Research On Intrusion Detection Based On Improved Fireworks Algorithm And SVM

With the popularity of Internet technology,the types and numbers of network intrusions are also increasing.Intrusion detection,as the second barrier of the Internet after the firewall,has been widely concerned by scholars since its birth.Constructing a practical and reliable intrusion detection system has also become a goal that network security engineers have been achieving.The existing intrusion detection algorithms have complex rules,low real-time performance and low detection efficiency.To further ensure the reliability of the network,designing a practical intrusion detection algorithm has good theoretical and practical significance.In this paper,the existing network intrusion detection methods and intrusion detection algorithms are deeply studied.By analyzing the advantages and disadvantages of various classifiers and applicable scenarios,SVM is selected as the core classifier of intrusion detection algorithm to discriminate network data.A slower problem is determined for the SVM optimal kernel function parameters and penalty factors,and the fireworks algorithm is used to accelerate the determination of the optimal SVM parameter combination.However,because the fireworks algorithm is easy to fall into the local optimal solution and the algorithm is slower in the later stage,in order to further overcome the shortcomings of the fireworks algorithm,the paper has improved the fireworks algorithm.Firstly,theoretically analyze the causes of the shortcomings of the fireworks algorithm,and then the corresponding improvement schemes—the initial fireworks population stretching technique and the fitness function stretching technique—are applied to the traditional fireworks algorithm and improved.Fireworks algorithm IFWA.Finally,IFWA is applied to the SVM optimal kernel function parameters and penalty factor selection,and the IFWA-SVM intrusion detection algorithm is obtained.In order to verify the effectiveness of the IFWA-SVM intrusion detection algorithm in practical applications,experimental tests were performed using the KDD CUP 99 and NSL-KDD data sets.The experimental results show that compared with the common SVM,traditional fireworks algorithm optimization SVM,BPNN and DNN algorithm,IFWA-SVM intrusion detection algorithm has faster model training speed and stable detection effect,which can classify intrusion detection data well.Certain availability and stability.