Study of Lightweight Virtualization Techniques Combined with Deep Learning in the Power Internet of Things (IoT)

Aiming at the problems of low efficiency and poor cross-platform compatibility in power load forecasting in the context of the power Internet of Things (IoT), this paper proposes a multi-hardware platform load forecasting program based on the lightweight container engine iSulaand plf BP-GA. By encapsulating a backpropagation neural network power load forecasting program improved by genetic algorithm and transfer learning through the iSula engine, the resource occupancy rate of the load forecasting program is optimized, enabling efficient running on domestic open-source platforms. Experimental results show that the proposed method has significantadvantages in prediction accuracy. Compared to running on physical hardware, containerized running can reduce resource occupancy while having high prediction accuracy, good compatibility and stability across multiple platforms.