Solving unbounded knapsack problem using evolutionary algorithms with bound constrained strategy

Solving unbounded knapsack problem using evolutionary algorithms with bound constrained strategy

Unbound Knapsack Problems (UKP) are important research topics in many fields like portfolio and asset selection, selection of minimum raw materials to reduce the waste, and generating keys for cryptosystems. Given the uncertainty in data, capacity, and time constraints, users have to look at the possible combination of data to get maximum benefit. This paper uses UKP as a numerical model to represent different industrial combination problems. It applies Evolutionary Algorithms (EA) with Bound Constrained Strategy (BCS) to construct a search space and algorithm parameters for finding the optimal solution. Evolutionary Algorithms (EA) like Genetic Algorithms (GA) and Particle Swarm Optimization (PSO) are designed based on reusable components for the algorithms to converge faster. Simulation for various objectives indicates that the GA and PSO can find the near-optimal solution in all cases. The execution time of GA and PSO for different goals and the variations in the algorithm parameters are measured. The measurement result shows the performance of GA and PSO is the same on an average for the differences in bounded constraints and parameter settings.