CAJ | 학술논문

针对传统时频原子在雷达信号分解中计算复杂度比较高的问题，提出了一种基于免疫克隆选择算法的时频原子快速分解方法。首先将Chirplet原子库分解为小原子库，然后并行地在每一个小原子库中搜索最佳原子，搜索过程建模为多参数寻优问题，通过免疫克隆选择算法的克隆、变异、记忆、替换等操作求解最优值，最后比较每一个小原子库中的最佳原子，将相似度最大的原子作为分解的最佳原子。仿真实验表明，该方法能够用较少的时频原子表示信号，在大幅减少时频原子搜索时间的同时，有效地抑制了噪声和交叉项的干扰。
침대전통시빈원자재뢰체신호분해중계산복잡도비교고적문제，제출료일충기우면역극륭선택산법적시빈원자쾌속분해방법。수선장Chirplet원자고분해위소원자고，연후병행지재매일개소원자고중수색최가원자，수색과정건모위다삼수심우문제，통과면역극륭선택산법적극륭、변이、기억、체환등조작구해최우치，최후비교매일개소원자고중적최가원자，장상사도최대적원자작위분해적최가원자。방진실험표명，해방법능구용교소적시빈원자표시신호，재대폭감소시빈원자수색시간적동시，유효지억제료조성화교차항적간우。
In view of high computational complexity in traditional time-frequency atom decomposition of ra-dar signals,a fast decomposition method based on immune clonal selection algorithm is proposed. Firstly, Chirplet atom dictionary is divided into small atom dictionaries. Then, the best atom is concurrently searched in every small atom dictionary and the searching process is built as an optimization problem, which is solved through cloning,mutation,memory and substitution in immune clonal selection algorithm. Finally,the best atom in every small atom dictionary is compared and the most similar atom is selected as the best decomposition atom. Simulation indicates that this method can extract the feature of signals through few atoms and it decreases the computational time while effectively suppressing noise and time-fre-quency cross-terms.