With the rapid development in the field of artificial intelligence, reinforcement learning has gained increasing attention as a vital learning paradigm. In optimization problems, the Island Algorithm, as a heuristic evolutionary algorithm, has shown promise in solving various practical problems. However, traditional Island Algorithm encounters challenges in global search and escaping local optima. To address this issue, this paper introduces an enhanced Island Algorithm that incorporates reinforcement learning. By introducing the Q-learning algorithm, each plant is considered as an intelligent agent, dynamically selecting evolutionary strategies based on environmental feedback to effectively guide the algorithm’s search space. We evaluate the improved algorithm using six benchmark test functions in 10 and 30 dimensions, demonstrating significant improvements in various metrics compared to the traditional Island Algorithm. However, the results also reveal challenges in the stability of the algorithm.