Day-Ahead Electricity Price Forecasting Based on MAMBA

The volatility and uncertainty of day-ahead electricity prices are increasing, and existing linear models and traditional methods show significant limitations in capturing the nonlinear relationships and temporal dynamics of price fluctuations, making them insufficient for practical forecasting needs. This paper proposes a day-ahead electricity price forecasting method based on the Multivariate Adaptive Model-based Algorithm (MAMBA). First, the method introduces a gating mechanism and implicit state modeling through the MAMBA framework, effectively capturing long-term dependencies in time series data while maintaining low computational cost. Then, the model performs feature extraction on the multivariate time series inputs and leverages an adaptive mechanism to identify nonlinear relationships and key importance among different features, thereby enhancing the model’s ability to represent complex market characteristics. Finally, based on the extracted temporal features and inter-variable dependencies, the model outputs high-precision electricity price forecasts, achieving accurate characterization and stable prediction of electricity market trends. Empirical analysis using data from the Singapore electricity spot market demonstrates that the MAMBA-based forecasting model outperforms ARIMA, BiLSTM, and Transformer models in terms of RMSE, MAE, and MAPE, validating its effectiveness in handling complex electricity price volatility scenarios.