Observer Designs For Sensorless Linear Servo Systems

Permanent magnetism linear synchronous motors (PMLSMs) have beenwidely used for servo systems with high speed and high precision, due to theirhigh efficiency, high reliability and no magnetizing current. Motion control ofPMLSMs requires accurate position and velocity signals to realize fieldorientation. In conventional servo systems, mechnical sensors are used for thispurpose. However, these additional sensors increase the costs of the systemand decrease the reliability. Therefore, it has been an attractive prospect ofavoiding position and velocity sensors for the purpose of sensorless servosystems with low cost and high reliability. The thesis focuses on severalrobust estimation methods on sensorless operation for PMLSM.Traditional sliding mode observer has the characteristics of adaptivityand robust to parameters perturbation and external disturbances, butchattering exists in the sliding mode. In order to weaken the chattering, speedis estimated by the second-order sliding mode super-twisting algorithmmodified. This method is of strong robust and fast convergence. The thesispresents two strategies of sensorless operation for PMLSM. Two-time-scaleof singular perturbation is applied to decompose the PMLSM's model in thestationaryαβcoordinate into fast subsystem and slow subsystem. Theposition estimated method is proposed based on sliding mode observer - backelectromagnetic force (SMO-BEMF). And speed is estimated based on second-order sliding mode super-twisting observer in the slow subsystem.Considering the nonlinear terms of PMLSM's model satisfying Lipschitzconditions in the rotationary dq coordinate, the observer design is discussedfor Lipschitz nonlinear systems in the presence of bounded disturbances. Newstable conditions have been found to guarantee observer error dynamicsconvergence expressed by means of LMI. And the position is estimatedthrough SMO-BEMF estimation method.With MATLAB/SIMULINK, a general simulation support is establishedfor PMLSM servo systems controlled by SPWM vector current sluggishtracking, according to PMLSM and vector control principles and computermodeling and simulation technologies. Simulations for estimation strategiesabove are performed on the support. Successful simulation results areobtained, and illustrate the feasibility of observer designs to realize sensorlesscontrol for PMLSM servo systems.