A Prediction Model And Its Application In Dynamic Multi-objective Optimization Based On Spline Interpolation

Dynamic multi-objective optimization problems(DMOP)as a basic optimization problems in scientific research and engineering practice is widespread and widely application prospect,solve the core problem,to promote the development of the field and application has important scientific and engineering significance.At present,the most dynamic multi-objective problem solving method(DMOEA)is certain to increase the algorithm search randomicity,are adopted to decrease the rate of convergence at the expense of auxiliary strategy to improve in some successful static multi-objective optimization evolutionary algorithm to deal with the problem,but most of these improvement strategy is not aimed at the characteristics of different issues the corresponding algorithm is improved,and therefore difficult to obtain satisfactory optimization performance.According to the characteristics of the dynamic multi-objective optimization problems,this paper proposes a new dynamic multi-objective optimization algorithm based on prediction model,using the heuristic knowledge to design a new prediction model,in order to solve dynamic multi-objective optimization problem,and apply the method to solve optimization problems in cloud computing task scheduling.This paper around the dynamic multi-objective optimization algorithm based on prediction model,the main contents include: first,dynamic multi-objective optimization problems are reviewed and the existing method,and introduces the related theory of function interpolation,establish dynamic multi-objective optimization algorithm based on prediction model of the basic concepts,and expounds the need for different reasons of design different forecasting model and points out the rationality of the forecasting problem;Second,in the general dynamic multi-objective problem and the forecasting model on the basis of,this paper proposes a new forecasting model based on dynamic multi-objective evolutionary algorithm(CSPM).This algorithm in the field of multi-objective(NSGA II)based on the classical algorithm,combined with the cubic spline interpolation function method,we design a new prediction model for the problem,and through the experiment proves that has good universality and convergence.Finally,on the basis of the algorithm designed a cloud computing network virtual machine scheduling method,the usage of the virtual machine by monitoring,calculation,to cope with different tasks obtained optimal allocation.