In practical industrial application, it is usually difficult to obtain sufficient labelled data due to the high cost of manual marking labels, which makes it hard to apply the traditional supervised learning in fault diagnosis. Unsupervised learning is a machine learning approach where the algorithm learns patterns and structures from unlabeled data without explicit input-output pairs, however, the trained model is often hard to be generalized without providing supervised information. While supervised learning relies solely on labeled data for training and unsupervised learning works with unlabeled data, semi-supervised learning leverages the benefits of both by using a smaller amount of labeled data along with a larger amount of unlabeled data. Semi-supervised learning in fault diagnosis helps to address the challenge of imbalanced data, where the number of faulty instances is often significantly fewer than that of normal instances. By utilizing unlabeled data, the model’s ability will be improved to generalize to various fault scenarios, including previously unseen ones. In this Chapter, semi-supervised intelligent fault diagnosis methods, such as Kernel Principal Component Analysis (KPCA), fuzzy kernel clustering algorithms, Self-organizing Map (SOM) neural networks, and Relevance Vector Machines (RVM), are introduced. These methods are validated and have suceeded in incipient fault diagnosis of transmissions and bearings.
Semi-supervised Learning Based Intelligent Fault Diagnosis Methods
Intelligent Fault Diagnosis and Health Assessment for Complex Electro-Mechanical Systems ; Chapter : 3 ; 95-191
2023-09-11
97 pages
Article/Chapter (Book)
Electronic Resource
English
Supervised SVM Based Intelligent Fault Diagnosis Methods
Springer Verlag | 2023
|Fault Diagnosis of Pumping Unit Based on Semi-Supervised Competitive Learning Process Neural Network
British Library Online Contents | 2014
|The Turnout Abnormality Diagnosis with Semi-supervised Learning Method
Springer Verlag | 2020
|