation analysis.

[0111] Optionally, step S2 is specifically:

[0112] Step S21: Perform finite element mesh partitioning on the multi-physical field topology snapshot, and adjust the element size to generate an initial finite element mesh data set;

[0113] In this embodiment, finite element mesh partitioning is performed based on the multi-physical field topology snapshot to decompose complex physical fields into multiple small, computable units. The process of mesh partitioning uses standard tetrahedral elements, and the selection of element size is crucial as it directly affects the calculation accuracy and efficiency. In this embodiment, the maximum size of the element is set to 5mm to ensure that details can be captured adequately without causing excessive computational burden. After partitioning, an initial finite element mesh data set is obtained. To improve accuracy, the boundary layer region is locally refined after partitioning to ensure that the physical behavior of the key region can be accurately described. Finally, the mesh partitioning result provides the preliminary data structure required for engineering simulation.

[0114] Step S22: Perform nonlinear contact area identification based on the initial finite element mesh data set, and construct a contact stiffness distribution matrix to generate a nonlinear contact area feature matrix;

[0115] In this embodiment, the contact area is determined by analyzing the points of interaction between physical fields and the geometric contact conditions between elements. The materials or surfaces in the contact area exhibit nonlinear behavior, so it is crucial to model them accurately. Then, a contact stiffness distribution matrix is constructed, and the element values of the matrix represent the stiffness relationship between the contact points. Here, the initial estimated value of the contact stiffness is 10 4 N / m 2, the contact stiffness is preliminarily estimated by analyzing the physical fields and material properties. The generated nonlinear contact area feature matrix provides basic data for subsequent optimization steps.

[0116] Step S23: Implement contact constraint optimization on the nonlinear contact area feature matrix, calculate the contact mechanics response, and obtain the nonlinear contact pair stiffness matrix; based on the nonlinear contact pair stiffness matrix, dynamically adjust the stiffness distribution to generate a contact-optimized finite element mesh;

[0117] In this embodiment, the nonlinear contact area feature matrix is subjected to contact constraint optimization. In this step, the optimization process is divided into two stages: the first stage is the calculation of the contact mechanics response, and the second stage is the dynamic adjustment of the stiffness matrix. In the first stage, the initial finite element mesh model is solved, and the displacement and stress response between the contact points are calculated using the finite element analysis method. In this process, different physical field interactions in the contact area are considered, such as friction, slip, and adhesion, etc. nonlinear phenomena. Specifically, the incremental method is used for iterative solution of displacement and stress. In each iteration, based on the initial contact stiffness matrix, the displacement and stress values of each contact point are calculated, and the contact stiffness matrix is updated according to the interaction force between adjacent contact points. In order to accurately capture the nonlinear behavior of the contact area, a nonlinear constitutive model based on contact mechanics theory is used. This model takes into account the micro-topography of the contact interface, adhesion, and the plasticity characteristics of the material. In the second stage, based on the calculated contact mechanics response, the contact stiffness matrix is dynamically adjusted to optimize the mechanical properties of the contact area. For this purpose, an iterative optimization algorithm is introduced. Each optimization is based on the results of the previous calculation to modify the contact stiffness matrix, and a target function that minimizes the contact stress is used for optimization. Specifically, the "adaptive gradient descent method" (Adaptive Gradient Descent, AGD) is used, and in each iteration, the optimization algorithm dynamically adjusts the stiffness distribution according to the stress distribution in the contact area. Through dynamic adjustment of the contact stiffness, the contact stress concentration phenomenon can be reduced while maintaining the stability of the contact. The initial value of the contact stiffness matrix can be set to 104N / m 2 , and the maximum allowable stress concentration is set to 50MPa to prevent plastic deformation of the contact points. After each optimization, the contact stiffness matrix is updated and provides new stiffness values for the next calculation. After multiple optimization iterations, the contact stiffness matrix converges to the optimal solution, generating a contact-optimized stiffness distribution, thereby ensuring that the mechanical response in the contact area meets the design requirements.

[0118] Step S24: Train the multi-agent reinforcement learning strategy based on the contact-optimized finite element mesh, optimize the strategy convergence, and generate a nonlinear contact strategy model;

[0119] In this embodiment, on the basis of the contact-optimized finite element mesh, the training of the multi-agent reinforcement learning strategy is carried out. In this step, the reinforcement learning method is used to dynamically adjust the contact optimization. Each agent represents a contact point and learns how to minimize the contact stress and maximize the contact stiffness under nonlinear contact constraints. In the training process, the deep Q-learning algorithm is used, and the goal is to train each agent to adjust its own strategy to optimize the mechanical response of the entire contact area. The training step of each agent is set to 0.01, and the total number of training rounds is set to 5000 to ensure that the final strategy has good convergence. After 5000 rounds of training, a converged nonlinear contact strategy model is obtained, which can effectively handle complex contact problems and optimize the contact mechanical response.

