Research On Parameter Identification Of PMSM And Application In Control

Permanent magnet synchronous motor (PMSM) has a simple structure, small size, high power density, high efficiency, fast response, reliable operation and many other advantages. It is widely used in high-performance and high-precision servo systems, especially for CNC lathe, aerospace equipment and other occasions. CNC lathe, winding equipment and some others have special requirements on the servo drives. Their diameter of load and moment of inertia usually change with the manufacturing process, this may affect the accuracy and stability of these kinds of equipment. The off-line identification of motor parameters before running is necessary for those high-precision equipments, then the corresponding controller can be properly initialized; and the parameters of the controller should be modified in time, when moment of inertia is changed. Nowadays, identifying the parameters of motor on line, then modifying the controllers is a hot issue of servo control technology.Currently, there are many parameter identification methods, but most of them are complex, difficult to achieve on and the identification accuracy is not high enough. To solve these problems, this paper proposes combining off-line identification and on-line identification together. First, the process of off-line parameter identification is researched, it is easy to control; then two kinds of on-line identification methods are researched and combined according to their advantages, finally, a more effective and less computational identification method is formed. Meanwhile, the above methods are simulated and validated, and then the application of parameter identification is further discussed. In the study, a mathematical model and a control system model of the motor are established, the parameters needed to be identified are pointed out; identify the parameters of motor off line through positional motion, even motion and accelerated motion, the results are used to initialize the controllers; the principles of least square method and gradient correction method are discussed, identification models are established, the inertia of motor is identified on line, according which the parameters of speed-loop's PID controller can be changed automatically. Through theoretical study and practical application to all the above identification methods, it is known that: the off-line identification method is fast and simple; least square method with forgetting factor can identify system parameters quickly at the beginning of running, But the tracking ability is not good enough later; however, The gradient correction method is just the opposite, at the beginning of running, the parameters identified fluctuate severely and slow in terms of speed, but it can track new value quickly when parameter is changed. Based on the study above, a new on-line identification method comes out, which combines the least squares identification method with the gradient correction method, highlights the advantages of each method and works well. At the end, the main reasons for generating the identification error in practical system are analyzed: the inverter has nonlinear factors. It is raised that, the dead time compensation method can be used to make up for them.