107 results about "Finite difference method" patented technology

The invention relates to the technical field of underground engineering safety monitoring and disaster simulation, in particular to a digital twinborn deduction platform for underground engineering disaster chain evolution simulation, which comprises a multi-disaster coupling numerical simulation module used for acquiring multi-source data including design drawing data, geological data and engineering monitoring data, and based on the multi-source data, establishing an initial stress field model by using a numerical calculation method combining a finite element and a finite difference, and carrying out leakage-settlement-structure failure multi-disaster coupling numerical simulation. According to the method, the partial differential equation for describing the evolution law of the underground engineering is used as a constraint term to be embedded into the deduction model, and online identification and dynamic extrapolation of parameters are executed by fusing real-time monitoring data, so that the accuracy of disaster evolution simulation under complex working conditions is remarkably improved; and the problem that the traditional numerical calculation method is difficult to meet the real-time requirement of digital twinning is solved.

The invention belongs to the technical field of nuclear reactor physical numerical calculation, and particularly relates to a method for improving precision and convergence of solving a diffusion equation based on a PINN, and the method comprises the following steps: S1, building a neural network algorithm model combining a physical information neural network PINN and source iteration; s2, aiming at a multi-region smooth transition neutron diffusion problem with small cross section difference between materials, optimizing the neural network algorithm model by adopting an optimization strategy of a training level; and S3, aiming at the multi-region neutron diffusion problem of large neutron flux gradient at the boundary due to large cross section difference between materials, improving the neural network algorithm model by adopting a domain-divided neural network architecture. According to the method, the problem that the calculation cost is exponentially increased in a high-dimensional or complex geometric scene due to the fact that a traditional finite difference method and a finite element method depend on grid division is avoided.

The invention discloses an internal wave electromagnetic response numerical simulation method based on a layered ocean model. The method comprises the following steps: acquiring geomagnetic field parameters; calculating distribution characteristics of seawater conductivity along with depth, and dispersing the distribution characteristics into a layered ocean conductivity model; an internal wave mechanical model is constructed, internal wave velocity field distribution is established through characteristic parameters of internal waves, and dynamic input is provided for electromagnetic field response calculation; substituting the internal wave velocity field and the geomagnetic field into a Maxwell equation set, and constructing an internal wave induction electromagnetic field control equation; and applying tangential electric field continuity and normal current continuity conditions on a layered interface, and performing discrete and iterative solution on an equation by adopting a finite difference method to obtain distribution characteristics of the electric field and the magnetic field at different depths and frequencies. According to the method, the problem of insufficient precision caused by a uniform conductivity hypothesis is solved, and application limitation caused by an ideal geomagnetic field hypothesis is avoided, so that accurate simulation of internal wave induction electromagnetic response under a complex ocean condition can be realized.

This invention provides a virtual-real fusion-based force feedback simulation control method for ship rudder control sticks, comprising: constructing a set of fluid control equations for the flow field on the rudder surface; discretizing the set of fluid control equations using a standard structured mesh to obtain a discretized set of fluid control equations; solving the discretized set of fluid control equations based on the finite difference method and boundary conditions to obtain the drag torque value of the fluid acting on the rudder plate; converting the drag torque value into a damping torque value; the control unit outputting a motor drive command based on the magnitude of the damping torque value; and the motor driver, based on the motor drive command output by the control unit, driving the force feedback simulation motor to generate a corresponding simulated damping torque value, which acts on the control stick to produce force feedback. This invention provides real-time control stick force feedback for the helmsman, allowing for direct perception of the ship's navigation status and enhancing steering stability.

The invention provides a physical field data fusion method, which belongs to the technical field of multi-source data fusion, and comprises the following steps: multiplying low-fidelity physical field data by a sensor mask matrix to obtain low-fidelity sensor data; constructing a multi-order difference format based on a finite difference method and a Taylor expansion theory, and processing the low-fidelity physical field data to obtain low-fidelity multi-order partial differential physical field data; training the pre-trained data fusion model through the low-fidelity multi-order partial differential physical field data and the low-fidelity sensor data to obtain a data fusion model; the data fusion model is finely adjusted and optimized by updating sensor data and updating multi-order partial differential physical field data, and an optimized data fusion model is obtained and used for fusion processing of multi-source heterogeneous multi-physical field data; the optimized data fusion model obtained by the method can perform feature extraction and fusion on the multi-source heterogeneous multi-physics field data to obtain high-fidelity data with consistent representation and high confidence.