[0120] Step S25: Encode the nonlinear contact strategy model into a strategy tensor to generate a nonlinear contact strategy tensor.

[0121] In this embodiment, the trained nonlinear contact strategy model is encoded into a strategy tensor. The process of tensor encoding converts the output of the strategy model into a high-dimensional tensor, which can more efficiently store and calculate the complex relationships of contact constraints. In this step, the tensor decomposition technique is used, and the tensor dimension is set to 4x4x4 to adapt to the multi-dimensional characteristics of multi-physical field contact behavior. Each dimension of the tensor encoding represents the interaction between the physical, mechanical, material and environmental relationships of the contact points. Specifically, the dimensions of the tensor correspond to the contact point position, contact force, contact stiffness and contact state, and each dimension contains 100 discrete points. After strategy tensor encoding, the nonlinear contact strategy is compressed into an efficient tensor format, which greatly improves the computational efficiency of the subsequent simulation software and lays the foundation for the rapid deployment of the model in different scenarios.

[0122] Optionally, the encoding structure dynamics smart contract in step S3 is specifically:

[0123] According to the nonlinear contact strategy tensor, perform tensor low-rank decomposition, set the tensor rank threshold r∈[5,15], and obtain a structure dynamics feature tensor;

[0124] In this embodiment, tensor low-rank decomposition is performed based on the nonlinear contact strategy tensor. First, the nonlinear contact strategy tensor is obtained, which contains the interaction of different physical fields in the contact area. Then, tensor low-rank decomposition is performed to reduce the computational complexity while preserving the key information in the tensor. The main goal of low-rank decomposition is to extract the main mode of the tensor. By setting a tensor rank threshold (for example, setting the rank threshold to 10), the tensor is decomposed into multiple sub-tensors with small ranks. Using the singular value decomposition (SVD) or high-order singular value decomposition (HOSVD) method, the nonlinear contact strategy tensor is reduced in dimension and most of the information is preserved. The final structural dynamics feature tensor is a simplified representation that can be more efficiently analyzed in subsequent steps.

[0125] Based on the structural dynamics feature tensor, a Lagrangian constraint optimization model is constructed, and the optimization iteration step size η = 0.01 and the convergence error threshold ε ≤ 10 -5 are set, and the generalized coordinate transformation matrix is calculated to obtain the structural mechanics optimization parameter set;

[0126] In this embodiment, a Lagrangian constraint optimization model is constructed based on the structural dynamics feature tensor. Based on the obtained structural dynamics feature tensor, a Lagrangian constraint optimization model is constructed, which handles the constraint conditions by introducing Lagrange multipliers. The constraints include mechanical constraints, geometric constraints, etc., to ensure that the physical and geometric limitations of the system are met during optimization, for example, the constraint conditions can include contact mechanics limitations between objects, displacement limitations, and balance requirements of the system. In the optimization process, the optimization iteration step size (such as setting the step size to 0.01) and the convergence error threshold (such as setting the convergence threshold to 10 -6 ) are set to ensure the accuracy and computational efficiency of the optimization results. In each iteration, the model parameters are updated and the generalized coordinate transformation matrix is calculated (using the least squares method), which is used to describe the transformation relationship of the dynamics system. Finally, the obtained structural mechanics optimization parameter set will be used in subsequent steps.

[0127] According to the structural dynamics optimization parameter set, a dynamics state transition equation is established, and a state-control coupled model is constructed by combining the pre-set multi-body system constraint equation, and the constraint weight matrix W c ∈ [0.1, 1.0] is set, and the smart contract basic framework is generated;

[0128] In this embodiment, the dynamic state transition equation is established according to the structure dynamics optimization parameter set. According to the obtained optimization parameter set, the dynamic state transition equation is established to describe the dynamic behavior of the system changing with time. On this basis, combined with the preset multi-body system constraint equation (such as momentum conservation equation, mechanical equilibrium equation, etc.), the state-control coupled model is constructed. In order to ensure the accuracy of the model, the constraint weight matrix is set (for example, the weight matrix W is set as a diagonal matrix, and the weight range is set as [0.1, 1.0]), and the relative importance of each constraint condition in the optimization process is adjusted. Through the joint solution of the state transition equation and the control equation, the basic framework of the intelligent contract is finally generated.

