Wind Turbine Failure Research Articles

Missing wind speed data reconstruction with improved context encoder network

Missing wind speed data are mainly caused by harsh weather, wind turbine failures, and data transmission errors, which have adverse effects on the performance of wind power forecasting, power curve modeling, and energy assessment. Inspired by context encoders (CE), this paper proposes an improved context encoder network (ICE) for missing wind speed data reconstruction. An auto-encoder architecture with multiple one-dimensional convolutional layers is established for data generation. During network training, a joint loss function that includes reconstruction loss and adversarial loss is presented to obtain the stable and near-real reconstructed wind speed data. We add an Inception layer to the generator network to automatically select the appropriate convolutional filters and then recalibrate the channel relationship between feature maps via the squeeze-and-excitation network. At last, this paper uses wind speed data collected from an on-shore wind farm to verify the effectiveness of the proposed network. The results show that the mean absolute error (MAE) and root mean square error (RMSE) of the ICE network in different data missing rates are 0.019–0.021 and 0.021–0.025, respectively. It has the lowest reconstruction errors compared with six typical data reconstruction methods.

Incorporating Machine Learning into Vibration Detection for Wind Turbines

With machine learning techniques, wind turbine components can be detected and diagnosed in advance, so degeneration can be prevented. Automatic and autonomous learning is used to predict, detect, and diagnose electrical and mechanical failures in wind turbines. Based on the implementation of machine learning algorithms adapted to the different components and faults of wind turbines, this study evaluates different methodologies for monitoring, supervision, and fault diagnosis.

An abnormal wind turbine data cleaning algorithm based on color space conversion and image feature detection

Wind power curve (WPC) is established through data collected from the Supervisory Control and Data Acquisition (SCADA) system of each wind turbine, which can be used to analyze the operation status. However, numerous outliers are contained in SCADA data caused by wind turbine failures, shutdown maintenance or other extreme conditions to deform the wind power curve. This paper proposes a data cleaning algorithm for wind turbine abnormal data based on wind power curve image by color space conversion and image feature detection. Considering wind speed, wind power and data frequency, a three-dimensional (3D) WPC image is constructed. The scattered outliers are cleared by their statistical characteristics. The Canny edge detection and Hough transform are introduced to extract image features of stacked outliers and locate them accurately. The proposed algorithm is compared with three common outlier detection algorithms, including two data-based algorithms and an image-based algorithm. Extensive experiments conducted on the data of 22 wind turbines from two different wind farms in China indicate the efficiency, stability and reliability of the proposed algorithm.

Vibration analysis for fault detection in wind turbines using machine learning techniques

The implementation of machine learning techniques allows to prevent in advance the degeneration of any component present in a wind turbine, as well as the detection and diagnosis of sudden failures. This methodology allows automatic and autonomous learning to predict, detect and diagnose electrical and mechanical failures in wind turbines. Four different failure states have been simulated due to bearing vibrations in wind turbines, comparing traditional techniques, such as frequency analysis, as well as the implementation of AI, using the KNN and SVM methodology. This contribution evaluates different methodologies for monitoring, supervision and fault diagnosis based on the implementation of machine learning algorithms adapted to the different components and faults of the wind turbine. Implementing these techniques, allows to anticipate a breakdown, reduce downtime and costs, especially if they are offshore.

Failure prediction of turbines using machine learning algorithms

Wind energy has become a significant component of the new trend, with wind turbines being frequently used to generate electricity. In recent decades, a number of big and small wind farms have been built in many nations as a part of the global push to increase the production of hygienic energy from sources of renewable energy. Failure of wind turbines will adversely affect the power production. In recent years, various analysis is conducted to understand the failure of turbines. Here, machine learning model approach is used to guess the failure of wind turbines. This model will help the engineering team to forecast the failure of turbines. Based on that, business activity and planning can be made to prevent power production loss. A model has been designed using Random Forest Algorithm and Extreme Gradient Boost algorithm (XGBOOST). On comparing Bias and Scores in the Receiver Operating Characteristic curve of both the algorithm, Random Forest algorithm gives the best result when compared to XGBoost algorithm.

