| The fast growing complexity and intelligence of today's electric drive controlsystem simulation necessitates real-time simulation to face the problems that off-linesimulation can't solve. Based on the analysis of advantages , characteristic andrealization method of real-time simulation, the author propose to design a real-timesimulation platform for electric drive control on dSPACE. The purpose is to create aspecial , functional and integrate platform for the field of electric drive control, whichcan lower the difficulty in use , help the user reuse some frame members and simplifythe operation process of real-time simulation and accelerate the progress of systemdevelopment.From requirement analysis, domain analysis, architecture design to anatomize ,realization and test , all of them can be modeled by UML graph. real-time modeling is aspecialty of the platform, so it's important to add some real-time extends for UML. Inthis paper, a new extends method is proposed, the introduction of QoS is helpful toanalyzing system time resource . Meanwhile, the time model is created, that includetime cell model, time mechanism model and dynamic time model. They can providesome service that satisfy the requirement of time.The other specialty of the platform is integrating intelligent algorithms ascontrollers, for example, fuzzy-PI control algorithm and artificial neural network controlalgorithm. The control theory , structure and the design process of each controller havebeen presented in detail. Then, these controller are simulated in the real-time platform,the result of simulation validate the algorithms effectivity and feasibility rapidly. Thesealgorithms have better characteristic both in dynamic performance and in staticperformance than traditional control strategy.After that, the author evaluate the whole research work objectively, and point outthe creativity and insufficiency. Finally, the prospective for further research is given andnew methods are also proposed in the end.
|
PDF Full Text Request