An Improved Genetic Algorithm for Thinning Acoustic Array.
He Li,Zhi-Hong Liu,Chui-Jie Yi,Zhe-Ming Lu
2017-01-01
Abstract:Optimization of large arrays, which improves the efficiency, plays an important role in acquiring signals in the acoustic field. Recently, artificial intelligence algorithms, such as genetic algorithms (GAs) were widely used in sparse array optimization 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 convergence 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 conducted 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.