An Adaptive and Efficient Simulation Method and System for Electric Propulsion Plasma Oscillations

Abstract

This invention provides an adaptive and efficient simulation method and system for electric propulsion plasma oscillations, relating to the field of electric propulsion plasma oscillation simulation technology. The method employs a high-precision numerical simulation model for electric propulsion plasma to ensure the accuracy of the simulation results. It decomposes the computational tasks of the simulation model through various adjustment parameters, i.e., it uses parallel computing to complete all computational tasks, thereby effectively improving simulation efficiency. Furthermore, when solving the simulation model, a multi-grid method is specifically used to solve the electromagnetic field model, which further improves the simulation computational efficiency. In addition, this invention utilizes a target neural network model to process the simulation results, updating all adjustment parameters in the above process, thereby continuously and adaptively optimizing multiple adjustment parameters used in the simulation process, ultimately achieving a technical effect that balances the simulation efficiency and accuracy of the electric propulsion plasma oscillation simulation model.

Description

Technical Field

[0001] This invention relates to the field of electric propulsion plasma oscillation simulation technology, and in particular to an adaptive and efficient simulation method and system for electric propulsion plasma oscillation. Background Technology

[0002] Current simulations of electric propulsion plasma oscillations typically employ the traditional particle method. This method simplifies the physical plasma discharge region of a real electric thruster into a numerical computational domain. The continuous physical space is divided into discrete geometric grids, on which discretized Maxwell's equations for the electromagnetic field or discretized Poisson's equations for the electrostatic field are solved to obtain the background electric and magnetic fields at each moment. Real plasma particles are considered as macroparticle clusters formed by the aggregation of numerous particles, possessing the same charge-to-mass ratio as real particles. After setting initial values, first-principles calculations are used to numerically solve for the dynamic characteristics of the formation and evolution of electric propulsion plasma oscillations.

[0003] However, in order to achieve high computational accuracy, hundreds of millions of particles and millions of grids are often required. The large number of particles and grids makes this method very inefficient. Increasing the weights and other acceleration methods may increase numerical noise, making it difficult for traditional electric propulsion plasma oscillation simulation to balance accuracy and efficiency. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive and efficient simulation method and system for electric propulsion plasma oscillations, so as to alleviate the technical problem in the prior art that it is impossible to balance the simulation efficiency and accuracy of electric propulsion plasma oscillation simulation.

[0005] In a first aspect, the present invention provides an adaptive and efficient simulation method for electric propulsion plasma oscillation, comprising: step 110, constructing a high-precision numerical simulation model of electric propulsion plasma; wherein the simulation model includes: a particle motion model, a charge deposition model, an electromagnetic field solution model, and a collision solution model; step 120, decomposing the computational tasks of the target model based on a first set of adjustment parameters, and decomposing the computational tasks of the electromagnetic field solution model based on a second set of adjustment parameters; wherein the first set of adjustment parameters includes: the number of first parallel computing tasks, the number of GPU cores required for each computing task, the workload of each GPU core, and the time step of each type of particle propulsion; the target model includes: a particle motion model, a charge deposition model, and a collision solution model; the second set of adjustment parameters includes: the number of second parallel computing tasks, the number of CPU cores required for each computing task, and the workload of each CPU core; step 130, performing simulation solution on the simulation model to obtain the simulation results. Results; Among them, the electromagnetic field solution model is solved using the multigrid method with the third set of adjustment parameters; the third set of adjustment parameters includes: multigrid cycle form, number of multigrid layers, smoothing times per layer, multigrid smoothing method, and grid spacing in each direction; Step 140, the simulation results are feature extracted using the target neural network model, and the updated first set of adjustment parameters, the updated second set of adjustment parameters, and the updated third set of adjustment parameters are determined based on the feature extraction results; the target neural network model is a model obtained after training with the goal of maximizing simulation efficiency; the feature extraction results include: electron density, electron temperature, elastic collision frequency between electrons and atoms, excitation collision frequency, ionization collision frequency, momentum exchange collision frequency and charge exchange collision frequency between ions and atoms, potential gradient, and plasma oscillation growth rate over time for each grid node; Step 150, Steps 120 to 140 are repeated until the preset simulation cutoff condition is reached.

[0006] Optionally, the simulation model is solved to obtain simulation results, including: initializing the number of particles in the electric propulsion plasma and the mass, charge, velocity, and position of each particle; depositing the charge of all charged particles onto the corresponding grid nodes of the finest grid in the multigrid based on the charge deposition model, obtaining the charge amount on each grid node; wherein, the grid node corresponding to the target charged particle is the 8 vertices of the nearest grid cube; the target charged particle represents any charged particle among all charged particles; calculating the charge density of the target grid node based on the charge amount of the target grid node and the unit grid volume in the finest grid; wherein, the target grid node represents Any grid node in the finest grid; based on the multigrid method and the charge density of each grid node in the finest grid, the electromagnetic field solution model is solved to obtain the potential distribution in the electric propulsion plasma; based on the potential distribution, the force on each particle is calculated, and the updated position and velocity of each particle are solved based on the particle motion model; based on the updated velocity of each particle and the collision solution model, the number of target collisions between particles is determined; among them, the target collisions include: elastic collisions between electrons and atoms, excited collisions between electrons and atoms, ionization collisions between electrons and atoms, momentum exchange collisions between ions and atoms, and charge exchange collisions between ions and atoms.

[0007] Optionally, based on the charge deposition model, the charge of all charged particles is deposited onto the corresponding grid nodes of the finest grid in the multigrid, obtaining the charge amount at each grid node, including: uniformly dividing all charged particles into... Group, and copy the mesh space of the finest mesh in the multigrid as This is done to establish a one-to-one correspondence between GPU cores, charged particle groups, and mesh space; where M represents the number of parallel computing tasks corresponding to the charge deposition model, and N represents the number of GPU cores required for each computing task; each GPU core deposits the charged particles in its corresponding charged particle group into the corresponding mesh space through the charge deposition model; all of these are then processed. The charge amounts in each grid space are accumulated to obtain the charge amount at each grid node.

[0008] Optionally, the particle motion model is as follows: ;in, Indicates the mass of the particle. Indicates the particle at the 1st The speed of motion at each time step Indicates the particle in the first... The speed of motion at each time step Indicates the time step of particle propulsion. Indicates the particle at the 1st The force at each time step Indicates the particle at the 1st The position of each time step. Indicates the particle at the 1st The position of each time step.

[0009] Alternatively, the charge deposition model is: ;in, Indicates the grid spacing in the x-direction. Indicates the grid spacing in the y-direction. Indicates the grid spacing in the z-direction. This represents the position coordinates of the particles undergoing charge deposition. Represents the coordinates of the grid nodes. Indicates particles at grid nodes Charge distribution weights on , , , This indicates the number of grid nodes in the x-direction. This indicates the number of grid nodes in the y-direction. This indicates the number of grid nodes in the z-direction. This represents the macro-particle weights.

[0010] Optionally, the electromagnetic field solution model is as follows: A discretized set of equations consisting of objective equations for each grid node; grid node The objective equation is expressed as: ;in, Represents grid nodes The potential on, Represents grid nodes charge density, It represents the vacuum permittivity.

