Research on Construction Method of Wavelet Telemetry Data with Improved Threshold

Abstract

Abstract In order to strengthen the applicability of data denoising algorithm, this thesis study the common telemetry data denoising algorithm based on the data of engine speed, flight space speed, cabin temperature and humidity, and establishes the evaluation model of error square sum and curve similarity to evaluate the denoising performance. Experiments show that the polynomial fitting has the greatest denoising error and slow convergence speed. The five-point cubic smoothing has the smallest overall denoising error, the median filtering algorithm can change the effect of smoothing effect by adjust it's moothing window, but ignores the authenticity of data. Therefore, the above three data denoising algorithms do not meet the requirements of telemetry data processing. In this thesis, an improved threshold function is proposed which effectively improves the data jump and excessive smoothing and reduce the denoising accuracy compared with the traditional thresholding function in order to makes the measured value closer to the true value. The algorithm is applied to the noise processing of four kinds of telemetry data, the results show that the denoising accuracy is improved significantly compared with the other three algorithms, which makes the measured value closer to the true value to reflect the changing trend of the original measurement data more truthfully, and the curve similarity is improved significantly, which are all above 80%.