Bias correction of wind power forecasts with SCADA data and continuous learning

Wind energy plays a critical role in the transition towards renewable energy sources. However, the uncertainty and variability of wind can impede its full potential and the necessary growth of wind power capacity. To mitigate these challenges, wind power forecasting methods are employed for applications in power management, energy trading, or maintenance scheduling. In this work, we present, evaluate, and compare four machine learning-based wind power forecasting models. Our models correct and improve 48-hour forecasts extracted from a numerical weather prediction (NWP) model. The models are evaluated on datasets from a wind park comprising 65 wind turbines. The best improvement in forecasting error and mean bias was achieved by a convolutional neural network, reducing the average NRMSE down to 22%, coupled with a significant reduction in mean bias, compared to a NRMSE of 35% from the strongly biased baseline model using uncorrected NWP forecasts. Our findings further indicate that changes to neural network architectures play a minor role in affecting the forecasting performance, and that future research should rather investigate changes in the model pipeline. Moreover, we introduce a continuous learning strategy, which is shown to achieve the highest forecasting performance improvements when new data is made available.

A review of federated learning in renewable energy applications: Potential, challenges, and future directions

Federated learning has recently emerged as a privacy-preserving distributed machine learning approach. Federated learning enables collaborative training of multiple clients and entire fleets without sharing the involved training datasets. By preserving data privacy, federated learning has the potential to overcome the lack of data sharing in the renewable energy sector which is inhibiting innovation, research and development. Our paper provides an overview of federated learning in renewable energy applications. We discuss federated learning algorithms and survey their applications and case studies in renewable energy generation and consumption. We also evaluate the potential and the challenges associated with federated learning applied in power and energy contexts. Finally, we outline promising future research directions in federated learning for applications in renewable energy.

Towards Fleet-wide Sharing of Wind Turbine Condition Information through Privacy-preserving Federated Learning

Terabytes of data are collected every day by wind turbine manufacturers from their fleets. The data contain valuable real-time information for turbine health diagnostics and performance monitoring, for predicting rare failures and the remaining service life of critical parts. And yet, this wealth of data from wind turbine fleets remains inaccessible to operators, utility companies, and researchers as manufacturing companies prefer the privacy of their fleets' turbine data for business strategic reasons. The lack of data access impedes the exploitation of opportunities, such as improving data-driven turbine operation and maintenance strategies and reducing downtimes. We present a distributed federated machine learning approach that leaves the data on the wind turbines to preserve the data privacy, as desired by manufacturers, while still enabling fleet-wide learning on those local data. We demonstrate in a case study that wind turbines which are scarce in representative training data benefit from more accurate fault detection models with federated learning, while no turbine experiences a loss in model performance by participating in the federated learning process. When comparing conventional and federated training processes, the average model training time rises significantly by a factor of 7 in the federated training due to increased communication and overhead operations. Thus, model training times might constitute an impediment that needs to be further explored and alleviated in federated learning applications, especially for large wind turbine fleets.

Vibration fault detection in wind turbines based on normal behaviour models without feature engineering

Most wind turbines are remotely monitored 24/7 to allow for an early detection of operation problems and developing damage. We present a new fault detection method for vibration-monitored drivetrains that does not require any feature engineering. Our method relies on a simple model architecture to enable a straightforward implementation in practice. We propose to apply convolutional autoencoders for identifying and extracting the most relevant features from the half spectrum in an automated manner, saving time and effort. Thereby, a spectral model of the normal vibration response is learnt for the monitored component from past measurements. We demonstrate that the model can successfully distinguish damaged from healthy components and detect a damaged generator bearing and damaged gearbox parts from their vibration responses. Using measurements from commercial wind turbines and a test rig, we show that vibration-based fault detection in wind turbine drivetrains can be performed without the usual upfront definition of spectral features. Another advantage of the presented method is that the entire half spectrum is monitored instead of the usual focus on monitoring individual frequencies and harmonics.

Vibration Fault Diagnosis in Wind Turbines based on Automated Feature Learning

A growing number of wind turbines are equipped with vibration measurement systems to enable a close monitoring and early detection of developing fault conditions. The vibration measurements are analyzed to continuously assess the component health and prevent failures that can result in downtimes. This study focuses on gearbox monitoring but is applicable also to other subsystems. The current state-of-the-art gearbox fault diagnosis algorithms rely on statistical or machine learning methods based on fault signatures that have been defined by human analysts. This has multiple disadvantages. Defining the fault signatures by human analysts is a time-intensive process that requires highly detailed knowledge of the gearbox composition. This effort needs to be repeated for every new turbine, so it does not scale well with the increasing number of monitored turbines, especially in fast growing portfolios. Moreover, fault signatures defined by human analysts can result in biased and imprecise decision boundaries that lead to imprecise and uncertain fault diagnosis decisions. We present a novel accurate fault diagnosis method for vibration-monitored wind turbine components that overcomes these disadvantages. Our approach combines autonomous data-driven learning of fault signatures and health state classification based on convolutional neural networks and isolation forests. We demonstrate its performance with vibration measurements from two wind turbine gearboxes. Unlike the state-of-the-art methods, our approach does not require gearbox-type specific diagnosis expertise and is not restricted to predefined frequencies or spectral ranges but can monitor the full spectrum at once.

Early fault detection with multi-target neural networks

Wind power is seeing a strong growth around the world. At the same time, shrinking profit margins in the energy markets let wind farm managers explore options for cost reductions in the turbine operation and maintenance. Sensor-based condition monitoring facilitates remote diagnostics of turbine subsystems, enabling faster responses when unforeseen maintenance is required. Condition monitoring with data from the turbines' supervisory control and data acquisition (SCADA) systems was proposed and SCADA-based fault detection and diagnosis approaches introduced based on single-task normal operation models of turbine state variables. As the number of SCADA channels has grown strongly, thousands of independent single-target models are in place today for monitoring a single turbine. Multi-target learning was recently proposed to limit the number of models. This study applied multi-target neural networks to the task of early fault detection in drive-train components. The accuracy and delay of detecting gear bearing faults were compared to state-of-the-art single-target approaches. We found that multi-target multi-layer perceptrons (MLPs) detected faults at least as early and in many cases earlier than single-target MLPs. The multi-target MLPs could detect faults up to several days earlier than the single-target models. This can deliver a significant advantage in the planning and performance of maintenance work. At the same time, the multi-target MLPs achieved the same level of prediction stability.

Multi-target normal behaviour models for wind farm condition monitoring

The trend towards larger wind turbines and remote locations of wind farms fuels the demand for automated condition monitoring and condition-based maintenance strategies that can reduce the operating cost and avoid unplanned downtime. Normal behaviour modelling has been introduced to automatically detect anomalous deviations from normal operation based on the turbine's SCADA data. A growing number of machine learning models and related threshold values of the normal behaviour of turbine subsystems are being developed by wind farm managers to this end. However, these models need to be kept track of, be maintained and require frequent updates. Every additional model increases the overall lifetime management effort in practice. This research explores and benchmarks multi-target machine learning models as a new approach to capturing a wind turbine's normal behaviour. We present an overview of multi-target regression methods and motivate their application and benefits in wind turbine condition monitoring. As a second contribution, we evaluate and benchmark the performance of multi-target normal behaviour models in a wind turbine case study. We find that multi-target models are advantageous in comparison to single-target modelling in that they can substantially reduce the lifecycle management effort of normal behaviour models without compromising on the accuracy of the models. Finally, we also outline some areas of future research.