[0011] Optionally, the collision solution model includes the following solution steps: using formulas Calculate the maximum collision frequency of all particles; where, Represents the number density of atoms. This represents the velocity of the i-th particle. , Let i represent the kinetic energy of particle i. This represents the collision cross section where particle i undergoes the j-th collision. Represents the maximum collision frequency of all particles; using the formula Calculate the probability of all particles experiencing an empty collision in the current time step; where, Indicates the time step of particle propulsion. This represents the probability that all particles will have an empty collision within the current time step; using the formula... Calculate the number of collisions between particles; where, Represents the total number of particles. Indicates the number of collisions between particles; randomly generated. A random number to correspond to A particle that collides; if If the k-th particle collides with the first type of collision, then the first type of collision occurs; where, This represents the random number corresponding to the k-th particle that collides; This represents the frequency at which the k-th particle experiences the j-th type of collision; if If the k-th particle collides with another particle, then the (j+1)-th collision occurs; if If the k-th particle that collides with the other particle has an empty collision, then the collision will occur.

[0012] Optionally, after determining the number of target collisions between particles, the method further includes: constructing a three-dimensional computational domain unit model and initializing the particle parameters in the unit model; wherein the particle parameters include: atomic density, electron density, ion density, electron kinetic energy, and ion kinetic energy; disabling the particle motion model and counting the number of target collisions between particles within the time step of particle propulsion; calculating the reaction rate error of each type of collision based on the number of collisions and particle parameters in the target collisions; and correcting the collision cross section of the corresponding collision and / or correcting the time step of particle propulsion if it is determined that there is a reaction rate error in the target collision that is greater than a preset threshold.

[0013] Optionally, before solving the electromagnetic field model using the multigrid method with a third set of adjustment parameters, the following steps are also included: rewriting the discretized equations into a linear equation system of the form AS=b; where S represents the array of unknowns, A represents the coefficient matrix of the array of unknowns S, and b represents the source term array. ,and .

[0014] Secondly, the present invention provides an adaptive and efficient simulation system for electric propulsion plasma oscillation, comprising: a construction module for constructing a high-precision numerical simulation model of electric propulsion plasma; wherein the simulation model includes: a particle motion model, a charge deposition model, an electromagnetic field solution model, and a collision solution model; a decomposition module for decomposing the computational tasks of the target model based on a first set of adjustment parameters, and decomposing the computational tasks of the electromagnetic field solution model based on a second set of adjustment parameters; wherein the first set of adjustment parameters includes: a first number of parallel computing tasks, the number of GPU cores required for each computing task, the workload of each GPU core, and the time step of each type of particle propulsion; the target model includes: a particle motion model, a charge deposition model, and a collision solution model; the second set of adjustment parameters includes: a second number of parallel computing tasks, the number of CPU cores required for each computing task, and the workload of each CPU core; and a simulation module for performing simulation solutions on the simulation model to obtain simulation results. The results are as follows: The electromagnetic field solution model is solved using a multigrid method with a third set of adjustment parameters. The third set of adjustment parameters includes: multigrid cycle form, number of multigrid layers, smoothing times per layer, multigrid smoothing method, and grid spacing in each direction. The processing module is used to extract features from the simulation results using the target neural network model, and to determine the updated first, second, and third set of adjustment parameters based on the feature extraction results. The target neural network model is a model trained with the goal of maximizing simulation efficiency. The feature extraction results include: electron density, electron temperature, elastic collision frequency between electrons and atoms, excitation collision frequency, ionization collision frequency, momentum exchange collision frequency and charge exchange collision frequency between ions and atoms, potential gradient, and plasma oscillation growth rate over time for each grid node. The optimization module is used to repeatedly call the decomposition module, simulation module, and processing module until the preset simulation cutoff conditions are reached.

[0015] This invention provides an adaptive and efficient simulation method for electric propulsion plasma oscillations. This method employs a high-precision numerical simulation model of electric propulsion plasma to ensure the accuracy of the simulation results. Furthermore, it decomposes the computational tasks of the simulation model through various adjustable parameters, i.e., it uses parallel computing to complete all computational tasks, thereby effectively improving simulation efficiency. In addition, a multi-grid method is specifically used to solve the electromagnetic field model during simulation, which further improves the computational efficiency. Moreover, this invention utilizes a target neural network model to process the simulation results, updating all adjustable parameters in the above process, thereby continuously and adaptively optimizing multiple adjustable parameters used in the simulation, ultimately achieving a technical effect that balances the simulation efficiency and accuracy of the electric propulsion plasma oscillation simulation model. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 A flowchart of an adaptive and efficient simulation method for electric propulsion plasma oscillation provided in an embodiment of the present invention;

[0018] Figure 2 A schematic diagram of a simulation loop calculation framework provided in an embodiment of the present invention;

[0019] Figure 3 A functional block diagram of an adaptive and efficient simulation system for electric propulsion plasma oscillation provided in an embodiment of the present invention;

[0020] Figure 4 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0022] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0023] The following embodiments of the present invention are described in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments may be combined with each other.

[0024] Example 1

[0025] Figure 1 A flowchart of an adaptive and efficient simulation method for electric propulsion plasma oscillations provided in an embodiment of the present invention is shown below. Figure 1 As shown, the method specifically includes the following steps:

[0026] Step 110: Construct a high-precision numerical simulation model of electric propulsion plasma.

[0027] For electric thruster plasma, its operation involves complex plasma reactions and wave-plasma interactions. Therefore, selecting a suitable simulation model is the theoretical basis for accurately describing the corresponding plasma processes. Thus, at the outset, a high-precision numerical simulation model of electric propulsion plasma needs to be established to ensure the accuracy of the simulation results. This simulation model includes: a particle motion model, a charge deposition model, an electromagnetic field solution model, and a collision solution model. The boundary conditions of each simulation model need to be defined according to the specific electric propulsion system. For example, non-metallic materials such as ceramics are defined as dielectric boundaries, and the anode and thruster wall are defined as first-type boundary conditions with a fixed potential. This embodiment of the invention does not specifically limit the boundary conditions of the simulation models; users need to adapt them to their actual simulation systems.

[0028] Step 120: Decompose the computational task of the target model based on the first set of adjustment parameters, and decompose the computational task of the electromagnetic field solution model based on the second set of adjustment parameters.

[0029] The method proposed in this invention is essentially an artificial intelligence-optimized simulation method for electric propulsion plasma oscillations. That is, it optimizes the simulation of electric propulsion plasma oscillations using artificial intelligence. After determining the simulation model, in order to optimize the simulation process of electric propulsion plasma oscillations through artificial intelligence, it is necessary to reserve scalable artificial intelligence optimization interfaces in each calculation module (i.e., the module used to solve the simulation model). This provides the conditions for achieving efficient simulation of electric propulsion plasma oscillations based on artificial intelligence optimization. Specifically, the relevant parameters affecting the solution of each simulation model are passed through the aforementioned interfaces.

[0030] This invention employs an adjustable-load parallel computing method to optimize the solution of the simulation model, thereby improving simulation computation efficiency. Specifically, the parallel task decomposition method for the simulation model needs to be divided into the following two categories:

[0031] 1. Particle Task Decomposition: Particle task decomposition involves dividing computational tasks related to particle motion and particle-mesh interactions into spatial regions or particle groups and allocating them to different computing units (CPU, GPU) for parallel computation. Since the computations involved in particle motion mainly include velocity and position updates, charge deposition, and particle collisions, these operations have high computational requirements but weak inter-task coupling, making them particularly suitable for accelerating computation using the multi-threaded architecture of GPUs. GPUs can efficiently process a large number of particle trajectories in parallel, significantly improving computational efficiency.

