Wind Turbine Condition Research Articles

Integrated Vibro-Acoustic Analysis and Empirical Mode Decomposition for Fault Diagnosis of Gears in a Wind Turbine

Over the last few years, wind turbine technology has experienced a rapid growth among the other renewable power developing technologies with respect to market share, size and technological design. Generally, wind turbines are subjected to harsh operating conditions which further yields to damage of the critical components. Hence, health monitoring of key components is a vital task which predicts the damage severity and gives the flexibility to plan the maintenance tasks. Wind turbine condition monitoring is a major area of interest in recent years aiming to improve the life of the machine components simultaneously reducing the operational and maintenance cost. Gearbox in wind turbines has the largest share of downtime among all other components affecting directly the cost of operation and maintenance.In this current investigation, an attempt has been made to diagnose the gear faults by using Empirical Mode Decomposition (EMD) methodology. Two condition monitoring techniques such as vibration analysis and acoustic signal analysis are integrated and the experiments are performed on a laboratory scaled three-stage gearbox having the speed ratio of 48:1. Local gear faults such as tooth chip and tooth root crack are seeded and the response is recorded in the form of vibration and acoustic signals. EMD analysis is implemented and the statistical features are extracted from the acquired data. The representative features are identified using a decision tree algorithm and these are classified using pattern recognition techniques – Support Vector Machine (SVM) to distinguish between the healthy and faulty classes. The challenges and the potential advantages are also discussed in this paper to establish the focus of integrated condition monitoring systems.

Condition Monitoring of Wind Turbine Gearbox Bearing Based on Deep Learning Model

Wind turbines condition monitoring and fault warning have important practical value for wind farms to reduce maintenance costs and improve operation levels. Due to the increase in the number of wind farms and turbines, the amount of data of wind turbines have increased dramatically. This problem has caused a need for efficiency and accuracy in monitoring the operating condition of the turbine. In this paper, the idea of deep learning is introduced into wind turbine condition monitoring. After selecting the variables by the method of the adaptive elastic network, the convolutional neural network (CNN) and the long and short term memory network (LSTM) are combined to establish the logical relationship between observed variables. Based on training data and hardware facilities, the method is used to process the temperature data of gearbox bearing. The purpose of artificial intelligence monitoring and over-temperature fault warning of the high-speed side of bearing is realized efficiently and conveniently. The example analysis experiments verify the high practicability and generalization of the proposed method.

A novel wind turbine condition monitoring method based on cloud computing

With the development of condition monitoring technology, the data collected by sensors are voluminous and much faster than before. The cloud computing technology is a good solution for big data processing, it is therefore very suitable to be applied in the condition monitoring of the wind turbine, especially for data-driven model-based condition monitoring methods. In order to solve this problem, a novel wind turbine condition monitoring method based on cloud computing is proposed in this paper. A data-driven model-based condition monitoring (CM) method by using hierarchical extreme learning machine (H-ELM) algorithm is adopted to achieve fault detection of the gearbox in the wind turbine, which has better performance than traditional ELM method. Then, compressed sensing (CS) method is applied to compress the first hidden layer output that will be uploaded to the cloud for further calculation. The proposed method is not only able to detect the faults effectively, but also considering data upload quantity reduction and data security. The case study validates the effectiveness of the proposed method. Consequently, it is effective and can also enhance economic benefit and operating efficiency of the wind farm.

Fault Diagnosis of Wind Turbine Based on Empirical Mode Decomposition

The structure of the gearbox of wind turbine is complex. The working environment is bad and the load is heavy which results in frequent failure of the gearbox parts and huge economic losses. In this paper, the signal processing method combined with empirical mode decomposition and improved bispectrum is used to diagnose the gearbox of wind turbine and the condition of wind turbine is monitored by a portable inspection. The feasibility and applicability of this method in the fault diagnosis of wind turbine gearbox are checked by an example.

