Prediction Method of Direct Normal Irradiance for Solar Thermal Power Plants Based on VMD-WOA-DELM
Siyuan Zhang, Dongsheng Niu, Zhi Zhou, Yanglong Duan, Jian Chen, Genben Yang
Abstract
The Direct Normal Irradiance (DNI), being the energy source for solar thermal power plants, can remarkably impact the reliability and efficiency of these plants because of its inherent randomness and fluctuations. In this view, we propose a prediction method for DNI based on the Variational Mode Decomposition-Whale Optimization Algorithm-Deep Extreme Learning Machine (VMD-WOA-DELM) to optimize the control and operation of such plants. Initially, the VMD technique is utilized to decompose the DNI into intrinsic mode function components, followed by the extraction of temporal and frequency domain characteristics to form feature vectors for each component. Subsequently, the WOA is employed for parameter optimization, enhancing algorithm stability, and yielding the optimal classification model. Finally, the solar DNI is determined by an improved Extreme Learning Machine algorithm, DELM. Taking a solar thermal power plant in Qinghai Province as a case study, an analysis of actual predictive performance and corresponding performance evaluation indicators concludes that the variations and numerical values of DNI can be accurately forecasted using the established prediction approach.