Multi-Step Forecasting of Global Horizontal Irradiance Using Long Short-Term Memory Network for Solving Economic Dispatch Problem

Renewable energy resources are very effective to meet energy needs due to growing cost of fuel and decline in fossil fuel holds. Therefore, focus has shifted towards affordable sources of electricity. For incorporating solar energy into power system, there are two major challenges: accurate forecasting of solar irradiance and solving economic dispatch problem. This research proposes application of a novel technique known as long short-term memory (LSTM) network, belonging to recurrent neural network (RNN) class, for optimum day ahead scheduling of three conventional thermal generators and a solar photovoltaic power plant as a cost minimization problem. Multivariate multi-step global horizontal irradiance (GHI) forecasting models have been established using LSTM depending upon meteorological variables and historical data of GHI. The simulations for GHI prediction are being performed on dataset gathered in Meteorological Station of National University of Sciences and Technology (NUST), Islamabad, Pakistan. The method of dividing the required load demand between the generators available in power system while indulging all units, equality and inequality constraints of system is the economic load dispatch (ELD). The ELD problem is solved by lambda iteration (LI) method and LSTM technique with and without solar energy incorporation. The training and testing patterns of powers, total costs and power losses for LSTM are obtained by using LI method, whereas dataset of load is taken from NUST grid. By comparing the results for one day, it is observed that solar energy integration results in saving of operating cost and reduction of power loss.