Seismic analysis of extended piles in sand considering effect of scouring and effect of water as added mass

Scouring is one of the major causes of the failure of bridges and offshore wind turbines. Piles under extreme scouring and earthquake loads are vulnerable to failure and need to be investigated thoroughly. In this study, the seismic responses of an extended concrete pile with no scour and four scouring depths conditions are investigated under an input motion based on a Kobe earthquake record. The effect of water during an earthquake has a significant impact on the structure as hydrodynamic pressure; however, it is generally ignored in the seismic analysis of piles. The present study uses the concept of added mass to consider the effects of water. The combined soilstructure-interaction and shear beam model is used to simulate pile and soil using the beam-on-nonlinear-Winkler foundation (BNWF) method. Soil is divided into multiple layers with increasing shear modulus and the mass of each soil layer is lumped into two nodes of the layer. Modal analysis and nonlinear time history analysis are performed on different models. Results of the modal analysis show the decrease in fundamental frequencies of the pile with increasing scouring depths. The time history analysis results are reported in the form of Arias and Housner intensities, envelopes of acceleration, velocity, displacement of pile top, and envelopes of shear force and bending moment of the pile. Results show significant changes with scouring depth and the effect of water.

Estimation of rotor and main bearing loads using artificial neural networks

Wind energy is one of the most important technologies for a climate-neutral energy supply. However, the premature failure of wind turbines due to unknown loads leads to a reduction in competitiveness compared to other energy sources. Here, load monitoring systems can make a significant contribution to the prevention of such failures. Most load monitoring systems for wind turbines focus on strain signals of structural components as the tower, main shaft or the rotor blade root. Based on these signals, axial forces, torsion or bending moments, which are acting on these components, are calculated. But this provides only partial information about the complete load situation, as transverse forces are not considered. Other methods use simulation or measurement data to train artificial neural networks (ANN) to estimate damage-equivalent loads acting on a wind turbine. However, this approach is accompanied by a loss of information, because the individual load components are condensed to an equivalent. In this work a method is presented that enables a measurement of rotor and main bearing loads considering all their individual load components. For this purpose, an ANN is trained with elastic multibody simulation (eMBS) data. Based on displacement signals, acting rotor and main bearing loads are estimated. The results show that even with consideration of nonlinearities, including nonlinear stiffness curves and bearing clearances, an appropriate accuracy can be achieved using the method presented.

Evaluation of Wind Turbine Failure Modes Using the Developed SWARA-CoCoSo Methods Based on the Spherical Fuzzy Environment

Accurately recognizing potential failures in the early stages of providing products or services can prevent the loss of investment and time and reduce the risk of safety hazards. Failure mode and effects analysis (FMEA) is a conventional approach for detecting and prioritizing the probable failures of a product’s design or production process. Nevertheless, the traditional risk priority number (RPN) method has come under criticism for its deficiencies. This paper proposes a modified FMEA method based on fuzzy Multi-Criteria Decision Making (MCDM) techniques to cope with the weaknesses of the previous methodologies and improve the primary method. The concept of spherical fuzzy sets (SFS) is utilized to address the vagueness and impreciseness of the information that allows the experts to have more freedom in making decisions by including membership, non-membership, and hesitation of fuzzy sets. Initially, the procedure of assigning weights to the RPN criteria is implemented with SFS step-wise weight assessment ratio analysis (SWARA). Then, the failure modes are ranked by the SFS combined compromise solution (CoCoSo) method. The effectiveness and practicality of the suggested approach are illustrated through a case study on the Manjil wind farm in Iran. Results show that the suggested model is more reliable and realistic to be utilized in the prioritization of failures than the common FMEA method or other integrated MCDM approaches.

Data Fusion Based on an Iterative Learning Algorithm for Fault Detection in Wind Turbine Pitch Control Systems.

