A Kind of Vibratory Isolation Algorithms Based on Neural Network.

Vibration isolation technology makes a significant effect in the high-precision instruments field, however, the anti-interference technology at low-frequency and ultra-low frequency becomes the bottleneck of high-precision instrument development obstructively. The regular vibratory used in oil and gas exploration has a good effect on controlling the interfering signals above 6 Hz, but it doesn't work well under 6 Hz. However, the Low-frequency excitation for hydrocarbon detection become a hotspot. In this paper, a hybrid vibration isolation method is proposed to suppress the interfering signal 6 Hz below and to improve the accuracy of the controllable vibratory excitation signal. A neural network (NN) with unique non-linear approximation capability is adopted to identify the vibration system and a NN predictive controller takes active control for the vibration systems. A simulation model is established using MATLAB/SIMULUNK. The simulation results showed that the proposed NN-based hybird isolation method can suppress the interference signals magnitude down by more than 92 % for 3-6 Hz interference signals, which put forward a novel effective anti-interference method for low-frequency vibration applications.