With the growing demand for offshore wind energy and the continued drive for reduced levelised cost of energy, it is necessary to make operation and maintenance activities more effective and reduce related costs. A key factor in achieving this aim is to more representatively model operation and maintenance activities, and to do this, simulation models should include more accurate weather forecasting algorithms. In this paper, three weather forecast modelling methods are used to generate projections of wind and wave values which are then used as inputs in an operation and maintenance simulation model. These methods include Markov Chains, gradient boosting and a novel hybrid regression/statistical approach which has been developed and is presented herein. The change in key performance indicators after the wind farm lifespan is simulated using the forecasting methods and then compared to one another. It is shown that the Markov Chain and hybrid models numerically perform similarly, although the hybrid method has some additional desirable features. Finally, it is shown that the effect of this type of modelling uncertainty leads to significantly differing performance estimates through the operation and maintenance model.

This paper presents the state-of-the-art technologies and development trends of wind turbine drivetrains – the system that converts kinetic energy of the wind to electrical energy – in different stages of their life cycle: design, manufacturing, installation, operation, lifetime extension, decommissioning and recycling. Offshore development and digitalization are also a focal point in this study. Drivetrain in this context includes the whole power conversion system: main bearing, shafts, gearbox, generator and power converter. The main aim of this article is to review the drivetrain technology development as well as to identify future challenges and research gaps. The main challenges in drivetrain research identified in this paper include drivetrain dynamic responses in large or floating turbines, aerodynamic and farm control effects, use of rare-earth material in generators, improving reliability through prognostics, and use of advances in digitalization. These challenges illustrate the multidisciplinary aspect of wind turbine drivetrains, which emphasizes the need for more interdisciplinary research and collaboration.

Condition monitoring systems are deployed in various industries for decades contributing to optimizing operational performance and maintenance efforts. Several publications address this potential for application in the offshore wind energy industry; however, none attempts to quantify the impact that longer warning times ahead of a failure would have on asset availability. The aim of this paper is to bridge this gap by considering particularly the access restrictions for offshore operations through a probabilistic model which simulates existence of different condition monitoring systems on offshore wind turbines in the time domain. Results of this study quantify the positive impact that a longer warning time of potential-to-functional failure (P-F interval) has on availability, highlighting that variation of maintenance strategy through transformation of unplanned activities into planned interventions that can be conducted during a suitable weather window ahead of a component failure can lead to reduced operation and maintenance (O&M) costs.

This research aims to bring together thermal modelling and machine learning approaches to improve the understanding on the operation and fault detection of a wind turbine gearbox. Recent fault detection research has focused on machine learning, black box approaches. Although it can be successful, it provides no indication of the physical behaviour. In this paper, thermal network modelling was applied to two datasets using SCADA (Supervisory Control and Data Acquisition) temperature data, with the aim of detecting a fault one month before failure. A machine learning approach was used on the same data to compare the results to thermal modelling. The results found that thermal network modelling could successfully detect a fault in many of the turbines examined and was validated by the machine learning approach for one of the datasets. For that same dataset, it was found that combining the thermal model losses and the machine learning approach by using the modelled losses as a feature in the classifier resulted in the engineered feature becoming the most important feature in the classifier. It was also found that the results from thermal modelling had a significantly greater effect on successfully classifying the health of a turbine compared to temperature data. The other dataset gave less conclusive results, suggesting that the location of the fault and the temperature sensors could impact the fault-detection ability.

This work presents a comprehensive review and discussion of the identification of critical components of the currently installed and next generation of offshore wind turbines. A systematic review on the reliability, availability, and maintainability data of both offshore and onshore wind turbines is initially performed, collecting the results from 24 initiatives, at system and subsystem level. Due to the scarcity of data from the offshore wind industry, the analysis is complemented with the extensive experience from onshore structures. Trends based on the deployment parameters for the influence of design characteristics and environmental conditions on the onshore wind turbines' reliability and availability are first investigated. The estimation of the operational availability for a set of offshore wind farm scenarios allowed a comparison with the recently published performance statistics and the discussion of the integrity of the data available to date. The failure statistics of the systems deployed offshore are then discussed and compared to the onshore ones, with regard to their normalised results. The availability calculations supported the hypothesis of the negative impact of the offshore environmental conditions on the reliability figures. Nonetheless, similarities in the reliability figures of the blade adjustment system and the maintainability of the power generation and the control systems are outlined. Finally, to improve the performance prediction of future offshore projects, recommendations on the effort worth putting into research and data collection are provided.