[0129] The intelligent contract basic framework is subjected to verifiable computation conversion, and the state consistency verification threshold δ is set s ≤0.05, a state consistency verification sub is constructed, and a dynamic credible intelligent contract is generated;

[0130] In this embodiment, the intelligent contract basic framework is subjected to verifiable computation conversion. According to the obtained intelligent contract basic framework, the contract is processed through verifiable computation conversion (VCT) to ensure that it meets the safety and performance requirements of the system. The state consistency verification threshold is set (for example, the verification threshold is set to 0.01), and under this threshold, the state consistency of the system is checked to ensure that the state information in the contract execution always meets the expectations. In this process, the state consistency verifier is constructed to verify each state of the system to ensure that the state change of the intelligent contract during execution is always within an acceptable range. If the state of the system deviates from the predetermined standard, it can be corrected by adjusting the parameters or model correction scheme. After verification, a credible dynamic intelligent contract is generated.

[0131] The dynamic credible intelligent contract is subjected to consensus mechanism test, and the number of consensus nodes N c =10 and the consensus verification round number R∈[50, 100] are set, and the intelligent contract execution time is optimized based on the consensus mechanism test result to obtain a structure dynamics intelligent contract.

[0132] In this embodiment, the consensus mechanism of the dynamic trusted smart contract is tested. The obtained dynamic trusted smart contract is tested for consensus mechanism to ensure that the smart contract can normally operate in a multi-node environment. The number of consensus nodes (for example, the number of consensus nodes is set to 20) and the number of consensus verification rounds (the number of verification rounds is set to 60) are set to simulate the consensus process in a multi-node environment. In the test, a hierarchical consensus mechanism (such as Raft or Tendermint) and a Practical Byzantine Fault Tolerance (PBFT) algorithm are used to ensure that the contract can work normally in a low-delay and high-throughput environment. Through the test, the execution efficiency and accuracy of the contract meet the expectations, and the execution time of the smart contract is optimized in the process. Finally, the maximum delay of the execution time is set to 100 ms, and the execution time of the smart contract is optimized through the consensus mechanism test results to minimize the delay of contract execution in a complex multi-physical field simulation system, thereby obtaining the final structural dynamics smart contract. Specifically, a distributed computing framework (such as Apache Spark or Apache Flink) is used to distribute the computing tasks of the smart contract, divide the processing logic of the contract into multiple sub-tasks, and assign them to different computing nodes for parallel execution. This method effectively reduces the delay caused by insufficient computing resources of a single node during contract execution and improves processing speed. In combination with a distributed storage system (such as HDFS) for data storage and access optimization, the data transmission delay between nodes is reduced, and the consistency and reliability of the data are ensured. Through the load balancing mechanism of distributed computing, the computing tasks of each node are automatically adjusted to ensure load balancing of each node, thereby avoiding the increase of delay caused by the overload of some nodes. Through this optimization method, the execution time of the smart contract is effectively shortened, and finally the maximum delay of the execution time is ensured to be not more than 100 ms in the test, thereby obtaining the final structural dynamics smart contract.

[0133] Optionally, the simulation software dynamic trusted verification in step S3 is specifically:

[0134] The structural dynamics smart contract is deployed in a smart contract environment, and the calculation precision threshold ∈ is set to 10 -5 and the maximum number of iterations I max = 500 to generate an initial environment for simulation software dynamics verification;

[0135] In this embodiment, the smart contract environment is set in the engineering simulation software to ensure that it can efficiently process multi-physical field simulation tasks. The calculation precision threshold is set to ∈ = 10 -5, the maximum number of iterations is 500 to ensure that the contract execution process can be completed within a reasonable calculation accuracy, while preventing excessive iterations in the calculation process. To this end, the initial environment includes a high-efficiency solver (e.g., a solver based on the Gauss-Seidel method) and high-performance computing resources to support dynamic calculation and simulation in the contract. The simulation software will also automatically adjust the solving process based on the maximum number of iterations set during deployment to avoid errors caused by non-convergent calculations. The environment generates a simulation software dynamics verification initial environment to provide a basis for contract verification.

[0136] Based on the simulation software dynamics verification initial environment, the dynamics state variables of contract execution are extracted to construct a state observation matrix.

[0137] In this embodiment, based on the simulation software dynamics verification initial environment, key dynamics state variables during contract execution are extracted, including but not limited to displacement, acceleration, stress, strain, and temperature, which reflect the changes of the structure under loading conditions. The minimum monitoring period of each variable is set to 0.01s to ensure that the state update in the simulation reflects the real dynamic behavior in real time. When extracting these data using the simulation system, all physical quantities are organized into a state observation matrix (e.g., a 6x6 matrix containing six main physical quantities such as displacement, velocity, and acceleration) for subsequent analysis and processing. The matrix is updated in real time and captures the interaction between different physical fields through multiple calculations during the simulation process.

[0138] Numerical stability analysis is performed on the state observation matrix to generate a state stability evaluation report.

