Research On Chaotic Phenomenon Of Underwater Vehicle's Thruster Motor Based On The Wavelet Neural Network

The brushless direct current thruster motor system is replacing the brush direct current motor system as the thruster motor of underwater vehicles. The underwater vehicle requires the thruster motor has high reliability, so it is important to research the brushless thruster motor system's reliability. The chaotic phenomenon of brushless thruster motor system is one reason of unreliability of motor system. This thesis researches the brushless thruster motor system chaotic model based on the brushless DC motor model, analyzes the chaotic features when the brushless thruster motor system run a moment with no load then lost its power. Then it analyzes the model and dynamic features of the brushless thruster motor system when it runs with no load, and gets the reasons of chaotic phenomenon when brushless thruster motor system run with no load. In this aspect this paper has developed the chaos theory.This thesis analyzes the existing phase-space reconstruction using one dimension time series, and points out its shortcoming. So it presents modified phase-space reconstruction that it increases computing speed and decreases the computing process against the correlation dimension and time delay. But the existing phase-space reconstruction using one dimension time series has redundant information no matter what any development, it can only use in the chaotic time series. For this problem, the paper put the wavelet analysis into the phase-space reconstruction, uses the wavelet phase-space reconstruction to reconstruct different time series. The simulate result proves that wavelet phase-space reconstruction can not only reconstruct original system's chaotic features, but also reconstruct original periodic features, quasi-periodic features, multi-periodic features etc, it improves the existing phase-space reconstruction's shortcoming.Finally, this thesis presents establishing wavelet neural network to analyze brushless thruster motor system's working state based on the wavelet phase-space reconstruction analysis. It can use the wavelet phase-space reconstruction to get original system's feature, then produce the control field that is suited to the this state using the neural network's memory. The simulate result indicates that this program's structure is simple, the operating speed is quick, the control field is precision. The control field can be used in the control system with the DSP or chip microcomputer. The method can supply the key technology preparation for chaotic control and anti-control.