Online health assessment of wind turbine based on operational condition recognition

To reduce the operation and maintenance (O&M) costs, the health assessment of wind turbine has received more and more attention in recent years. However, it is difficult to evaluate the health condition of wind turbine due to the complex and non-stationary operation environment. This paper proposes a data-driven approach for online health assessment of wind turbine based on operational condition recognition. First, the operational condition parameters are selected by analyzing the monitoring data of wind turbine. Considering the time-varying of operation environment, the operational conditions are divided into four subspaces utilizing the K-means clustering algorithm. Then, using the historical state parameters data under normal operation, a health benchmark model is constructed in each operational condition space based on Gaussian Mixture Model (GMM). Further, a Softmax model is trained according to the results of operational condition classification, which is used to identify the online operational condition of wind turbine. Moreover, an overall health index (HI) based on Mahalanobis distance is developed to assess the health condition of wind turbine. Finally, the method is verified by the actual supervisory control and data acquisition (SCADA) data of a wind field in northwestern China. The test results show that the proposed approach can track the running state of the wind turbine accurately and play a good role in early fault warning.

The Research of Wind Turbine Condition Monitoring Based on ANFIS

With the continuous development of wind power industry, the number of installed wind power generator is increasing. Due to its complicated structure and poor working conditions, easily happened all kinds of faults. For the nonlinear problems and difficulty in modeling of wind turbine system fault diagnosis , this paper proposes a wind turbine monitoring system, using 20 different SCADA normal data developed 20 adaptive fuzzy neural inference system (ANFIS) model, and proposes a kind of can apply to all models of anomaly detection method, using the prediction error to evaluate the practicability of model in a simulated SCADA signal, comprehensive 20 anomaly detection results of the model, get the final decision to the conclusion that the simulation results show that the system can accurately diagnose the wind turbine system fault.

Retracted: Manifold learning‐based fuzzy k‐principal curve similarity evaluation for wind turbine condition monitoring

AbstractAs wind energy is experiencing an unprecedented development in today's world, the condition monitoring of wind turbine systems which can avoid serious accidents and economic losses are gathering more and more attentions. Considering the evaluation methods based on machine learning are complicated and unstable in terms of model training and parameter selection, this paper proposed a novel assessment algorithm based on similarity analysis of fuzzy k‐principal curves (FKPCs) in manifold space. Initially, 38 fusion features containing time‐domain, frequency‐domain, and wavelet node energy features are extracted from the vibration signal of the wind turbine bearing. Then, the nonlinear local algorithm Laplacian eigenmaps was introduced to transform the original features to the projected lower dimensional space and obtain the more typical parameters. Combining fuzzy clustering with the k‐principal curve creatively, smooth principal curves of the feature sets were then extracted according to point distribution in three‐dimensional manifold space. Finally, the Hausdorff distance algorithm was employed to calculate the similarity between the principal curves of healthy and test datasets to assess the running condition of wind turbine. To validate the present method, the experiments of accelerated bearing degradation were carried out. A series of comprehensive and persuasive comparisons and analysis with different feature conversion methods (principal component analysis [PCA] and Isomap), different distance analysis algorithms (Euclidean distance and Mahalanobis distance), and different assessment models (hidden Markov model [HMM] and deep belief network [DBN]) verified that the proposed FKPC technique can monitor the operating status of the machine more accurately without complex model training and parameter selection processes.

Comparative assessments of binned and support vector regression-based blade pitch curve of a wind turbine for the purpose of condition monitoring

The unexpected failure of wind turbine components leads to significant downtime and loss of revenue. To prevent this, supervisory control and data acquisition (SCADA) based condition monitoring is considered as a cost-effective approach. In several studies, the wind turbine power curve has been used as a critical indicator for power performance assessment. In contrast, the application of the blade pitch angle curve has hardly been explored for wind turbine condition monitoring purposes. The blade pitch angle curve describes the nonlinear relationship between pitch angle and hub height wind speed and can be used for the detection of faults. A support vector machine (SVM) is an improved version of an artificial neural networks (ANN) and is widely used for classification- and regression-related problems. Support vector regression is a data-driven approach based on statistical learning theory and a structural risk minimization principle which provides useful nonlinear system modeling. In this paper, a support vector regression (a nonparametric machine learning approach)-based pitch curve is presented and its application to anomaly detection explored for wind turbine condition monitoring. A radial basis function (RBF) was used as the kernel function for effective SVR blade pitch curve modeling. This approach is then compared with a binned pitch curve in the identification of operational anomalies. The paper will outline the advantages and limitations of these techniques.

