摘要
针对能源问题和风力发电机组噪声检测过程复杂的现状,研究IEC 61400-11风力发电机组噪声测量技术标准,提出一种回归分析和BP神经网络相结合的方法,对风电机组噪声的A计权声压级进行预测。由风电现场采集的数据建立多元线性回归方程,求取回归系数,分析简化后,用较少的输入量训练BP神经网络,建立机组的噪声预测模型。将模型应用于新疆某风电场的实际测试过程中,效果良好,验证该方法的可行性和有效性。
Aiming at the complicated process of wind turbine noise detection, the IEC 61400-11 technology standard for noise measurement was studied, and a kind of forecasting method combining regression analysis with BP neural network was put forward. The A-weighted sound pressure level of wind turbine noise was forecasted. The multi-variable linear regression equation was established according to the in-situ collected noise data, and the regression coefficients were obtained. Then the simplified equation was used to train the BP neural network with less input data. Finally, the noise prediction model was established. This model was applied to noise measurement in a wind farm in Xinjiang efficiently. And the feasibility and effectiveness of this method was verified.
出处
《噪声与振动控制》
CSCD
2013年第6期49-52,123,共5页
Noise and Vibration Control
基金
国家自然科学基金(基金编号:51267017)
新疆大学自然科学基金(基金编号:XY110129)
关键词
声学
风力发电
噪声预测
回归分析
BP神经网络
acoustics
wind turbine generation
noise prediction
regression analysis
BP neural network
