Application of velocity adaptive shuffled frog leaping bat algorithm in ICS intrusion detection

摘要

In this paper, a velocity adaptive shuffled frog leaping bat algorithm (VASFLBA) is proposed to solve the problem that the bat algorithm (BA) is easy to fall into local optimum and a lack of deep local search ability. Firstly, the influence of the current stochastic local optimal solution on the search of the algorithm is considered. Two adaptive proportional regulation factors are introduced to balance global and local search. Then, the locally deep search ability is enhanced by using the meme transfer mechanism of shuffled frog leaping algorithm (SFLA). In addition, stochastic population competition is introduced to improve the global search ability and when the algorithm trapped in the local optimum, differential mutation operation is performed on the current global optimal bat so that the algorithm can jump out of the local optimum. The superiority of VASFLBA is verified by benchmark test functions. On this basis, VASFLBA is used to optimize the parameters of support vector machine (SVM) in intrusion detection of industrial control system (ICS), and the standard dataset for ICS intrusion detection is used for simulation. The results show that, compared with BA, SFLA and other algorithms, VASFLBA can better solve the problem of SVM parameters selection.