基于机器学习的风电机组机械传动系统故障诊断研究
Research on Fault Diagnosis of Wind Turbine Mechanical Transmission System Based on Machine Learning
宾世杨 1张振 1唐俊杰 1唐惜春1
作者信息
- 1. 国家电投集团广西兴安风电有限公司,广西 桂林 541300
- 折叠
摘要
为准确诊断风电机组机械传动系统故障,提出一种基于机器学习的风电机组机械传动系统故障诊断方法.通过经验模态分解(EMD)方法分解风电机组机械传动系统振动信号,获取不同频率下的固有模态函数(IMF),经过对比分析获取可以描述故障特征频率的IMF分量,经过重构得到故障信号,使用自相关分析法去除故障信号中的噪声.通过机器学习中的Lasso正则化自编码神经网络提取风电机组机械传动系统故障特征,采用改进的粒子群算法对最小二乘支持向量机优化处理,构建分类器,将提取到的样本输入到分类器中,完成风电机组机械传动系统故障诊断.经实验测试证明,所提方法能够高效率、高精度地完成故障诊断处理.
Abstract
In order to accurately diagnose the faults of the mechanical transmission system of wind tur-bines,a fault diagnosis method of the mechanical transmission system of wind turbines based on machine learning is proposed.The vibration signal of the mechanical transmission system of the wind turbine is de-composed by the EMD method,and the IMF at different frequencies is obtained.After comparative analy-sis,the IMF component that can describe the characteristic frequency of the fault is obtained,and the fault signal is obtained through reconstruction and autocorrelation analysis is used to remove noise from faulty signals.The fault features of the mechanical transmission system of wind turbines are extracted through the Lasso regularized self-encoding neural network in machine learning,and the improved particle swarm algorithm is used to optimize the least squares support vector machine,and a classifier is constructed.The extracted samples are input into the classifier to accomplish fault diagnosis of wind turbine mechanical transmission systems.The experimental test proves that the proposed method can complete the fault diag-nosis and processing with high efficiency and high precision.
关键词
机器学习/风电机组/机械传动系统故障诊断/EMDKey words
machine learning/wind turbines/mechanical transmission system fault diagnosis/EMD引用本文复制引用
基金项目
广西电网有限责任公司科技项目(040600KK52100012)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据