With the rapid development of Internet of Things technology and the increasing demand for electricity, intelligent analysis and processing of electricity data based on deep learning has important practical significance for user electricity consumption behavior and improving energy efficiency. This article took the power Internet of Things (PIoT) as the background, based on deep learning, to analyze and predict its intelligent electricity consumption behavior. On this basis, a multi-layer neural network-based method for identifying electricity consumption behavior was adopted. This article intended to conduct research in four stages: data preprocessing, feature engineering, model training, and result validation. On January 1, 2023, the electricity consumption was 120.5kWh, with a temperature of 10 ℃ and a humidity of 60%; the electricity consumption from January to February 2023 was 118.2kWh, with a temperature of 12 ℃ and a humidity of 55%. The research results of this article can help to gain a deeper understanding of user electricity consumption behavior and provide scientific basis for optimizing power grid operation and intelligent decision-making.