Research On Multi-objective Optimization Algorithm And Its Application Based On Gradient Information

Most of current studies are about solving issues concerning multi-objective optimization based on the multi-objective optimization algorithm of evolutionary algorithms. This kind of algorithm is suitable to solve the problem because of it is a kind of random search algorithm based on the concept of Pareto optimal. However, it is limited in the application of engineering practice, especially the large amount of calculation in solving the large, complex engineering optimization problem. Therefore, finding the effective multi-objective optimization method to raise efficiency is urgently needed.According to the slowly calculation speed of the traditional evolutionary algorithm, the paper try to use the gradient information to solve the multi-objective optimization problems. Firstly it introduces the concepts of the gradient information, and generalizes the concepts of the direction of interest in the multi-objective optimization problems. The paper is applied the steepest descent method to multi-objective optimization problems which is based on the derivative. Its search direction is the DOI instead of the steepest descent direction. The simulation result to the GenMED functions and the DAM optimization process shows that the multi-objective optimization algorithm based on the steepest descent method can solve this kind of problem effectively.The algorithm based on gradient information which is used in dynamic multi-objective optimization problems is less. We use the direction derivative instead of the gradient direction consider of the difficulty in solving the DOI. The HSMGOA algorithm proposed in this paper is use the selection and collocation method and finds two gradient optimization branches to iterative optimization with the selected individuals based on the thought of decomposition and aggregation. It makes the population reach the optimal frontier quickly. The simulation result to the GenMED functions and the DAM optimization process shows that the HSMGOA algorithm for solving unconstrained multi-objective optimization problems has the faster speed. Finally we blended HSMGOA with NSGA2to solve the supplementary food biochemical reaction process effectually.