Research on Auxiliary Decision Making Algorithm for New Energy Participation in Multi Level and Multi Variety Electricity Market Based on 1DCNN-SLSTM

In response to the volatility and nonlinearity of new energy power data, this paper presents a 1DCNN-SLSTM-based auxiliary decision-making algorithm for the new energy power market. 1DCNN extracts short-term local power data features, and SLSTM models long-term temporal dependencies, capturing seasonal and daily cycles. Multi-source feature fusion enables high-precision predictions. Experiments show the model surpasses ARIMA, LSTM, and SVM in RMSE (0.035), MAE (0.028), and $\mathrm{R}^{2}(0.92)$. Despite a 25 -minute training time, the 0.12second prediction time meets real-time needs and can be optimized with hardware. The findings offer reliable algorithmic support for new energy market scheduling and risk control.