摘要
针对传统的克隆选择算法存在的不足,提出一种改进的克隆选择算法ICSA。该算法在克隆选择算法的基础上,利用负选择算法优化了克隆初始抗体群的生成方式,加入对抗原性质的评判环节,引入克隆选择动力学模型来模拟生物免疫系统中抗体增殖的动态行为,用以指导ICSA中的抗体增殖,并针对盾构地下工程风险实时识别的要求,采用了在线和增量式的学习方式,做到边学习、边识别、边更新。ICSA在标准数据集与盾构地下工程数据的仿真实验表明,在二分类模式识别上具有很高的分类性能。
An improved clonal selection algorithm named ICSA presented is inspired by some deficiencies of the traditional clonal selection algorithm. In our ICSA, the way of producing the initial clonal antibody set is optimized by using the negative selection algorithm. Then a key procedure is added to judge the nature of the antigen, and the clonal selection dynamical model is also introduced in order to simulate the dynamic behavior of the proliferation of the antibody in biological immune system, which is used to guide the proliferation of,the antibody in ICSA. Finally, to meet the requirement of the real-time risk recognition in the shield underground project, an online learning and incremental algorithm in ICSA is designed so that it can do learning and recognizing as well as updating. The simulation experiment of the ICSA in the standard data sets and shield underground engineering data show thatit has a high performance to solve the problem of two- classification pattern recognition.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第11期2741-2744,2748,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(50778109)
上海市科技攻关计划基金项目(08511501702)
上海市重点学科建设基金项目(J50103)
关键词
克隆选择
抗原评判
动力学模型
免疫算法
二分类模式识别
clonal selection
antigen judgment
dynamical model
immune algorithm
two-classification pattern recognition