Abstract: To ensure the stable operation of school surveillance systems, research on the design and optimization of power distribution schemes for school surveillance systems is carried out. First, a hierarchical distributed architecture design is adopted, in which the perception and collection layer uses intelligent devices to collect power and environmental data. The data transmission layer builds a redundant ring network with the help of industrial Ethernet and unifies the data format. The intelligent control layer develops corresponding software and uses relevant formulas to analyze energy consumption trends. Then, an intelligent power distribution early warning system is constructed. Its front-end perception module is equipped with a fault arc detector, and the intelligent analysis module uses anomaly detection algorithms and anomaly scoring formulas to evaluate anomalies. The central management platform is responsible for receiving early warnings, implementing remote control, and generating reports. Finally, the load distribution is optimized. The power quality is monitored by leveraging the Internet of Things and big data. Data is transmitted back through wireless communication to construct a deep learning load forecasting model. Genetic algorithms are introduced to generate the optimal solution, enabling the system to adaptively adjust its strategy. The experimental results show that in the voltage test, after optimization, the voltage fluctuation in the teaching building was reduced from within ±5% to within ±2%. The library has been reduced from ±6% to within ±2.5%. The playground has been reduced from around ±7% to within ±3%. In terms of energy consumption comparison, the energy consumption of each monitoring point has decreased after optimization, with an overall average reduction of 10% to 15%. Moreover, the fluctuation of energy consumption data has become smaller.