The invention discloses a forward and inverse kinematics calculation method for a 2-RSS-1U parallel mechanism of a humanoid robot, and relates to the technical field of humanoid robots, and the inverse kinematics calculation method comprises the following steps: establishing an ankle shaft seat point coordinate system, and obtaining a pitch angle and a roll angle of a sole; calculating coordinates of the bottom coordinate points of the ankle connecting rods relative to the initial coordinate system based on a spatial triangular relation; converting to a driving motor coordinate system by combining the known position relation of the coordinate system; and calculating the angle of the driving motor according to the connecting rod length and vertex coordinates. The forward kinematics resolving method comprises the following steps: determining a fixed translation relationship between a mechanism constant parameter and a coordinate system; constructing a nonlinear equation set by taking the roll angle and the pitch angle as unknown quantities; and solving by combining a two-dimensional residual vector with a Newton iteration method, estimating a Jacobian matrix by using a finite difference method in iteration, judging convergence according to a residual threshold, and outputting an attitude angle. According to the forward and inverse kinematics calculation method, the calculation stability and adaptability can be improved.

The application belongs to the field of gear transmission system squeeze film damper damping, especially relates to a kind of gear squeeze film damper damping quick calculation method and system, obtains the structural parameter of squeeze film damper, carries out velocity calculation and constructs the Reynolds equation of squeeze film damper oil film pressure;Based on half-step center finite difference method, the Reynolds equation is solved, and the oil film distribution pressure of oil film damper is obtained;Based on oil film distribution pressure, the oil film counterforce of oil film damper is calculated using double integral algorithm.Compared with the classical short bearing approximate solution and long bearing approximate solution, its use range is wide, and compared with super relaxation iteration method, its calculation efficiency is high.Make it can be used in gear dynamics model to carry out damping analysis.When solving fluid Reynolds equation by finite difference method, only the working condition with regular boundary condition can be processed.Based on variational principle and Galerkin method, a new more general finite element method is proposed to calculate the oil film counterforce of SFD.

This invention discloses a simulation method for determining different snow storage material cover options. The method includes: a simulation method for snow storage experiments, including mathematical methods for calculating the total amount of solar radiation on any slope and mathematical methods for unsteady-state heat transfer; and a numerical method for simulating the performance of snow storage cover options, including finite difference methods and economic analysis methods. This method solves the existing problem of selecting snow storage cover material options for different regions and climate conditions.

The invention discloses a triangular mesh-based high-order electromagnetic field simulation method, and belongs to the field of computational electromagnetism. In order to solve the problems of significant numerical dispersion and phase error accumulation when an existing triangular mesh time domain finite difference method is used for processing a complex geometric structure, the method comprises the following steps: firstly, introducing a first additional loop on a basic triangular mesh, and combining with a first optimization coefficient to construct a high-order model; and a second additional loop is further introduced, and an optimization model is constructed in combination with a second optimization coefficient, so that the phase error is effectively inhibited under the coarse mesh, and high-precision electromagnetic simulation of a large-scale polygon structure is realized.

The invention provides a comprehensive pipe gallery safety state evaluation method and system, and relates to the technical field of comprehensive pipe gallery safety state evaluation methods, and the method comprises the steps: employing a steady-state heat release method to calculate the simulation temperature of each three-dimensional grid node; a finite difference method is adopted to calculate a temperature field distribution coefficient of each three-dimensional grid node, and a temperature distribution coefficient of a three-dimensional grid center point is obtained through an arithmetic average method; a coupling damage method is adopted to construct a coupling coefficient matrix of any grid position, equivalent damage energy of any grid position is calculated through an inner product method, and the equivalent damage energy is mapped into a safety coefficient of any grid center point through normalization. By adopting three-dimensional grid discretization, complex geometric characteristics of the pipe gallery structure can be more accurately captured, the position of a potential fire source is clearly marked, the hidden danger recognition efficiency is improved through targeted monitoring, high-precision calculation of a simulated temperature field under the complex boundary condition is achieved, equivalent damage energy is mapped into a dimensionless safety coefficient, and cross-regional risk comparability is achieved.