Wind turbine pitch system condition monitoring is an active area of research, and this paper investigates the use of the Isolation Forest Machine Learning model and Supervisory Control and Data Acquisition system data for this task. This paper examines two case studies, turbines with hydraulic or electric pitch systems, and uses an Isolation Forest to predict failure ahead of time. This novel technique compared several models per turbine, each trained on a different number of months of data. An anomaly proportion for three different time-series window lengths was compared, to observe trends and peaks before failure. The two cases were compared, and it was found that this technique could detect abnormal activity roughly 12 to 18 months before failure for both the hydraulic and electric pitch systems for all unhealthy turbines, and a trend upwards in anomalies could be found in the immediate run up to failure. These peaks in anomalous behaviour could indicate a future failure and this would allow for on-site maintenance to be scheduled. Therefore, this method could improve scheduling planned maintenance activity for pitch systems, regardless of the pitch system employed.

Anomaly detection for Wind Turbine pitch system fault detection is an active area of research within the Wind Energy community. In this paper a novel condition monitoring technique was presented with the aim to reduce data storage requirements and computational intensity. The technique utilised an Isolation Forest Anomaly Detection model. This new technique was trained on a period of data per turbine to then test on each subsequent month of data individually. The anomaly count is then plotted to observe a trend before failure. This was done as a tool to assist in planning maintenance actions. The number of training months were compared for each turbine to find the minimum appropriate. It was found that one month was appropriate in most cases, with others requiring 3 or 4 months. The technique was able to detect failure up to 3 months before, and abnormal activity roughly 10 to 12 months before failure. Therefore this could be very beneficial to wind farm operators for scheduling maintenance activity around weather windows and vessel rate fluctuations.

Predicting car ownership patterns at high spatial resolution is key to understanding pathways for decarbonisation—via electrification and demand reduction—of the private vehicle fleet. As the factors widely understood to influence car ownership are highly interdependent, linearised regression models, which dominate previous work on spatially explicit car ownership modelling in the UK, have shortcomings in accurately predicting the relationship. This paper presents predictions of spatially disaggregated car ownership—and change in car ownership over time—in Great Britain (GB) using deep neural networks (NNs) with hyperparameter tuning. The inputs to the models are demographic, socio-economic and geographic datasets compiled at the level of Census Lower Super Output Areas (LSOAs)—areas covering between 300 and 600 households. It was found that when optimal hyperparameters are selected, these neural networks can predict car ownership with a mean absolute error of up to 29% lower than when formulating the same problem as a linear regression; the results from NN regression are also shown to outperform three other artificial intelligence (AI)-based methods: random forest, stochastic gradient descent and support vector regression. The methods presented in this paper could enhance the capability of transport/energy modelling frameworks in predicting the spatial distribution of vehicle fleets, particularly as demographics, socio-economics and the built environment—such as public transport availability and the provision of local amenities—evolve over time. A particularly relevant contribution of this method is that by coupling it with a technology dissipation model, it could be used to explore the possible effects of changing policy, behaviour and socio-economics on uptake pathways for electric vehicles —cited as a vital technology for meeting Net Zero greenhouse gas emissions by 2050.

Anomaly detection for wind turbine condition monitoring is an active area of research within the wind energy operations and maintenance (O & M) community. In this paper three models were compared for multi-megawatt operational wind turbine SCADA data. The models used for comparison were One-Class Support Vector Machine (OCSVM), Isolation Forest (IF), and Elliptical Envelope (EE). Each of these were compared for the same fault, and tested under various different data configurations. IF and EE have not previously been used for fault detection for wind turbines, and OCSVM has not been used for SCADA data. This paper presents a novel method of condition monitoring that only requires two months of data per turbine. These months were separated by a year, the first being healthy and the second unhealthy. The number of anomalies is compared, with a greater number in the unhealthy month being considered correct. It was found that for accuracy IF and OCSVM had similar performances in both training regimes presented. OCSVM performed better for generic training, and IF performed better for specific training. Overall, IF and OCSVM had an average accuracy of 82% for all configurations considered, compared to 77% for EE.

