基于人工蜂群遗传算法的稀疏全聚集成像方法研究

稀疏阵列设计是一种可提高相控阵成像实时性的有效途径。遗传算法能较好解决线性阵列稀疏这种典型的约束优化问题。但此算法的局部搜寻能力较弱，且在后期搜寻效率较差。为此，论文通过构建人工蜂群-遗传算法的阵列稀疏方法，把蜂群寻找最优解的过程引入到传统遗传算法中，增加全局最优解的搜索能力。结果显示，人工蜂群-遗传算法优化后得到的稀疏阵列比遗传算法优化后得到的稀疏阵列具有更高的旁瓣抑制力，阵列的峰值旁瓣水平达到-11.40 dB。而在阈值为-6 dB时，两种算法优化得到的稀疏阵列主瓣宽度都等同于全阵列2.8°的主瓣宽度。最后，论文通过相控阵检测系统在钢轨试样上采集超声信号，利用人工蜂群-遗传算法设计得到的稀疏矩阵进行全聚焦成像。实验结果显示，当阵元数为32的线性阵列在稀疏率达到75%时，稀疏阵列的阵列性能指标分辨率、信噪比与满阵相差不大，但成像效率却提高了53.04%。

Sparse TFM imaging by artificial bee colony-genetic algorithm

Sparse array design is an effective way to improve the real-time performance of ultrasonic phased array imaging. Genetic algorithm can solve the typical constrained optimization problem of linear array sparsity well. However, the local search ability of genetic algorithm is poor, and the search efficiency is low in the late evolution period. Therefore, in this paper, the process of finding the optimal solution of bees is introduced into the traditional genetic algorithm by constructing the array sparse method of artificial bee colony-genetic (ABCG) algorithm to increase the searching ability of global optimal solution. Simulation results show that the peak sidelobe level (PSL) of the sparse array optimized by ABCG algorithm reaches -11.40 dB, which shows higher sidelobes suppression ability than the sparse array optimized by genetic algorithm. The main lobe width (MLW) optimized by the two algorithms is basically equivalent to 2.8°, the main lobe width of full array at the threshold of -6 dB. Finally, a phased array detection system is used to collect ultrasonic signals from rail samples, and total focus method (TFM) is performed with the sparse matrix designed by ABCG algorithm. Experimental results show that when the sparsity rate of 32-element linear array reaches 75%, the array performance indicator (API) and signal to noise ratio (SNR) of the sparse array are not much different from that of full array, but the imaging efficiency is improved by 53.04%.