Machine learning methods for wind turbine condition monitoring: A review

This paper reviews the recent literature on machine learning (ML) models that have been used for condition monitoring in wind turbines (e.g. blade fault detection or generator temperature monitoring). We classify these models by typical ML steps, including data sources, feature selection and extraction, model selection (classification, regression), validation and decision-making. Our findings show that most models use SCADA or simulated data, with almost two-thirds of methods using classification and the rest relying on regression. Neural networks, support vector machines and decision trees are most commonly used. We conclude with a discussion of the main areas for future work in this domain.

Comparative analysis of binning and Gaussian Process based blade pitch angle curve of a wind turbine for the purpose of condition monitoring

Several studies have used the power curve as a critical indicator to assess the performance of wind turbines. However, the wind turbine internal operation is affected by various parameters, particularly by blade pitch angle. Continuous monitoring of blade pitch angle can be useful for power performance assessment of wind turbines. The blade pitch curve describes the nonlinear relationship between pitch angle and hub height wind speed which to date has been little explored for wind turbine condition monitoring. Gaussian Process models are nonlinear and nonparametric technique, based on Bayesian probability theory. Such models have the potential give results quickly and efficiently. In this paper, we propose a Gaussian Process model to predict blade pitch curve of a wind turbine for condition monitoring purposes. The obtained Gaussian Process based blade pitch curve is then compared with a conventional approach based on a binned blade pitch curve for identifying operational anomalies purposes. Finally, the weaknesses and strengths of these methods are summarised. SCADA data from healthy wind turbines are used to train and evaluate the performance of these techniques.

RETRACTED ARTICLE: Fault classification and detection in wind turbine using Cuckoo-optimized support vector machine

Fault detection in wind turbine which is identified with complete system monitoring under multi-fault scenario is proposed. When a fault is detected, its types and location are recognized for easy maintenance. Fault in wind turbines is caused due to the high speed of gearbox, generator bearing and the failures occurred in various parts. In wind farm, wind turbine condition monitoring is used to reduce the maintenance cost and also improves the accuracy. Generally, in wind turbine gearbox condition monitoring using sensor is a gainful method to monitor wind turbine performance and fault. This paper nominates a method to decide the parameters for support vector machine (SVM) in wind turbine called Cuckoo search optimization (CSO). The combination of optimization technique with classification technique is evaluated. MATLAB platform was used to evaluate the various faults under fixed value and gain factor conditions. Comparing the accuracy with SVM, particle swarm optimized SVM and k-nearest neighbor, the proposed fault detection and fault isolation technique (CSO-SVM) is improved by 2.5%, 3.5% and 6.5%, respectively. The result shows the CSO model based on SVM algorithm accomplishes the most accurate fault detection than the past models.

Maintenance and remanufacturing strategy: using sensors to predict the status of wind turbines

The costs associated with inspecting wind turbines are high due to their size and complexity. One potential method by which such costs can be reduced, is through the development of robust systems that can monitor the conditions of wind turbines remotely. This study proposes embedding sensors into wind turbines to monitor the conditions of the wind turbines throughout their life cycles. The information retrieved from these sensors could be helpful in two ways: It could facilitate the provision of predictive maintenance for the turbines and enhance the performance of end-of-life (EOL) processing operations. During the maintenance phase, sensors can help to predict failures before they occur because they provide condition information about the products. During the EOL processing phase, they help to improve disassembly and inspection operations. Therefore, the use of embedded sensors in wind turbines could potentially reduce maintenance costs and increase EOL profit. This study compares regular and sensor-embedded wind turbine systems, which are modeled using discrete event simulation. A design of experiments study was carried out on the models. During the experimental stage, while conducting experiments, key variables, such as maintenance cost, disassembly cost, inspection cost, and EOL profit, were monitored. At the analysis stage, pairwise t-tests were performed to determine the statistical significance of the results. The results stage revealed that sensors can provide significant benefits to closed-loop supply chain systems when they are embedded into wind turbines.

