Research Of Adaptive Control System For Spinning Machine Based On Neural Network

With the development of textile industry in China, the demands of top-gradespinning machine is becoming larger and larger, it will cost a lot of money to importtop-grade spinning machine. It requires us to develop top-grade spinning machine, en-hances rapidly the performance cost ration of spinning machine. The spinning machinecontrol system is nonlinear and indefinite so that conventional PID method can't get agood control result in spinning machine. But neural network technology is applied tospinning machine, which gives a new method to accomplish better control of spinningmachine.Due to the control problem in spinning machine control, the neural network PIDcontrol is applied to the spinning machine control system, in order to get a bettercontrol result and adapt top-grade spinning machine. On the basis of the analysis ofspinning machine control system principle, this research takes BLDCM as the mainmotor, taking the place of the tradition model of AC motor plus speed regulation viafrequency conversion, in order to improve the e?ciency and e?ectiveness of energysaving system of spinning machine.The intelligent control method based on RBF neural network is introduced onthe basis of the analysis of conventional spinning machine control system. A new im-proved arithmetic is proposed in allusion to the defect of familiar RBF neural networkarithmetic. In the proposed method, the reference model is constructed by using aninteractive human-machine platform which is designed based on MATLAB neural net-work toolbox to identify the main motor. Then, then, RBF neural network identifiesthe main motor by on-line means and acquired on-line tuning information of controllerparameters, and implements self-learning of controller's parameters by single neuroncontroller, thus achieve on-line regulation of controller's parameters. Because MAT-LAB -based RBF realization method has such advantages as self-adapting networkarchitecture and non-artificially selected initial values, the random of network trainingis decreased. The results show that with the proposed method, and higher controlprecision and robustness is obtained.On the basis of theoretic analysis and simulation research,this paper explores thedesign of intelligent control system, in which PLC and DSP are utilized as the corecontroller. This paper expounds the composition and function of the hardware of thesystem, and the design method of the software architecture of the system.