Studies On The Characteristic Analysis And Processing Of The Plant Electrical Signal

From 2004 to 2006, this study firstly monitored the electrical signals in 5 species in compositae (Dendranthema morifolium, Dahlia pinnata, Senecio cruentus, Bellis perennis and Chrysanthemum coronarium) and the phloem, cambium, xylem of Osmanthus Jragrans in Oleaceae systematically. The obtained data of the plant electrical signal was processed by using the method of time domain analysis, frequency domain analysis, wavelet analysis, ARIMA model and neural network, and the plant→electrical signal obtaining→computer information data processing→plant electrical signal characteristic analyzing and forecasting system was set up primarily. The significant achievements are as follows:1. The facts that the amplitudes of the electrical signals in 5 compositae and the phloem, cambium, xylem of Osmanthus jragrans are mainly inμV magnitude, the mean-square values are between 10^-5-1mV² approximately, and the power spectrum distributions are below 1Hz, which determine that the plant electrical signal is a sort of low frequency micro-signals are confirmed farther. These are the important basis for the further study.2. For the first time, the relation model of the form y = f(m,t)c + g(m,t) was set up based on the studies on the relationship between the electrical signals in the cambium and the electrical signals in the phloem and the xylem. The biological significance of the model was discussed simply, and the mathematic logical token about the lengthening and thickening of the stem of the woody plants was added newly in botany.3. The db3 wavelet was chosen as the basis function to decompose the plant electrical signal in 3 scales and reconstruct the coefficient in each layer. With the scale changing, the detail components of the plant electrical signal are displayed evidently. The detail coefficients are relative to the noise in the signal. The low frequency scales (approximate coefficients) reflect the main character of the plant electrical signal, and the high frequency scales (detail coefficients) reflect the jump spot of the plant electrical signal. Through analyzing the detail coefficients, we can detect the arisen position of the action wave accurately. The plant electrical signal was also de-noised by the wavelet soft-threshold de-noising method, and the effect of the de-noising is good.4. The good fitting and forecasting effect of the plant electrical signal were obtained fistly by using the ARIMA model. The time series fitting and forecasting of the plant electrical signal by using of Radial Basis Funtion (RBF) neural network was also carried out primarily; the result shows that the effect of the examination of inner coincidence is very good, and the fitting errors are quite small. The method of improving the effect of the extrapolation (forecasting) was also discussed primarily.