Wavelet Denoising Based On Genetic Algorithm In Attitude Sensor

Posture Angle is the key point for the attitude measurement system, its accuracy will directly affect the weapon system’s ability to hit target. In the complex battlefield environment, there are all kinds of jamming signal, such as acceleration, vibration, electromagnetic and gaussian white noise, etc. Then we considered the theory of wavelet analysis which can analyze signal in time and frequency domain to deal with the noise signal.This paper describes the corresponding circuit design part of the new high precision attitude sensor, and analyzed the vulnerability in system structure for external interference. Then introduces the latest development of the theory of wavelet analysis, and illustrates the theory and method of wavelet denoising in detail. In view of the defect of hard threshold and soft threshold function in traditional denoising method, we proposed a new wavelet threshold function based on exponential function model, and compared the denoising effect of different threshold function.Because the wavelet function is a collection, in order to denoise the signal effectively and ensure that the original signal loss less as far as possible which requires to select suitable wavelet function to decompose and reconstruct the original signal, we both denoise the original signal and the noise signal with same method, then use a coefficient which based on SNR to evaluate the applicability of various wavelet functions.Finally, in consideration of the multi-scale and multi-resolution characteristics of wavelet transform, we put forward a method which combined with generalized cross validation in wavelet theory and the advantage of global optimization in genetic algorithm to solve the optimal threshold value of each scale so that value can be adaptive adjustment, and apply this method to the signal distortion caused by acceleration in obliquity sensor. For the output voltage signal of angle sensor, the experimental result shows that compared with the traditional wavelet thresholding denoising method, the proposed method has more effective in signal distortion inhibition, and the oscillation of signal voltage does not exceed the range of IV, the maximum oscillation peak voltage also has been reduced from 5V to 3.5V, and the processing of noise detail for the signal is still more useful.