分数阶傅里叶变换自适应滤波算法及应用-物理电子学专业论文.docx

II II Abstract The research of adaptive filtering algorithms is one of most active subjects in the modern signal processing. The adaptive filtering algorithms are applied widely in the areas such as: communication, radar, sonar, biomedical engineering and so on. The objective of researchers is to find an adaptive algorithm that reduces convergence time and computational complexity. The LMS algorithm converges slowly, so its applications are limited. Against the shortage of LMS, the transform domain LMS algorithm is proposed. New approach of transform domain adaptive filtering algorithm is exploited by applying with the fractional Fourier transform. The fractional Fourier transform adaptive filtering algorithm and its application is present in this paper. The main contents are summarized as follows: The principle of adaptive filtering algorithm and its improved algorithms are summarized. Based on pretty performance of fractional Fourier transform, discrete fractional cosine transform adaptive filtering algorithm is proposed. The solution of optimal fractional order is to establish the relation curve between misadjustment and fractional order. Derivation and realization of discrete cosine transform adaptive filtering algorithm are elaborated in this paper. In the system identification model, the EVR and the characteristics of the learning curve prove the superiority of this algorithm. Compared to the LMS algorithm, block LMS (BLMS) algorithm speeds up obviously. Combining transform domain LMS algorithm and BLMS algorithm, the transform domain block LMS adaptive filter algorithm is proposed. The orthogonal transforms such as: DCT, DFRFT, DFRCT have been used in the algorithm. So the DCT-BLMS algorithm, DFRFT-BLMS algorithm, DFRCT-BLMS algorithm are presented. The simulations show that: transform domain block adaptive filtering algorithms performance better than block adaptive filtering algorithm. The discrete fractional cosine transform adaptive filtering algorithm was