Model parameter parallel simulation optimization method and device based on proxy model

Abstract

The invention discloses a model parameter parallel simulation optimization method and device based on an agent model, and relates to the field of simulation optimizing.The method comprises the steps that a plurality of different parameter combinations are input into a plurality of simulation environments respectively, and each simulation environment executes a simulation process based on one parameter combination and outputs a corresponding simulation result; a plurality of simulation environments run in parallel; adding the parameter combination of the current iteration process and the corresponding simulation result into the accumulated data set, and training by using the data in the accumulated data set to obtain an agent model; generating a plurality of parameter combinations of the next iteration by using the agent model and the optimization algorithm; according to the method, the proxy model is utilized, so that the optimization algorithm can quickly realize iteration, meanwhile, a parallel simulation mechanism is provided, and the total simulation execution frequency can be reduced, so that the time consumed in the simulation process is greatly reduced, the parameter optimization time consumption is reduced, and the parameter optimization efficiency is improved.

Description

technical field [0001] The invention relates to the field of simulation optimization, in particular to a method and device for parallel simulation optimization of model parameters based on an agent model. Background technique [0002] Simulation optimization is an important part of industrial intelligent design. Its core link is to use optimization algorithms to find a set of input parameters of the simulation environment under the constraint conditions, so that one or some output performances of the simulation environment can be optimized, that is, Optimize the model parameters. [0003] The current parameter optimization process for the simulation environment is as follows: input the parameter combination into the simulation environment, the simulation environment executes the simulation process based on the input parameter combination and outputs the corresponding simulation results, and then inputs the simulation results to the optimization algorithm for iteration to for...

AI Technical Summary

Problems solved by technology

For the black-box optimization process, meta-heuristic optimization methods (including genetic algorithm, ant colony algorithm and particle swarm algorithm, etc.) can be used as optimization algorithms for optimization, but there are following defects in using meta-heuristic optimization methods as optimization algorithms: , a simulation process takes a long time (greater than several hours or days, etc.), and the iterative process of the meta-heuristic optimization method requires a large number of fitness evaluation processes, which will greatly increase the running time of the entire parameter optimization process. even unacceptable

Images