基于改进蝙蝠算法的工业控制系统入侵检测

针对蝙蝠算法（BA）易陷入局部极小的缺点，提出了两点改进：（1）在蝙蝠位置更新时考虑了当前局部最优解分布对算法的影响；（2）将差分进化算法（DE）中的变异操作迁移到蝙蝠算法中，采用随机性变异的方式增加了种群多样性，提升了算法局部搜索能力，并通过典型测试函数验证了本文算法的优越性。将该算法用于工业控制系统（ICS）入侵检测中支持向量机（SVM）分类器的参数优化，使用工控入侵检测标准数据集进行仿真研究。结果表明，与DE、粒子群算法（PSO）和遗传算法（GA）等优化算法相比，其优化的SVM入侵检测模型在检测率、漏报率和误报率等指标上都有显著提升。

Intrusion Detection of Industrial Control System Based on Improved Bat Algorithm

Aiming at the local minima problem of the standard bat algorithm (BA),this paper makes two improvements.Firstly,the current local optimal solution distribution is considered during the updating of bats'' positions.Secondly,the random variation operation in differential evolution (DE) algorithm is introduced into BA to increase the diversity of the population and enhance the local search ability of the BA algorithm.Besides,the superiority of the proposed algorithm is illustrated by means of typical test functions.Moreover,the proposed algorithm is applied to the parameters optimization of support vector machine (SVM) classifier in industrial control system (ICS) intrusion detection model.The simulation results from the standard dataset for industrial system intrusion detection show that,compared with DE,particle swarm optimization (PSO) and genetic algorithm (GA),the optimized SVM intrusion detection model via the proposed algorithm can effectively improve the detection rate,false negative rate,and false alarm rate.