The heterogeneous networks formed by the integration of VLC (visible light communication) and RF (radio frequency) communication could fully leverage their complementary advantages. Coordinating the access process of different networks and designing efficient resource allocation scheme are the core issues for improving system performance in heterogeneous networks. The resource allocation algorithm was designed for VLC-RF networks, which aims to fulfill the communication requirements of terminal devices and efficiently guarantee heterogeneous QoS (quality of service). First, to address the susceptibility of interruptions in VLC channel transmission, the cooperative service mechanism of VLC-RF is established, and the system service capability was evaluated based on the effective capacity theory. MMOO (Markov modulated on-off) process was adopted to model the service traffic, and the effective bandwidth theory was utilized to analyze the bandwidth requirements of differentiated service arrivals. Then, with the objective of minimizing the RF service rate, the resource allocation optimization problem was formulated. This problem incorporates the constraints of user QoS requirements, allocation factors, and transmission probabilities. The reverse learning strategy and the adaptive mutation mechanism were introduced to improve ALA (artificial lemming algorithm). Based on IALA (improved artificial lemming algorithm), the optimization problem was solved. Subsequently, the resource allocation algorithm was designed, and the optimal allocation scheme was determined. Finally, the proposed algorithm was verified and analyzed by the MATLAB simulation platform. The results demonstrate that the proposed algorithm could achieve efficient resource utilization while ensuring the QoS requirements of the terminals.