Improved Multiobjective Cross Entropy Algorithm And Its Application To Emergency Scheduling Optimization

Multiobjective optimization problem is common in science and engineering practice.Evolutionary multiobjective optimization is one of the hotspots in the field of evolutionary computation nowadays,which mainly uses evolutionary computation to solve multiobjective optimization problems.Multiobjective cross entropy algorithm MOO CEM algorithm is a very effective multiobjective evolutionary algorithm,which has strong global search ability and faster convergence speed.However,the elite population extraction strategy based on a certain threshold results in low efficiency and poor distribution of the solution,and leads to the phenomenon of over convergence or uneven distribution.In view of the shortcomings of the MOO CEM algorithm,this paper proposes an improved multiobjective cross entropy algorithm,MOCE-S.MOCE-S is mainly improved in three aspects: the first is the introduction of the concentration based immune selection mechanism,which is based on the principle of promoting low concentration and high affinity antibody and inhibiting high concentration and low affinity antibody.The elite population is extracted to improve the distribution of the algorithm.The second aspect is to prevent the new population extraction mechanism to reduce the convergence of the algorithm.In this paper,the K means clustering and histogram method are used to classify the new sample group,so as to improve the convergence performance of the improved algorithm.The third aspect is to improve the processing methods of constraint conditions in multiobjective optimization problems,to guide the evolution of sample groups to Pareto optimal solutions in a better way.This paper compares the four algorithms of MOCE-S,MOO CEM,NSGA-II and NNIA on four test functions of ZDT1,ZDT2,ZDT3 and OSY.The results of the test functions and related performance indicators show the superiority of the MOCE-S algorithm in solving multiobjective optimization problems.Finally,aiming at the problem of emergency scheduling optimization for hot spots,a multiobjective emergency scheduling optimization model is established in this paper,and the MOCE-S algorithm is used to solve the multiobjective emergency scheduling optimization problem with the actual case as the background.