[0139] In this embodiment, numerical stability analysis is performed on the state observation matrix. The eigenvalue analysis method is used to calculate the eigenvalues of the state matrix, and the Jacobi method is used to obtain the maximum eigenvalue. It is assumed that the modulus of the maximum eigenvalue |λ_max| must be less than 1 to maintain numerical stability. To ensure the stability of matrix calculation, we set the stability threshold as |λ_max| < 0.99. If the maximum eigenvalue of the state matrix exceeds this threshold, adjustments are made by adjusting the initial conditions, modifying the step size, or introducing damping, and the state matrix is recalculated. Based on this analysis result, a state stability evaluation report is generated, which includes the stability evaluation of each physical quantity, stability optimization schemes, and provides corresponding adjustment suggestions.

[0140] According to the state stability evaluation report, the time step of the dynamics state variable is adaptively adjusted, with an initial time step Δt0 = 0.001s and a maximum step size Δt max = 0.1s, and the step size is dynamically updated in combination with the pre-set error control factor γ ∈ [0.8, 1.2] to generate time step optimization dynamics data

[0141] In this embodiment, the time step of the dynamic state variable is adaptively adjusted according to the state stability evaluation report. The initial time step is set to 0.001s, the maximum step is set to 0.1s, and the time step is automatically adjusted based on the error control factor (set to 0.1). When the error exceeds 1e-5 during calculation, the time step is automatically reduced to ensure calculation accuracy; if the error is below the threshold, the system will increase the time step to improve calculation efficiency. In this way, the goal of the step is to optimize the balance between accuracy and efficiency in the calculation process by dynamically adjusting the time step. The dynamic data set after time step optimization is generated according to these optimizations, and this data is used as the basis for subsequent simulation and analysis.

[0142] The credibility of the time step optimized dynamic data is calculated, and the credibility lower limit P min is set to 95%, the consistency error of the smart contract execution is calculated, and the consistency error control criterion E cons ≤10 -4 The consistency error calculation result is optimized for credibility, and a dynamic credibility verification report is generated;

[0143] In this embodiment, the credibility of the time step optimized dynamic data is calculated. First, set the credibility lower limit to 0.95, and calculate the consistency error during the execution of the smart contract. For this purpose, a model based on Gaussian Process Regression (GPR) is used to correct potential abnormal measurement points. The abnormal threshold is set to 10 -4 . Specifically, the consistency error control criterion is set as follows: when the consistency error is greater than 10 -4 , a weighted correction method is used to reduce the impact of the error, and the corrected error should be restored to within 10 -4 . At this time, a weighted linear interpolation method is used to correct the abnormal points and their surrounding data points in the calculation process of the smart contract, and the weight coefficient is determined by the relative difference between the abnormal points and the surrounding data points. In this way, the corrected data points can maintain consistency while minimizing the impact of errors on the overall calculation process. In this process, interpolation coefficients and weights are dynamically adjusted based on historical data and current measurement results to ensure the consistency of each corrected data point with the contract state. After these steps, the dynamic data optimized by the consistency error control criterion is generated, thereby improving the credibility of the contract execution, and the final dynamic credibility verification report is generated, which details the data stability after error correction, optimization effect, and accuracy of the smart contract execution.

[0144] The dynamic credibility verification report is verified by the consensus mechanism, and the number of consensus nodes N c= 10, consensus verification round number R = 100, and generate a dynamic and reliable chain.

[0145] In this embodiment, the dynamic and reliable verification report is verified by the consensus mechanism. The number of consensus nodes is set to 10, and the number of consensus verification rounds is set to 100 rounds, simulating a multi-node environment to verify the consensus process of the smart contract among nodes. Each node synchronously checks the contract execution state during runtime, and based on the latest state information, a decision is made together to ensure the synchronous execution of the entire contract in different node environments. In the consensus process, the verification node will calculate based on the state consistency error and the preset consensus rules, and finally obtain the consistency result of the contract execution. Based on the results of these consensus verifications, the execution time of the smart contract is further optimized, and the maximum delay is set to not more than 100 ms. The stability of the contract execution is tested through the consensus mechanism, and the final dynamic and reliable chain is generated to ensure that the contract can be efficiently and accurately executed during engineering simulation.

[0146] Optionally, step S4 is specifically:

[0147] Step S41: Based on the dynamic and reliable chain, the structural dynamic state variables are extracted, and the contact stress distribution data in the nonlinear contact strategy tensor are combined to construct an initial dynamic-contact feature model;

[0148] In this embodiment, the key dynamic state variables of the structure, such as displacement, velocity, and acceleration, are extracted through the dynamic and reliable chain. Then, the stress distribution information of the contact area is obtained from the nonlinear contact strategy tensor. In order to construct the initial dynamic-contact feature model, first, the dynamic state variables and the contact stress distribution data are synchronized at the same time point, and then the two sets of data are jointly mapped through the weighted summation method based on the adjacency matrix to obtain a joint feature matrix containing structural dynamics and contact stress information. This matrix is used for subsequent contact dynamics analysis. For the accuracy requirement of time alignment, the time step is set to 10 -4 s to ensure data synchronization.