In this article, we propose a recent iterative learning algorithm for sensor data fusion to detect pitch actuator failures in wind turbines. The development of this proposed approach is based on iterative learning control and Lyapunov’s theories. Numerical experiments were carried out to support our main contribution. These experiments consist of using a well-known wind turbine hydraulic pitch actuator model with some common faults, such as high oil content in the air, hydraulic leaks, and pump wear.

Analyzing a Decade of Wind Turbine Accident News with Topic Modeling

Despite the significance and growth of wind energy as a major source of renewable energy, research on the risks of wind turbines in the form of accidents and failures has attracted limited attention. Research that applies data analytics methodologically in this context is scarce. The research presented here, upon construction of a text corpus of 721 selected wind turbine accident and failure news reports, develops and applies a custom-developed data analytics framework that integrates tabular analysis, visualization, text mining, and machine learning. Topic modeling was applied for the first time to identify and classify recurring themes in wind turbine accident news, and association mining was applied to identify contextual terms associated with death and injury. The tabular and visual analyses relate accidents to location (offshore vs. onshore), wind turbine life cycle phases (transportation, construction, operation, and maintenance), and the incidence of death and injury. As one of the insights, more incidents were found to occur during operation and transportation. Through topic modeling, topics associated most with deaths and injuries were revealed. The results could benefit wind turbine manufacturers, service providers, energy companies, insurance companies, government bodies, non-profit organizations, researchers, and other stakeholders in the wind energy sector.

Optimizing the Cost and Reliability of Shared Anchors in an Array of Floating Offshore Wind Turbines

Abstract Commercial floating offshore wind projects are expected to emerge in the U.S. by the end of this decade. Currently, however, high costs for the technology limit its commercial viability, and a lack of data regarding system reliability heightens project risk. This work presents an optimization algorithm to examine the tradeoffs between cost and reliability for a floating offshore wind array that uses shared anchoring. Combining a multivariable genetic algorithm with elements of Bayesian optimization, the optimization algorithm selectively increases anchor strengths to minimize the added costs of failure for a large floating wind farm in the Gulf of Maine under survival load conditions. The algorithm uses an evaluation function that computes the probability of mooring system failure, then calculates the expected maintenance costs of a failure via a Monte Carlo method. A cost sensitivity analysis is also performed to compare results for a range of maintenance cost profiles. The results indicate that virtually all of the farm's anchors are strengthened in the minimum cost solution. Anchor strength is increased between 5 and 35% depending on farm location, with anchor strength nearest the export cable being increased the most. The optimal solutions maintain a failure probability of 1.25%, demonstrating the tradeoff point between cost and reliability. System reliability was found to be particularly sensitive to changes in turbine costs and downtime, suggesting further research into floating offshore wind turbine failure modes in extreme loading conditions could be particularly impactful in reducing project uncertainty.

Wind turbine failure prediction and health assessment based on adaptive maximum mean discrepancy

Wind turbines are being designed with increasingly sophisticated electrical and mechanical components, leading to more complicated maintenance procedures and higher failure costs. Much of the research in this area has provided reliable early warning of wind turbine failures in an effort to reduce downtime and maintenance costs. Recently, data-driven approaches have been introduced into the health status assessment of wind turbines. However, obtaining high-precision failure predictions and interpretable health status assessments is still challenging. In this research, we propose a unified framework for predicting failures in and assessing the health of wind turbines. First, we empirically grouped wind turbine failures into four categories. Low frequency failures involving extensive downtime are critical, due to their tendency to gradually deteriorate and the severe consequences they introduce; thus, these were adopted here for use in the predictive maintenance strategy. The adaptive maximum mean discrepancy algorithm, a novel similarity metric for distributions, was used as a means of capturing the dynamic weights of sensor attributes and predicting potential failures. A convolutional neural network was then employed to identify false alarms and improve the precision of the model. This approach will significantly improve the prediction rate and reflect the relationship between failures and features/sensors, which is essential to further improve the performance of wind turbines. Experiments showed that our health assessment method had a higher failure prediction rate and a lower false alarm rate than the traditional method. The process also provides the gradual degradation trend for wind turbines, as influenced by specific failures. An empirical study of one-year of SCADA data from six wind turbines was further used to demonstrate the interpretability of the approach.

