CAJ | 학술논문

在对基于异常的入侵检测进行训练时，缺少一个实时有效的训练集，提出了一种融合自控粒子群和免疫进化的入侵数据分类方法，对网络数据进行聚类分析，生成可靠的训练数据。粒子群模糊C均值聚类算法需要提前确定聚类数目，这在网络数据分析处理中是很难把握的，引入自控粒子群的方法根据迭代演算情况自动调节不同聚类数目的粒子群规模，使数据最后聚合在一个数目最优的聚类集中，同时为了克服陷入局部最优的问题，引入免疫进化机制，使部分粒子在当前最优指导下进行合理变异和替换，跳出局部最优解。
재대기우이상적입침검측진행훈련시，결소일개실시유효적훈련집，제출료일충융합자공입자군화면역진화적입침수거분류방법，대망락수거진행취류분석，생성가고적훈련수거。입자군모호C균치취류산법수요제전학정취류수목，저재망락수거분석처리중시흔난파악적，인입자공입자군적방법근거질대연산정황자동조절불동취류수목적입자군규모，사수거최후취합재일개수목최우적취류집중，동시위료극복함입국부최우적문제，인입면역진화궤제，사부분입자재당전최우지도하진행합리변이화체환，도출국부최우해。
The intrusion detection based on abnormity needs real-time and effective sets, a kind of intrusion data classification algorithm combining self-control particle swarm and immune evolutionary is proposed to cluster the network data and get the reliable training sets. The number of clustering needs to be confirmed before the Particle Swarm Optimization(PSO)and Fuzzy C-Means clustering(FCM)algorithm and this is difficult for the network data. The proportion of the particle swarm with different number of clustering is adjusted according to the information from the working and it can make the sets to an optimization clus-tering. The immune evolutionary is imposed to solve the problem that the PSO-FCM is easier to fall into local optimization and this can make the particle to mutate and displace under surveillance.