<p>Fish feeding decision plays an essential role in controlling fish growth and ensuring water quality in aquaponic systems. The feeding decisions need to be optimized to satisfy production objectives. The aim of this paper is to develop a new feeding control optimization method based on reinforcement learning (RL) combined with background knowledge by simulation analysis. Firstly, to address the feeding optimization problem, a fundamental RL-based feeding control method is proposed. The objective of the fish feeding control problem is to minimize the error between end individual fish weight at harvest time and target value. Furthermore, a fish feeding control method based on RL combined with background knowledge is proposed, which integrates the feeding control decision of model predictive control (MPC) into RL. Specifically, MPC is initially applied to the known model of the aquaponic system to determine an optimal feeding control strategy. The obtained feeding control decision is considered the background knowledge of aquaponic systems to determine initial parameters of RL. Finally, the proposed methods are validated via simulation of an aquaponic system model. Different feeding control methods are compared to show performance with respect to end individual fish weight and feed conversion ratio (FCR). The simulation results show that the proposed feeding control method based on RL combined with background knowledge can make effective feeding decisions and decreases 10.04% in FCR compared with MPC method.</p>