Modeling greenhouse temperature by means of PLSR and BPNN

To achieve the precise control of the greenhouse, further enhance energy efficiency, it is necessary to clearly understand the greenhouse temperature system's characteristics. This work is concerned with the greenhouse temperature system in the typical summer climate under east-central China conditions. Firstly, fuzzy control algorithm is used to obtain reasonable greenhouse inside temperature range on the Android client. Secondly, two modeling approaches which are partial least-square regression (PLSR) and back-propagation neural network (BPNN) are respectively adopted to predict greenhouse inside temperature based on the environmental information obtained including outside temperature, humidity(inside and outside), solar radiation,wind speed and direction. At last, the predicted inside temperature value is compared with the measured value, and the experiment results show that the prediction accuracy of BPNN is better than PLSR around the premise of both methods achieving the ideal effects.