Robust and sensorless control of linear switched reluctance motors

Abstract

Linear Switched Reluctance Motor (LSRM) combines the features of Switched Reluctance Motor (SRM) and linear motor; it has certain distinct advantages in high speed and precision positioning applications. As a type of SRM, LSRM has the advantages of simple structures and low cost. It also has the capabilities of withstanding harsh environments, fault tolerance characteristics and maintenance free operation. As a type of linear motor, LSRM has a direct-drive mechanism which enables quick response, high sensitivity and precise position tracking. Moreover, LSRM can reduce the overall installation space, due to its straight forward and small size structure. On the other hand, there is a major disadvantage in LSRM. The magnetic circuit of an LSRM is inherently nonlinear and the real-time position information is required for discrete commutation and control. This makes inconvenient in high-precision position control. As a result, a high precision position sensor needs to be installed. This not only increases complexity and cost to the system, but also reduces the reliability of the drive system. The ultimate objective of this project is to investigate and propose an effective robust controller and a novel sensorless control scheme for a high-performance LSRM. In this way, the expensive position sensor can be eliminated, and the LSRM can enlarge its scope of applications with the overall cost reduced by as mush as 40%. To achieve this target, the first step is to fully understand the behaviours of the LSRM. For this matter, detailed and accurate magnetic characteristics are obtained by using a novel measurement method which incorporates an online estimation of winding resistor. After investigating various modeling techniques, a full-order model and a two-time-scale based reduced-order model are proposed for the purpose of simulations, controller design and position estimation. By observing the responses for sinusoidal waveforms under different frequencies, the effectiveness of the reduced-order model is verified. On the basis of the reduced-order model, a Self-Tuning Regulator (STR) for high precision position control of the LSRM is developed. Simulations and experimental results demonstrate that, under the control of the proposed STR, the LSRM drive system can achieve a high-precision position tracking. The proposed control method has almost no influence under the variations of system parameters, external load disturbances and saturation effects of power drivers. The next research step is to review the passivity theory and access its suitability for robust control of LSRM. After that, a full-order model based Passivity-Based Controller (PBC) and a reduced-order model based PBC are designed for the LSRM drive system, respectively. To improve the precision of position tracking, two mechanisms (i.e. the robust and adaptive mechanisms) are incorporated into the controller. Both algorithms guarantee the global stability of the system and the insensitivity to system uncertainties and external disturbances. The third stage of this project is to investigate and propose a continuous position estimation scheme for the LSRM. For this purpose, a novel position estimation scheme which can operate down to zero speed is developed. The scheme uses diagnostic current injection into an unenergized phase to detect the motor position. Also, a new current integration index is proposed to quantify the measurement of current waveform. Results of hardware implementation demonstrate the effectiveness of the proposed estimation scheme for the LSRM. Finally, based on the previous knowledge of robust control and position estimation, a system design of sensorless position control for LSRMs is proposed. In this research stage, the main considerations for a sensorless control system are presented firstly and the excitation scheme for power drive and position estimation is discussed and analyzed. And then a sensitivity based optimal data selection scheme is proposed to calculate the relative position in a pole pitch by using the relative positions in a pole width. At last, a novel predictive filter and a method of jump detecting of regions are proposed to process the estimated position signals and achieve the estimation of absolute position. The effectiveness and feasibility of the proposed scheme are confirmed by the hardware implementation of the sensorless position control system on the proposed LSRM.