Abstract—The uneven distribution of single-phase loads in low-voltage (LV) distribution network and the uncertainty of consumers’ electricity behavior aggravate the three-phase imbalance issue, which further influence the safe and economical operation of power system. This paper proposes an optimization method based on load forecasting technique for mitigating three-phase imbalance issue in LV distribution network. Firstly, clustering algorithm is used to analyze the consumers’ electricity consumption behavior. Secondly, a deep learning algorithm based on long short-term memory (LSTM) network is applied to predict future load data of distribution network. Compared to traditional optimization method using historical load data, this method is designed to achieve better optimization performance by establishing a new three-phase imbalance model based on predicted load data. Simulation is conducted using realistic load data. Results indicate that the proposed optimization method can provide various phase-sequence adjustment strategies from different perspectives to meet the specific operation requirement of certain area. Consequently, the safety and stability of distribution network can be enhanced.

distribution network;load forecasting;long short-term memory network;three-phase imbalance