Multiobjective Local Search Flower Pollination Algorithm Based on Decomposition

Flower Pollination Algorithm is a single objective optimization algorithm such that the aim of this algorithm is to get the minimum (or maximum) value of the given problem. However, it is easy to get engineering problems with objectives more than one. Therefore multiobjective optimization algorithms are needed to solve these problems. One of the possible approach to propose multiobjective optimization algorithm is to alter the single objective optimization algorithm, improved them with some operators. In this research one of the recently proposed algorithm named as Flower Pollination Algorithm (FPA) is converted to multiobjective optimization algorithm. The idea of this research is to integrate FPA algorithm to the Multiobjective Evolutionary Algorithm Based on Decomposition (MOEA/D) as crossover operator. For this purpose tree different update formulation is compared with MOEA/D algorithm. The proposed formulation evaluate random number generators and Levy fly dynamics as described in the original FPA. The proposed multiobjective optimization algorithm is applied to 15 benchmark problems (MaF) with respect to two metrics. These two metrics are used to compare algorithms with respect to the solution accuracy and distribution of the solutions on objective space.