基于风电场数据特征的功率预测与经济运行研究  

Research on Power Prediction and Economic Operation Based on Wind Farm Data Characteristics

作　　者：朱振军 ZHU Zhenjun(Wuling Power Corporation,Ltd.,Changsha 410004,China)

机构地区：[1]五凌电力有限公司,湖南长沙410004

出　　处：《微型电脑应用》2025年第2期222-225,248,共5页Microcomputer Applications

摘　　要：由于风力具有不确定性,导致风电波动性较强。为了对风电功率进行准确的预测,建立基于数据特征的功率预测系统。根据风电场数据特征对数值天气预报(NWP)风速数据进行修正,对异常风电数据进行滤波和补齐处理,并对风电机组进行实时监测,实现风电场经济性运行。风电场电能采集模块完成对风电场龙风线、主变线路高压侧、变压器等信息的采集。采用电压互感器(PT)二次回路压降的方法提高测量精度。利用拉丁超立方抽样策略(LHS)对灰狼寻优算法进行改进,保证种群初始参数抽样过程的稳定性,建立基于门控循环单元网络的功率预测系统,实现对风电功率的预测。实验结果显示,风电功率预测系统的训练时间仅为845 s,平均预测误差低于5%,预测精度最大。Because the wind is uncertain,the volatility of wind power is stronger.For accurate projections of wind power,power prediction system based on data characteristics is established.According to the characteristic of wind data on NWP,wind speed data are modified,the abnormal wind data are filtered and filled by data processing,and the wind turbines are monitored in real time to achieve the economic operation of the wind farm.The power acquisition module of wind farm collects the information of wind farm dragon line,high voltage side and transformer.It adopts the method of PT secondary loop pressure drop to improve the measurement accuracy.Latin hypercube sampling strategy(LHS)is used to improve the grey wolf optimization algorithm to ensure the stability of sampling process of initial parameters of the population.A power prediction model based on gate recurrent unit is established to predict wind power.The experimental results show that the training time of the wind power prediction model is only 845 s,and the average prediction error is less than 5%,which indicates the maximum prediction accuracy.

关 键 词：风电场数据特征 功率预测 PT二次回路 灰狼寻优算法 门控循环单元 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]