针对海量风电数据下风电机组异常数据处理、功率曲线模型构建精度低等问题,提出一种基于IR-BIN算法的风功率曲线模型构建算法。首先,为了降低风电机组异常数据对风功率曲线模型准确性,提高风功率曲线对数据分布表征水平,提出了一种基于四分位法、准则的异常数据处理方法。其次,提出一种IR-BIN算法的风功率曲线构建方法,以提高风功率曲线构建准确性、减小利用风功率曲线构建误差,综合分析风功率曲线模型评价指标、曲线偏差、契合率等指标,评估所提方法构建曲线模型准确性。最后,以2019年5月份内蒙古塞罕坝风电机组数据进行分析,实验结果表明:与建模效果较好的IR方法相比所提方法R2提高了0.005,其余3种评价指标均处于最好值;所提IR-BIN方法构建曲线与理论曲线契合率约为0.9214,高于其余4种方法构建曲线契合率,应用于风电机组健康性能评估,评估结果与实际情况基本一致。In response to the challenges of abnormal data processing and the low accuracy in constructing power curve models for wind power units amidst extensive wind power big data, we propose an algorithm for wind power curve model construction based on the IR-BIN algorithm. Firstly, to enhance the accuracy of the wind power curve model and improve its representation of data distribution, we introduce an anomaly detection method utilizing quartile analysis and a three-sigma criterion. Secondly, we present a construction methodology for wind power curves grounded in the IR-BIN algorithm aimed at increasing modelling precision while minimizing errors in curve construction. The evaluation index, curve deviation and fit rate of the wind power curve model are analyzed comprehensively, and the accuracy of the curve model constructed by the proposed method is evaluated. Finally, based on the data of Saihanba Wind turbine in Inner Mongolia in May 2019, and the experimental results show that compared to the super
