【摘要】 为准确评估风电机组运行状态，保障机组正常运行，文章提出了一种基于风电机组数据采集与监视控制系统（SCADA）正常状态下数据和深度学习的机组运行状态评估方法。首先，在对SCADA原始数据进行清洗和归一化处理的基础上，利用最大相关最小冗余方法度量输出功率与多个采集量之间的关系，提取相关特征变量；然后，基于门控循环单元网络，智能提取正常数据间的分布规则，进而构建风电机组运行状态评估模型，采用预测功率与实际功率间的残差判定机组运行状态，即当机组发生故障时，残差将偏离原有的稳定状态；最后，选用某风电场发生故障时的实际数据对所提出的运行状态评估模型进行验证。结果表明，该模型可以提前预报风电机组的异常状态，从而为风电场及早安排预防性检修提供参考。更多还原

【Abstract】 In order to accurately evaluate the operating condition of wind turbines and ensure consistent operation, a wind turbine operating condition evaluation method based on data of wind turbine supervisory control and data acquisition(SCADA) in normal state and deep learning is proposed. First, on the basis of cleaning and normalizing the original SCADA data, max-relevance and min-redundancy(m RMR) is used to measure the relationship between the output power and multiple variables, and the relevant features are extracted. Then, based on the gated recurrent unit(GRU) network, the distribution rules between normal data are extracted to construct the operating condition evaluation model. The residual between the predicted power and the actual power is used to evaluate the operating condition of the wind turbine. When the wind turbine fails, the residual will deviate from the original steady condition. Finally, the actual SCADA data before and after the failure of a wind farm is used to verify the proposed method. The results show that the model can warn the abnormal condition of the wind turbine in advance, so as to provide a reference for the early preventive maintenance of the wind farm.更多还原