A simulation-driven online scheduling algorithm for the maintenance and operation of wind farm systems

Wind turbines experience stochastic loading due to seasonal variations in wind speed and direction. These harsh operational conditions lead to failures of wind turbines, which are difficult to predict. Consequently, it is challenging to schedule maintenance actions that will avoid failures. In this article, a simulation-driven online maintenance scheduling algorithm for wind farm operational planning is derived. Online scheduling is a suitable framework for this problem since it integrates data that evolve over time into the maintenance scheduling decisions. The computational study presented in this article compares the performance of the simulation-driven online scheduling algorithm against two benchmark algorithms commonly used in practice: scheduled maintenance and condition-based monitoring maintenance. An existing discrete event system specification simulation model was used to test and study the benefits of the proposed algorithm. The computational study demonstrates the importance of avoiding over-simplistic assumptions when making maintenance decisions for wind farms. For instance, most literature assumes maintenance lead times are constant. The computational results show that allowing lead times to be adjusted in an online fashion improves the performance of wind farm operations in terms of the number of turbine failures, availability capacity, and power generation.

Remote Monitoring and Diagnostics of Pitch-Bearing Defects in an MW-Scale Wind Turbine Using Pitch Symmetrical-Component Analysis

Recently, multiple wind turbine failure databases have reviewed that the pitch system is one of the subassemblies with the highest failure rates and largest contributors to the overall downtime. Therefore, there has been an increasing interest to provide remote health monitoring for wind turbine pitch system. While most of the research articles are discussing pitch actuation system (hydraulic or electric actuator) faults only, there is very limited research on pitch-bearing-defect detection. This article provides a remote and hardware-free solution to monitor multiaxis pitch-bearing health condition called pitch symmetrical-component analysis. It leverages readily available low-resolution (100 Hz) electrical measurements, mechanical measurements, and control signals from the existing pitch control platform, and innovatively applies symmetrical-component analysis in multiphase ac system to multiaxis pitch control system and introduces multiaxis pitch-bearing degradation trending curves. This hardware-free solution can be directly applied to the existing wind turbines and successfully give the wind farm operator an early warning before multiaxis pitch bearing fails. It has been proved to be accurate, low cost, and has minimum impacts on turbine normal operation, and has been validated by field data from several North America MW-scale wind farms. This approach turns out to be the first hardware-free (no additional hardware needed) method to remotely monitor and diagnose multiaxis wind turbine pitch-bearing condition.

A failure analysis of floating offshore wind turbines using AHP-FMEA methodology

An AHP-FMEA methodology is proposed to analyse the floating offshore wind turbines failure causes. The method introduces the expert opinions to generate a risk index through the Analytical Hierarchy Process criteria weighting technique. Main components, critical systems, failure modes, and failure causes are analysed. Fifteen failure scenarios are identified and corresponding preventive and corrective actions are suggested to cut off paths of failure propagations and to reduce the failure impacts. This method allows to minimize the catastrophic failures on floating offshore wind turbines. Moreover, a comparison study is conducted to remove the uncertainty when implementing the presented method, which validates the proposed methodology's creativeness and feasibility.

