菊花微弱电信号的神经网络预测

Analysis and forecast of neural networks on weak electrical signals in chrysanthemum

依据菊花(Dendranthema morifolium)植物电信号小波软阈值消噪后的数据,进行了其电信号时间序列的高斯径向基函数(RBF)神经网络预测.菊花植物电信号是一种微弱低频非平稳信号,最大幅值1 093.44μV,最小为-605.35μV,均值-11.94μV;功率谱分布为0-0.2 Hz.该结果说明,利用RBF人工神经网络对植物微弱电信号进行短期预测是可行的,其预测数据可作为以节能为目标依据植物自适应电信号特性建立温室和/或塑料大棚智能自动化控制系统的重要参数.

Taking an electrical signal in the chrysanthemum （Dendranthema morifolium） as the time series and using the Gaussian radial base function （RBF） and a delayed input window chosen at 50, an intelligent RBF forecasting system is set up to forecast the signal after the wavelet soft-threshold de-noised backward, It is obvious that the electrical signal in chrysanthemum is a sort of the weak, unstable and low frequency signal. There is the maximum amplitude at 1 093.44 μV, minimum -605.35 μV, average value -11. 94μV; and below 0.2 Hz at frequency in the chrysanthemum respectively. The result shows that it is feasible to forecast the plant electrical signal for a short period by using RBF neural networks. The forecast data can be used as important preferences for intelligent automatic control systems based on the adaptive characteristic of plants to achieve energy saving on agricultural production in greenhouses and/or plastic lookums.