The calculation method of the uplift displacement of a lower tunnel caused by a layered foundation pit excavation comprises the following steps: (1) according to the relative position relationship between the foundation pit and the lower tunnel, relevant parameters are determined, and a calculation analysis model is established; (2) according to the load acting on the inside of the soil, a stress-strain solution below the load acting surface is derived, the tunnel and the foundation are respectively equivalent to an Euler-Bernoulli long beam and a Lifshin foundation model, the change of the additional stress of the soil caused by the unloading of the foundation pit excavation is calculated by using the Mindlin solution of the elastic layered foundation, and thus a corresponding mechanical model is obtained, and a longitudinal deformation control differential equation of the tunnel under the unloading of the foundation pit excavation is derived; and (3) finally, the finite difference method is used to calculate the uplift deformation of the lower existing tunnel caused by the foundation pit excavation. The calculation method can more reasonably simulate the mechanical model of the layered foundation tunnel, and makes the calculation of the uplift amount of the lower tunnel caused by the foundation pit excavation more accurate and reliable.

The application discloses a numerical simulation method for tsunamis and lake surges based on Boussinesq type equations, and comprises the following steps: obtaining digital elevation and initial disturbance data of a target area; constructing a high-order Boussinesq type control equation containing adjustable dispersion parameters and a frequency dispersion term to adjust the dispersion approximation precision; under a unified explicit time advancing system, discretizing the conservation term and the frequency dispersion term by using a finite volume method and a finite difference method respectively, and loading data to a calculation grid; judging the dry and wet states based on a preset water depth threshold, and truncating the frequency dispersion term for solving in the dry grid area to obtain simulation results of a flow field and a water level field. By constructing the control equation containing the adjustable dispersion parameters and the mixed numerical discrete framework, combining the explicit time advancing and the dry and wet grid truncation processing, the method does not need iteration and is simple to calculate; the dispersion description precision is improved, and the numerical stability problem of the dry and wet boundary is solved.

The application discloses a boundary plasma simulation method under a unit magnetic surface coordinate system and belongs to the technical field of magnetic confinement tokamak boundary plasma simulation. Firstly, a physical model is expressed by using a unit magnetic surface coordinate system, physical quantity and geometric effect are separated by using unitized base vectors, and singularity near an X point is avoided; secondly, a toroidal Fourier series representation is combined with difference along a magnetic field line, problem dimension is reduced, and numerical dissipation is reduced; thirdly, a generalized finite difference method and a transformed coordinate system are adopted to process the X point and adjacent points, and the X point can be directly contained in a grid; finally, a grid screening method is used to pre-calculate geometric quantities and difference coefficients under high-density grids, and the calculation precision of geometric effect is improved. The application can accurately simulate the influence of the X point on the stability of the boundary plasma, significantly reduces the consumption of computing resources, and improves the simulation efficiency and accuracy.

This embodiment discloses a dynamic evaluation method for the reliability of railway tunnels along the time axis. By considering the time-effect control equations for the initial lining of railway tunnels, as well as initial and boundary conditions, a loss function for a physical driving model is established using the finite difference method to train the model. This solves the problem that the solutions of traditional AD sampling points are independent and it is difficult to effectively utilize information from adjacent areas. After training the physical driving model, the initial lining displacement values ​​of the railway tunnel are predicted, and a system limit state function is established based on the predicted initial lining displacement values ​​to obtain the time-varying failure probability of the railway tunnel. By establishing a time-effect control equation for the initial lining of railway tunnels, this invention can fully consider the spatial and temporal variation characteristics of surrounding rock parameters, improve the efficient fusion capability of real-time monitoring data, and make the evaluation results close to the actual state of the tunnel, greatly reducing potential railway traffic hazards.

The invention provides a crossed ball guide rail workbench five-degree-of-freedom motion error analysis method, which comprises the steps of establishing a skin surface model based on part tolerance information and plane feature information, obtaining displacement spinor through constraint registration simulation, establishing a Jacobian matrix based on an assembly structure, and constructing a Jacobian skin surface model in combination with spinor information, calculating to obtain pose offset information of left and right guide rails and a slide rail in the guide rail pair, determining a space pose of a guide rail system in a global coordinate system by combining an actually measured raceway shape error, substituting the space pose into a crossed ball guide rail error model established based on a Hertz contact theory and static balance, and obtaining a five-degree-of-freedom motion error of the slide table through iterative solution. And finally, carrying out key error sensitivity analysis by adopting Monte Carlo simulation and a difference method. According to the method, the error transmission relation from the base to the working table function point can be accurately analyzed, the influence degree of the part tolerance on the motion precision is revealed, and an important reference basis is provided for design and assembly of the precise working table.

