基于风电场功率预测的数据价值研究

Research on Data Value Based on Wind Farm Power Prediction

风电系统的运行伴随着海量数据的生成,是数据应用的重要场景。风电系统的输出受天气等因素影响,波动较大,因此风电输出的预测对于电力系统平衡、经济调度意义重大。为了探索风电系统中的数据价值,本文基于风电场的发电功率数据建立了小时级的多输入-多输出日前预测模型。进一步,本文对数据-模型-经济收益的链路进行梳理,结合合作博弈中的沙普利值思想,对模型中的数据点和特征进行价值评估。本文通过算例分析探索高价值数据的特点,结果表明筛选高价值数据能够实现风电系统降本增效,在此基础上为风电系统中数据的管理与共享提出了相应的建议。

The operation of the wind power system is accompanied by the generation of massive data,which is an essen-tial scenario for data application.The output of the wind power system is affected by factors such as weather and fluctu-ates wildly.Therefore,predicting wind power output is significant for power system balance and economic dispatch.In order to explore the value of data in wind power systems,this paper establishes an hour-level multi-input-multiple-output day-ahead prediction model based on the power generation data of wind farms.Further,this paper sorts out the link of data-model-economic benefits.It combines Shapley value in game theory to evaluate the model's value of data points and features.This paper explores the characteristics of high-value data through case analysis.The results show that screening high-value data can reduce costs and increase the efficiency of wind power systems.On this basis,correspond-ing suggestions are put forward for managing and sharing data in wind power systems.