[0032] However, the parallelization of particle motion requires consideration of the load balancing problem among particles to avoid some computational units becoming bottlenecks due to excessive load. Furthermore, by adjusting the time step of each particle propulsion method to match the plasma frequency, optimal computational efficiency can be achieved while maintaining computational accuracy. Therefore, when decomposing the particle task, various parameters in the first set of adjustment parameters should be optimized in conjunction with the task characteristics of the target model to achieve efficient particle simulation. The first set of adjustment parameters includes: the number of first parallel computing tasks, the number of GPU cores required for each computing task, the workload of each GPU core, and the time step of each particle propulsion method; the target model includes: a particle motion model, a charge deposition model, and a collision solution model.

[0033] 2. Field Solution Task Decomposition: Field solution task decomposition refers to dividing the task involving electromagnetic field distribution calculations into smaller parts and allocating them to different computational units for parallel solving. Unlike particle tasks, field solving often involves numerical computations on a global grid, with strong coupling between tasks. In particular, field smoothing and residual calculations in multigrid methods require information exchange across multiple grid levels. Due to the advantages of CPUs in handling complex task dependencies and memory management, field solution task decomposition is more suitable for CPU implementation. Specifically, parallelization of field solving can be achieved by dividing the grid into regions.

[0034] However, the data communication overhead between CPUs must be optimized appropriately. If the task granularity is not properly allocated, the increased communication overhead may offset the performance improvement brought by parallelization. Therefore, when decomposing the computational tasks of the electromagnetic field solution model, the parameters in the second set of adjustment parameters below need to be optimized to ensure the improvement of overall simulation efficiency. The second set of adjustment parameters includes: the number of second parallel computing tasks, the number of CPU cores required for each computing task, and the workload of each CPU core.

[0035] Step 130: Perform simulation and solve the simulation model to obtain the simulation results.

[0036] The electromagnetic field solution model is solved using a multigrid method with a third set of adjustment parameters. The third set of adjustment parameters includes: the multigrid cycle form (e.g., V cycle, W cycle, F cycle), the number of multigrid layers, the number of smoothing operations per layer, the multigrid smoothing method (Jacobi iteration or successive over-relaxation iteration, etc.), and the grid spacing in each direction.

[0037] Specifically, after decomposing the computational tasks of each simulation model, considering that the data communication overhead between particles and the mesh cannot be ignored, especially in high-resolution meshes, where a large amount of particle charge deposition and field information updates may lead to frequent host-device data transmission, thereby reducing the overall simulation efficiency, this embodiment of the invention adopts a multi-mesh method to improve the simulation solution speed when solving the electromagnetic field model, and includes the relevant parameters in the algorithm (i.e., the aforementioned third set of adjustment parameters) in the list of parameters that need to be adjusted by artificial intelligence.

[0038] The impact of the parameters included in the aforementioned first, second, and third sets of adjustment parameters on simulation efficiency is highly dependent on the computational conditions and the current state of the plasma simulation. Improper selection not only fails to improve the simulation efficiency of electric propulsion plasma oscillations but can also severely reduce it. Furthermore, due to the complexity of the simulation parameters and the large magnitude of change in plasma parameters over simulation time, it is difficult to manually select simulation parameters in a timely manner. Therefore, this embodiment of the invention requires artificial intelligence optimization of the electric propulsion plasma numerical simulation model.

[0039] Step 140: Use the target neural network model to extract features from the simulation results, and determine the updated first set of adjustment parameters, the updated second set of adjustment parameters, and the updated third set of adjustment parameters based on the feature extraction results.

[0040] Among them, the target neural network model is a model obtained after training with the goal of maximizing simulation efficiency; the feature extraction results include: electron density, electron temperature, elastic collision frequency between electrons and atoms, excitation collision frequency, ionization collision frequency, momentum exchange collision frequency and charge exchange collision frequency between ions and atoms, potential gradient and plasma oscillation growth rate over time for each grid node.

[0041] The simulation results of the simulation model include: the amount of charge on each grid node, the charge density of each grid node, the potential distribution in the electric propulsion plasma, the velocity and position of each particle, and the number of target collisions between particles. The target collisions include: elastic collisions between electrons and atoms, excited collisions between electrons and atoms, ionization collisions between electrons and atoms, momentum exchange collisions between ions and atoms, and charge exchange collisions between ions and atoms.

[0042] For the following reasons, after obtaining the simulation results, this embodiment of the invention needs to use the target neural network model to extract features from the simulation results, so as to update the first adjustment parameter set, the second adjustment parameter set, and the third adjustment parameter set based on the feature extraction results:

[0043] 1. When the particle weights are equal, the electron density affects the number of macroparticles, which in turn affects the solution efficiency of the particle motion model, charge deposition model, and collision solution model. The parallel workload of each needs to be allocated through the first set of adjustment parameters. In addition, the electron density directly determines the plasma frequency. The plasma frequency needs to be matched by adjusting the time step of each particle in the first set of adjustment parameters in order to obtain the optimal computational efficiency while maintaining the computational accuracy.

[0044] 2. Electron density and electron temperature directly determine the plasma Debye length. The plasma Debye length needs to be matched by adjusting the grid spacing in each direction through the second adjustment parameter set in order to obtain the optimal computational efficiency while maintaining computational accuracy.

[0045] 3. Electron temperature, the elastic collision frequency between electrons and atoms, the excitation collision frequency, the ionization collision frequency, the momentum exchange collision frequency between ions and atoms, and the charge exchange collision frequency will affect the solution efficiency of the collision solution model. Parallel workload needs to be allocated through the first set of adjustment parameters.

[0046] 4. Electric potential gradient: A strong electric potential gradient will reduce the efficiency of the electromagnetic field solution model. It is necessary to increase the computational resources (number of computational cores) occupied by the electromagnetic field solution model by adjusting the second and third adjustment parameter sets in order to improve computational efficiency.

[0047] 5. The rate of increase of plasma oscillation over time may trigger more particle dynamic behaviors, increasing computational complexity. Therefore, it is also used as one of the input parameters. The robustness of the neural network model needs to be improved by adjusting the first set of adjustment parameters, the second set of adjustment parameters, and the third set of adjustment parameters.

[0048] 6. Various parameters included in the feature extraction results, and spatially non-uniform regions where parameters change drastically (such as the plasma sheath region), may lead to uneven load. It is necessary to increase the allocation of computing resources (number of computing cores) in the computing density region by adjusting the first set of parameters and the second set of parameters.

[0049] The following section details the methods for determining each parameter in the above feature extraction results:

[0050] Based on the method for solving the charge density of grid nodes, by replacing the deposited charge with "1" or the directional components of particle velocity or particle kinetic energy, the particle density, directional components of particle velocity, or particle kinetic energy of any type of particle in the target grid node can be calculated. Here, "any type of particle" refers to electrons, ions, and atoms in the calculation. Based on the particle density, directional components of particle velocity, and particle kinetic energy of the electrons in the target grid node, the electron temperature of the target grid node can be calculated using the following algorithm: [2×(electron kinetic energy - electron directional motion kinetic energy) / (3×electron density)]. Here, the electron directional motion kinetic energy is calculated from the directional components of electron velocity in the target grid node. This method can accurately solve for the electron temperature and obtain the electron thermal motion state by eliminating the contribution of electron directional drift to the electron temperature by removing the electron directional motion kinetic energy.

