Due to the dynamic nature of opportunistic networks, the network environment undergoes frequent changes where efficient forwarding is crucial to alleviate the occurrences of situations such as node buffer overflow or inefficient utilization of network resources. In this research, we apply reinforcement learning by treating each node in the network as a reinforcement learning agent. Firstly, we utilize the encounter probability between nodes and the remaining cache ratio to construct a forwarding utility learning model based on node cache optimization. Then, integrating the contextual information of nodes, we propose the Opportunistic Routing using Q-Learning with Context Information (ORQLCI) algorithm. We validated the ORQLCI algorithm proposed in this paper using simulations and the results demonstrate that compared to opportunistic network routing algorithms assisted by node information, our scheme effectively improves message delivery ratio, reduces routing overhead, and decreases network latency.