Design and implementation of an algorithm for optimizing swarm intelligence

Abstract

quotExact and optimal solutions for the combinatorial and Non-deterministic Polynomial time (NP) Hard problems are found difficult to achieve using traditional evolutionary algorithms. The biological swarms like ants, birds, fishes and bees exhibits the self organization behaviour and algorithmic techniques inspired from swarming nature of these lives were termed Swarm Intelligence. Ant Colony Optimization (ACO) and Particle Swarm Optimization (PSO) are existing Swarm Intelligence (SI) approach, which are used for optimization of many problems which are combinatorial in nature. Travelling Sales Person (TSP) is one such problem which is taken into consideration for implementation of the proposed designed algorithm for optimizing the swarm intelligence behaviour. The investigation focuses on performance optimization of Ant Colony Optimization (ACO) and Particle Swarm Optimization (PSO). newlineVarious hybrid approaches are also there to optimize the performance of ACO and PSO. The 2-opt local search technique is one such technique which was implemented to improve the performance of ACO and PSO. The self organizing and stigmergic behaviour of real ants are implemented using probabilistic transient rule and pheromone evaporation rule. The velocity update rule is used to implement PSO algorithm. A combinatorial problem, Travelling Sales Person (TSP) is chosen as a case study to implement the basic approach of ACO and PSO and hybrid as well. During the implementation the problem found is that the local optimal solution is achieved but after that it stuck into stagnation phase where solution cannot be optimized more. newlineTo overcome this stagnant behaviour and to optimize the swarming behaviour of Ant Colony Optimization (ACO) there arise a need to design and implement new swarm component for TSP problem. So a New Swarm Component (NSC) is introduced to control the probabilistic rule behaviour for the best node selection for TSP problem. NSC helps to choose the best node for shortest path traversal and eliminate the duplicate node sel