Experimental investigation into the state estimation of a forestry crane using the unscented Kalman filter and a multiphysics model

Abstract

To increase productivity, reduce energy use, and minimize unplanned maintenance, manufacturers of heavy machinery must instrument their products. As explained in the literature, state and parameter estimators can successfully integrate machine sensor signals with simulation results from computational models. This leads to comparable or improved observations even when fewer sensors are being used. This study introduces a state observer based on the unscented Kalman filter for the coupled mechanical and hydraulic systems. The resulting reality-driven simulation procedure is applied to a hydraulically actuated forestry crane that has been instrumented to provide the necessary sensor information. This study analyzes the performance of state observer in four different scenarios and recommends an optimal sensor configuration for the application. Estimation accuracy of observer in the simulation of the mechanics and hydraulics components is evaluated using the percent normalized root mean square error (PN-RMSE) and 95% confidence interval.