The Construction Of Multi-Objective Differential Evolution Algorithms And Its Application

Differential Evolution algorithm (DE) was introduced by R.storn and K.price in 1995. And it has already been applied successively to many areas such as multimodal function optimization, neural network learning, digital filter design, multi-objective optimization and so on.In this paper, we have done a series of research on Differential Evolution algorithm in multi-objective optimization and constraint handing areas. Moreover, we have applied it to three real application fields. Firstly, we described the relevant background, basic principle and implementation details of Differential Evolution algorithm. Secondly, we did some research on the construction of Multi-objective Differential Evolution algorithm, and analyzed four Multi-objective Differential Evolution algorithms which were published in recent years, such as Pareto Differential Evolution (PDE), Pareto Differential Evolution Approach (PDEA), Multi-objective Differential Evolution (MODE) and especially Differential Evolution for Multi-objective Optimization (DEMO). Thirdly, we proposed a novel multi-objective optimization concept to handle constraints. Experimental results from eight constrained test functions show that the proposed method is capable of successfully optimizing constrained single- and multi-objective problems. Finally, we applied the Differential Evolution algorithm and constraints handling method proposed in this paper to three hot application fields, such as the Geometry Optimization of Argon Atom Clusters, the Security-Aware Real-Time Scheduling and the Constrained Layout Optimization.The main contributions in this paper are listed as follows:1) We analyzed and summarized the basic principle and algorithm implementation of Differential Evolution algorithm in detail.2) We analyzed and compared four Multi-objective Differential Evolution algorithms which were published in recent years, such as PDE (2002), PDEA (2002), MODE (2003) and especially DEMO (2005).3) In this paper, we proposed a novel multi-objective optimization concept to handle constraints. Experimental results from eight constrained test functions show that the proposed method is capable of successfully optimizing constrained single- and multi-objective problems.4) In this paper we describe a new search method that uses Differential Evolution algorithm (DE) to optimize the geometry of small argon atom clusters. Experimental results show that the exact global optimal configuration of argon clusters with atom number N≤16 can be found in a reasonable computing time, and approximate optimization can also be obtained for clusters with N.=30. From their 3-D geometry structures, we can see that their optimal energy structures are highly symmetrical.5) In this paper, on the one hand, we abstract two mathematic models for the real-time scheduling problems considering security requirements. On the other hand, we propose a new security-aware real-time scheduling algorithm based on Differential Evolution algorithm (SAREC-DE), which can improve overall security level of the system by up to approximately 25% on the base of SAREC-EDF when real-time requirement is guaranteed. Experiment data and simulated results show the feasibility and availability of the proposed models and method.6) We applied Multi-objective Differential Evolution (DEMO) algorithm to the Constrained Layout Problem, and experimental results show that Multi-objective Differential Evolution (DEMO) can successively optimize Layout Problem.