基于运行数据和高斯过程回归的风电机组发电性能分析与监测

Performance analysis and monitoring based on SCADA data and Gaussian process regression for wind turbine power generation

采用SCADA运行数据,结合风电机组的运行原理,详细分析了对发电性能有密切影响的因素,包括环境因素及机组各个子部件如变桨系统、偏航系统、控制系统的运行状态。采用适合风电机组运行数据强随机性和高噪声特点的高斯过程回归方法建立了发电性能模型。该模型表征了机组发电性能正常时风能利用系数与其影响因素之间的复杂关系,将实时运行数据作为发电性能模型输入,通过分析模型预测残差能够实时监测风电机组发电性能的异常变化。通过风电场实际运行数据仿真,验证了所提方法的可行性。

According to the operational theories of wind turbine and based on the SCADA data,the major influencing factors of its power generation performance are analyzed,including the environmental factors and the operating conditions of wind turbine components,such as pitch system,yaw system and control system. Gaussian process regression is applied in the construction of power generation performance model to adapt to the high randomness and strong noise of wind turbine operating data. The constructed model describes the relationship between the wind power utilization coefficient and its influencing factors ,which is applied to monitor the abnormal change of wind turbine power generation performance in real time by analyzing the residual of model prediction with the real time operating data as the model inputs. The feasibility of the proposed method is verified by the simulation based on the actual operating data of a wind farm.