Identification of a Moss Growth System using an Artificial Neural Network Model
2006
Ushada, M. | Murase, H.
A moss (Rhacomitrium canescens) growth system was identified in order to clarify the causal relationship between input parameters and output parameters utilising artificial neural network (ANN) model. Input parameters were defined as three cardinal temperatures in addition to ambient temperature while output parameters defined as heat unit accumulation, relative rate of growth, leaf area index, moss height, moss mass, and temperature stress factor. The physical properties of moss were determined using a non-destructive imaging method and a conventional destructive method. Some of the data needed for the system identification were obtained from the literature. The model performance was tested and successfully described the relationship between input and output parameters which can be used for moss growth control. A minimum learning error of 6·96x10⁻² was attained at convergence of the ANN training. The most notable experimentally determined values were specific leaf area of 1·498 m2/kg and ground area of 28 mm2/plant in the leaf area index sub-model.
显示更多 [+] 显示较少 [-]