Using Transfer Learning to Build Physics-Informed Machine Learning Models for Improved Wind Farm Monitoring

This paper introduces a novel, transfer-learning-based approach to include physics into data-driven normal behavior monitoring models which are used for detecting turbine anomalies. For this purpose, a normal behavior model is pretrained on a large simulation database and is recalibrated on the available SCADA data via transfer learning. For two methods, a feed-forward artificial neural network (ANN) and an autoencoder, it is investigated under which conditions it can be helpful to include simulations into SCADA-based monitoring systems. The results show that when only one month of SCADA data is available, both the prediction accuracy as well as the prediction robustness of an ANN are significantly improved by adding physics constraints from a pretrained model. As the autoencoder reconstructs the power from itself, it is already able to accurately model the normal behavior power. Therefore, including simulations into the model does not improve its prediction performance and robustness significantly. The validation of the physics-informed ANN on one month of raw SCADA data shows that it is able to successfully detect a recorded blade angle anomaly with an improved precision due to fewer false positives compared to its purely SCADA data-based counterpart.

[1]  Niklas Renström,et al.  System-wide anomaly detection in wind turbines using deep autoencoders , 2020 .

[2]  N. Dimitrov,et al.  A surrogate model approach for associating wind farm load variations with turbine failures , 2020, Wind Energy Science.

[3]  Jian Li,et al.  Anomaly detection for wind turbines based on the reconstruction of condition parameters using stacked denoising autoencoders , 2020 .

[4]  Nikolay Krasimirov Dimitrov,et al.  Surrogate models for parameterized representation of wake‐induced loads in wind farms , 2019, Wind Energy.

[5]  Frederik Zahle,et al.  Implementation of the blade element momentum model on a polar grid and its aeroelastic load impact , 2019, Wind Energy Science.

[6]  Eric Simley,et al.  OpenOA: An Open-Source Code Base for Operational Analysis of Wind Power Plants , 2019 .

[7]  Goran Nenadic,et al.  Machine learning methods for wind turbine condition monitoring: A review , 2019, Renewable Energy.

[8]  Christian Bauckhage,et al.  Informed Machine Learning - Towards a Taxonomy of Explicit Integration of Knowledge into Machine Learning , 2019, ArXiv.

[9]  Felipe A. C. Viana,et al.  Fleet Prognosis with Physics-informed Recurrent Neural Networks , 2019, Structural Health Monitoring 2019.

[10]  Ramakrishna Tipireddy,et al.  Physics-informed Machine Learning Method for Forecasting and Uncertainty Quantification of Partially Observed and Unobserved States in Power Grids , 2018, HICSS.

[11]  John Aasted Sørensen,et al.  Wind turbine site-specific load estimation using artificial neural networks calibrated by means of high-fidelity load simulations , 2018, Journal of Physics: Conference Series.

[12]  Simon J. Watson,et al.  Using SCADA data for wind turbine condition monitoring – a review , 2017 .

[13]  Julio J. Melero,et al.  Wind Turbine Failures - Tackling current Problems in Failure Data Analysis , 2016 .

[14]  Xiandong Ma,et al.  Model-based and fuzzy logic approaches to condition monitoring of operational wind turbines , 2015, Int. J. Autom. Comput..

[15]  Meik Schlechtingen,et al.  Comparative analysis of neural network and regression based condition monitoring approaches for wind turbine fault detection , 2011 .

[16]  David Infield,et al.  Online wind turbine fault detection through automated SCADA data analysis , 2009 .

[17]  Andrew Kusiak,et al.  Models for monitoring wind farm power , 2009 .

[18]  Torben J. Larsen,et al.  Wake meandering: a pragmatic approach , 2008 .

[19]  Miguel A. Sanz-Bobi,et al.  SIMAP: Intelligent System for Predictive Maintenance: Application to the health condition monitoring of a windturbine gearbox , 2006, Comput. Ind..

[20]  Sofiane Achiche,et al.  Wind turbine condition monitoring based on SCADA data using normal behavior models. Part 1: System description , 2013, Appl. Soft Comput..