Hybrid Adaptive Ant System Algorithm and Its Application Research

摘要

In order to solve the problem of pheromone shortage and slow convergent speed of existing ant system algorithm (AS) in its application to complex optimal searching, this paper presents a new hybrid adaptive ant system algorithm with pheromone weight multiplier and pheromone balance operator, which can adaptively adjust select probabilities and pheromone strength. The simulation results — the global optimum value searching of multi-extremum continuous functions and nonlinear continuous function, solving complex TSP problem and neural network parameter learning with ant system algorithm -- present a good characteristic with lesser probability at local optimization, more strong global optimization capability and some good characteristics on reducing CPU time, and demonstrate the effectiveness of algorithm in improving optimization speed and adding result diversity.