Transfer learning applications for anomaly detection in wind turbines

Cyriana M.A. Roelofs,Christian Gück,Stefan Faulstich

DOI: https://doi.org/10.1016/j.egyai.2024.100373

2024-04-04

Abstract:Anomaly detection in wind turbines typically involves using normal behaviour models to detect faults early. However, training autoencoder models for each turbine is time-consuming and resource intensive. Thus, transfer learning becomes essential for wind turbines with limited data or applications with limited computational resources. This study examines how cross-turbine transfer learning can be applied to autoencoder-based anomaly detection. Here, autoencoders are combined with constant thresholds for the reconstruction error to determine if input data contains an anomaly. The models are initially trained on one year's worth of data from one or more source wind turbines. They are then fine-tuned using smaller amounts of data from another turbine. Three methods for fine-tuning are investigated: adjusting the entire autoencoder, only the decoder, or only the threshold of the model. The performance of the transfer learning models is compared to baseline models that were trained on one year's worth of data from the target wind turbine. The results of the tests conducted in this study indicate that models trained on data of multiple wind turbines do not improve the anomaly detection capability compared to models trained on data of one source wind turbine. In addition, modifying the model's threshold can lead to comparable or even superior performance compared to the baseline, whereas fine-tuning the decoder or autoencoder further enhances the models' performance.

Machine Learning,Artificial Intelligence

What problem does this paper attempt to address?

This paper investigates the application of transfer learning for anomaly detection in wind turbines. Typically, anomaly detection in wind turbines relies on normal behavior models to detect faults early. However, training an autoencoder model for each turbine is both time-consuming and resource-intensive. Therefore, for wind turbines with limited data or computational resources, transfer learning becomes crucial. The study focuses primarily on how transfer learning across turbines can be applied to autoencoder-based anomaly detection. The paper first trains a turbine with one year of data and then fine-tunes it with a small amount of data from another turbine. Three fine-tuning methods are studied: adjusting the entire autoencoder, adjusting only the decoder, or adjusting only the model's threshold. These transfer learning models are compared with baseline models trained using one year of data from the target turbine. The results show that models trained using data from multiple turbines do not significantly improve anomaly detection performance and do not have an advantage over models trained using data from a single source turbine. Additionally, adjusting the threshold may result in comparable or even better performance than the baseline model, while fine-tuning the decoder or the entire autoencoder further enhances model performance. The paper also validates the ability to achieve early fault detection even with limited training data through three case studies. The conclusion suggests that transfer learning models can effectively perform anomaly detection even with fine-tuning data as short as 1 to 3 months, reducing the required data and resources, thus showing potential in fault detection for wind turbines.

Transfer learning applications for anomaly detection in wind turbines

Vibration Anomaly Detection of Wind Turbine Based on Temporal Convolutional Network and Support Vector Data Description

Anomaly Detection on Small Wind Turbine Blades Using Deep Learning Algorithms

Transfer Learning for Fault Detection with Application to Wind Turbine SCADA Data

Supervised Transfer Learning Framework for Fault Diagnosis in Wind Turbines

Anomaly Detection for Wind Turbines Using Long Short-Term Memory-Based Variational Autoencoder Wasserstein Generation Adversarial Network under Semi-Supervised Training

Anomaly Detection of Wind Turbines Based on Deep Small-World Neural Network

Anomaly detection of wind turbine based on Norm-Linear-ConvNeXt-TCN

Diagnosis of wind turbine faults with transfer learning algorithms

Anomaly detection and diagnosis for wind turbines using long short-term memory-based stacked denoising autoencoders and XGBoost

Anomaly detection for wind turbines based on the reconstruction of condition parameters using stacked denoising autoencoders

Anomaly Prediction for Wind Turbines Using an Autoencoder with Vibration Data Supported by Power-Curve Filtering

Supervisory Control and Data Acquisition for Fault Diagnosis of Wind Turbines via Deep Transfer Learning

Machine Learning for Wind Turbine Fault Prediction through the Combination of Datasets from Same Type Turbines

LSTM-Autoencoder Based Anomaly Detection Using Vibration Data of Wind Turbines

Exploring the Limits of Early Predictive Maintenance in Wind Turbines Applying an Anomaly Detection Technique

Anomaly detection and critical SCADA parameters identification for wind turbines based on LSTM-AE neural network

A Novel Transfer Learning Method Based on Conditional Variational Generative Adversarial Networks for Fault Diagnosis of Wind Turbine Gearboxes under Variable Working Conditions

Anomaly Detection and Explanation of Wind Turbine Main Bearings using Autoencoder and Bayesian Network Models

CARE to Compare: A real-world dataset for anomaly detection in wind turbine data

Unsupervised anomaly detection using graph neural networks integrated with physical-statistical feature fusion and local-global learning