A Stroke Estimation Method for Linear Oscillatory Machine Using an Improved Sliding Mode Observer

Abstract

For the linear compressor system driven by a linear oscillatory machine (LOM), the piston stroke is very important to realize the high working efficiency for the whole system. In order to obtain precise and reliable piston stroke signals without using the position sensor, this article proposes a stroke estimation method for the LOM using an improved sliding mode observer (SMO). For the improved SMO, the integral sliding surface is chosen to reduce the chattering phenomenon and the steady-state error of the system, and a novel reaching law is introduced to construct the sliding control law, which can further reduce the influence of the sliding chattering. Due to the piston stroke being obtained by integrating the back electromotive force (EMF) as estimated by the improved SMO, the pure integration saturation problem will occur due to that the measured current and voltage signals always contain some dc components. In order to cope with this problem, a self-adaptive bandpass filter is designed, which can obtain a filtered signal with almost no amplitude and phase deviation by adaptively adjusting its center frequency. Simulation and experimental results confirm that the proposed improved SMO has higher stroke estimation accuracy than the back EMF integration and the conventional SMO.