The application discloses an online roll temperature and thermal crown prediction method based on finite difference and polynomial regression, aiming at the technical pain point that the accuracy of the traditional analytical model is insufficient and numerical simulation cannot meet the online real-time calculation demand, a plurality of groups of working conditions are designed through an orthogonal experiment method, and a finite difference method is used to calculate roll steady-state temperature field data to construct a training data set, based on the data set, a polynomial regression algorithm with elastic network regularization is used to fit an axial temperature distribution general model with the roll temperature field as an input variable, after obtaining the online surface axial temperature distribution of the roll, the general model is substituted into the internal temperature field to restore the internal temperature field and calculate the radial thermal expansion amount at each position, so that compensation information is provided for a shape control system of a rolling mill. The application simplifies complex physical field simulation into lightweight algebraic formula operation, retains the high analytical accuracy of the finite difference method, realizes real-time prediction, and can significantly improve the rolling quality of silicon steel products.

This invention discloses a method for identifying dynamic loads on structures based on dynamic displacement inversion and motion control equations. The method includes: dividing the structure under test into several four-node inverse shell elements; deploying sensors at the centroid positions of the upper and lower surfaces of each element; reconstructing the displacement values ​​of each element node based on the collected strain data; predicting the displacement values ​​at any position of the structure under test using a high-resolution spatial domain reconstruction algorithm based on the reconstructed displacement values ​​of each element node; discretizing the fourth-order spatial deflection derivative of the displacement field of the structure under test using the finite difference method; approximating the second-order derivative of the strain data using the second-order central difference method to obtain the higher-order spatiotemporal deflection derivative; and establishing a model of the relationship between the dynamic displacement and load distribution of the structure under test by combining the motion control equations of the structure under test, thereby achieving dynamic load identification of the structure under test. Given the known geometry and material properties of the structure, this method can not only reconstruct the time history of the dynamic loads on the structure but also simultaneously identify the load application positions.

The invention discloses a Leidenfrost effect three-dimensional numerical simulation method based on a double-distribution function lattice Boltzmann method coupled with a finite difference method. The method comprises the following steps: 1, constructing a three-dimensional computational domain and defining lattice point types; 2, initializing a lattice point type, a fluid domain density field, a velocity field, a temperature field and a distribution function; 3, calculating fluid domain density, pressure and thermal diffusion coefficient according to the distribution function; 4, calculating resultant force of molecular interaction force and volume force among different phase states based on a pseudo-potential model; 5, calculating a fluid domain collision process; 6, calculating a migration process; 7, treating wall surface boundary conditions; 8, updating the fluid domain temperature field by using the temperature distribution function, and updating the solid domain temperature field by using a finite difference method; and 9, carrying out loop iteration on the steps 3-8 until the preset simulation time is reached. According to the invention, the whole process of the Leidenfrost phenomenon generated when the liquid drops impact the high-temperature wall surface can be stably and accurately simulated, and an accurate and reliable numerical simulation tool is provided for the design of efficient heat dissipation schemes such as spray cooling and the like.

The invention relates to a universal low numerical dispersion time domain finite difference method, which comprises the following steps of: based on a physical model structure of electromagnetic simulation engineering, selecting a time domain finite difference FDTD implicit or explicit subclass method which needs to be applied, and solving an optimal weighting factor and a correction factor; a stability CFL condition is verified; reconstructing an iterative formula that the electric field and the magnetic field advance along with time steps; the calculation main body carries out calculation solution, and the calculation main body involving implicit solution carries out calculation by using a tridiagonal matrix; on the basis of the steps, electric field parameters and magnetic field parameters in the whole area under a specific simulation instance are obtained, and therefore expected physical parameters are obtained through calculation. The isotropic sampling scheme is carried out in the stability condition limitation direction, the traditional sampling scheme is still kept in the remaining direction, and integration is carried out through a weighting factor and a correction factor. Wherein the values of the weighting factor and the correction factor are in one-to-one correspondence with the dispersion characteristics, so that the algorithm is ensured to have accurate parameter adaptability.

