Application Study On Difference Mind Evolutionary Algorithm In Generalized Predictive Control And High-speed Electric Multiple Unit Trains Braking System

Generalized Predictive Control(GPC)is an efficient control method it applied to linear system.With strong adaptability and robustness,it is used widely in industry.Its algorithms improvement and application to nonlinear system have been studied.High-speed Electric Multiple Unit(EMU)trains are used widely in the world.Because of its increasing speed the study on EMU trains breaking system is more and more significant.The Automatic Train Operation(ATO)used in EMU trains will be a key step to the development and intelligentization of the high-speed railway.So studying the ATO of the EMU breaking system is of momentous current significance and application prospect.Mind Evolutionary Algorithm(MEA)which imitates the evolution process of human thinking is an advanced intelligent optimization algorithm.It has fast convergence speed and will not degenerate.It is widely applied in function optimization,model identification and image processing.In this paper,the theory of Mind Evolutionary Algorithm(MEA)is perfected and its algorithm is revised helps mind evolutionary idea succeeding in improving GPC and building model of EMU breaking system.The GPC based on the model is put forward.The primary contents of the paper include the following contents:1.Pefection and improvement of the theory of MEA1)To analyze the transfer probabilities of individual and subpopulation and prove that MEA is almost sure convergent.2)To debate the mind contents and their influence on the convergence and velocity of MEA.And the internal reflection mechanism of MEA and propose structure frame of MEA based on subpopulation reflection will be elaborated.3)To advance Difference Mind Evolutionary Algorithm(DMEA)based on subpopulation information sharing mechanism.The sub-population whose fitness value is maximum participates directly in evolving in the next generation,and the rest shares information according to fitness values.All subpopulations are sole.The almost sure convergence of DMEA is proved strictly.Five representative functions are simulated and studied.2.The application of DMEA in GPC1)To present a scheme that DMEA tunes adaptively parameters of GPC.The value domain of parameters constitutes the solution space of DMEA.The cost function of GPC,the maximum value of the system output and its decay speed constitute the fitness function of DMEA.DMEA adjusts constantly parameters online.Minimum phase system and time-varying model are simulated and studied.2)To build a kind of adaptive GPC of nonlinear system based on which DMEA revises error.The author will identify nonlinear system with Controlled Auto-Regressive Integrated Moving Average(CARIMA)model that describes line system,and discover there is the error that is caused by the nonlinearity of the system.So the non-line system is considered to linear subsystem and nonlinear subsystem.DLS identifies CARIMA model that describes linear system.DMEA optimizes the output of the nonlinear subsystem.The sum of two outputs is the output of nonlinear system.The differential pressure system of.import and export that is a control circle of the coal-pulverizing system with ball mill is simulated and studied.3)To put forward a kind of GPC with some constraint based on the DMEA.DMEA acts as an optimization way and find the input of the system with some constraint for GPC.The reheat steam temperature system in power plant is simulated and studied.3.GPC of EMU trains breaking system based on DMEA1)To propose the Hammerstein model of EMU train braking system considered to its nonlinearity and importance in ATO.It is a kinetic model,accord with control law.Firstly,the working process of EMU train brakingsystem is introduced according to the transport mechanism of the braking instruction.Secondly,the author uses the non-line function to depict the braking characteristics table and uses the delay system to describe the delay characteristics of the braking instruction transmission and the braking controller working.The first-order linear system represents the feedback process of the break.The other first-order linear system describes the remission process of EMU train acceleration impulse.The three parts constitute the Hammerstein model.Then,the way that DMEA identifies parameters of the model is introduced.Finally,simulation results of China Railway High-speed 2(CRH2)is simulated and studied.2)To apply two steps GPC to EMU train braking system according to the Hammerstein model.Intermediate variable is revised in order to make it correspond to gears.The stability of controller lies on only GPC.CRH2 is simulated and studied for comparison two steps GPC with PID.The primary innovation achievements of the paper include the following contents:1)To analyze transfer probabilities of individual and subpopulation;to prove that MEA is almost sure convergence.2)To elaborate the internal reflection mechanism of MEA and put forward DMEA based on subpopulation information sharing mechanism.3)To present a scheme that DMEA tunes adaptively parameters of GPC.4)To build a kind of adaptive GPC of nonlinear system based on which DMEA revises error.5)To advance a kind of GPC with some constraint based on the DMEA.6)To propose the Hammerstein model of EMU train braking system and introduce the way that DMEA identifies parameters of the model.7)To apply two steps GPC to EMU train braking system according to the Hammerstein model.