Abstract:

In order to improve the global search ability of the Adaptive Genetic Algorithm（AGA） under high-selection-pressures, a hybrid algorithm entitled Beetle Antennae-Genetic Algorithm（BAGA） is proposed based on Beetle Antennae Search（BAS）. In order to enhance the functional guiding and local searching abilities of AGA, Beetle Antennae Operator（BA） is utilized to improve the new individuals produced by AGA. The data-driven strategy is adopted to reduce the complexity caused by hybridizing of algorithms, and the sensitivity analysis is carried out regarding different dimensional variables with the objective function so as to optimize the evolution path and improve the algorithmic efficiency. The application effect of the algorithm in truss size optimization is studied through quantitative experiments, and the advantages and characteristics of the algorithm are demonstrated through qualitative analysis of the reasons behind the data. The results achieved from the truss size optimization case study show that the lowest economic benefit scheme of steel is 2 490.56 kg, which is consistent with the results from other meta-heuristic algorithms, confirming the accuracy and effectiveness of the proposed BAGA. The average steel amount of 40 000 economic benefit schemes is 2 491.43 kg with the standard deviation 8.05 and the convergence rate 98%, compared with other meta-heuristic algorithms, demonstrating the high stability of BAGA.

Key words: genetic algorithm, beetle antennae search algorithm, hybrid algorithm, size optimization, sensitivity analysis

摘要：

为加强自适应遗传算法在高压选择下的全局搜索能力，提出了一种结合天牛须搜索的杂交算法。利用天牛须搜索算子对遗传算法产生的新个体进行局部改良，以增强导向作用和局部搜索能力。采用数据驱动策略改善算法杂交引起的复杂度问题，对不同维度变量进行基于目标函数的灵敏度分析，优化其进化路径从而达到提高算法运行效率的目的。通过定量实验研究算法在桁架尺寸优化问题上的应用效果，并定性分析数据背后的原因展示算法的优点和特点。研究结果表明：在桁架结构尺寸优化研究中，用钢量最低的经济效益方案为2 490.56?kg，与现有元启发式算法研究结果吻合，证实了算法的准确性及有效性；40 000个经济效益方案用钢量平均值为2 491.43 kg，标准差为8.05，收敛率达到98%，与其他元启发式算法相比证实了该算法较高的稳定性。

关键词: 遗传算法, 天牛须搜索算法, 杂交算法, 尺寸优化, 灵敏度分析