[0149] Step S42: Energy conservation analysis is performed on the initial dynamic-contact feature model, and the stress gradient and deformation response of the contact area are calculated to establish a contact stiffness dynamic adjustment function;

[0150] In this embodiment, the energy flow in the dynamic-contact feature model is analyzed based on the law of conservation of energy. In the model, the energy transfer at each contact point is determined by the contact stress and the amount of structural deformation. By analyzing the stress gradient and local deformation response in the contact area, the variation law of contact stiffness is calculated. The finite element method (such as setting the grid size to 0.5 mm) is used to accurately solve the stress and deformation field in the contact area, and further derive the dynamic adjustment function of the contact stiffness, ensuring the balance of energy between the contact points. This function can automatically adjust the contact stiffness according to real-time simulation data, optimizing the structural response.

[0151] Step S43: Perform contact dynamics parameter optimization according to the contact stiffness dynamic adjustment function to obtain an optimized set of contact dynamics parameters;

[0152] In this embodiment, based on the established contact stiffness dynamic adjustment function, the contact dynamics parameters in the initial dynamic-contact feature model are optimized and adjusted. Through optimization algorithms (such as particle swarm optimization PSO), within a given contact stiffness range, the friction coefficient, viscoelastic model parameters, and other dynamics parameters of the contact surface are adjusted. To improve optimization accuracy, the optimization step size is set to 0.01, and the maximum number of iterations is set to 500. Through multiple iterations, the optimal set of contact dynamics parameters is found, making the stress and deformation of the contact area more consistent with physical reality, and parallel processing is used in calculations to improve efficiency.

[0153] Step S44: Obtain historical fluid simulation data and extract fluid-structure coupling relationships from the historical fluid simulation data to obtain a set of fluid coupling parameters;

[0154] In this embodiment, fluid mechanics data related to the structure is extracted from the simulation database in the simulation software to obtain key parameters of the interaction between the structure and the fluid. The velocity field, pressure field, vorticity distribution, and other data in the CFD simulation results are used, and the coupling relationship between the fluid and the structure is extracted in combination with the structural deformation data. Through data fitting and regression analysis (such as polynomial regression), the fluid-structure coupling parameters are determined, such as the influence coefficient of fluid flow rate on structural stress, to obtain a set of fluid coupling parameters. The data processing accuracy is set to 5x10-3 to ensure the accuracy of the coupling relationship.

[0155] Step S45: Construct a fluid mechanics vortex model based on the optimized set of contact dynamics parameters and the set of fluid coupling parameters;

[0156] In this embodiment, the optimized contact dynamics parameter set and fluid coupling parameter set are used to construct a fluid mechanics vortex model in combination with the Navier-Stokes equation. This model can simulate the flow of fluid on the surface of a complex structure and vortex phenomena. Numerical calculation is performed using fluid dynamics simulation software (such as OpenFOAM), and the time step is set to Δt = 10 -6 s during the calculation process, and the simulation of fluid vortex behavior is ensured to accurately reflect the interaction between the structure and the fluid by optimizing the grid division accuracy (for example, the grid size is 1 mm).

[0157] Step S46: Perform simulation credibility verification on the fluid mechanics vortex model, and optimize the dynamics-fluid interaction characteristic parameters based on the credibility verification results to finally generate a vortex flow field map.

[0158] In this embodiment, the reliability of the fluid mechanics vortex model under different physical conditions is evaluated by performing large-scale simulation. The reliability of the simulation results is evaluated using a reliability analysis based on the Monte Carlo method to ensure that the vortex model can accurately reflect the interaction between the structure and the fluid in actual applications. On this basis, the dynamics-fluid interaction characteristic parameters (such as the fluid-structure coupling coefficient and the vortex generation position) are adjusted, the fluid mechanics vortex model is optimized through a back propagation algorithm, and the final fluid mechanics vortex model is generated. Subsequently, the fluid mechanics parameters (such as the vortex intensity and the vortex position) output by the model are presented in the form of a map using a map technology, thereby obtaining a vortex flow field map. This map can be used in actual engineering to predict the complex interaction between fluid and structure and provide a basis for structure optimization.