Systems and methods are disclosed. The method includes, for each of a plurality of spatial partial differential equations (sPDEs) within a wave equation, determining a linear system of equations using an approximate solution at a plurality of grid nodes. The linear system of equations includes an inverse matrix, a first vector, and a second vector. The method further includes, for each of the plurality of sPDEs, evaluating the inverse matrix by evaluating a first portion of the inverse matrix using a first deconvolution filter and evaluating a second portion of the inverse matrix using a second deconvolution filter. The method further still includes, for each of the plurality of sPDEs, evaluating the first vector using the evaluated inverse matrix and the second vector as well as determining the wavefield using the evaluated first vector for each of the plurality of sPDEs, a velocity model, and the wave equation.

This invention discloses a method for calculating the ultimate stress of surrounding rock in a non-circular tunnel, belonging to the field of tunnel engineering. The method includes the following steps: establishing a coordinate system with the center of the invert arch as the origin and defining the plastic zone based on fundamental parameters, wherein the fundamental parameters include tunnel geometric parameters, altered rock physical and mechanical parameters, and in-situ stress parameters; deriving the characteristic stresses and rotation angles of the invert arch boundary and the side arch boundary based on the fundamental parameters, the coordinate system, and the Mohr-Coulomb yield criterion to obtain the boundary conditions; deriving the differential and integral equations of the slip line in the plastic zone based on the altered rock physical and mechanical parameters to obtain the basic equations; constructing a computational grid using the finite difference method based on the boundary conditions and the basic equations and iteratively solving the problem to obtain the converged characteristic stresses and rotation angles of the grid nodes; and calculating the ultimate stress distribution of the plastic zone based on the converged characteristic stresses and rotation angles of the grid nodes.

The application discloses a nonlinear random structure dynamic reliability calculation method, and belongs to the field of structure health monitoring, comprising the following steps: collecting structure dynamic response data, and extracting high-order component signals of the structure dynamic response; correcting a random structure nonlinear model of an actual structure, and determining the probability distribution of the random structure nonlinear model parameters; applying external excitation load to the random structure nonlinear model, establishing a nonlinear random structure dynamic response state equation, and obtaining a probability density evolution equation; applying an absorbing boundary condition to the structure dynamic response according to a first exceeding failure criterion, combining a finite difference method and a nonlinear time-history integration method, obtaining the structure dynamic reliability changing with time under the dynamic load, and realizing the quantification of the reliability evolution process of the actual structure under the external load, avoiding the response assumption in the traditional reliability calculation process, and improving the calculation precision of the structure reliability.

The application discloses a closed space transient sound field simulation method based on modal expansion. The wave equation is solved based on the modal expansion theory, and finally the sound field is represented as superposition of a series of sound modes, that is, represented as a linear combination of the product of modal expansion coefficients and characteristic functions, wherein the modal expansion coefficients are quantities related to time information, and the characteristic functions are quantities related to space information. The application establishes a calculation format fusing boundaries and internal spaces based on the decomposition of the time-domain wave equation, thereby avoiding the large calculation amount problem of the traditional finite difference method in time and space. A time scale stable model is established through the solution of the transient modal expansion coefficient, thereby avoiding the instability problem of the time-domain boundary element method in the time scale. The application realizes efficient and accurate simulation of the closed space transient sound field through accurate modeling of the propagation and attenuation process of the sound field in the closed space.

The invention discloses a forward and inverse kinematics calculation method for a 2-SPS-1U parallel mechanism of a humanoid robot, and relates to the technical field of humanoid robots. Inverse kinematics calculation includes the steps that the pitch angle and the roll angle of a wrist, the inherent position of a spherical hinge point at the tail end of an SPS branched chain and the fixed translation vector of a wrist rotating shaft base coordinate system relative to a driving motor coordinate system are obtained; and constructing an attitude rotation matrix to convert the tail end spherical hinge point to a real-time coordinate, converting the real-time coordinate to a driving motor coordinate system by combining a fixed translation vector, and calculating a coordinate vector modulus length to obtain a branched chain telescopic length. The forward kinematics solution comprises the steps of obtaining the actually measured length and constant parameters of the branched chain, and constructing a nonlinear equation set by taking a roll angle and a pitch angle as unknown quantities; and solving by combining a two-dimensional residual vector with a Newton iteration method, estimating a Jacobian matrix by using a finite difference method in iteration, and outputting an attitude angle when the residual is smaller than a threshold value. The method is accurate and stable in calculation and adapts to the robot wrist joint control requirement.