Pitch fault diagnosis of wind turbines in multiple operational states using supervisory control and data acquisition data

Supervisory control and data acquisition data including comprehensive signal information have been widely applied to fault diagnosis. However, because of the complex operational condition of wind turbines, supervisory control and data acquisition data become complicated and abstract to study. This article proposes a pitch fault diagnosis method of wind turbines in multiple operational states using supervisory control and data acquisition data. According to the performance of characteristic parameters in nine operational states of wind turbines, Gaussian mixture model clustering and the analysis of normal performance curves are applied to model the relationship of pitch angle, rotor speed, and wind speed. Four cases have been studied to demonstrate the feasibility of the proposed method. The advantages of the proposed approach are as follows: (1) simplifying the analysis of supervisory control and data acquisition data through dividing the data into nine parts; (2) detecting pitch faults earlier than supervisory control and data acquisition monitoring system; (3) visualizing the abnormal behavior of the pitch system; and (4) improving the interpretability of the method with the incorporation of domain knowledge.

SCADA‐based wind turbine anomaly detection using Gaussian process models for wind turbine condition monitoring purposes

The penetration of wind energy into power systems is steadily increasing; this highlights the importance of operations and maintenance, and specifically the role of condition monitoring. Wind turbine power curves based on supervisory control and data acquisition data provide a cost-effective approach to wind turbine health monitoring. This study proposes a Gaussian process (a non-parametric machine learning approach) based algorithm for condition monitoring. The standard IEC binned power curve together with individual bin probability distributions can be used to identify operational anomalies. The IEC approach can also be modified to create a form of real-time power curve. Both of these approaches will be compared with a Gaussian process model to assess both speed and accuracy of anomaly detection. Significant yaw misalignment, reflecting a yaw control error or fault, results in a loss of power. Such a fault is quite common and early detection is important to prevent loss of power generation. Yaw control error provides a useful case study to demonstrate the effectiveness of the proposed algorithms and allows the advantages and limitations of the proposed methods to be determined.

Numerical investigations of the effect of rotating and non-rotating shaft on aerodynamic performance of small scale urban vertical axis wind turbines

The shaft is a very important part of small scale vertical axis wind turbines (VAWTs) being used in urban environments. The shedding vortices will be formed as wind passes around the shaft, the shedding vortices released by the shaft have a negative effect on the aerodynamic performance of the blades passing the wake of the wind shaft, and this will lead to lower power output of the wind turbine. The objective of this study is to explore the differences of the effect of the rotating and non-rotating shafts of VAWTs on the aerodynamic performance of small scale urban VAWTs. In addition, the effect of surface roughness on the performance of the rotating and non-rotating shaft wind turbines is also investigated. The dynamic analysis of the VAWT is carried out by the method of Computational Fluid Dynamics. The aerodynamic loads in the study are obtained by solving the two-dimensional (2D) unsteady Navier-Stokes equations with the Transition SST turbulence model. The results are validated with experimental results. The results show that when the ratio of the shaft diameter to the wind turbine diameter (α) is 3.9%, the power coefficient (Cp) of the rotating shaft wind turbine is 2.2% higher than that of the non-rotating shaft wind turbine. Increasing the shaft diameter-based Reynolds number (Res) will increase the Cp of rotating and non-rotating shaft wind turbines, when Res is 2.6 × 104 and the Cp of the non-rotating shaft wind turbine is 5.7% higher than that of the rotating shaft wind turbine. The Cp of the non-rotating and rotating shaft wind turbines is reduced by 6.9% and 6.3%, respectively, by increasing the turbulence intensity from 4% to 16%. The optimal power output occurs under the condition of the non-rotating shaft wind turbine, and the relative surface roughness (ks/ds) is 0.005, where it is observed that Cp increased by 2.1% in comparison to that of the smooth shaft wind turbine.

Electrical & mechanical diagnostic indicators of wind turbine induction generator rotor faults

In MW-sized wind turbines, the most widely-used generator is the wound rotor induction machine, with a partially-rated voltage source converter connected to the rotor. This generator is a significant cause of wind turbine fault modes. In this paper, a harmonic time-stepped generator model is applied to derive wound rotor induction generator electrical & mechanical signals for fault measurement, and propose simple closed-form analytical expressions to describe them. Predictions are then validated with tests on a 30 kW induction generator test rig. Results show that generator rotor unbalance produces substantial increases in the side-bands of supply frequency and slotting harmonic frequencies in the spectra of current, power, speed, mechanical torque and vibration measurements. It is believed that this is the first occasion in which such comprehensive approach has been presented for this type of machine, with healthy & faulty conditions at varying loads and rotor faults. Clear recommendations of the relative merits of various electrical & mechanical signals for detecting rotor faults are given, and reliable fault indicators are identified for incorporation into wind turbine condition monitoring systems. Finally, the paper proposes that fault detectability and reliability could be improved by data fusion of some of these electrical & mechanical signals.

