PREDICTION MODEL OF PITCH ANGLE OF GREENHOUSE ELECTRIC TRACTORS BASED ON TIME SERIES ANALYSIS

Abstract

The pitch angle of greenhouse tractors changes when operating on rough soil pavement. As a result, the feedback signal lags behind the tractor motion attitude signal, thereby affecting the real-time control of tilling depth. In this study, a pitch angle prediction model of greenhouse electric tractor was proposed based on extended Kalman filter (EKF) and time series analysis to improve the dynamic response speed of tilling depth regulation by providing predictive information for advance control. EKF was used to track the tilling depth of greenhouse electric tractor in real time, and an auto-regressive moving average model (ARMA) was established for the obtained time series data. ARMA (2, 1) was designed as the pitch angle prediction model of greenhouse electric tractors by constructing a simulation model. Inertia measurement unit (IMU) of tractor was used to construct the extended Kalman estimation model of the pitch angle. Actual vehicle tests were carried out under different working conditions. Results show that the estimated values obtained under two operating conditions have a high correlation with the measured values, with correlation coefficients(R) of 0.9504 and 0.9734, root mean square error (RMSE) of 0.2355 and 0.2173, and maximum absolute error (MAE) of 0.1929 and 0.1703, respectively. And ,the MAE and the RMSE of the predicted and measured values of ARMA (2,1) model approximately have the same value under the two conditions, with with the R of 0.9665 and 0.9755, the RMSE of 0.2002 and 0.1812, and the MAE of 0.1578 and 0.1387, respectively. The effectiveness of ARMA (2, 1) as the pitch angle estimation and prediction model of greenhouse electric tractors is verified. This study provides theoretical reference for designing the control law of tilling depth stability in subsequent greenhouse operation. Keywords: Time series, prediction, pitch angle,electric tractor