Availability and LCOE Analysis Considering Failure Rate and Downtime for Onshore Wind Turbines in Japan

In this study, the availability and the levelized cost of energy (LCOE) are investigated considering failure rate and downtime for onshore wind turbines in Japan. The failure mode effect analysis is conducted using the wind turbine failure database collected by the New Energy and Industrial Technology Department Organization (NEDO). The normalized failure rate and downtime between Europe and Japan are comparable. The occurrence rate is similar between Europe and Japan, but the downtime in Japan is much longer than that of Europe. Three cost-reduction scenarios are then proposed to improve availability and to reduce LCOE using assumed failure rate and downtime in each mode based on the industry interview and best practices in Japan. The availability is improved from 87.4% for the baseline scenario to 92.7%, 95.5% and 96.4% for the three scenarios, and LCOE is also reduced from 13.7 Yen/kWh to 11.9, 11.0 and 10.7 Yen/kWh. Finally, the probability distributions of downtime and repair cost are obtained for each failure mode. It is found that the probability distributions of the failure modes with the shortest downtime show similar probability distributions regardless of the size of the assembly. The effects of downtime and repair-cost uncertainties on LCOE are also evaluated.

Early Fault Diagnostic System for Rolling Bearing Faults in Wind Turbines

Abstract The operation and maintenance costs of wind farms are always high due to high labor costs and the high replacement cost of parts. Thus, it is of great importance to have real-time monitoring and an early fault diagnostic system to prevent major events, reduce time-based maintenance, and minimize the cost. In this paper, such a two-step system for early stage rolling bearing failures in offshore wind turbines is introduced. First, empirical mode decomposition is applied to minimize the effect of ambient noise. Next, correlation coefficients between a reference signal and test signals are obtained and incipient fault detection is achieved by comparing the results with a threshold value. Through further analysis of the envelope spectrum, sample entropy for selected intrinsic mode functions is obtained, which is further used to train a support vector machine classifier to achieve fault classification and degradation state recognition. The proposed diagnostic approach is verified by experimental tests, and an accuracy of 98% in identifying and classifying rolling bearing failures under various loading conditions is obtained.

Wind turbine tower failure analysis and wind-induced vibration control

Wind energy has great potential for development in clean energy. The economic loss caused by the destruction of the wind turbine tower in a wind turbine failure is very huge. Therefore, this paper analyzes the gray correlation between the tower damage type and the cause of the damage. The results show that storms, pitch system failure, and wind speed exceeding the design wind speed have a good correlation with the tower damage type. In a storm environment, the dynamic response of the tower top of the wind turbine tower is relatively large, even leading to the overall collapse of the tower. The tuned mass damper (TMD) has an obvious damping effect on the wind vibration response of the tower.

Imbalance Detection of Rotor Rotational Speed Using Doppler Radar

In operating a wind turbine, both predictive and condition-based maintenances are required to minimize the downtime caused by maintenance. The imbalance of rotor rotational speed is an important factor for diagnosing wind turbine failures. The rotational speed imbalance can be caused by accumulated damage or the accumulation of ice, dust, and moisture. In this paper, we proposed a method for detecting the rotational speed imbalance of a wind turbine using a Doppler radar. We calculated the difference in the rotational speed for different times using spectrograms obtained by observing the wind turbine with a Doppler radar and determined the rotational speed imbalance using the fast Fourier transform. The performance of the proposed algorithm was verified using both synthetic and numerical data.

Optimal Temperature-Based Condition Monitoring System for Wind Turbines

With the increasing demand for the efficiency of wind energy projects due to challenging market conditions, the challenges related to maintenance planning are increasing. In this paper, a condition-based monitoring system for wind turbines (WTs) based on data-driven modeling is proposed. First, the normal condition of the WTs key components is estimated using a tailor-made artificial neural network. Then, the deviation of the real-time measurement data from the estimated values is calculated, indicating abnormal conditions. One of the main contributions of the paper is to propose an optimization problem for calculating the safe band, to maximize the accuracy of abnormal condition identification. During abnormal conditions or hazardous conditions of the WTs, an alarm is triggered and a proposed risk indicator is updated. The effectiveness of the model is demonstrated using real data from an offshore wind farm in Germany. By experimenting with the proposed model on the real-world data, it is shown that the proposed risk indicator is fully consistent with upcoming wind turbine failures.