[0159] Alternatively, the phase field damage evolution in step S5 is specifically:

[0160] Based on the convergent optimization field, the initial damage field is extracted, and the time evolution analysis of the damage field is performed to generate a preliminary damage evolution model;

[0161] In this embodiment, the potential damage area in the structure is preliminarily identified by using the convergent optimization field data. The convergent optimization field (for example, an optimization result that is converged to an error of less than 1 × 10 -4 5after multiple iterations) obtained through previous multi-physical field coupling simulation is combined with the historical simulation data of temperature, stress, and vibration to extract the initial damage field in the LS-DYNA environment using the phase field damage model. In the specific operation, the area where the local stress of the material exceeds 150 MPa is defined as the initial damage area, and the time step is set to 0.01 seconds. The damage evolution in each time step is iteratively simulated. After dynamic time evolution analysis, a preliminary damage evolution model reflecting the generation and expansion of local damage in the structure is formed. This model can be used to predict the initial damage trend of the material under actual working conditions.

[0162] The preliminary damage evolution model is discretized by finite elements to obtain a local damage evolution characteristic map;

[0163] In this embodiment, the preliminary damage evolution model is discretized by finite elements to achieve more detailed damage prediction. Eight-node cubic elements are used for mesh partitioning, and local encryption is performed in the damage area. The discretization of the damage model considers the nonlinear characteristics of different material properties and damage evolution stages. By performing more accurate finite element discretization on the local area, a local damage evolution characteristic map is obtained, which shows the damage changes at different positions and different time points.

[0164] Time evolution analysis is performed based on the local damage evolution characteristic map to obtain a damage propagation path;

[0165] In this embodiment, time evolution analysis is performed based on the local damage evolution characteristic map, and the damage propagation path is simulated by step loading and load unloading cycles. According to the damage propagation path, a directional and regional model of damage propagation is established, focusing on the crack propagation speed and expansion direction, and the maximum damage propagation rate is set to 0.01 mm / s. Using this path information, key damage areas can be identified, and the model can be optimized through a feedback mechanism to determine the future evolution trend of damage.

[0166] According to the damage propagation path, the damage field parameters of the preliminary damage evolution model are optimized, and the damage evolution process is updated to obtain an updated damage evolution model;

[0167] In this embodiment, according to the damage propagation path, the damage field parameters of the preliminary damage evolution model are optimized, and the optimization algorithm is set to particle swarm optimization (PSO), with a maximum particle number of 50 and an iteration number of 100. This process optimizes the material constitutive relationship, damage threshold, and fatigue parameters of the damage model, thereby improving the accuracy of the model and updating the damage evolution process to generate an updated damage evolution model.

[0168] According to the multi-physics field topology snapshot and the updated damage evolution model, multi-physics field interaction integration is performed to obtain a multi-physics field evolution twin.

[0169] In this embodiment, the multi-physics field interaction integration is performed based on the multi-physics field topology snapshot and the updated damage evolution model. This process integrates temperature, pressure, stress, and other multi-physics field data using a multi-field coupling method to combine different field influencing factors. During the calculation process, a three-dimensional scene and a multi-scale method are used to ensure the coupling accuracy between different scales and different physical fields, and finally a multi-physics field evolution twin is obtained, which provides a powerful tool for dynamic monitoring, prediction, and optimization in practical applications.

[0170] Optionally, the present specification provides a software operation and maintenance system for implementing the software operation and maintenance method described above, and the software operation and maintenance system comprises:

[0171] The coupling tensor compression module is configured to acquire the engineering simulation heterogeneous hardware data, and perform multi-physics field coupling tensor compression on the engineering simulation heterogeneous hardware data to obtain a multi-physics field topology snapshot.

[0172] The nonlinear constraint module is configured to perform finite element grid nonlinear constraint based on the multi-physics field topology snapshot to obtain a nonlinear contact strategy tensor.

[0173] The dynamics credible verification module is configured to encode a structure dynamics smart contract according to the nonlinear contact strategy tensor, and perform simulation software dynamics credible verification on the structure dynamics smart contract to obtain a dynamics credible chain.

[0174] The vortex flow field construction module is configured to construct a fluid mechanics vortex model based on the dynamics credible chain, and perform boundary layer separation simulation flow optimization on the fluid mechanics vortex model to obtain a vortex flow field atlas.

[0175] The multi-physics field interactive evolution module is configured to perform explicit dynamics explicit integration on the vortex flow field atlas to obtain a converged optimization field, perform phase field damage evolution on the converged optimization field to obtain a multi-physics field evolution twin, and deploy the multi-physics field evolution twin through a preset engineering simulation cloud architecture.

[0176] Optionally, the present specification also provides a computer readable storage medium, wherein a computer program is stored, and the computer program is executed to implement the software operation and maintenance method described above.

[0177] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, the scope of the present application being defined by the appended claims and not by the above description, and it is intended to encompass all variations falling within the meaning and the scope of the equivalent elements of the application file.

