Research Of Intelligent PID Control Method In Permanent Magnet Servo System

As the microelectronics, power electronics, sensor technology, permanent magnetic materials and control theory develops, the AC servo system is widely used in CNC,industrial robots,etc. AC servo technology has now become one of the support technologies of industrial automation. PMSM(permanent magnet synchronous motor), as the most important one among AC servo motor, however, is a non-linear charged object.Parameter changes and non-linear characteristics often make the linear constant PID control unsatisfactory in PMSM control.As to these problems, domestic and foreign experts has conduct in-depth study and put forward a variety of solutions in recent years.Solutions include magnetic field trajectory control method and direct torque control method based on electric magnetic field theory and so on.These methods make the system performance improved, but they are still based on accurate mathematical model of the object,some of them need a lot of sensors and observers and therefore complex, and some even can not escape the impact of changes in non-linear motor parameters.With the recent development of intelligent control, many scholars began to introduce intelligent control into PMSM Servo System,aiming at overcoming these shortcomings.Intelligent Control requires no accurate model of an object, and it has non-linear, variable structure, self-learning qualities and characteristics in itself, which help to overcome the impact of the varying parameters and non-linear factors in PMSM Servo System effectively and improving the system performance.Thereby,the intelligent control technology has a vast applications in the permanent magnet AC servo systems.This paper first started with the mathematical model of permanent magnet synchronous motor, and then systematically analysed the vector control theory which is the most widely used at present. Vector control can be achieved for electrical decoupling between the AC-DC axis, with a linear torque control characteristics, have access to a relatively steady output of torque and a wide speed range, which theoretically solved the AC motor speed control problem, making the control of AC motors easy as DC motors', and getting comparable dynamic performance with the DC Governor System.This paper also describes the features and principles of the traditional PID controller. Then we probed into neural network PID control,one of the most widely used intelligent PID control algorithm,to understand the general intelligent PID control process, as well as the problems to be solved.We find that intelligent PID controller is indeed able to overcome the affects of varying parameters and non-linear factors in the PMSM Servo System effectively, and able to improve system performance.And then introduced a PID neural network evolved under the neural network ,which simplifies the structure of PID parameters when using neural networks to learning, and is more suitable for online self-learning control system,and is a great complement to current PID neural network controller.Finally, the paper proposes a PSO-BP neural network control algorithm as to research studying sample space and learn the connection weights of BP neural network according to PSO's utstanding performance, and I also applicated the algorithm to PMSM control system whose controller chip is TI's TMS320F2812 DSP, and thus achieved good results.While the PSO-BP neural network PID controller has a good control effect for PMSM control, the BP neural network has a inevitable defect, that is the necessity of studying sample space. To solve this problem, this paper also proposed SPIDNN controller to control PMSM and also performed experiments on it, although the control performance of this algorithm is no better than a sample-studied PSO-BP neural network, the algorithm can conduct online learning, so this control algorithm is adapted to control systems which can not use PSO-BP neural network control algorithm for it's difficult if not impossible to find space to study, and this control algorithm has played better control effect than the traditional PID control algorithm.