Multiple ant colony optimization using both novel LSTM network and adaptive Tanimoto communication strategy

摘要

Ant Colony Optimization (ACO) tends to fall into local optima and has insufficient convergence when solving the Traveling Salesman Problem (TSP). To overcome this problem, this paper proposes a multiple ant colony optimization (LDTACO) based on novel Long Short-Term Memory network and adaptive Tanimoto communication strategy. Firstly, we introduce an Artificial Bee Colony-based Ant Colony System (ABC-ACS), which along with the classic Ant Colony System (ACS) and Max-Min Ant System (MMAS), form the final proposed algorithm. These three types of subpopulations complement each other to improve overall optimization performance. Secondly, the evaluation reward mechanism is proposed to enhance the guiding role of the Recommended paths, which can effectively accelerate convergence speed. Besides, an adaptive Tanimoto communication strategy is put forward for interspecific communication. When the algorithm is stagnant, the homogenized information communication method is activated to help the algorithm jump out of the local optima, thus improving solution accuracy. Finally, the experimental results show that the proposed algorithm can lead to more accurate solution accuracy and faster convergence speed.