Improved Resonant Frequency Tracking Strategy based on Load Parameter Prediction for Linear Oscillatory Machine

As the key power part of the linear compressor, the linear oscillatory machine (LOM) has been paid more attention by both academia and industry recently for its compact size, high efficiency, low vibration, etc. In practice, the efficiency of LOM is significantly impacted by the changes in load parameters, of which the best working point can be tracked by the resonant frequency tracking control. However, the traditional resonant frequency tracking control with the idea of stroke-current phase suffers some serious problems, like slow tracking speed, low accuracy, large current peak during starting, fluctuate amplitude of piston, and so on. In this paper, a Kalman filter for the mechanical system of LOM is proposed, which avoids the impact of environmental noise and obtains the observed values of load parameters. On this basis, a parameter correction scheme based on gradient descent method is proposed. By calculating the corresponding gradient matrix, iterative adjustment and correction of the covariance matrix in the Kalman filter are realized. The combination of the two realizes the load parameter prediction of the LOM, and on this basis, completes the frequency control of the LOM. Comprehensive simulation results have fully demonstrated the advantages of proposed control strategy compared to those of traditional methods.