This paper proposes a signal processing approach for wind turbine gearbox vibration signals based on employing multiple analysis pipelines. These so-called pipelines consist of combinations of various advanced signal processing methods that have been proven to be effective in literature when applied to wind turbine vibration signals. The performance of the pipelines is examined on vibration data containing different wind turbine gearbox faults in the planetary stages. Condition indicators are extracted from every pipeline to evaluate the fault detection capability for such incipient failures. The results indicate that the multipronged approach with the different pipelines increases the reliability of successfully detecting incipient planetary stage gearbox faults. The type, location, and severity of the fault influences the choice for the appropriate processing method combination. It is therefore often insufficient to only utilize a single processing pipeline for vibration analysis of wind turbine gearbox faults. Besides investigating the performance of the different processing techniques, the main outcome and recommendation of this paper is thus to employ a diversified analysis methodology which is not limited to a sole method combination, to improve the early detection rate of planetary stage gearbox faults.

Operations and Maintenance (O&M) can make up a significant proportion of lifetime costs associated with any wind farm, with up to 30% reported for some offshore developments. It is increasingly important for wind farm owners and operators to optimise their assets in order to reduce the levelised cost of energy (LCoE). Reducing downtime through condition-based maintenance is a promising strategy of realising these goals. This is made possible through increased monitoring and gathering of operational data. SCADA data are useful in terms of wind turbine condition monitoring. This paper aims to perform a comprehensive comparison between two types of normal behaviour modelling: full signal reconstruction (FSRC) and autoregressive models with exogenous inputs (ARX). At the same time, the effects of the training time period on model performance are explored by considering models trained with both 12 and 6 months of data. Finally, the effects of time resolution are analysed for each algorithm by considering models trained and tested with both 10 and 60 min averaged data. Two different cases of wind turbine faults are examined. In both cases, the NARX model trained with 12 months of 10 min average Supervisory Control And Data Acquisition (SCADA) data had the best training performance.

Photovoltaic generation systems are widely used around the electrical network. There is a large variety of photovoltaic models for power production estimation. However, the use of machine learning algorithms in this application is yet to be explored. In this paper, a comparison between statistical and deterministic approaches is presented. The statistical models proposed use machine learning methodologies with neural network algorithms and automatic strategies to clean the data. To validate the model a comparison with two different deterministic models in different situations and locations is analyzed. The accuracy of the neural network algorithms when trained and tested in the same location makes the proposed model a good performance ratio tool for a photovoltaic power station.

Condition monitoring (CM) systems are installed in wind turbines (WTs) in order to avoid component downtime and reduce maintenance costs. Vibration monitoring is widely used for the WT gearbox, which is a component with a significant downtime. Given that the installed wind capacity grows, the volume of CM data increases, making manual interpretation of vibration signals challenging. Therefore, there is a need for an efficient and automated maintenance decision support system. The aim to this paper is to propose an automated framework for gearbox incipient failure diagnosis. The framework utilises vibration signals and performs health estimation and fault isolation based on signal processing and artificial intelligence (AI) techniques. The methodology is demonstrated through a case study of vibration data from operating WTs with known gearbox failures. The study can be used to optimise wind turbine maintenance actions.

Reducing wind turbine downtime through innovations surrounding asset management has the potential to greatly influence the levelised cost of energy (LCoE) for large wind farm developments. Focusing on generator bearing failure and vibration data, this paper presents a two-stage methodology to predict failure within 1 to 2 months of occurrence. Results are obtained by building up a database of failures and training machine learning algorithms to classify the bearing as healthy or unhealthy. This is achieved by first using clustering techniques to produce subpopulations of data based on operating conditions, which this paper demonstrates can greatly influence the ability to diagnose a fault. Secondly, this work classifies individual clusters as healthy or unhealthy from vibration-based condition monitoring systems by applying order analysis techniques to extract features. Using the methodology explained in the report, an accuracy of up to 81.6% correct failure prediction was achieved.

Health state assessment of wind turbine components has become a vital aspect of wind farm operations in order to reduce maintenance costs. The gearbox is one of the most costly components to replace and it is usually monitored through vibration condition monitoring. This study aims to present a review of the most popular existing gear vibration diagnostic methods. Features are extracted from the vibration signals based on each method and are used as input in pattern recognition algorithms. Classification of each signal is achieved based on its health state. This is demonstrated in a case study using historic vibration data acquired from operational wind turbines. The data collection starts from a healthy operating condition and leads towards a gear failure. The results of various diagnostic algorithms are compared based on their classification accuracy.

