Research on Power Probability Forecasting Method for Photovoltaic Cluster Based on MIC and MQ-WaveNet

View at publisher Summarise with VIRAT
Authors: Tianjiao Ma, Qing Li, Zhaoxiang Deng
Cited by: 0
References: 19

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

To quantify photovoltaic (PV) power uncertainty in a region, a direct multi-step probabilistic forecasting method for PV clusters is proposed, based on key influence variables from representative field stations. First, representative stations and key explanatory variables are selected using the Maximal Information Coefficient (MIC). Second, a multi-view quantile regression WaveNet (MQ-WaveNet) is developed, integrating quantile regression with WaveNet's long-range temporal modeling capability; it employs a residual-connected encoder and a dual-MLP decoder to directly predict multiple future steps and quantiles for the PV cluster. Case studies use data from 14 adjacent PV stations in a northwest China base. Results show the method significantly improves modeling efficiency, with both deterministic and probabilistic forecasts outperforming mainstream models.

Fields of Study & Subjects

Solar Radiation and Photovoltaics Energy Load and Power Forecasting Photovoltaic System Optimization Techniques

Publication details

Journal / source
International Conference on Electrical Automation and Artificial Intelligence (ICEAAI)
Pages
363–368
Publication date
16 January 2026
Language
EN