基于混合变异GOA优化极大似然的DOA估计研究

DOA Estimation Based on Mixed Mutation GOA Optimization Maximum Likelihood

选择极大似然(ML)方法进行DOA估计时,由于进行了多维搜索,导致运算效率较慢,所以提出一种改进的蝗虫优化算法优化极大似然,引入了柯西变异算子和重心反向解,将迭代后的位置再次更新.一方面加速了极大似然估计的搜索效率,另一方面避免了常规蝗虫优化算法容易陷入局部最优的问题.仿真实验表明,与现有的大多数优化方法相比,提出的方法对信噪比的泛化能力更强;在不同信源个数时都可以保证其估计效果的精确度,将真实值与预测值的误差控制在±1°内;在相干信号下,混合变异GOA-ML方法也具有较强竞争力.综合各类评价指标来看,混合变异蝗虫(GOA)极大似然方法拟合优度更好,稳定性更高.

In order to solve the maximum likelihood estimation method for DOA estimation,due to the multidimensional search result in the shortage of the computing efficiency is slow,we proposed an improved grasshopper optimization algorithm to optimize the maximum likelihood. Which was introduced the Cauchy mutation operator and centroid opposition solution,and updated the location again after interation. On the one hand,it accelerates the search efficiency of maximum likelihood estimation;on the other hand,it avoids the problem that the conventional grasshopper optimization algorithm is prone to fall into the local optimal. Simulation results show that compared with most existing optimization methods,the proposed method has better generalization ability to SNR. The accuracy of the estimation effect can be guaranteed within ±1° when the number of sources is different,including the error control between the real value and the predicted value. According to all kinds of evaluation indexes,the mixed mutation GOA maximum likelihood method has better goodness of fit and higher stability.