基于优化极限学习机的工业控制系统入侵检测

Intrusion detection of industrial control system based on optimized extreme learning machine

为解决极限学习机(ELM)随机给定输入权值和隐含层结点的阈值,导致泛化能力和精度不理想的问题,提出混合自适应量子粒子群(HAQPSO)优化算法对输入权值和隐含层结点的阈值进行参数寻优。在量子粒子群优化算法的基础上,加入差分策略和Levy飞行策略,采用自适应改变的控制方法控制收缩-扩张系数,有效避免算法的早熟,增强算法全局寻优能力,通过对典型函数的测试验证了该算法的优越性。构建基于HAQPSO优化ELM的工控入侵检测模型,将仿真结果与其它算法进行比较,HAQPSO优化后的ELM在准确率、精确率和召回率等指标上都有明显提升。

To deal with the problem of poor generalization ability and low accuracy caused by randomly given input weights and thresholds of hidden layer nodes of extreme learning machine(ELM),a hybrid adaptive quantum particle swarm optimization(HAQPSO)algorithm was proposed to optimize the parameters of input weights and thresholds of hidden layer nodes.To prevent the premature convergence problem and improve the global-optimization capability,based on the quantum particle swarm optimization(QPSO)algorithm,the difference strategy and Levy flight strategy were added,and the contraction-expansion coefficient was controlled using adaptive control method,the superiority of the algorithm was verified by the test of typical functions.The industrial control system intrusion detection model based on HAQPSO optimized ELM was constructed,and the simulation results were compared with other algorithms.The ELM optimized by HAQPSO has obvious improvement in detection rate,precision rate,and recall rate.