An Improved Genetic Algorithm for Thinning
An Improved Genetic Algorithm for Thinning
复制标题
一种改进的稀疏遗传算法
DOI：
--
复制
发表时间：
2017
影响因子：
--
通讯作者：
Zhe-Ming Lu
中科院分区：
文献类型：
--
作者：
He Li;Zhi-Hong Liu;Chui-Jie Yi;Zhe-Ming Lu
对大阵列的优化，对提高效率起着重要的作用。在声场信号采集中起着重要作用。最近，人工智能。遗传算法（GAs）等算法在稀疏阵列优化中得到了广泛应用。感谢他们在全局优化方面的出色表现。本文提出了一种新的。初始部署形成方法：“随机位置分配”，并提出新的Cross。以及遗传算法中的突变规则，以提高声阵列的性能。的。改进的遗传算法减少了计算量，提高了转换率。gence率。代数从30代[1,12]减少到20代以下。同样的条件。此外，最大旁瓣电平（最大SLL）被降低。模拟是骗人的。由200个各向同性或方向元素组成的线性阵列上的管道。各向同性。元素残留77%时，最大SLL为-27.06dB，低于[1]（-22.09 dB）、[3]（-24.03 dB）、[8]（- 0.23.05 dB）和[13]（-22.13 dB）。而对于定向。元件时，最大SLL为-27.15dB，低于[3]（-23.69 dB）和[3]（-25.25 dB）。剩下75%的元素。改进的遗传算法比以前的方法性能更好。非对称线阵列的优化条件。此外，我们的算法具有鲁棒性。对元件的指向性和线阵的转向角。
Optimization of large arrays, which improves the eciency, plays an im-.portant role in acquiring signals in the acoustic eld. Recently, articial intelligence.algorithms, such as genetic algorithms (GAs) were widely used in sparse array optimiza-.tion for their excellent performance in global optimization. This paper proposes a new.initial deployment forming method: \random place assignment" and proposes new Cross.and Mutation rules in GA algorithm to improve the performance of acoustic array. The.improved GA algorithm reduces the amount of computation and increases the conver-.gence rate. The number of generations decrease from 30 [1,12] to less than 20 under the.same condition. Also the max side lobe level (max SLL) is reduced. Simulations are con-.ducted on a linear array composed of 200 isotropy or directional elements. For isotropy.elements, the max SLL is -27.06dB which is lower than [1](-22.09 dB),[3](-24.03),[8](-.23.05 dB) and [13] (-22.13 dB) when 77% of elements are left. While for directional.elements, the max SLL is -27.15dB, lower than [1](-23.69 dB)and [3](-25.25 dB when.75% of elements are left.The improved GA performs better than the previous methods in.terms of optimization for asymmetric line arrays. Furthermore, our algorithm is robust.to the directivity of elements and the steering angle of the linear array.