Fangzhou Cheng

Quantitative Evaluation of Wind Turbine Faults Under Variable Operational Conditions

Wind turbines have been widely used for clean and renewable electricity generation. The maintenance costs of wind turbines constitute a significant portion of the total cost of the generated electricity. Thus, health management systems are increasingly needed to reduce the maintenance costs and improve the reliability of wind turbines. This paper proposes a novel framework for the quantitative evaluation of faults and health conditions of wind turbines using generator current signals. A synchronous resampling algorithm is designed to handle nonstationary current signals for fault feature extraction. The extracted fault features are used to reconstruct new signals, whose correlation dimensions are then calculated by using the Grassberger–Procaccia (G–P) algorithm for fault and health condition evaluation of the wind turbines. Experimental studies are carried out for a direct-drive wind turbine equipped with a permanent-magnet synchronous generator (PMSG) in the healthy condition and two faulty conditions. Results show that the proposed framework can not only detect the faults but also quantify different health conditions for the wind turbine.

Current-Based Fault Detection and Identification for Wind Turbine Drivetrain Gearboxes

This paper proposes a new fault detection and identification framework for drivetrain gearboxes of wind turbines equipped with doubly-fed induction generators (DFIGs) based on the fusion of DFIG stator and rotor current signals. First, the characteristic frequencies of gearbox faults in DFIG stator and rotor currents are analyzed. Different time- and frequency-domain features of gearbox faults in DFIG stator and rotor current signals are then defined, and the methods to extract these features are introduced. These features are used as the inputs of multiclass support vector machines with probabilistic outputs for fault mode identification. Different schemes that use a single stator or rotor current signal or both stator and rotor current signals for the feature- or decision-level information fusion are designed. Experimental results obtained from a DFIG wind turbine drivetrain test rig are provided to validate the proposed current-based fault detection and identification framework.

Current-based fault detection and identification for wind turbine drivetrain gearboxes

This paper proposes a new fault detection and identification framework for drivetrain gearboxes of wind turbines equipped with doubly-fed induction generators (DFIGs) based on the fusion of DFIG stator and rotor current signals. First, the characteristic frequencies of gearbox faults in DFIG stator and rotor currents are analyzed. Different time- and frequency-domain features of gearbox faults in DFIG stator and rotor current signals are then defined, and the methods to extract these features are introduced. These features are used as the inputs of multiclass support vector machines with probabilistic outputs for fault mode identification. Different schemes that use a single stator or rotor current signal or both stator and rotor current signals for the feature- or decision-level information fusion are designed. Experimental results obtained from a DFIG wind turbine drivetrain test rig are provided to validate the proposed current-based fault detection and identification framework.

A case-based data-driven prediction framework for machine fault prognostics

Fault prognosis is an important step to achieve condition-based maintenance for machinery systems. The existing fault prognostic methods can generally be categorized into three major classes: case-based, data-driven, and model-based methods. This paper proposes a novel case-based data-driven prognostic framework based on the adaptive neuro-fuzzy inference system (ANFIS) and big data concept. The framework contains two phases. One is an offline learning phase, in which big historical data are used to build an ANFIS model-case library. The other is the online prognostic phase, in which the fault prognosis of a new machinery system (i.e., a new case) is accomplished by using the proper ANFIS model(s) chosen from the model-case library. The proposed framework is tested by using the experimental data of bearing faults collected from a bearing test rig. Result shows that it has better fault prognostic accuracy than the traditional data-driven method.

A comparative study on Vibration- and current-based approaches for drivetrain gearbox fault diagnosis

Gearboxes are widely used in rotary machines, such as wind turbines, automobiles, and helicopters. Failures of gearboxes contribute to a significant portion of the total failures and downtime in these machines. Gearbox fault diagnosis is an effective means to prevent catastrophic failures, improve the reliability, and reduce the downtime and maintenance cost of these machines. Vibration-based approaches have been employed in most commercially available condition monitoring systems (CMSs); while current-based approaches have received increasing attentions in the industry and academia. This paper presents a comparative study on the vibration- and current-based approaches for gearbox fault diagnosis. Theoretical analysis and experimental tests are carried out to show that: in vibration signals, ghost frequencies appear when the gearbox is healthy and disappear when a gear fault occurs; vibration signals of a faulty gearbox are modulated by shaft rotating frequencies at gear meshing frequencies (GMFs); in current signals, the fundamental frequency component is dominant when the gearbox is healthy; and current signals are modulated by gearbox characteristic frequencies at the fundamental frequency when gear faults appear.

Fault Prognosis and Remaining Useful Life Prediction of Wind Turbine Gearboxes Using Current Signal Analysis

Gearbox faults constitute a significant portion of all faults and contribute to a significant portion of the downtime of wind turbines. Thus, an accurate prediction of the gearbox remaining useful life (RUL) is important to achieve condition-based maintenance to ensure secure and reliable operations of wind turbines and reduce the cost of wind power. This, however, is a challenging work due to the lack of accurate physical degradation models and limited data. This paper proposes a new fault prognostic and RUL prediction method for gearboxes based on the adaptive neuro-fuzzy inference system (ANFIS) and particle filtering (PF) approaches. In the proposed method, the fault feature is extracted from the measured one phase stator current of the generator connected with the gearbox; the ANFIS learns the state transition function of the extracted fault feature; the PF algorithm then predicts the RUL of the gearbox based on the learned state transition function and new information of the fault feature. Experimental results on a gearbox run-to-failure test are provided to show the effectiveness of the proposed method.