[0178] The above description is merely one specific implementation of the present application, which enables those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A software operation and maintenance method, characterized in that, Includes the following steps: Step S1: Acquire heterogeneous hardware data for engineering simulation, and perform multiphysics coupling tensor compression on the heterogeneous hardware data to obtain a multiphysics topology snapshot; Step S1 specifically includes: Step S11: Obtain multiphysics data of heterogeneous hardware for engineering simulation, and perform time-sensitive synchronization processing on the multiphysics data to generate an initial multiphysics dataset; Step S12: Perform non-Euclidean topological mapping on the initial multiphysics dataset, and perform data standardization and temporal consistency correction to generate standardized multiphysics data. Step S13: Perform tensor principal component analysis based on standardized multiphysics data to extract material nonlinear features including strain-stress coupling parameters, non-uniform thermal expansion coefficients, and contact stiffness evolution parameters, and generate a nonlinear material feature matrix; Step S14: Construct a multi-physics topology structure across spatiotemporal scales based on the nonlinear material characteristic matrix, and generate an initial multi-physics topology model; Step S15: Perform nonlinear tensor low-rank approximation dimensionality reduction on the initial multiphysics topology model, compress and preserve physical interaction relationships, and generate a multiphysics topology snapshot; Step S2: Perform nonlinear constraints on the finite element mesh based on the multiphysics topology snapshot to obtain the nonlinear contact strategy tensor; Step S3: Encode the structural dynamics smart contract using the nonlinear contact strategy tensor, and perform simulation software dynamics trust verification on the structural dynamics smart contract to obtain the dynamics trust chain; Step S4: Construct a fluid dynamics vortex model based on the dynamics trust chain, and perform boundary layer separation simulation flow optimization on the fluid dynamics vortex model to obtain the vortex flow field map; Step S5: Perform explicit dynamics integration on the vortex flow field map to obtain the convergent optimization field; perform phase field damage evolution based on the convergent optimization field to obtain a multiphysics evolution twin, and deploy the multiphysics evolution twin through the preset engineering simulation cloud architecture.

2. The software operation and maintenance method according to claim 1, characterized in that, Step S12 is as follows: Step S121: Perform discrete-time data alignment on the initial multiphysics dataset, at a time step size of [missing information]. Impute missing data within s, and set the maximum offset correction threshold to s. Adjust the time offset of adjacent measurement points to generate a time-aligned multiphysics dataset; Step S122: Perform non-Euclidean graph topological mapping on the time-aligned multiphysics dataset and construct a local physical interaction graph to generate a non-Euclidean initial multiphysics graph. Step S123: Perform hierarchical spectral clustering analysis on the initial non-Euclidean geometry physics diagram, select the first N=10 eigenvectors for Laplace eigenvalue decomposition, and apply a threshold. Feature truncation is performed to generate a spectrally smoothed multiphysics map; Step S124: Normalize the spectral smoothing multiphysics plot and apply the normalization data based on the sliding window length. Perform time series trend correction to generate standardized multiphysics data; Step S125: Perform Bayesian optimal interpolation prediction based on standardized multiphysics data and set anomaly threshold. Outlier removal is performed within the confidence interval. Interpolation is then performed to complete the data and generate standardized multiphysics data.

3. The software operation and maintenance method according to claim 1, characterized in that, Step S2 is as follows: Step S21: Perform finite element mesh generation on the multiphysics topology snapshot and adjust the element size to generate an initial finite element mesh dataset; Step S22: Identify nonlinear contact regions based on the initial finite element mesh dataset, construct the contact stiffness distribution matrix, and generate the nonlinear contact region feature matrix; Step S23: Perform contact constraint optimization on the characteristic matrix of the nonlinear contact region, calculate the contact mechanical response, and obtain the stiffness matrix of the nonlinear contact pair; Based on the dynamic adjustment of stiffness distribution by nonlinear contact stiffness matrix, a contact-optimized finite element mesh is generated. Step S24: Train a multi-agent reinforcement learning strategy based on the contact optimization finite element mesh, optimize the policy convergence, and generate a nonlinear contact policy model; Step S25: Encode the nonlinear contact strategy model into a policy tensor to generate a nonlinear contact strategy tensor.

4. The software operation and maintenance method according to claim 1, characterized in that, The coded structure dynamics smart contract mentioned in step S3 specifically refers to: Tensor low-rank decomposition is performed based on the nonlinear contact strategy tensor, and a tensor rank threshold is set. Thus, the structural dynamics characteristic tensor is obtained; A Lagrangian-constrained optimization model is constructed based on the structural dynamics characteristic tensor, and the optimization iteration step size is set. With convergence error threshold And calculate the generalized coordinate transformation matrix to obtain the set of structural mechanics optimization parameters; Based on the structural dynamics optimization parameter set, a dynamic state transition equation is established, and a state-control coupled model is constructed by combining it with the pre-set multibody system constraint equations, and the constraint weight matrix is ​​set. Generate the basic framework for smart contracts; Perform verifiable computational transformation on the basic framework of smart contracts and set a threshold for state consistency verification. Construct a state consistency verifier and generate a dynamically trusted smart contract; Test the consensus mechanism of the dynamic trusted smart contract and set the number of consensus nodes. Consensus verification rounds Based on the consensus mechanism test results, the execution time of the smart contract was optimized to obtain the structural dynamics smart contract.

