A Novel Method to Predict The Power Demand of Consumers with Ev using Encoder Decoder LSTM
Prajeesha, Shreepad Udgire, Siddharth Vivek Kamath, Vishal Singh S, Srinidhi Srinidhi, M. Anuradha · 2024
Ahstract-LSTM is a recurrent neural network model used for capturing long-term dependencies in sequence prediction functions. This paper proposes an LSTM encoder-decoder net-work model to predict the power load demand of a certain area under a Data Concentrator Unit (DCU)and optimize the power generation and consumption. The traditional meters are replaced with smart meters that are capable of capturing the fluctuations and power usage patterns of consumers. The data captured by the smart meter includes the consumption by normal household appliances that come under inductive load, capacitors load, resistive load, charging points of electric vehicles, etc. The involvement of electric vehicle charging points gives high fluctuations in the consumption of power depending on the vehicle, charging time, the time of use, etc. This data collected from the smart meters for every 15 minutes is used to train the model to understand and predict the usage pattern of consumers. The predicted pattern can be used for controlling the power generation and its distribution effectively from a substation which avoids outages. Our model outperforms the standard LSTMs in terms of RMSE which in turn predicts the efficiency of the entire system.