Machine learning methods for wind turbine condition monitoring: A review
风力发电机监控系统中常用的一些机器学习的方法
Typical ML steps:
- data sources
- feature selection and extraction
- model selection (classification 2/3,regression 1/3)
- validation
- decision-making
Commonly use: neural networks, support vector machines, decision tree
Feature selection and extraction
Pre-process:(filtering methods)
- quantile filter, extremestudentized deviate test and the Hampel identifier
Feature selection
- Wrapper methods:需翻墙
- Embedded methods
- Filter methods