改进引力搜索算法用于工控系统入侵检测

Improved gravitational search algorithm for industrial control system intrusion detection

为提高引力搜索算法的全局搜索能力和收敛速度,提出改进引力搜索算法(IGSA)。为引力常量嵌入混沌映射,使其在减小的同时可以混沌地变化,快速地跳出局部极小值,扩展搜索区域;引入细菌觅食算法(BFA)的趋化算子,利用最优个体信息对当前最佳粒子进行调整,提高收敛速度。4种基准函数的测试结果对比表明,IGSA有着更好的搜索能力和收敛速度。利用IGSA对孪生支持向量机(TWSVM)的参数进行寻优,将寻优后的TWSVM分类器应用于工控标准入侵检测数据集。实验结果表明,IGSA-TWSVM对整体入侵的误报率、漏报率和对各类入侵的检测率都优于其它算法。

To improve the global search ability and convergence speed of gravitation search algorithm,an improved gravity search algorithm(IGSA)was proposed.The chaotic map was embedded into the gravitational constant to make it change chaotically while decreasing,so that it quickly jumped out of the local minimum to expand the search area.The chemotactic operator of the bacterial foraging algorithm(BFA)was introduced,and the optimal individual information was used to adjust the current optimal particle to improve the convergence rate.The test results of the four benchmark functions show that IGSA has better search abi-lity and convergence speed.The IGSA was used to optimize the parameters of the twin support vector machine(TWSVM),and the optimized TWSVM classifier was applied to the industrial control standard intrusion detection data set.The results show that IGSA-TWSVM has better false positive rate and false negative rate of the overall intrusion and the detection rate of various intrusions than other algorithms.