Research On The Adjustable Fuzzy Matching Negative Selection Immune Algorithm

Biology immune system is a self-adaptation and self-organization system with high intelligence. It contains abundant information processing mechanism. Immune algorithm is an engineering application algorithm used to simulate biology immune system. It has many merits as a rising bionics algorithm and shows strong life in the filed of information science. It has become an important research subject. This paper aims at the existent problems of original immune negative algorithm. An adjustable fuzzy matching negative selection algorithm is brought forward based on immune negative selection mechanism. The related theory and contents of the algorithm are deeply analyzed. Through simulation and application examples, the performance of the designed algorithm is improved.The main research contents are summed up as follows.1. Based on the negative selection mechanism of biology immune system, an adjustable fuzzy matching negative selection algorithm is brought forward. This algorithm detects adjustable matching thresholds continuously. Compared with traditional negative selection algorithm, this algorithm can decrease holes number obviously. At the same time, considered the matching uncertainty and illegibility between antibodies and antigens, the concepts of continuous similarity and deviation are presented. On the base of meeting continuous similarity, the fuzzy matching with controlled similarity is realized, so the detecting ratio and capability are improved.2. Based on adjustable fuzzy matching negative selection algorithm, an effective detector set generated algorithm is presented. This algorithm has effective detector set. It can eliminate the redundancy phenomenon existed in original detector set. The detecting efficiency is increased by smaller detector set to detect nonself action in bigger range. At the same time, the method of setting up detector set is analyzed. It uses model set and the concept of right complete model to set up detector corpora. The self-adaptation calculation formula of detector number is given. The minimal detector set is analyzed by probability theory. The relative system parameters are confirmed and discussed.3. An immune negative algorithm which can detect whether some Nonself character is a hole or not is presented by the idea of pattern. This algorithm uses state tree method to memorize various states which possibly appeared during searching process. It uses trial method and backdate method to detect and judge the states. In this thesis, the holes generated in immune negative selection algorithm which possesses effective detector is studied by simulation.4. Based on the vaccine theory of biology immune system, an improved design scheme of adjustable fuzzy matching negative selection algorithm is presented. The vaccine operator and positive selection operator are introduced to this algorithm. The algorithm has adaptation immune response function and its response period is shortened when the same antigen enters body again. The improved algorithm can update vaccine data-base and has self-adaptation function for setting up secondary response.