Immune cultural algorithm for fault feature selection
基于免疫文化算法的故障特征选择方法

This paper proposed a new cultural algorithm. It applied the principle of immune clone selection as population space of cultural algorithm, meanwhile designed a new knowledge of history and its influence function. This paper proposed a wrapper feature selection algorithm based on immune cultural algorithm to remove redundant variable in the industrial process fault diagnosis process, reduced the data dimension and improved the performance of fault diagnosis. The method used antibody populations to global search and retained the best individual through the belief space of cultural algorithms. And it used high dimensional data of UCI data sets to test this method. The results show that compared with diagnose fault directly, the algorithm effectively reduces the dimension of the feature space and improves the classification accuracy in fault diagnosis of Temnnessee-EastmanTE process.