[0051] Based on the collision solution model, the frequency of each type of collision can be calculated by counting the number of collisions per unit time. After obtaining the potential distribution in the electrically propelled plasma, the potential gradient can be calculated based on the potential distribution.

[0052] Build as Figure 2 The illustrated iterative computational framework, consisting of a charge deposition model -> electromagnetic field solution model -> particle motion model -> collision solution model -> charge deposition model, yields simulation results after a certain number of iterations. The number of iterations is determined by the required simulation time. Based on these simulation results, the electron density sequence at each spatial point over a given time period is taken. The absolute value is then low-pass filtered to obtain the envelope signal. Numerical differentiation allows for the calculation of the plasma oscillation rate over time.

[0053] In this embodiment of the invention, the target neural network model is a model obtained by training an initial neural network model with the goal of maximizing simulation efficiency. Although the input data of the neural network model is the simulation result of the aforementioned simulation model, the input features used during model training are essentially the feature extraction results corresponding to the simulation results. The output labels are the first set of adjustment parameters, the second set of adjustment parameters, and the third set of adjustment parameters. Simulation efficiency is inversely proportional to the simulation time of the simulation cycle (charge deposition model -> electromagnetic field solution model -> particle motion model -> collision solution model), that is, the longer the simulation time, the lower the simulation efficiency. During model training, the total time of each simulation cycle can be statistically analyzed, and the simulation efficiency can be determined based on the average of the total time of multiple adjacent simulation cycles to reduce fluctuation errors.

[0054] The specific training process for a neural network model is as follows:

[0055] 1) Data acquisition and preparation: For a given input feature, uniform sampling is performed within the parameter range of the output label, and an electric propulsion plasma oscillation simulation without adaptive adjustment is run to record the plasma parameters (feature extraction results) and calculation time data at each time step.

[0056] 2) Data analysis and feature extraction: Analyze the data obtained in 1), extract the input features, the shortest computation time and its corresponding output label, and establish the mapping relationship between the input features and the optimal output label.

[0057] 3) Based on the data obtained in 2), train the initial neural network model until a preset termination condition is met, enabling it to accurately predict the relationship between input features and the optimal output label. The preset termination condition, for example, is reaching a specified number of training iterations and achieving stable convergence in simulation efficiency.

[0058] Step 150: Repeat steps 120 to 140 until the preset simulation cutoff condition is reached.

[0059] After obtaining the target neural network model through training, the trained target neural network model is integrated into the electric propulsion plasma oscillation simulation system. During the electric propulsion plasma oscillation simulation, the target neural network model can predict and dynamically adjust the output label by collecting simulation results in real time and calculating feature extraction results, thereby realizing adaptive and efficient simulation of electric propulsion plasma oscillation.

[0060] This invention provides an adaptive and efficient simulation method for electric propulsion plasma oscillations. This method employs a high-precision numerical simulation model of electric propulsion plasma to ensure the accuracy of the simulation results. Furthermore, it decomposes the computational tasks of the simulation model through various adjustment parameters, i.e., it uses parallel computing to complete all computational tasks, thereby effectively improving simulation efficiency. In addition, a multi-grid method is specifically used to solve the electromagnetic field model during simulation, which further improves the computational efficiency. Moreover, this invention utilizes a target neural network model to process the simulation results, updating all adjustment parameters in the above process, thereby continuously and adaptively optimizing multiple adjustment parameters used in the simulation process, ultimately achieving a technical effect that balances the simulation efficiency and accuracy of the electric propulsion plasma oscillation simulation model.

[0061] In one optional implementation, step 130 above, which involves performing a simulation solution on the simulation model to obtain simulation results, specifically includes the following steps:

[0062] Step 1301: Initialize the number of particles in the electric propulsion plasma and the mass, charge, velocity, and position of each particle.

[0063] In this embodiment of the invention, the particles in the electrically propelled plasma include electrons, ions, and atoms. Before simulation, it is necessary to initialize the quantity, mass, charge, velocity, and position of each of the above particles.

[0064] Step 1302: Based on the charge deposition model, the charge of all charged particles is deposited onto the corresponding grid nodes of the finest grid in the multigrid to obtain the charge amount on each grid node; wherein, the grid node corresponding to the target charged particle is the 8 vertices of the grid cube closest to it; the target charged particle represents any charged particle among all charged particles.

[0065] In this embodiment of the invention, the charge deposition model is as follows: ;in, Indicates the grid spacing in the x-direction. Indicates the grid spacing in the y-direction. Indicates the grid spacing in the z-direction. This represents the position coordinates of the particles undergoing charge deposition. Represents the coordinates of the grid nodes. Indicates particles at grid nodes Charge distribution weights on , , , This indicates the number of grid nodes in the x-direction. This indicates the number of grid nodes in the y-direction. This indicates the number of grid nodes in the z-direction. This represents the macro-particle weights.

[0066] Based on the above charge deposition model, the charge of each charged particle can be deposited onto the grid nodes in a multigrid. This charge deposition model is applicable to three-dimensional multigrids; therefore, the grid nodes corresponding to the target charged particle are the eight vertices of the nearest grid cube. If a two-dimensional multigrid is used, the grid nodes corresponding to the target charged particle are the four vertices of the nearest grid. Multiple particles may have charge deposition on each grid node, and the charge amount of any grid node is the net charge obtained by subtracting all negative charges from all positive charges on that grid node. In this embodiment of the invention, this charge deposition model is also applicable to handling particle density, average particle velocity, and average particle energy.

[0067] Step 1303: Calculate the charge density of the target grid node based on the charge amount of the target grid node and the unit grid volume in the finest grid; where the target grid node represents any grid node in the finest grid.

[0068] Specifically, the charge density of a target grid node is obtained by dividing the charge of the target grid node by the unit grid volume in the finest grid.

[0069] Step 1304: Based on the multigrid method and the charge density of each grid node in the finest grid, the electromagnetic field solution model is solved to obtain the potential distribution in the electric propulsion plasma.

[0070] Generally, solving for electromagnetic fields can be broadly divided into two parts: solving for the electric field and solving for the magnetic field. Both involve using finite difference solutions to Maxwell's equations to obtain the spatial distribution of the electric and magnetic fields. In electric thrusters, especially Hall thrusters, the applied magnetic field is much larger than the magnetic field generated by plasma coupling, so it can be considered a static magnetic model. In the static magnetic model, particles are acted upon by both electric and magnetic fields, but the induced magnetic field generated by particle motion is very small and can be neglected. In this case, the magnetic field can be considered a static magnetic field, and the electric field can be solved using Poisson's equations during the calculation.

[0071] After statistically obtaining the charge value at each grid node in the computational domain, the spatial potential distribution can be obtained using the Poisson equation: ,in, Represents electric potential, This represents the charge value at the grid node. Represents the vacuum permittivity. Represents the number density of ions. This represents the electron number density. However, due to the high dimensionality of the Poisson equation, an analytical solution cannot be obtained directly. Therefore, it is necessary to process it into a system of linear equations using the difference discretization method.

[0072] In one alternative implementation, the electromagnetic field solution model is: A discretized system of equations consisting of objective equations for grid nodes; where the grid nodes... The objective equation is expressed as: ;in, Represents grid nodes The potential on, Represents grid nodes charge density, It represents the vacuum permittivity.

[0073] The spatial distribution of the electric potential can be obtained by solving the discretized equations. The spatial distribution of the electric field can be further obtained using differential potential. The formula for calculating the electric field strength is: .