Multiscale Filtering Reconstruction for Wind Turbine Gearbox Fault Diagnosis Under Varying-Speed and Noisy Conditions

One challenge in the gearbox fault diagnosis of variable-speed wind turbines via the popular vibration signal analysis is that, the signal spectrum is smeared by the time-varying shaft rotating speed and contaminated by some speed-unrelated components and noise. This paper proposes a new method, called multiscale filtering reconstruction, for wind turbine gearbox fault diagnosis under varying-speed and noisy conditions. The major contribution is that the new method can solve the spectrum smearing problem of the speed-related components while suppressing the speed-unrelated components and noise simultaneously. First, the time-variant vibration envelope signal is decomposed into time-variant monocomponent signals via a Vold–Kalman filter-based multiscale filtering algorithm. Then, the monocomponent signals are converted to be time-invariant by using a signal reform formula. Finally, a purified signal is reconstructed by synthesizing the reformed signals with different weighting factors. A gearbox fault diagnostic scheme based on the proposed method is developed for wind turbines equipped with doubly fed induction generators, in which the rotating frequency of the selected shaft used for multiscale filtering is estimated from the generator rotor current signals. Experimental studies on different gearbox faults are conducted to validate the proposed method and its superiority over the traditional angular resampling method.

Rotor current-based fault diagnosis for DFIG wind turbine drivetrain gearboxes using frequency analysis and a deep classifier

Fault diagnosis of drivetrain gearboxes is a prominent challenge in wind turbine condition monitoring. Many machine learning algorithms have been applied to gearbox fault diagnosis. However, many of the current machine learning algorithms did not provide satisfactory fault diagnosis results due to their shallow architectures. Recently, a class of machine learning models with deep architectures called deep learning has received more attention, because it can learn high-level features of inputs. This paper proposes a new fault diagnosis method for the drivetrain gearboxes of the wind turbines equipped with doubly-fed induction generators (DFIGs) using DFIG rotor current signal analysis. In the proposed method, the instantaneous fundamental frequency of the rotor current signal is first estimated to obtain the instantaneous shaft rotating frequency. Then, the Hilbert transform is used to demodulate the rotor current signal to obtain its envelope, and the resultant envelope signal contains fault characteristic frequencies that are in proportion to the varying DFIG shaft rotating frequency. Next, an angular resampling algorithm is designed to resample the nonstationary envelope signal to be stationary based on the estimated instantaneous shaft rotating frequency. After that, the power spectral density analysis is performed on the resampled envelope signal for the gearbox fault detection. Finally, a classifier with a deep architecture that consists of a stacked autoencoder and a support vector machine is proposed for gearbox fault classification using extracted fault features. Experimental results obtained from a DFIG wind turbine drivetrain test rig are provided to verify the effectiveness of the proposed method.

Quantitative evaluation of wind turbine faults under variable operational conditions

Wind turbines have been widely used for clean and renewable electricity generation. The maintenance costs of wind turbines constitute a significant portion of the total cost of the generated electricity. Thus, health management systems are increasingly needed to reduce the maintenance costs and improve the reliability of wind turbines. This paper proposes a novel framework for the quantitative evaluation of faults and health conditions of wind turbines using generator current signals. A synchronous resampling algorithm is designed to handle nonstationary current signals for fault feature extraction. The extracted fault features are used to reconstruct new signals, whose correlation dimensions are then calculated by using the Grassberger-Procaccia (G-P) algorithm for fault and health condition evaluation of the wind turbines. Experimental studies are carried out for a direct-drive wind turbine equipped with a permanent-magnet synchronous generator (PMSG) in the healthy condition and two faulty conditions. Results show that the proposed framework can not only detect the faults but also quantify different health conditions for the wind turbine.

Fault prognosis of drivetrain gearbox based on a recurrent neural network

Prognostic health management is an effective approach to enhance the reliability and reduce downtime of rotating machines, such as wind turbines. To achieve effective health management, fault prognosis is indispensable. This paper proposes a fault prognostic method for drivetrain gearboxes, which are one of the most troublesome subassemblies in wind turbines. The proposed method consists of signal collection, health index extraction, health index prediction, and decision making for maintenance. The signal used for fault prognosis is a phase current measured from the generator connected with the gearbox. A health index called noise-to-signal ratio (NSR) of the current signal is proposed to reflect the health condition of the gearbox. A recurrent neural network (RNN) is designed to predict the health index online and maintenance is scheduled when the predicted health index reaches a predetermined threshold. The proposed fault prognostic method is validated by the data obtained from an accelerated gearbox run-to-failure experiment for an emulated wind turbine drivetrain consisting of a gearbox and a generator.