Abstract: Aiming at the low stability and poor detection performance of the traditional immune algorithm in the network fault detection, an artificial immune system algorithm based on the clonal selection and immune memory mechanism is proposed. In the algorithm, the way that immature detectors added to is adjusted. A method for evaluating the effectiveness of the memory detectors is designed, and the classification strategy is implemented according to the effectiveness of the detectors. Different sub group is imposed different evolutionary strategy. The experiments show that the stability and the detection performance have obvious improvement.

Key words: effectiveness evaluation, grading memory, clonal selection, evolution strategy, detection performance

摘要： 针对传统免疫算法在网络故障检测中存在的稳定性低、检测性能差等问题，提出一种基于克隆选择和免疫记忆机理的人工免疫系统算法。该算法调整未成熟检测器的补入方式，设计对检测器进行有效性评估的机制。给出依据评估结果对记忆检测器实施分级的策略，对各级别的检测器子群体采用不同的进化策略。实验结果表明，与传统算法相比，该算法的稳定性和检测性能都有一定改善。

关键词: 有效性评估, 分级记忆, 克隆选择, 进化策略, 检测性能

CLC Number: 

• TP311.52