5. The software operation and maintenance method according to claim 1, characterized in that, The specific steps of the simulation software dynamics credibility verification in step S3 are as follows: Deploy the smart contract environment for the structural dynamics smart contract and set a computational accuracy threshold. and maximum number of iterations =500, generate the initial environment for dynamic verification of the simulation software; Based on the initial environment of the simulation software dynamics verification, the dynamic state variables of contract execution are extracted to construct the state observation matrix; Perform numerical stability analysis on the state observation matrix and generate a state stability assessment report; Based on the state stability assessment report, the time step of the dynamic state variables is adaptively adjusted, and an initial time step is set. Maximum step size And combined with a preset error control factor Perform dynamic step size updates to generate time step optimization dynamic data. Calculate the reliability of the time-step optimization dynamics data and set a lower reliability limit. Calculate the consistency error of smart contract execution and set consistency error control criteria. The consistency error calculation results are optimized for reliability, and a dynamic reliability verification report is generated. The consensus mechanism is used to verify the dynamics trustworthiness verification report, and the number of consensus nodes is set. Consensus verification rounds Generate a dynamic trust chain.

6. The software operation and maintenance method according to claim 1, characterized in that, Step S4 is as follows: Step S41: Extract the structural dynamic state variables based on the dynamic trust chain, and combine them with the contact stress distribution data in the nonlinear contact strategy tensor to construct an initial dynamic-contact feature model; Step S42: Perform energy conservation analysis on the initial dynamic-contact characteristic model, and calculate the stress gradient and deformation response in the contact area to establish a dynamic adjustment function for contact stiffness; Step S43: Optimize the contact dynamic parameters according to the contact stiffness dynamic adjustment function to obtain the optimized contact dynamic parameter set; Step S44: Obtain historical fluid simulation data and extract the fluid-structure coupling relationship from the historical fluid simulation data to obtain the fluid coupling parameter set; Step S45: Construct a hydrodynamic vortex model based on the optimized contact dynamics parameter set and the fluid coupling parameter set; Step S46: Perform a simulation to verify the credibility of the fluid dynamics vortex model, optimize the dynamic-fluid interaction characteristic parameters based on the credibility verification results, and finally generate the vortex flow field map.

7. The software operation and maintenance method according to claim 1, characterized in that, The phase field damage evolution described in step S5 is specifically as follows: The initial damage field is extracted based on the convergent optimization field, and the time evolution analysis of the damage field is performed to generate a preliminary damage evolution model. The preliminary damage evolution model was discretized using the finite element method to obtain a local damage evolution characteristic map; Based on the local damage evolution feature map, time evolution analysis is performed to obtain the damage propagation path; The damage field parameters of the preliminary damage evolution model are optimized based on the damage propagation path, and the damage evolution process is updated to obtain the updated damage evolution model. Multiphysics interaction integration is performed based on multiphysics topological snapshots and updated damage evolution models to obtain multiphysics evolution twins.

8. A software operation and maintenance system, characterized in that, The software operation and maintenance system is used to execute the software operation and maintenance method as described in claim 1, and includes: The coupling tensor compression module is used to acquire heterogeneous hardware data for engineering simulation and perform multiphysics coupling tensor compression on the heterogeneous hardware data for engineering simulation to obtain a multiphysics topology snapshot. The nonlinear constraint module is used to perform nonlinear constraints on finite element meshes based on multiphysics topology snapshots to obtain nonlinear contact strategy tensors. The dynamics trust verification module is used to encode the structural dynamics smart contract according to the nonlinear contact strategy tensor, and to perform simulation software dynamics trust verification on the structural dynamics smart contract to obtain the dynamics trust chain; The vortex flow field construction module is used to construct a hydrodynamic vortex model based on a dynamic credibility chain, and to perform boundary layer separation simulation flow optimization on the hydrodynamic vortex model to obtain the vortex flow field map; The multiphysics interactive evolution module is used to perform explicit dynamics integration on the vortex flow field map to obtain a convergent optimization field; based on the convergent optimization field, phase field damage evolution is performed to obtain a multiphysics evolution twin, and the multiphysics evolution twin is deployed through a preset engineering simulation cloud architecture.

9. A computer-readable storage medium, characterized in that, It contains a computer program that, when executed, implements the software operation and maintenance method as described in any one of claims 1-7.

Images