中文摘要：

差分进化（DE）算法具有操作简单，控制参数少，鲁棒性好等特点，但在对某些连续空间复杂函数进行优化时存在搜索盲目性较大、效率不高的问题。为此提出一种基于最小二乘支持向量机（LS—SVM）的自适应DE算法，该算法改进了标准DE算法的差分变异和交叉等关键遗传操作，引入了基于LS-SVM的种群进化引导策略，基于LS-SVM对种群”最优训练集数据进行回归函数逼近和优化，分析了种群进化引导策略的自适应应用条件，给出了算法的整体流程及各关键步骤的复杂度。对标准测试函数的对比优化结果表明，改进算法相比标准DE算法具有更好的全局寻优能力和更高的优化效率，可以满足对连续空间复杂函数优化问题的可靠、高效求解。

英文摘要：

Differential evolution （DE） is characterized by its simple operation, few control parameters and fine robustness. However, DE yet still usually has difficulty with some complicated function optimization in continuous search space for its searching blindness and inefficiency. An adaptive differential evolution algorithm based on LS-SVM （Least Square Support Vector Machine） was proposed. The key genetic operators like differential mutation and crossover were improved; Adaptive orientation strategy for population evolution based on LS-SVM function n-best training dataset regression, approximation and optimization was deployed,＂ With applying condition discussed, the procedure and complexity of the proposed strategy were summarized, The results of comparative tests of the optimized algorithm with conventional one based on various standard functions effectively prove that the high requirements of global optimization accuracy and convergence efficiency for continuous multimodal functions are well fulfilled with the proposed algorithm.