A Hybrid KAN-LSTM Model for Day-Ahead Price Forecasting in the Indian short term power market

The necessity of handling sudden rises in electricity demand, coping with unplanned maintenance of generating stations, or managing unforeseen circumstances has led to a gradual increase in participants and the volume of trade in the Day-Ahead Market over the years in the overall Indian short-term power market industry.The price sequence in the Day-Ahead Market (DAM) is highly volatile, non-linear, and influenced by various uncertainty parameters. To address this complexity, a hybrid model was developed that combines the Kolmogorov-Arnold Network (KAN) and LSTM networks with time series embeddings to forecast DAM price values in the short-term power market of the Indian Energy Exchange (IEX), providing support to all market participants.This model leverages historical data, including prices, weather conditions, and energy generation for the corresponding period. Statistical methods such as the Diebold-Mariano (DM) test and the Paired t-test are employed to assess the predictive accuracy of the proposed model compared to the LSTM model. Additionally, evaluation metrics such as RMSE, MAE, and R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> are used to validate the model’s performance.