[0074] To improve the solution speed and simulation efficiency of the equation system, this embodiment of the invention includes the following steps before solving the electromagnetic field solution model using the multigrid method with a third set of adjustment parameters:

[0075] The discretized system of equations is rewritten as a linear system of equations in the form AS = b; where S represents the array of unknowns, A represents the coefficient matrix of the array of unknowns S, and b represents the array of source terms. ,and .

[0076] For numerical solutions to the aforementioned linear equations, traditional direct solutions (Gaussian elimination, LU decomposition, etc.) and iterative solutions (Jacobi iteration, successive over-relaxation iteration, etc.) are too slow. Therefore, this embodiment of the invention employs a multigrid method to improve computational efficiency. The multigrid method yields the electric potential of each grid node in the spatial multigrid, and then interpolation is used to obtain the electric potential at the location of each particle, thus obtaining the potential distribution in the electrically propelled plasma.

[0077] Step 1305: Calculate the force on each particle based on the potential distribution, and solve for the updated position and velocity of each particle based on the particle motion model.

[0078] Taking a single particle as an example, after obtaining the electric potential at the particle's location, according to... The electric field strength at the location of the particle can be obtained. The force on the charged particle in the electric field is F=qE, where q is the particle charge.

[0079] During the simulation, the velocity and position of each particle need to be recorded, and calculations and data updates need to be performed at each time step. When the number of particles is large, it places high demands on the computer's storage space. In order to ensure computational accuracy while minimizing storage space, the particle motion model in this embodiment of the invention is as follows: ;in, Indicates the mass of the particle. Indicates the particle at the 1st The speed of motion at each time step Indicates the particle at the 1st The speed of motion at each time step Indicates the time step of particle propulsion. Indicates the particle at the 1st The force at each time step Indicates the particle at the 1st The position of each time step. Indicates the particle at the 1st The position is determined by each time step. That is, the particle's velocity *v* and position *x* are at different time points. The time points for position calculation are chosen as integer multiples of the time step, while the time points for velocity calculation are chosen as half-integer multiples of the time step. The calculation times for *x* and *v* are staggered by 0.5 seconds. The two methods are solved alternately and cyclically, which effectively reduces the storage space requirements of the computer.

[0080] Step 1306: Based on the updated motion velocity and collision solution model of each particle, determine the number of target collisions between particles; wherein, the target collisions include: elastic collisions between electrons and atoms, excitation collisions between electrons and atoms, ionization collisions between electrons and atoms, momentum exchange collisions between ions and atoms, and charge exchange collisions between ions and atoms.

[0081] In an electric thruster, various collisions occur between particles. Different types of collisions occur between electrons and atoms, ions and atoms, electrons and ions, atoms and atoms. Due to the small size of the electric thruster and the large mean free path of most collisions, these collisions occur relatively infrequently. The dominant collision processes in an electric thruster are elastic, excited, and ionizing collisions between electrons and atoms, and momentum-exchange and charge-exchange collisions between ions and atoms.

[0082] In this embodiment of the invention, the collision solution model includes the following solution steps:

[0083] Step A: Using the formula Calculate the maximum collision frequency of all particles; where, Represents the number density of atoms. This represents the velocity of the i-th particle. , Let i represent the kinetic energy of particle i. This represents the collision cross-section of particle i undergoing the j-th collision. The collision cross-section data was obtained by consulting publicly available experimental data and was pre-set in the simulation. The maximum collision frequency of all particles is represented by N. In this embodiment of the invention, based on the number of collision types included in the target collision, N is 5, representing 5 types of collisions.

[0084] The atomic number density is obtained as follows: referring to the method of calculating charge distribution weights using the charge deposition model, the aforementioned charge distribution weights can be used as atomic distribution weights when calculating the atomic number density. All "atoms" are deposited onto each grid node to obtain the number of atoms on each grid node. This number is then divided by the volume of the finest grid cell to obtain the atomic number density of each grid node. This represents the maximum atomic number density among all grid nodes in the space.

[0085] Step B: Using the formula Calculate the probability of all particles experiencing an empty collision in the current time step; where, Indicates the time step of particle propulsion. This represents the probability that all particles will have an empty collision within the current time step.

[0086] Step C: Using the formula Calculate the number of collisions between particles; where, Represents the total number of particles. This indicates the number of collisions between particles.

[0087] Step D: Randomly generate A random number to correspond to A particle that collides.

[0088] Step E: If If the k-th particle collides with the first type of collision, then the first type of collision occurs; where, This represents the random number corresponding to the k-th particle that collides; This represents the frequency at which the k-th particle experiences the j-th type of collision.

[0089] Step F: If If the k-th particle collides with another particle, then the (j+1)-th collision will occur.

[0090] Step G: If If the k-th particle that collides with the other particle has an empty collision, then the collision will occur.

[0091] To handle particle collisions using the methods described above, it is only necessary to... By calculating collisions with individual particles, the number of particles involved in the collision calculation is greatly reduced, thus improving computational efficiency.

[0092] In an optional implementation, after determining the number of target collisions between particles, the present invention further includes the following steps:

[0093] Step 201: Construct a three-dimensional computational domain unit model and initialize the particle parameters in the unit model; wherein, the particle parameters include: atomic density, electron density, ion density, electron kinetic energy and ion kinetic energy.

[0094] Step 202: Disable the particle motion model and count the number of target collisions between particles within the time step of particle propulsion.

[0095] Step 203: Calculate the reaction rate error for each type of collision based on the number of collisions and particle parameters in the target collision.

[0096] Step 204: If it is determined that there is a reaction rate error greater than a preset threshold in the target collision, the collision cross section of the corresponding collision is corrected, and / or the time step of particle propulsion is corrected.

[0097] Specifically, after improving the efficiency of electric propulsion plasma oscillation simulation by employing parallel computing, this embodiment of the invention proposes to calibrate the collision model to further enhance simulation accuracy. Optionally, a three-dimensional computational domain unit model with a size of 1mm × 1mm × 1mm (1mm is an exemplary value) is constructed, and the atomic density in the unit model is initialized. electron density Ion density Electron kinetic energy kinetic energy of ions .

[0098] Next, in order to disable the change in particle velocity, as well as loss and generation, after the collision model occurs, this embodiment of the invention disables the aforementioned particle motion model and counts the number of events of each type of collision that occur within the given time step of particle propulsion in the simulation.

[0099] Based on the collision types included in the embodiments of this invention, target collisions can be divided into two main categories: collisions between electrons and atoms (Type I collisions) and collisions between ions and atoms (Type II collisions). Therefore, the following methods should be referenced when calibrating the collision solution model:

[0100] If the j-th type of collision (j=1,2,3) occurs in the first type of collision, then the actual reaction rate of the j-th type of collision (the number of collision events occurring per unit time and per unit volume) is calculated using the formula... The theoretical reaction rate for the j-th collision is calculated using the formula... Calculations are performed, in which, This represents the number of the j-th collision. This represents the macroparticle weight (the weight of the real particle represented by a macroparticle in the simulation of the macroparticle method). Represents the elementary charge. Indicates electron mass. and The error is the reaction rate error of the j-th collision in the first type of collision.

[0101] If the j-th type of collision (j=1,2) occurs in the second type of collision, then the actual reaction rate of the j-th type of collision (the number of collision events occurring per unit time and per unit volume) is calculated using the formula... The theoretical reaction rate for the j-th collision is calculated using the formula... Calculations are performed, in which, Indicates the mass of the ion. and The error is the reaction rate error of the j-th collision in the second type of collision.

