Energy Load Forecasting using Deep Learning Approach-LSTM and GRU in Spark Cluster

Electric load forecasting has a significant role in power grids in order to facilitate the decision making process of energy generation & consumption. Long term forecasting is not feasible as there might be an uncertainty in the prediction because of irregular increase in the demand for power with the growing population and dependency on electric power. Since the behaviour of electric load time series is very much non-linear and seasonal, Neural Networks are best suited model for learning the Non-Linear behaviour within the data and for forecasting purpose. This paper deals with the Recurrent Neural Networks based Models: Long-Short-Term-Memory (LSTM) and Gated-Recurrent-Unit (GRU) to deal with this challenge. Observations have been made based on the distributed implementation of various configurations of LSTM-RNN & GRU-RNN on spark clusters for hyper parameter tuning purpose and deploying best suited configuration with least RMSE value using apache memos resource management.