1. This article discusses the condition monitoring of wind turbines, which can be divided into three main categories: physical model, signal driven and data driven.
2. The data-driven methods can be further divided into classification, operation curve and normal behavior model (NBM).
3. This article proposes an incremental learning strategy for NBM based on multivariate state estimation technique (MSET), which includes a similarity-based sample selection method and a dynamic down-sampling method to ensure the normality of incremental data and real-time remove the redundant samples in MM.
This article provides a comprehensive overview of condition monitoring techniques for wind turbines, discussing various approaches such as physical models, signal driven methods, and data driven methods. The authors then focus on the data driven approach, providing an in-depth analysis of different algorithms such as relevance vector machine, shallow neural networks, deep learning, and multivariate state estimation technique (MSET). They propose an incremental learning strategy for MSET that includes a similarity-based sample selection method and a dynamic down-sampling method to ensure the normality of incremental data and real-time removal of redundant samples in MM.
The article is generally reliable and trustworthy due to its comprehensive coverage of the topic at hand. It provides detailed explanations of each approach discussed as well as relevant references to back up its claims. Furthermore, it presents both advantages and disadvantages of each approach discussed in order to provide readers with a balanced view on the topic.
However, there are some potential biases that should be noted when reading this article. For example, while it does discuss various approaches to condition monitoring for wind turbines, it does not explore any counterarguments or alternative approaches that may exist outside those discussed in detail within the paper. Additionally, while it does provide references for each claim made throughout the paper, some sources are outdated or lack sufficient detail to fully support all claims made by the authors.