This invention discloses a phase retrieval method for bright-field microscopy based on uniform light intensity, belonging to the technical field of phase retrieval in optical measurement. The method includes: acquiring focused light intensity images, positive defocus images, and negative defocus images along the optical axis; calculating the axial differential of light intensity using the finite difference method; simplifying the light intensity transmission equation to a Poisson equation and solving the Poisson equation to obtain the initial phase; constructing a positive higher-order error compensation model for the axial differential of light intensity and solving for the phase; and using a gradient acceleration strategy to iteratively calculate the phase to obtain the final phase result. This invention significantly improves the stability and accuracy of phase retrieval. Simultaneously, by combining angular spectrum theory to construct an axial differential higher-order error compensation model and introducing a gradient acceleration strategy, it effectively corrects defocus errors, greatly improves iterative convergence efficiency, and enhances the applicability of the method in practical microscopic imaging scenarios.

The invention relates to a method for calculating the average temperature of the cross section of a main cable of a suspension bridge, and the method comprises the steps: carrying out the dimension reduction based on a two-dimensional plane heat transfer model, so as to obtain a one-dimensional radial heat transfer model; performing space and time discretization on the one-dimensional radial heat transfer model to discretize the radial continuous temperature field of the main cable into the temperature of space-time grid points; finite difference decomposition of the one-dimensional radial heat transfer model is obtained through difference, and a difference format is determined; gradually solving the temperature of each node in each time step to form a temperature field matrix; and calculating the area weighted average temperature to obtain the average temperature of the cross section of the main cable. According to the method, a two-dimensional plane heat transfer model is simplified into a one-dimensional radial heat transfer model based on a cross section heat energy change equivalence principle, a computational domain is reduced to a radius line from a whole main cable cross section, and consumption and requirements on computational resources are greatly reduced; and a differential equation is converted into a linear equation set based on a finite difference method, so that the calculation efficiency is greatly improved while the same precision is maintained, and the calculation speed is improved from an hour level to a sub-second level.

The application provides a large-inclination well / horizontal well array induction logging resistivity anisotropy correction method and relates to the field of geological exploration. The method comprises the following steps: modeling by adopting a three-dimensional finite difference method, performing array induction logging response simulation, and simulating the variation law of apparent resistivity under different inclination angles; establishing a resistivity correction chart according to the variation law of the apparent resistivity under different inclination angles obtained by simulation; if the anisotropy coefficient of the measured target well is known, a correction formula is obtained by chart fitting, and the resistivity is corrected through the correction formula; if the anisotropy coefficient of the measured target well is unknown, the anisotropy coefficient is calculated through a build-up section, a correction formula is obtained by chart fitting, and the resistivity is corrected through the correction formula. The method can effectively eliminate the resistivity'spike' phenomenon of the logging result, obtain a more approximate real formation resistivity value, and improve the accuracy of the reservoir evaluation of the large-inclination well and the horizontal well.

The invention discloses an underground pipeline displacement analysis and correction system and method, and relates to the field of pipeline displacement monitoring. Based on the Winkler foundation beam theory, after a pipeline displacement control equation is established, a finite difference method is used for solving, and a linear equation set with the displacement of the center point of each pipe section as a solving target can be obtained; a matrix in the equation set has obvious regularity and can be solved by means of MATLAB programming; the solving process does not depend on a specific distribution form of a soil displacement field, and can be suitable for any displacement field form; vector superposition of vertical settlement and horizontal displacement is synthesized, the space deformation state of the pipeline is reflected more truly, and the safety assessment reliability of the existing pipeline is improved. In addition, soil body rigidity characteristics are correlated through a foundation bed coefficient, and meanwhile, vertical and horizontal pipe-soil interaction is considered, so that the problem of prediction deviation caused by neglect of a pipe-soil separation effect in a traditional method is solved.

The present application belongs to the technical field of geophysical exploration and basin analysis, and provides a lithospheric thermal structure inversion method based on physical constraints and event identification. The method builds physical constraint conditions including rheological constraints of the lithosphere, isostatic compensation constraints and thermodynamic consistency constraints, establishes a layered lithospheric thermal conduction model, solves the transient heat conduction equation by using the finite difference method, realizes automatic identification of tectonic events and parameter initialization by using a machine learning regression model, builds a multi-objective loss function and performs global optimization by using a differential evolution algorithm, and finally evaluates the parameter uncertainty by using the Monte Carlo method. The method solves the problems of strong multi-solution, lack of physical rationality and low calculation efficiency of traditional inversion methods, and realizes efficient and reliable inversion of lithospheric thermal structure parameters.