中文摘要：

RBF（Radical Basis Function）神经网络是一种典型的三层前向神经网络。虽然RBF神经网络的非线性逼近能力、分类能力以及学习速度都要好于其他的神经网络，但是RBF神经网络在实际应用中隐含层中心点难求，不能被广泛地应用于入侵检测系统中。免疫算法是基于免疫系统的学习算法，免疫算法不仅对干扰具有较强维持系统平衡的能力，而且具有较强的模式分类能力。为了得到最优的RBF神经网络并将其应用到入侵检测系统中，提出了一种免疫算法优化的基于最小均方差的联合RBF神经网络，即IA-LMS-RBF算法。仿真实验结果表明，与传统的K-means和随机法选取基函数中心点相比，基于免疫算法求取中心点的LMS-RBF神经网络，不仅能明显地提高对已知攻击的检测能力，并且对于未知的攻击行为也能很好地进行识别。IA-LMS～RBF算法有效提高了入侵检测系统的效率，保证了计算机系统的安全性。

英文摘要：

RBF neural network is a typical three-layer feed forward neural network. Although approximation capacity, classification and learning speed of RBF neural network is superior to others, it is difficult to find the optimal value of the center point which is not used widely in intrusion detection system. Immune algorithm is a learning algorithm based on the immune system. It not only owns a strong a- bility to maintain system balance,but also has strong pattern classification. In order to get the optimal RBF neural network and apply it to the intrusion detection system, an immune algorithm has been proposed to optimize the LMS-RBF neural network, which is based on the minimum mean square, called as associated IA-LMS-RBF algorithm. Simulation results shows that compared with the traditional K - means and randomly to select the basis function center,the immune algorithm to strike the center of the LMS-RBF neural network not only significantly improves the ability to detect the known attacks,but also has a good recognition to the unknown attacks,IA-LMS-RBF algorithm can effectively improve the efficiency of intrusion detection system and make sure computer system is becoming more secure.