Forecasting of Market Power using LSTM in day ahead market

In a non-competitive power market, the market participants such as generation companies (GENCOs) objective is to maximize profit by submitting their bids to the Independent System Operator (ISO). The GENCOs strategic behavior exercises market power and it becomes difficult for the market to achieve perfect competitiveness. The market power is evaluated through various indices like Herfindahl–Hirschmann Index (HHI), and the Residual Supplier Index (RSI). The paper presents an approach to forecast the market power by proposed an economic-based market power index (MPI) for the GENCOs using long short-term memory (LSTM) with uncertainties such as line contingencies. The methodology provides insight to the financial perspective rather than the structural perspective of market power. With 100% loading, HHI is 2038 and RSI is 0.722, 87.5% loading HHI is 2122 and RSI is 0.722, 75% loading HHI is 2226 and RSI is 0.665, 62.5% loading HHI is 2550 and RSI is 0.722, 50% loading HHI is 3038 and RSI is 0.584. The proposed methodology is applied to the IEEE 30 bus system and calculated for the supplier side for effectively monitoring the behaviour of GENCOs under uncertainties.