GEP-based Prediction Model Of Medal Fatigue Time

Metal fatigue has become to face the problem and focus of the study in the modern world. Predicting the time of the metal fatigue issues, which become the most popular and most concern problem. At present time, metal fatigue prediction method mainly depend on physical methods, first through the metal materials and controlling the test environment, to the need for test data, finally through according to a number of theorems and rules for data processing to get concludes. In the course of data processing, deviation is inevitable, and evolutionary computation with self-organizing, adaptive and self-learning intelligent features, the data obtained using the known function of the unknown. The paper introduces an evolutionary computation method-the basic structure of gene expression programming, then the algorithm realization of their specific research. By systematic study, gene expression programming in the probability of genetic operator which rely heavily on artificial set of parameters, affecting the accuracy of the results. This article presents a modified GEP algorithm, and in this new approach based on the corresponding algorithm process. Finally, it applied to the prediction of metal fatigue time, GEP algorithm compared with in the merits of the original and the improved algorithm, obtained more satisfactory results. This mainly includes the following:(1) The paper research in the basic principle and process of the GEP algorithm. Detailed study of the gene expression programming encoding, fitness function, genetic factors, it gave the specific algorithm flow. The paper summarizes the characteristics of GEP and compared with other genetic algorithm differences.(2) Proposed a new improve algorithm of the GEP, which is based on the original algorithm. The genetic factor of the GEP is changed by way of a hierarchical principle, the GEP expression divided into parts. By different parts of the corresponding policies, more individuals good develop and ultimately achieve fitness function in the range. This new gene expression programming compared with the traditional algorithm, which is achieved to the global optimal solution, can inhibit premature emerging and convergence to local optimum. The overall structure of new GEP algorithm is more simple and easier to understand, and the algorithm gets more reliable.(3) The use of the improved GEP algorithm to predict the time of metal fatigue. Firstly, this model is described in detail, making it more suitable for gene expression programming and solve the problem of requirements; Secondly, GEP algorithm using improved prediction of metal fatigue parameters derived the expression K; Finally, the result of metal fatigue time prediction are obtained by programming. The innovation of this paper that advance a new genetic algorithm, which combined with the principle of the original gene expression programming, using the new GEP algorithm to predict the time of metal fatigue, and obtaining good results. Improved GEP algorithm to improve the efficiency of the algorithm, the optimal solution of the problem is easier to access and avoid local optimum. Experimental results show that the new GEP algorithm to metal fatigue time prediction is more accurate than traditional methods, which has great practical value.