This work proposes a multi-source mating selection based multi-objective evolutionary algorithm（MMSEA）. In MMSEA, the spectral clustering algorithm is used to exploit the property of the multi-objective optimization problems. Based on the obtained population structure information, a multi-source mating selection strategy is designed to guide the algorithm search. The convergence of the algorithm is accelerated and the diversity of the population is maintained by setting multiple mating selections for each individual and using similar-based reproduction. The experimental results show that the proposed reproduction operator can effectively improve the performance of the algorithm. MMSEA is experimentally compared with variety of mainstream multi-objective evolutionary algorithms, and parameter sensitivity is also performed. In these experiments, MMSEA demonstrates strong competitiveness over the other approaches in solving typical multi-objective optimization problems with complex characteristics.

In order to solve the problem that the multi-objective workflow scheduling is difficult to optimize in the cloud computing environment, this paper proposes a differential flower pollination algorithm for the multi-objective workflow scheduling. The algorithm models the tasks and virtual machines in the workflow into pollen and models the complete scheduling sequence into flowers. Then it adopts a discrete flower pollination process according to the partial order relationship of the task. The simulation results show that compared with the algorithms NSGA-Ⅱ and MEOA/D, the algorithm can have higher resource utilization under the limited deadline and budget.

Most of the current constrained many-objective evolutionary algorithms focus on the convergence accuracy, but the convergence speed is relatively slow. In order to improve the convergence speed, a constrained many-objective evolutionary algorithm based on angle information (CMaOEA-AI) is proposed. In the algorithm, a selection operation based on the angle violation function is proposed to improve the convergence speed, which directly selects the superior individuals according to the dynamic convergence and diversity. Thereafter a crossover operation based on the differential evolutionary algorithm is proposed, which can select the infeasible solutions to participate in the crossover operation at different evolutionary stages. Simulation experiments are performed on the standard test function sets C-DTLZ. Compared with four state-of-the-art constrained many-objective evolutionary algorithms, the proposed algorithm shows good convergence accuracy while the convergence speed is greatly improved, and the higher the objective dimension, the better the effect.

In recent years, the multi-objective evolutionary methods have been widely used for solving overlapping community detection problem and have achieved good community division performance. To design appropriate individual encoding and evolution strategies is important to improve the performance of multi-objective overlapping community detection evolutionary algorithm. To this end, a dual representation method is designed to encode the non-overlapping community structures and overlapping nodes respectively, which can effectively obtain the overlapping community structures. On the basis of the dual representation, this paper proposes a dual representation-based multi-objective evolutionary algorithm for overlapping community detection (DRMOEA). In DRMOEA, an initialization strategy based on community boundary nodes is suggested to obtain good initial individuals，with the aim to improve the detection performance of the algorithm. In addition, for the overlapping part of the dual-representation, this paper proposes a crossover strategy according to the boundary nodes of elite individuals, which uses community boundary information to guide the evolution of the population towards a better direction. Finally, the experimental results on nine real-world networks show that the proposed DRMOEA is better than five representative baseline overlapping community detection algorithms.

To improve the distribution performance of multiobjective particle swarm optimization algorithm, an adaptive multiobjective particle swarm optimization algorithm, based on the decomposed archive, named AMOPSO-DA, is developed in this paper. First,an external archive update strategy, based on the spatial distribution information of optimal solutions, is designed to improve the searching ability of AMOPSO-DA. Second, an adaptive flying parameter adjustment strategy, based on the evolutionary direction information of each particle, is proposed to balance the exploration ability and the exploitation ability. Finally, this proposed AMOPSO-DA is applied to some multiobjective optimization problems. The experiment results demonstrate that AMOPSO-DA can obtain well-distributed optimal solutions.

Visualization technology is conducive to the evaluation and analysis of the solution sets obtained by solving many-objective optimization problem,but the existing many-objective visualization technology cannot effectively preserve Pareto dominance relation,maintain frontier distribution and retain shape.To solve the above problems,this paper presents quasi-circular mapping visualization.Many-objective are uniformly distributed in order on a unit arc according to their correlation.Based on the fitness function value,the solution sets are mapped into a polygon in quasi-circular space.So 3 dimensional visualization of the solution set is achieved through the geometric center and area of polygons.On the basis of this,the quasi-circular domination and equilibrium are defined.The dominance relation and mapping occlusion under quasi-circular mapping are theoretically analyzed and proved.Compared with parallel coordinates,principal component analysis and radial visualization,this method can preserve the Pareto dominance.In addition,it can also reflect frontier distribution and shape in the original space and effectively avoid data blocking.It helps decision makers to evaluate and select many-objective solution sets visually.

More and more complex multi-objective optimization problems have emerged in the real world, and the novel heuristic algorithms need to be developed to meet the challenge. A multi-objective firefly algorithm based on multiply cooperative strategies (MOFA-MCS) is proposed in the paper. MOFA-MCS uses the method of homogenization and randomization to generate the initial population, adopts the elite solutions in the external archive to lead the firefly to move, exerts Lévy flights to add random disturbance in the moving process, and finally, the ε-three-point shortest path strategy is also applied to maintain the diversity of the archive solutions. MOFA-MCS is compared with other six representative multi-objective evolutionary algorithms on 12 benchmark multi-objective test problems, and the experimental results show that MOFA-MCS has significant performance advantages in terms of convergence and diversity.

Unmanned aerial vehicle system (UAVS) cooperative combat model with temporal constraint of task type is insufficient making decision by single objective optimization.The multi-objective multi-strategy fusion quantum particle swarm optimization (MSQPSO) algorithm was proposed.To establish the task allocation model more accord with the actual operation situation,adding temporal constraint of task type and multi-UAV cooperative constraint,and abstracting the various capabilities of UAV.The quantum particle swarm optimization was improved by good-point set theory,scale chaos factor,quantum mutation and dynamic inertia weight.The multi-objective optimization was adopted to make decision.The final simulation results verify the effectiveness and superiority of the proposed MSQPSO algorithm.

In expensive multi-objective evolutionary algorithms,the evaluation of a large number of objective vectors spend a lot of time or experimental cost and lead to the cost of disaster.According to the fact that Pareto dominance relationships among candidate solutions are depended on the rank relationships of objective components,this paper proposes a predict method of rank equivalent to determine Pareto dominance.A decision vector and object vector rank matrix is established,and rank correlation analysis is used to calculate the correlation coefficient matrix R.Under the assumption of linear correlation,a prediction equation is established to predict rank relationships.Testing results on typical multi-objective optimization problems show that the proposed method only requires establishing a linear prediction model,which can remarkably improve the prediction accuracy and reduce the calculation of original expensive target function.Finally,the prediction method is integrated into the NSGA-II,it can avoid reconstruction the model in the process of evolution,then effectively decrease the number of evaluation for expensive objective vectors.