Research On Multi-Objective Optimization Algorithm Based On IGD~+ Indicator

Multi-objective optimization problem(MOP)is a basic problem in the research field of science and engineering and is widely exist in practical applications.For example,in the production process of gears,producers need to take into account its weight,cost,noise and life.Then in the design process of the aircraft wing,it is necessary to simultaneously meet the aerodynamic requirements in both vertical and high-speed flight conditions.Due to the parallelism,fast calculation and excellent general adaptation of the swarm intelligence optimization algorithm,the research on the swarm intelligence optimization algorithm has received more and more attention.In this thesis,we focus on researching on multi-objective optimization algorithm based on a excellent performance indicator:IGD+indicator for solving multi-objective optimization problems.The main contents are summarized as follows:The descent of the selection pressure due to the increasing dimension of the objective space of the algorithm is difficult for multi-objective evolutionary algorithm.We propose a multi-objective evolutionary algorithm based on IGD+indicator.Aiming at the construction of the reference point set for calculating the IGD+indicator,the systematic sampling method is adopted to generate a set of uniformly distributed weight vectors as the reference point set to ensure the diversity of the non-dominated solution set.In addition,an environment selection mechanism based on IGD+index and reference vector clustering is proposed to guarantee the uniqueness of the selected solution and fully exploit the candidate solutions with good convergence and diversity.Experimental results demonstrate that our proposed algorithm has a strong capability of searching for dominance optimal solution compared with a variety of state-of-the-art optimal methods.Aiming at how to effectively balance the convergence and distribution of external archive and how to assist the particles in jumping out of the local optimum in multi-objective particle swarm optimization algorithm,we propose a multi-objective particle swarm optimization algorithm based on IGD+indicator and objective space decomposition.Balancing convergence and diversity is difficult for particle swarm optimization algorithm when updating the external archives,we propose two strategies for updating the external archives:(1)An external archive update strategy based on the target space decomposition.(2)External file update strategy based on IGD+indicator.The above two strategies are used to screen the integrated candidate solution set to take into account both convergence and diversity.In addition,the algorithm uses a pBest updating strategy based on IGD+indicator to update each particle according to the IGD+indicator value between the particle and the corresponding reference point.Finally,because the particle swarm optimization algorithm is easy to fall into the local optimal problem,we introduce the simulated binary crossover(SBX)and polynomial mutation(PM)operators to re-specify the position of the particle to improve global search capabilitiy of the particle.The experimental results have validated that the proposed algorithm is suitable for MOPs with different Pareto front and has a good universality.