基于多元角域指标离群检测的风电齿轮箱故障预警方法

针对风速和载荷随机波动造成的早期故障特征难以提取和量化的特点,提出基于多元角域指标离群检测的风电齿轮箱故障预警方法。首先利用阶比重采样技术将非平稳的时域振动信号转化为具有平稳或准平稳特性的角域信号,提取角域无量纲指标反映风电齿轮箱早期故障趋势;其次利用多元相关度测量建立风电齿轮箱早期故障识别模型;最后采用多元离群检测方法实现风电齿轮箱早期故障预警。实验表明,该方法能够实现较为准确地的风电齿轮箱早期故障预警,具有较好的工程实际意义。

Early warning method for wind turbine gearbox based on multivariate outlier detection of angle domain parameters

According to the characters of difficulties for extracting and quantifying the early faults features under fluctuations working conditions，proposing the wind turbine gearbox fault warning method based on multivariate．Firstly，transforming the non-stationary time domain signals into stationary angle domain signals and extracting the angle domain dimensionless parameters，reflecting the early wind turbine gearbox faults by extracting the angle domain dimensionless；Secondly，establishing the early faults cognition model based on correlation measure analysis；Finally，realizing the wind turbine gearbox faults early warning by means of Multivariate outlier detection．Experimental results show that the method can achieve early warning more accurate and has good practical significance.