| Due to uncertain disturbances,especially non-Gaussian stochastic disturbances,it is difficult for the traditional control theory to provide proper control laws to meet the requirements of modern industry.In non-Gaussian systems,the control object is usually the probability density functions(PDFs)and the control law is obtained through optimizing the performance index.The gradient descent was often used to optimize the performance index.However,the gradient descent is complex and time-consuming.Generally,there are no optimal solutions and it is not suitable for complex practical systems.In this paper,Genetic algorithm is applied to optimize the performance indexes and search for the global solution.This paper also makes a preliminary exploration of this optimal control algorithm combined with simulation,and applies the control algorithm to grinding,ORC system,fiber vibration detection.The main research work is listed as follows: 1.PDF control on grinding systemFor one-dimensional single-input single-output nonlinear systems,the output PDFs are estimated by using the kernel density estimation(KDE)method,and then the performance index function is established based on the stochastic distributed control theory of the functional operator model.The established performance index function is optimized by using a properly adjusted genetic algorithm to find the optimal control input.2.PDF control based the Organic Rankine Cycle(ORC)systemFor the two-dimensional state nonlinear systems with uncertain external disturbances,the output PDFs are estimated by the two-dimensional KDE method after choosing appropriate bandwidth matrix and kernel functions.Then,according to the stochastic distributed control theory of the functional operator model,the distance between the output PDF and the target PDF established as a part the performance index function.The established performance index function is optimized by using a properly adjusted genetic algorithm to find the optimal control input.3.PDF control based optical fiber vibration detectionAccording to the optical fiber data collected during trotting on the ground with buried optical fibers,a dynamic model is established,which is used as the output PDF model to be controlled.According to the control objective,the performance index function is established,and the genetic algorithm based optimal control algorithm is used to find the optimal control input of the model to be controlled. |
