This study investigates the dynamic processes of epidemic spread and policy optimization in heterogeneous Barabasi-Albert networks using the susceptible-infectious-recovered-susceptible model. By systematically adjusting key parameters, including the number of inter-subnetwork connections, the number of sub-networks, and node connection strategies, we conducted a comprehensive analysis of the impact of these factors on epidemic transmission. This analysis revealed the critical role of network structure in disease spread. The findings indicate that augmenting inter-subnetwork connections and node connections expedites the propagation of the epidemic, culminating in elevated infection peaks but diminished overall epidemic durations. The model validation is further substantiated by the use of real-world data from the outbreak of the novel coronavirus disease in Wuhan, China, with the simulation results demonstrating a close alignment with the observed trends. This finding serves to substantiate the model's efficacy. Studies on policy optimization have indicated that the premature relaxation of control measures can result in elevated infection peaks. Conversely, the easing of measures at opportune times can facilitate more effective epidemic control. This research establishes a theoretical framework for public health decision-making, particularly in terms of balancing epidemic control with socio-economic recovery.