الگوریتم انتخاب منفی ترکیبی مبتنی بر بی نظمی و کاربردهای آن در خطا و تشخیص ناهنجاری

This paper proposes a new negative selection algorithm method that uses chaotic maps for parameter selection. This has been done by using of chaotic number generators each time a random number is needed by the original negative selection for mutation and generation of initial population. The coverage of negative selection algorithm has been improved by using chaotic maps. The proposed algorithm utilizes from clonal selection to obtain optimal non-overlapping detectors. In many anomaly or fault detection systems, training data don’t represent all normal data and self/non-self space often varies over the time. In the testing stage, when any test data cannot be detected by any self or non-self detector, the nearest detectors are found by K-Nearest Neighbor (K-NN) method and the nearest detector is mutated as a new detector to detect this new sample. Proposed chaotic-based hybrid negative selection algorithm (CHNSA) has been analyzed in the broken rotor bar fault detection and Fisher Iris datasets.