Constructing distributed photovoltaic systems on industrial building rooftops and establishing adaptive microgrid energy flow scheduling models are effective means of achieving full deployment and consumption of photovoltaics. The operational modes and stakeholders involved in shared energy storage and peer-to-peer trading differ significantly, influencing both the energy flow scheduling and on-site consumption rates of microgrids. In this study, a dual-objective function model with multiple constraints was designed, and particle swarm optimization was applied to seek optimal energy storage capacity. Pareto optimal solution sets under dual objectives with conflicting interests were explored, and comparative analyses of energy flow scheduling under shared energy storage and peer-to-peer trading microgrid modes were conducted. Furthermore, the daily cost-effectiveness of these two modes was evaluated. The results indicated that stakeholders with conflicting interests difficult to achieve win–win outcomes in energy scheduling. The integration of peer-to-peer trading not only reduced shared energy storage capacity by 18% but also achieved local consumption rates of 62% and 100% in summer and winter, respectively. Peer-to-peer trading had a minimal impact on user power costs, yet it increases power revenues by 32% and 235% in summer and winter, respectively, thereby reducing the net expenditure of the users. The results of this study provide a reference for microgrid energy flow scheduling models involving entities of uncommon interest in industrial parks.