[0102] In other words, the theoretical reaction rate formula for a target collision can be expressed as: If a collision of the first kind occurs, then x = e; if a collision of the second kind occurs, then x = i.

[0103] The above method can be used to calculate the reaction rate error of each type of collision in the target collision. If the error is less than a preset threshold (e.g., 5%), the calculation accuracy of the parallelized collision model is considered to meet the requirements. If there is a reaction rate error greater than the preset threshold, the collision cross-section of the corresponding collision is corrected, and / or the time step of particle propulsion is corrected (because an excessively large time step may cause collisions that should occur to not occur). For example, if the reaction rate error of the elastic collision between electrons and atoms is greater than the preset threshold, the collision cross-section of the elastic collision between electrons and atoms needs to be corrected, and / or the time step of electron and atom propulsion needs to be corrected.

[0104] To ensure efficient computation of the charge deposition model, directly converting the serial code to parallel processing would lead to read / write conflicts between different computational cores processing the same mesh node, ultimately causing severe distortion of the simulation results. A common approach in existing technologies is for computational cores to wait for each other. If two computational cores need to read / write the same mesh node simultaneously, one core reads / writes first, while the other waits for the former to complete. While this method ensures the accuracy of the simulation results, when numerous conflicts occur between different computational cores, the waiting time between them will slow down the overall efficiency of the charge deposition model.

[0105] Therefore, in this embodiment of the invention, step 1302, which involves depositing the charge of all charged particles onto the corresponding grid nodes of the finest grid in the multigrid based on the charge deposition model, to obtain the charge amount on each grid node, specifically includes the following steps:

[0106] Step 13021, uniformly divide all charged particles into Group, and copy the mesh space of the finest mesh in the multigrid as The system is divided into units to ensure a one-to-one correspondence between GPU cores, charged particle groups, and mesh space; where M represents the number of the first parallel computing tasks corresponding to the charge deposition model, and N represents the number of GPU cores required for each computing task.

[0107] Step 13022: Each GPU core deposits the charge of the charged particles in its corresponding charged particle group into the corresponding grid space using a charge deposition model.

[0108] Step 13023, all The charge amounts in each grid space are accumulated to obtain the charge amount at each grid node.

[0109] Based on the method described above, in actual computation, each GPU core deposits its own charged particle group onto the corresponding grid space using the charge deposition model. At this point, the computations of each GPU core do not conflict. After the computation is complete, the charge amounts in the grid space of each GPU core are summed, thus obtaining an accurate charge distribution while improving computational efficiency. This method can effectively improve the computational efficiency of the charge deposition model when there are many conflicts between different cores.

[0110] In summary, the embodiments of the present invention can greatly improve the simulation efficiency of electric propulsion plasma oscillation simulation models while ensuring simulation accuracy, and can adaptively optimize the adjustment parameters of multiple key links in the simulation process by combining artificial intelligence, thereby reducing the consumption of computing resources.

[0111] Example 2

[0112] This invention also provides an adaptive and efficient simulation system for electric propulsion plasma oscillations. This system is mainly used to execute the adaptive and efficient simulation method for electric propulsion plasma oscillations provided in Embodiment 1 above. The following is a detailed description of the adaptive and efficient simulation system for electric propulsion plasma oscillations provided in this invention.

[0113] Figure 3 A functional block diagram of an adaptive and efficient simulation system for electric propulsion plasma oscillation provided in an embodiment of the present invention is shown below. Figure 3 As shown, the system mainly includes: a construction module 10, a decomposition module 20, a simulation module 30, a processing module 40, and an optimization module 50, wherein:

[0114] Module 10 is used to construct a high-precision numerical simulation model of electric propulsion plasma; the simulation model includes: particle motion model, charge deposition model, electromagnetic field solution model and collision solution model.

[0115] The decomposition module 20 is used to decompose the computational tasks of the target model based on a first set of adjustment parameters, and to decompose the computational tasks of the electromagnetic field solution model based on a second set of adjustment parameters. The first set of adjustment parameters includes: the number of first parallel computing tasks, the number of GPU cores required for each computing task, the workload of each GPU core, and the time step of each type of particle propulsion. The target model includes: a particle motion model, a charge deposition model, and a collision solution model. The second set of adjustment parameters includes: the number of second parallel computing tasks, the number of CPU cores required for each computing task, and the workload of each CPU core.

[0116] The simulation module 30 is used to simulate and solve the simulation model to obtain simulation results. The electromagnetic field solution model is solved using the multigrid method with the third set of adjustment parameters. The third set of adjustment parameters includes: multigrid cycle form, number of multigrid layers, number of smoothing times per layer, multigrid smoothing method, and grid spacing in each direction.

[0117] The processing module 40 is used to extract features from the simulation results using the target neural network model, and to determine the updated first set of adjustment parameters, the updated second set of adjustment parameters, and the updated third set of adjustment parameters based on the feature extraction results. The target neural network model is a model obtained after training with the goal of maximizing simulation efficiency. The feature extraction results include: electron density, electron temperature, elastic collision frequency between electrons and atoms, excitation collision frequency, ionization collision frequency, momentum exchange collision frequency and charge exchange collision frequency between ions and atoms, potential gradient, and the growth rate of plasma oscillation over time for each grid node.

[0118] Optimization module 50 is used to repeatedly call the decomposition module, simulation module, and processing module until the preset simulation cutoff condition is reached.

[0119] This invention provides an adaptive and efficient simulation system for electric propulsion plasma oscillations. The system employs a high-precision numerical simulation model of electric propulsion plasma to ensure the accuracy of the simulation results. Furthermore, it decomposes the computational tasks of the simulation model through various adjustment parameters, i.e., it uses parallel computing to complete all computational tasks, thereby effectively improving simulation efficiency. Specifically, when solving the simulation model, a multi-grid method is used to solve the electromagnetic field model, which further enhances the simulation computational efficiency. In addition, this invention utilizes a target neural network model to process the simulation results, updating all adjustment parameters in the above process, thereby continuously and adaptively optimizing the multiple adjustment parameters used in the simulation, ultimately achieving a technical effect that balances the simulation efficiency and accuracy of the electric propulsion plasma oscillation simulation model.

[0120] Optionally, the simulation module 30 includes:

[0121] The initialization unit is used to initialize the number of particles in the electric propulsion plasma and the mass, charge, velocity, and position of each particle.

[0122] The charge deposition unit is used to deposit the charge of all charged particles onto the corresponding grid nodes of the finest grid in the multigrid based on the charge deposition model, so as to obtain the charge amount on each grid node; wherein, the grid node corresponding to the target charged particle is the 8 vertices of the grid cube closest to it; the target charged particle represents any charged particle among all charged particles.

[0123] The first calculation unit is used to calculate the charge density of the target grid node based on the charge amount of the target grid node and the unit grid volume in the finest grid; wherein, the target grid node represents any grid node in the finest grid.

[0124] The electromagnetic field solving unit is used to solve the electromagnetic field solving model based on the multigrid method and the charge density of each grid node in the finest grid, so as to obtain the potential distribution in the electric propulsion plasma.

[0125] The second calculation unit is used to calculate the force on each particle based on the potential distribution, so as to solve the updated position and velocity of each particle based on the particle motion model.

