A Model-Based Evolutionary Algorithm using FCM-clustering and PCA

The structure of operators in most traditional multi-objective evolutionary algorithms are based on fixed heuristics such as intersection and mutation, which are unable to learn the structures or properties of problems to create. To use evolutionary algorithms to learnability, news of evolutionary algorithms with the model is presented. In model evolutionary algorithms, innovative operators are replaced by machine learning models such as instructional and sample models. In this paper, a multi-objective evolutionary algorithm with a model is presented in which in each generation, a probable area of the search space deserves the model as a possible model. In the decision area in the search space, which are the dominant points, they are better ranked, we have clustered or different fuzzy methods, or on other points with the first order, it is a contest selection action. If you are missing, you take the form to be removed at close range, and the result is considered to be the center of the clusters, and then, clustering is done based on the nearest neighbors. The principal component analysis algorithm, which is the best method for reducing the given dimensions linearly, has been used for the models. New solutions are obtained from the model if it is a normal distribution. The proposed method is tested and the results are compared with the method of non-dominated sorting genetic algorithms. The results show that this method is faster than earlier methods and with fewer repetitions and evaluation of functions, the results are better and bold.