Sensorless Control of a PMSM Based on an Adaptive Sliding Mode Observer

Abstract

This thesis proposes new strategies to improve the speed control performance of the sensorless field oriented control (FOC) of the permanent magnet synchronous motor (PMSM) over the wide speed range based on the sliding mode observer (SMO). There are two major approaches proposed to escalate the sensorless speed control performance in this thesis. The first approach focuses on tuning the gain of the PI controller of the sensorless control and the second approach focuses on improvements of the SMO for the sensorless control to work over a wide speed range. Thus far, many researches focusing on sensorless control of the FOC have been proposed to realize the requirement of the high-speed control performance of the PMSM, because the FOC can provide faster and more robust speed control. To this end, PID controller has been widely used for the FOC algorithm. A simple PI controller allows to achieve very high performances when tuned optimally. However, it is a tough task to find the PI gain value using the conventional analytical method in a complex system such as speed control of the PMSM, and it takes tremendous mathematical process. The present study proposes the usage of genetic algorithm (GA) for tuning the six gains of the three PI controllers for the FOC algorithm to obtain high-speed control performance of the PMSM. The GA can provide fast and accurate result for deciding the PI gain because the search area is on the broad area. Furthermore, the GA is very effective for the complex system, and it can solve multivariable problem simultaneously. In sensorless speed control of the FOC of a PMSM, rotor angular position information is needed to achieve high control performance in addition to the PI controller gains. There are many sensorless control methods for the PMSM that were already developed by the researcher such as Luenberger observer, Kalman filter, etc. In the present study, an SMO is chosen as the main observer for the sensorless control of the PMSM among various observer types. However, due to some issues such as the operational speed range, the SMO is only suitable for medium to the high-speed range. In particular, the chattering problem becomes the major drawback of the SMO. The speed and rotor angular position become deteriorated because of the chattering problem, especially in the low-speed area. Thus, a thorough investigation on the conventional SMO was done by analyzing the simulations and experimental results to solve the chattering problem and to improve its estimation performance. The present study proposes and implements an adaptive observer gain for the SMO and a cascade low-pass filter (LPF) with a variable cut-off frequency (VCF), in order to decrease the chattering problem and to improve the sensorless control performance. The adaptive observer gain for the SMO was designed based on the Lyapunov stability method and applied to lessen the chattering problem. The cascade LPF with a VCF was proposed to reduce the chattering problem and to strengthen the filtering capability in the SMO. A variable phase delay compensation method was also implemented to counterbalance the phase shift caused by the filter over the wide operational speed range of the variable speed drive applications. The sensorless speed control of the PMSM based on the proposed SMO was implemented on a 32-bit fixed-point DSP. The experimental results confirm the effectiveness of the proposed sensorless control method. The proposed SMO was able to estimate the angular position and speed of the rotor precisely over the wide speed range which leads to a good sensorless control performance. Moreover, the good dynamic performance of the sensorless control of the PMSM was obtained even when the PMSM operates at 0.005 p.u. of the rated speed value.