[0126] The determination unit is used to determine the number of target collisions between particles based on the updated motion velocity and collision solution model of each particle; among which, the target collisions include: elastic collisions between electrons and atoms, excited collisions between electrons and atoms, ionization collisions between electrons and atoms, momentum exchange collisions between ions and atoms, and charge exchange collisions between ions and atoms.

[0127] Optionally, the charge deposition unit is specifically used for:

[0128] All charged particles are uniformly divided into Group, and copy the mesh space of the finest mesh in the multigrid as The system is divided into units to ensure a one-to-one correspondence between GPU cores, charged particle groups, and mesh space; where M represents the number of the first parallel computing tasks corresponding to the charge deposition model, and N represents the number of GPU cores required for each computing task.

[0129] Each GPU core deposits the charge of its corresponding charged particle group into the corresponding grid space using a charge deposition model.

[0130] All The charge amounts in each grid space are accumulated to obtain the charge amount at each grid node.

[0131] Alternatively, the particle motion model is as follows: ;in, Indicates the mass of the particle. Indicates the particle at the 1st The speed of motion at each time step Indicates the particle at the 1st The speed of motion at each time step Indicates the time step of particle propulsion. Indicates the particle at the 1st The force at each time step Indicates the particle at the 1st The position of each time step. Indicates the particle at the 1st The position of each time step.

[0132] Alternatively, the charge deposition model is: ;in, Indicates the grid spacing in the x-direction. Indicates the grid spacing in the y-direction. Indicates the grid spacing in the z-direction. This represents the position coordinates of the particles undergoing charge deposition. Represents the coordinates of the grid nodes. Indicates particles at grid nodes Charge distribution weights on , , , This indicates the number of grid nodes in the x-direction. This indicates the number of grid nodes in the y-direction. This indicates the number of grid nodes in the z-direction. This represents the macro-particle weights.

[0133] Optionally, the electromagnetic field solution model is as follows: A discretized set of equations consisting of objective equations for each grid node; grid node The objective equation is expressed as: ;in, Represents grid nodes The potential on, Represents grid nodes charge density, It represents the vacuum permittivity.

[0134] Optionally, the collision solution model includes the following solution steps:

[0135] Using formulas Calculate the maximum collision frequency of all particles; where, Represents the number density of atoms. This represents the velocity of the i-th particle. , Let i represent the kinetic energy of particle i. This represents the collision cross section where particle i undergoes the j-th collision. This represents the maximum collision frequency of all particles.

[0136] Using formulas Calculate the probability of all particles experiencing an empty collision in the current time step; where, Indicates the time step of particle propulsion. This represents the probability that all particles will have an empty collision within the current time step.

[0137] Using formulas Calculate the number of collisions between particles; where, Represents the total number of particles. This indicates the number of collisions between particles.

[0138] Randomly generated A random number to correspond to A particle that collides.

[0139] if If the k-th particle collides with the first type of collision, then the first type of collision occurs; where, This represents the random number corresponding to the k-th particle that collides; This represents the frequency at which the k-th particle experiences the j-th type of collision.

[0140] if If the k-th particle collides with another particle, then the (j+1)-th collision will occur.

[0141] if If the k-th particle that collides with the other particle has an empty collision, then the collision will occur.

[0142] Optionally, after determining the number of target collisions between particles, the device is also used to:

[0143] Construct a three-dimensional computational domain unit model and initialize the particle parameters in the unit model; the particle parameters include: atomic density, electron density, ion density, electron kinetic energy, and ion kinetic energy.

[0144] Disable the particle motion model and count the number of target collisions between particles within the time step of particle propulsion.

[0145] The reaction rate error for each type of collision is calculated based on the number of collisions and particle parameters in the target collision.

[0146] If it is determined that there is a reaction rate error greater than a preset threshold in the target collision, the collision cross section of the corresponding collision is corrected, and / or the time step of particle propulsion is corrected.

[0147] Optionally, before solving the electromagnetic field solution model using the multigrid method with a third set of adjusted parameters, the device is also used for:

[0148] The discretized system of equations is rewritten as a linear system of equations in the form AS = b; where S represents the array of unknowns, A represents the coefficient matrix of the array of unknowns S, and b represents the array of source terms. ,and .

[0149] Example 3

[0150] See Figure 4 This invention provides an electronic device, which includes a processor 60, a memory 61, a bus 62, and a communication interface 63. The processor 60, the communication interface 63, and the memory 61 are connected via the bus 62. The processor 60 is used to execute executable modules, such as computer programs, stored in the memory 61.

[0151] The memory 61 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 63 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc.

[0152] Bus 62 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 4 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0153] The memory 61 is used to store programs. After receiving an execution instruction, the processor 60 executes the program. The method executed by the apparatus defined by the process disclosed in any of the foregoing embodiments of the present invention can be applied to the processor 60 or implemented by the processor 60.

[0154] Processor 60 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 60 or by instructions in software form. Processor 60 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 61. Processor 60 reads the information in memory 61 and, in conjunction with its hardware, completes the steps of the above method.

[0155] The computer program product of the adaptive and efficient simulation method and system for electric propulsion plasma oscillation provided in this embodiment of the invention includes a computer-readable storage medium storing non-volatile program code executable by a processor. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation, please refer to the method embodiments, which will not be repeated here.

[0156] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0157] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0158] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0159] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," "third," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0160] Furthermore, terms such as "horizontal," "vertical," and "overhanging" do not necessarily imply that a component must be absolutely horizontal or overhanging, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but rather that it can be slightly tilted.

[0161] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0162] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive and efficient simulation method for electric propulsion plasma oscillation, characterized in that, include: Step 110: Construct a high-precision numerical simulation model of electric propulsion plasma; wherein the simulation model includes: a particle motion model, a charge deposition model, an electromagnetic field solution model, and a collision solution model; Step 120: Decompose the computational task of the target model based on a first set of adjustment parameters, and decompose the computational task of the electromagnetic field solution model based on a second set of adjustment parameters; wherein, the first set of adjustment parameters includes: the number of first parallel computing tasks, the number of GPU cores required for each computing task, the workload of each GPU core, and the time step of each type of particle propulsion; the target model includes: the particle motion model, the charge deposition model, and the collision solution model; the second set of adjustment parameters includes: the number of second parallel computing tasks, the number of CPU cores required for each computing task, and the workload of each CPU core; Step 130: Perform simulation and solve the simulation model to obtain simulation results; wherein, the electromagnetic field solution model is solved using a multigrid method with a third set of adjustment parameters; the third set of adjustment parameters includes: multigrid cycle form, number of multigrid layers, number of smoothing times per layer, multigrid smoothing method, and grid spacing in each direction; Step 140: Feature extraction is performed on the simulation results using the target neural network model to determine the updated first set of adjustment parameters, the updated second set of adjustment parameters, and the updated third set of adjustment parameters based on the feature extraction results; wherein, the target neural network model is a model obtained after training with the goal of maximizing simulation efficiency; the feature extraction results include: electron density, electron temperature, elastic collision frequency between electrons and atoms, excitation collision frequency, ionization collision frequency, momentum exchange collision frequency and charge exchange collision frequency between ions and atoms, potential gradient, and plasma oscillation growth rate over time for each grid node; Step 150: Repeat steps 120 to 140 until the preset simulation cutoff condition is reached.

