Condition monitoring of an axial piston pump utilizing the Kalman filter

Abstract

Condition monitoring of a hydraulic pump is an essential process for maximum operational time and pump life longevity. One method of condition monitoring is to estimate parameters characterized by flow losses. Even in a lab environment, direct flow loss measurement may be impractical. Therefore, a method of estimation must be utilized. Filters, such as the Kalman Filter have been implemented in numerous engineering applications to estimate states and parameters. A first order, nonlinear model of the pump discharge pressure, along with several measurements from experimental tests, have been utilized to implement and compare five filters: a pole placement filter (PPF), Kalman Filter (KF), Extended Kalman Filter (EKF), particle filter (PF), and Unscented Kalman Filter (UKF). The discharge pressure, swash plate angle, and a flow loss parameter within the model are estimated using these filters. An eighth order model has also been utilized to test estimation of multiple flow loss parameters (low and high Reynolds number flow losses) simultaneously. The KF, EKF, and PF is utilized for this model, using simulation data. A least squares fit, utilizing the volumetric efficiency of the pump and considering low and high Reynolds number flow losses, has also been calculated and compared to the filter results for the experimental data. Results show that flow losses can be tracked utilizing a simple first or second order model utilizing standard filtering techniques such as the EKF.