控制与信息技术Issue(5)：79-86,8.DOI:10.13889/j.issn.2096-5427.2025.05.400
基于发电功率数据挖掘的光伏电站动态清洁周期算法
Dynamic Cleaning Cycle Algorithm for PV Stations Based on Power Generation Data Mining
陈炳煌 1黄海滨 1陈敏龙 1庄琛 1康孝种1
作者信息
- 1. 国能(泉州)热电有限公司,福建 泉州 362804
- 折叠
摘要
Abstract
Excessive dust accumulation on the surface of photovoltaic(PV)modules can significantly reduce power generation efficiency and revenues of PV plants,and in severe cases may even lead to safety incidents.Therefore,implementing a rational and reliable cleaning decision strategy is crucial.This study proposes an algorithm that optimizes support vector regression(SVR)using an improved grid search(GS)method.A predictive model for ideal power under dust-free conditions is developed based on historical data of optimal cleaning cycles from power plants.This model accurately forecasts ideal power outputs under dust-free conditions,enabling the quantification of generation losses by comparing these forecasts with actual outputs.Subsequent comprehensive analysis,incorporating calculated generation losses and cleaning costs,facilitates dynamic optimization of cleaning cycles to achieve an optimal balance between operational costs and revenues.This algorithm enhances traditional GS methods by incorporating binary search operations in the parameter search space and step size scaling,significantly improving optimization efficiency for the penalty coefficient c and kernel parameter g in the SVR model while reducing overall computation time.The proposed algorithm achieved an average absolute error of 0.944 and root mean square error(RMSE)of 1.213,demonstrating excellent prediction accuracy.Not only does it efficiently and accurately predict generation losses in PV plants,but it also provides reliable technical support for intelligent and refined decision-making in cleaning operations,effectively ensuring both power generation revenue and operational safety.关键词
光伏组件/清洁决策/积灰清洗/发电量损失/改进网格搜索/支持向量回归Key words
photovoltaic module/cleaning decision/dust cleaning/power generation loss/improved grid search(GS)/support vector regression(SVR)分类
电子信息工程引用本文复制引用
陈炳煌,黄海滨,陈敏龙,庄琛,康孝种..基于发电功率数据挖掘的光伏电站动态清洁周期算法[J].控制与信息技术,2025,(5):79-86,8.
