On Pole and L2-Sensitivity Minimization Using Gradient Decent Algorithm for the Finite Precision Digital Filter Implementations (Draft)

• Main Authors: TSIA, WEN-SHIANG, 蔡文翔
• Other Authors: KO, HSIEN-JU
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/02274828597972498391

碩士 === 亞洲大學 === 資訊工程學系碩士在職專班 === 105 === This thesisaims to develop an infinite impulse filter synthesis method by using gradient decent algorithm for minimizing L2-sensitivity measure subject to minimal pole sensitivity measure. Based on the derived result of the so called sparse normal-form filter realization, the optimal filter structure can be obtained by considering the L2-sensitivity minimization problem subject to the pole-sensitivity function which is summing up all unweighted pole sensitivity measure. By the technique of trace inequalities, a mathematically tractable cost function can be minimized by using gradient decent algorithm to adjust the degree of freedom in the cost function for the minimization problem. Finally, numerical examples are performed to illustrate the effectiveness of the proposed approach.