2. The adaptive and efficient simulation method for electric propulsion plasma oscillation according to claim 1, characterized in that, The simulation model is solved to obtain simulation results, including: Initialize the number of particles in the electrically propelled plasma and the mass, charge, velocity, and position of each particle; Based on the charge deposition model, the charge of all charged particles is deposited onto the corresponding grid nodes of the finest grid in the multigrid, thus obtaining the charge amount on each grid node; wherein, the grid node corresponding to the target charged particle is the 8 vertices of the nearest grid cube; the target charged particle represents any charged particle among all the charged particles; The charge density of the target grid node is calculated based on the charge amount of the target grid node and the unit grid volume in the finest grid; wherein, the target grid node represents any grid node in the finest grid; Based on the multigrid method and the charge density of each grid node in the finest grid, the electromagnetic field solution model is solved to obtain the potential distribution in the electric propulsion plasma; The force on each particle is calculated based on the potential distribution, and the updated position and velocity of each particle are solved based on the particle motion model. Based on the updated velocity of each particle and the collision solution model, the number of target collisions between particles is determined; wherein, the target collisions include: elastic collisions between electrons and atoms, excitation collisions between electrons and atoms, ionization collisions between electrons and atoms, momentum exchange collisions between ions and atoms, and charge exchange collisions between ions and atoms.

3. The adaptive and efficient simulation method for electric propulsion plasma oscillation according to claim 2, characterized in that, Based on the charge deposition model, the charge of all charged particles is deposited onto the corresponding grid nodes of the finest grid in the multigrid, resulting in the charge amount at each grid node, including: All the charged particles are uniformly divided into Group, and copy the grid space of the finest grid in the multigrid as The components are arranged in a way that makes a one-to-one correspondence between GPU cores, charged particle groups, and mesh space; where M represents the number of the first parallel computing tasks corresponding to the charge deposition model, and N represents the number of GPU cores required for each computing task. Each GPU core deposits the charge of its corresponding charged particle group into the corresponding grid space through the charge deposition model; All The charge amounts in each grid space are accumulated to obtain the charge amount at each grid node.

4. The adaptive and efficient simulation method for electric propulsion plasma oscillation according to claim 1, characterized in that, The particle motion model is as follows: ;in, Indicates the mass of the particle. Indicates the particle at the 1st The speed of motion at each time step Indicates the particle at the 1st The speed of motion at each time step Indicates the time step of particle propulsion. Indicates the particle at the 1st The force at each time step, Indicates the particle at the 1st The position of each time step. Indicates the particle at the 1st The position of each time step.

5. The adaptive and efficient simulation method for electric propulsion plasma oscillation according to claim 1, characterized in that, The charge deposition model is as follows: ; in, Indicates the grid spacing in the x-direction. Indicates the grid spacing in the y-direction. Indicates the grid spacing in the z-direction. This represents the position coordinates of the particles undergoing charge deposition. Represents the coordinates of the grid nodes. Indicates particles at grid nodes Charge distribution weights on , , , This indicates the number of grid nodes in the x-direction. This indicates the number of grid nodes in the y-direction. This indicates the number of grid nodes in the z-direction. This represents the macro-particle weights.

6. The adaptive and efficient simulation method for electric propulsion plasma oscillation according to claim 5, characterized in that, The electromagnetic field solution model is as follows: A discretized set of equations consisting of objective equations for each grid node; Grid nodes The objective equation is expressed as: ;in, Represents grid nodes The potential on, Represents grid nodes charge density, It represents the vacuum permittivity.

7. The adaptive and efficient simulation method for electric propulsion plasma oscillation according to claim 1, characterized in that, The collision solution model includes the following solution steps: Using formulas Calculate the maximum collision frequency of all particles; where, Represents the number density of atoms. This represents the velocity of the i-th particle. , Let i represent the kinetic energy of particle i. This represents the collision cross section where particle i undergoes the j-th type of collision. This represents the maximum collision frequency of all particles; Using formulas Calculate the probability of all particles experiencing an empty collision in the current time step; where, Indicates the time step of particle propulsion. This represents the probability that all particles will have an empty collision within the current time step; Using formulas Calculate the number of collisions between particles; where, Represents the total number of particles. This indicates the number of collisions between particles; Randomly generated A random number to correspond to One particle colliding; if If the k-th particle collides with the first type of collision, then the first type of collision occurs; where, This represents the random number corresponding to the k-th particle that collides; This represents the frequency at which the k-th particle undergoes the j-th collision. if If the k-th particle collides with another particle, then the (j+1)-th particle will undergo the (j+1)-th collision. if If the k-th particle that collides with the other particle has an empty collision, then the collision will occur.

8. The adaptive and efficient simulation method for electric propulsion plasma oscillation according to claim 7, characterized in that, After determining the number of target collisions between particles, the following is also included: A three-dimensional computational domain unit model is constructed, and the particle parameters in the unit model are initialized; wherein, the particle parameters include: atomic density, electron density, ion density, electron kinetic energy, and ion kinetic energy; Disable the particle motion model and count the number of times the target collision occurs between particles within the time step of particle propulsion; Based on the number of each collision in the target collision and the particle parameters, the reaction rate error of each collision is calculated; If it is determined that there is a reaction rate error greater than a preset threshold in the target collision, the collision cross section of the corresponding collision is corrected, and / or the time step of the particle propulsion is corrected.

9. The adaptive and efficient simulation method for electric propulsion plasma oscillation according to claim 6, characterized in that, Before solving the electromagnetic field model using the multigrid method with a third set of adjusted parameters, the following steps are also included: The discretized system of equations is rewritten as a linear system of equations in the form AS=b; where S represents the array of unknowns, A represents the coefficient matrix of the array of unknowns S, and b represents the source term array. ,and .

10. An adaptive and efficient simulation system for electric propulsion plasma oscillation, characterized in that, include: A construction module is used to build a high-precision numerical simulation model of electric propulsion plasma; wherein, the simulation model includes: a particle motion model, a charge deposition model, an electromagnetic field solution model, and a collision solution model; The decomposition module is used to decompose the computational tasks of the target model based on a first set of adjustment parameters, and to decompose the computational tasks of the electromagnetic field solution model based on a second set of adjustment parameters; wherein, the first set of adjustment parameters includes: the number of first parallel computing tasks, the number of GPU cores required for each computing task, the workload of each GPU core, and the time step of each type of particle propulsion; the target model includes: the particle motion model, the charge deposition model, and the collision solution model; the second set of adjustment parameters includes: the number of second parallel computing tasks, the number of CPU cores required for each computing task, and the workload of each CPU core; The simulation module is used to simulate and solve the simulation model to obtain simulation results; wherein, the electromagnetic field solution model is solved using a multigrid method with a third set of adjustment parameters; the third set of adjustment parameters includes: multigrid cycle form, number of multigrid layers, number of smoothing times per layer, multigrid smoothing method, and grid spacing in each direction; The processing module is used to extract features from the simulation results using a target neural network model, and to determine updated first, second, and third adjustment parameter sets based on the feature extraction results; wherein, the target neural network model is a model obtained after training with the goal of maximizing simulation efficiency; the feature extraction results include: electron density, electron temperature, elastic collision frequency between electrons and atoms, excitation collision frequency, ionization collision frequency, momentum exchange collision frequency and charge exchange collision frequency between ions and atoms, potential gradient, and plasma oscillation growth rate over time for each grid node; The optimization module is used to repeatedly call the decomposition module, the simulation module, and the processing module until the preset simulation cutoff condition is reached.

Images