Constrained Single-objective And Multi-objective Evolutionary Algorithms And Their Applications

Constrained single-objective and multi-objective optimization problems have been studied in the fields of science,economy and engineering during the past few decades.Due to the presence of different types of constraints,the interference among constraints,the objective functions involved and the interrelationship between the constraints and the objective functions,these constrained optimization problems are usually very difficult to be solved.Algorithms for solving them can be grouped into two categories,including traditional mathematical methods and evolutionary algorithms with constraint-handling techniques.Traditional mathematical methods have been applied to solve constrained optimization problems.However,they usually require gradient information of a given problem,which might be impractical for solving real-world problems,since we cannot always obtain gradient information.The need of developing other methods without taking gradient information into account for constrained optimization problems is very important.Evolutionary algorithm,inspired by nature,with constraint-handling techniques is such a method for solving constrained problems.It is a powerful optimization tool which only requires little domain knowledge and is easy to implement.Evolutionary algorithms have stimulated the interests of many researchers during the past few decades.At the same time,a few constraint-handling techniques are also carried out.By combining them,some constrained optimization evolutionary algorithms for solving constrained optimization problems have been proposed.However,constrained single-objective and multi-objective evolutionary algorithms still have not been fully explored.This thesis mainly focuses on the algorithm design and application of constrained single-objective and multi-objective optimization.The main contributions of this thesis are summarized as follows:1)We proposed a novel constraint-handling technique based on biased dynamic weights for constrained single-objective evolutionary algorithm.2)We proposed a decomposition-based evolutionary algorithm with boundary search and archive for constrained multi-objective optimization problems.3)We proposed a novel constraint-handling technique based on directed weights to deal with constrained multi-objective optimization problems.4)We designed an unbalanced constrained multi-objective test suite with three different types of biased constraints,yielding three different types of constrained test problems in which the degree of imbalance is scalable via a set of parameters introduced for each problem.5)We designed a set of constrained multi-objective optimization problems with deceptive constraints in this thesis and proposed a cooperative framework with an improved version of directed weights to solve constrained multi-objective optimization problems.