With O&M costs accounting between 25-30% of life-cycle costs, it becomes pertinent to model related activities analytically, accounting for all downtime-contributing factors and at the same time incorporating in the analysis practicalities of operations. Related analysis should be able to account for the accurate prediction of weather data, classification of maintenance interventions and modelling of failure rates, and finally, apply realistic strategies with respect to planned and unplanned maintenance activities. This paper reports the development of the initial version of an open-access tool, which allows the estimation of availability of a given wind farm with specified characteristics throughout its service life, allowing for the simulation of a number of scenarios related to reliability parameters, vessels specifications and availability, number of technicians etc, towards optimising a wind farm maintenance strategy. Here, we present initial results for a reference case study, showing applicability and responsiveness of the tool and sensitivity analysis of the jack-up vessel mobilisation time as a varying parameter as it was found to have a significant impact in the estimated availability.

Vibration signals are widely used in wind turbine drivetrain condition monitoring with the aim of fault detection, optimization of maintenance actions and therefore reduction of operating costs. Signals are most commonly sampled by accelerometers at high frequency for a few seconds. The behavior of these signals varies significantly, even within the same turbine and depends on different parameters. The aim of this paper is to explore the effect of operational and environmental conditions on the vibration signals of wind turbine gearboxes. Parameters such as speed, power and yaw angle are taken into account and the change in vibration signals is examined. The study includes examples from real wind turbines of both normal operation and operation with known gearbox faults. The effects of varying operating conditions are removed using kalman filtering as a state observer. The findings of this paper will aid in understanding wind turbine gearbox vibration signals, making more informed decisions in the presence of faults and improving maintenance decisions.

Operation and maintenance costs of wind turbines are highlydriven by gearbox failures, especially offshore were the logisticsof replacements are more demanding. It is therefore verycritical to foresee incipient gearbox faults before they becomecatastrophic failures. Wind turbine gearbox condition monitoringis usually performed using vibration signals comingfrom accelerometers installed on the gearbox surface. Thecurrent monitoring practice is a rule-based approach, wherealarms are activated based on thresholds. However, too muchmanual analysis may be required for some failure modes andthis can become quite challenging as the installed wind capacitygrows. Also, since false alarms have to be avoided,these thresholds are set quite high, resulting in late stage diagnosisof components. Given the fact there is a large amountof historic operating data with confirmed gearbox failure incidents,this paper proposes a framework that uses a machinelearning approach. Vibration signals are used from the gearboxsensors and processed in the frequency domain. Featuresare extracted from the processed signals based on the fault locationsand failure modes, using domain knowledge. Thesefeatures are used as inputs in a layer of pattern recognitionmodels that can determine a potential component fault locationand failure mode. The proposed framework is illustratedusing failure examples from operating offshore wind turbines.

Innovations which help facilitate predictive maintenance strategies have the potential to greatly reduce costs associated with wind turbine O&M by driving efficiency and increasing wind turbine availability. This study uses multiple examples of the same generator bearing failure to provide insight into how condition monitoring systems can be used in to train machine learning algorithms with the ultimate goal of predicting failure and remaining useful life. Results show that by analysing high-frequency vibration data and extracting key features to train support vector machine algorithms, an accuracy of 67% can be achieved in successfully predicting failure 1–2 months before occurrence. This study reflects on the limitations surrounding a generalised training approach, taking advantage of all available data, showing that if too many different examples are considered of different wind turbines and operating conditions, the overall accuracy can be diminished.

The goal of the present paper is to achieve the diagnosis of an in-service 1.5 MW wind turbine equipped with a doubly-fed induction generator through current signature and vibration analyses. Real data from operating machines have rarely been analysed in the scientific literature through current signature analysis supported by vibrations. The wind turbine under study was originally misdiagnosed by the operator, where a healthy component was replaced and the actual failure continued progressing. The chronological evolution of both the electrical current and vibration spectra is presented to conduct an in-depth tracking of the fault. The diagnosis is achieved through spectral analysis of the stator currents, where fault frequency components related to rotor mechanical unbalance are identified. This is confirmed by the vibration analysis, which provides insightful information on the health of the drive train. These results can be implemented in condition monitoring strategies, which is of great interest to optimise operation and maintenance costs of wind farms.