State Estimation of Integrated Energy System Based on Deep Neural Network

The integrated energy system (IES) can improve the efficiency of energy utilization, so it is widely concerned. In order to realize the accurate perception of IES, it is necessary to conduct state estimation (SE). In this paper, a deep neural network (DNN) based SE method is proposed for IES with a large number of historical data. In this method, the spatiotemporal relationship between the data of the IES is established by screening the data with high correlation degree, and then the DNN model is trained with these data. The DNN algorithm assisted by the data screening (DS-DNN) solves the problem that the SE of gas system and thermal system in the dynamic process of IES is difficult to obtain accurate result quickly. The accuracy of the proposed SE algorithm is verified by the example simulation analysis of the IES.