Exploratory Development of Parameters Identify Based on Reformative PSO Algorithm in DTC System

摘要

In order to improve the stability and dependability of the Direct Torque Control (DTC)system in low speed state of asynchronous dynamo, a kind of reformative Particle Swarm Optimization(PSO)algorithm, which optimizes the Wavelet Neural Network(WNN),is used for observing parameters which contain rev, magnetic likage and stator resistance, insteading of conventional velocity generator and magnetic likage sight. Relativing to the problems, such as easily sinking into the part optimal value, low speed in astringency, short exactitude in precision and so on, this reformative method divides the optimize particles into two teams, one of them adopts part Particle Swarm algorithm which adhibits compressibility factor, the others adopts global Particle Swarm algorithm which adhibits inertial weighting, both the local value and the global value can be compromised for improving the astringency speed and precision through combining these two teams. At the same time the network can constitute the link between wavelet transform and network quotiety, which is used for observing parameters. Passed by the validate of experimental result, this kind of reformative PSO algorithm which optimal WNN can fast converge, and it has good exactitude in precision. So it can make the asynchronous dynamo keep high capability, especially in low speed state.