Short-term prediction of disturbance inputs of the greenhouse system: A case study on outside humidity (ICCAS 2015)

Abstract

Greenhouse microclimate control system is a typical hybrid system, in which the discrete (or logical) variables and continuous variables interact. The existence of outside measurable but uncontrollable disturbance inputs including outside temperature, humidity, wind speed, solar radiation, and et al, makes the control problem of greenhouse microclimate a challenging one and some conventional control methods not applicable, which motivates us to investigate the prediction problem of disturbance inputs of greenhouse control system. First, grey prediction model GM (1, 1) and time series model ARIMA (p, d, q) are adopted to predict the outside humidity over the next four hours, respectively. Then, considering the nonstationary property of the humidity sequence, wavelet analysis theory is applied to decompose the humidity sequence into different scales in order to reduce the randomness of the original sequence. Furthermore, the low frequency signal and high frequency one are predicted by GM (1, 1) and time series model ARIMA (p, d, q), respectively. Finally, simulation studies are carried out to compare the three prediction methods.