Gaussian Process Operational Curves for Wind Turbine Condition Monitoring

Due to the presence of an abundant resource, wind energy is one of the most promising renewable energy resources for power generation globally, and there is constant need to reduce operation and maintenance costs to make the wind industry more profitable. Unexpected failures of turbine components make operation and maintenance (O&M) expensive, and because of transport and availability issues, the O&M cost is much higher in offshore wind farms (typically 30% of the levelized cost). To overcome this, supervisory control and data acquisition (SCADA) based predictive condition monitoring can be applied to remotely identify early failures and limit downtime, boost production and decrease the cost of energy (COE). A Gaussian Process is a nonlinear, nonparametric machine learning approach which is widely used in modelling complex nonlinear systems. In this paper, a Gaussian Process algorithm is proposed to estimate operational curves based on key turbine critical variables which can be used as a reference model in order to identify critical wind turbine failures and improve power performance. Three operational curves, namely, the power curve, rotor speed curve and blade pitch angle curve, are constructed using the Gaussian Process approach for continuous monitoring of the performance of a wind turbine. These developed GP operational curves can be useful for recognizing failures that force the turbines to underperform and result in downtime. Historical 10-min SCADA data are used for the model training and validation.

Aerodynamic effects of compressibility for wind turbines at high tip speeds

In the present work two dimensional airfoil computations are used to investigate the effects of compressibility in the tip region of large scale wind turbines of 20 MW+ size. In the past application of incompressible CFD solvers have been wide spread for wind turbine aerodynamics, due to their efficiency and robustness at the near incompressible conditions experienced near the rotor center. With the increasing size of modern wind turbines and the desire to approach high tip speeds, the incompressible assumption might be violated in the tip region of the turbine.To investigate the effects of compressibility and the possibility of correcting incompressible flow solutions using explicit compressibility corrections, a CFD study of 2D airfoil aerodynamics at conditions of a large scale wind turbine is performed. The present study show that classical compressibility corrections can be successfully applied as a post-processing step to incompressible solutions, reducing the error in the predicted lift and drag to within a few percent for attached flow conditions where viscous effects are limited at Mach numbers upto 0.3.

Detached Eddy Simulations of the local Atmospheric Flow Field within a Forested Wind Energy Test Site located in Complex Terrain

Within a current project a wind energy test site will be installed in complex terrain in Southern Germany. On such complex orographies various inflow conditions influence the power gain and the blade loads of wind turbines. For example, forest vegetation modifies the wind direction and velocity, in addition to inclination angles caused by the complex escarpment. The scope of the present work is to characterize the influence of a forested area in complex topographies on the local flow field by means of CFD. The results will be compared to inflow conditions of wind turbines in complex terrain without consideration of forests in a wind tunnel model. Objective is to analyze the degree of influence, a forested hill introduces to the flow field on top of the terrain near the designated wind turbine position and to observe flow separation due to the vegetation. To clarify the results, simulations of flows within forests in flat terrain will be shown and compared to numerical results and measurements from literature.

Statistical Evaluation of SCADA data for Wind Turbine Condition Monitoring and Farm Assessment

Operational data from wind farms is crucial for wind turbine condition monitoring and performance assessment. In this paper, we analyse three wind farms with the aim to monitor environmental and operational conditions that might result in underperformance or failures. The assessment includes a simple wind speed characterisation and wake analysis. The evolution of statistical parameters is used to identify anomalous turbine behaviour. In total, 88 turbines and 12 failures are analysed, covering different component failures. Notwithstanding the short period of data available, several operational parameters are found to deviate from the farm trend in some turbines affected by failures. As a result, some parameters show better monitoring capabilities than others, for the detection of certain failures. However, the limitations of SCADA statistics are also shown as not all failures showed anomalies in the observed parameters.