Correlational broad learning for optimal scheduling of integrated energy systems considering distributed ground source heat pump heat storage systems

Renewable energies play an irreplaceable role in building an environment-friendly society; nevertheless, the curtailment phenomenon of renewable energies is serious. To improve the application of renewable energies and save economic costs, an optimal scheduling model based on distributed ground source heat pump heat storage system (DGSHPHSS) is established in this paper. The DGSHPHSS operates as an electrical load to store thermal energy during the valley load period, reducing the curtailment of wind power generation; DGSHPHSS provides thermal energy to the users during the peak load period, reducing the power cost. Besides, to more accurately and faster predict the load curve, this paper proposes a correlational broad learning (CBL) prediction model. The maximum wind power curtailment and economic costs with DGSHPHSSs under the low wind power curves are reduced by 50% and $254,500 than no DGSHPHSS, respectively; the peak values of wind power curtailment and economic costs with DGSHPHSSs under the high wind power curves are reduced by 40% and $65,300, respectively. The prediction model is inspired by sequence characteristics and external factors such as ambient temperature and time-of-use electrical prices. The prediction error obtained by the CBL prediction model can be reduced to 4.52% in simulating integrated energy systems (IESs).