基于小生境遗传算法的分布式OS-CFAR检测系统优化与性能分

利用小生境遗传算法, 对不同检测窗长度和检测信噪比的三传感器分布式O S2CFAR 检测系统进行了优化设计, 给出了一组针对不同检测环境与融合方式的搜索结果。分析表明, 对于非一致环境下分布式O S2CFAR 检测系统, 小生境遗传算法是一种良好的优化算法。利用搜索结果, 研究了不同融合方式下环境变化对分布式O S2CFAR 检测系统的性能影响, 结果表明,“或”融合对检测环境的非一致变化具有较强的鲁棒性, 而“3 选2”融合和“与”融合对检测环境的变化比较敏感。
关键词: 遗传算法; 小生境; 非一致环境; 分布式恒虚警检测; 数据融合
Abstract: In th is paper, niche genet ic algo rithm s are used to op t im ize the perfo rmance of a dist ributed 32sen2 so r O S2CFAR detect ion system w ith different reference length s and signal2to2no ise rat io s. A set of quasi2op2 t imum results is given and analyzed, w h ich p roves that niche genet ic algo rithm s are efficient fo r th is op t im iza2 t ion. The influences of the nonident ity in environment on the perfo rmance of the dist ributedO S2CFAR detec2
to r fo r different fusion rules are discussed subsequent ly. The results indicate that w hen OR fusion is em2 p loyed, the system is robust to the nonident ical variety of detect ion environment, w h ile 2 of 3 and AND fu2 sion are sensit ive in the same situat ion.
Key words: genet ic algo rithm s; niche; nonident ical environments; dist ributed CFAR detect ion; data fusion