摘要： 结合遗传算法的群体搜索性和拟牛顿迭代法的局部细致搜索性，提出一种基于遗传-拟牛顿混合算法的到达时间差定位方法。该方法利用遗传算法进行全局迭代，当收敛结果达到满意值后将其作为拟牛顿迭代的初始值继续迭代，直至得到精确解，由此克服遗传算法后期搜索效率低以及拟牛顿法对初始值敏感的缺陷。仿真结果表明，在参数设置合理的前提下，相比遗传算法和拟牛顿法，该混合算法性能稳定，具有较快的定位速度和较高的定位精度。

关键词: 定位, 遗传算法, 拟牛顿算法, 到达时间差, 测量误差

Abstract: By combining group searching characteristic of Genetic Algorithm(GA) and local searching characteristic of quasi-Newton method, this paper proposes a hybrid localization algorithm for Time Difference of Arrival(TDOA). GA iterates globally until the satisfactory convergence value is achieved, and the value is used as the initial value of quasi-Newton iteration. The iteration process continues until the exact solution is get. So the hybrid algorithm overcomes the shortcomings that quasi-Newton method has high sensitivity to initial point and GA has low search efficiency in later period. Experimental results show that if the parameters are assumed reasonably, the hybrid algorithm has higher stability, localization rate and localization precision than GA and quasi-Newton method.

Key words: localization, Genetic Algorithm(GA), quasi-Newton algorithm, Time Difference of Arrival(TDOA), measurement error

中图分类号: 

• TP393