Fault diagnosis is an effective tool to guarantee safe operations in gearboxes. Acoustic and vibratory measurements in such mechanical devices are all sensitive to the existence of faults. This work ...addresses the use of a deep random forest fusion (DRFF) technique to improve fault diagnosis performance for gearboxes by using measurements of an acoustic emission (AE) sensor and an accelerometer that are used for monitoring the gearbox condition simultaneously. The statistical parameters of the wavelet packet transform (WPT) are first produced from the AE signal and the vibratory signal, respectively. Two deep Boltzmann machines (DBMs) are then developed for deep representations of the WPT statistical parameters. A random forest is finally suggested to fuse the outputs of the two DBMs as the integrated DRFF model. The proposed DRFF technique is evaluated using gearbox fault diagnosis experiments under different operational conditions, and achieves 97.68% of the classification rate for 11 different condition patterns. Compared to other peer algorithms, the addressed method exhibits the best performance. The results indicate that the deep learning fusion of acoustic and vibratory signals may improve fault diagnosis capabilities for gearboxes.
•A deep random forest fusion technique is proposed.•Both acoustic and vibratory signals are used for gearbox fault diagnosis.•Fault diagnosis capability can be improved by deep learning fusion.
This paper proposes a novel intelligent fault diagnosis method to automatically identify different health conditions of wind turbine (WT) gearbox. Unlike traditional approaches, where feature ...extraction and classification are separately designed and performed, this paper aims to automatically learn effective fault features directly from raw vibration signals while classify the type of faults in a single framework, thus providing an end-to-end learning-based fault diagnosis system for WT gearbox without additional signal processing and diagnostic expertise. Considering the multiscale characteristics inherent in vibration signals of a gearbox, a new multiscale convolutional neural network (MSCNN) architecture is proposed to perform multiscale feature extraction and classification simultaneously. The proposed MSCNN incorporates multiscale learning into the traditional CNN architecture, which has two merits: 1) high-level fault features can be effectively learned by the hierarchical learning structure with multiple pairs of convolutional and pooling layers; and 2) multiscale learning scheme can capture complementary and rich diagnosis information at different scales. This greatly improves the feature learning ability and enables better diagnosis performance. The proposed MSCNN approach is evaluated through experiments on a WT gearbox test rig. Experimental results and comprehensive comparison analysis with respect to the traditional CNN and traditional multiscale feature extractors have demonstrated the superiority of the proposed method.
A primary challenge in fault diagnosis is to extract multiple components entangled within a noisy observation. Therefore, this paper describes and analyzes a novel framework, based on convex ...optimization, for simultaneously identifying multiple features from superimposed signals. This work adequately exploits the underlying prior information that multiple faults with similar frequency spectrum have different morphological waveforms that can be sparsely represented over the union of redundant dictionaries. Within this framework, prior information is formulated into regularization terms, and a sparse optimization problem, which can be solved through the alternating direction method of multipliers (ADMM), is proposed. Meanwhile, the convergence and computational complexity of the proposed iterative framework are profoundly investigated. Moreover, sensitivity analyses and adaptive selection rules for the regularization parameters are described in detail through a set of comprehensive numerical studies. The proposed framework is validated through performing the diagnosis of multiple faults for gearbox in a wind farm. The comparison with respect to the state of the art in the field is illustrated in detail, which highlights the superiority of the proposed framework.
•A multi-sensor fusion method that combines vibration and rotational speed signals.•A novel deep learning algorithm (BiConvLSTM) to classifying planetary gear faults.•Superior to existing deep ...learning algorithms such as CNN, LSTM, and CNN-LSTM.
Gearbox fault diagnosis is expected to significantly improve the reliability, safety and efficiency of power transmission systems. However, planetary gearbox fault diagnosis remains a challenge due to complex responses caused by multiple planetary gears. Model-based gearbox fault diagnosis techniques extract hand-crafted features from sensor data based on underlying physics and statistical analysis, which are not effective in extracting spatial and temporal features automatically. While deep learning methods such as convolutional neural network (CNN) enable automatic feature extraction from multiple sensor sources, they are not capable of extracting spatial and temporal features simultaneously without losing critical feature information. To address this issue, we introduce a novel deep neural network based on bidirectional-convolutional long short-term memory (BiConvLSTM) networks to determine the type, location, and direction of planetary gearbox faults by extracting spatial and temporal features from both vibration and rotational speed measurements automatically and simultaneously. In particular, a CNN determines spatial correlations between two measurements within one time step automatically by combining signals collected from three accelerometers and one tachometer. Long short-term memory (LSTM) networks identify temporal dependencies between two adjacent time steps. By replacing input-to-state and state-to-state operations in the LSTM cell with convolutional operations, the BiConvLSTM can learn spatial correlations and temporal dependencies without losing critical features. Experimental results have shown that the BiConvLSTM network can detect the type, location, and direction of gearbox faults with higher accuracy than conventional deep learning approaches such as CNN, LSTM, and CNN-LSTM.
•Review vibration based fault diagnosis methods of wind turbine planetary gearbox.•Fundamental analysis of the planetary gearbox vibration response are reviewed.•Fault enhancement and operation ...pattern effect elimination are reviewed.•Feature extraction and fault diagnostics strategy are reviewed.•Prospects of vibration based fault diagnosis methods are discussed.
As one of the most immensely growing renewable energies, the wind power industry also experiences a high failure rate and operation & maintenance cost. Therefore, the condition monitoring and fault diagnosis of a wind turbine (WT) generator set are highly needed. Among different components of a WT generator set, WT planetary gearbox plays a crucial role in transmission and leads to relatively higher failure rate and longer downtime. Towards this, a number of studies have been reported in both the academic journals and conference proceedings. This paper provides a systemic and pertinent state-of-art review on WT planetary gearbox condition monitoring techniques on the topics of fundamental analysis, signal processing, feature extraction, and fault detection. Moreover, a few valuable open issues are pointed out and potential research directions are suggested.
A situation often encountered in the condition monitoring (CM) and health management of gearboxes is that a large volume of CM data (e.g., vibration signal) collected from a healthy state is ...available but CM data from a faulty state unavailable. Fault detection under such a situation is usually tackled by modeling the baseline CM data and then detect the fault by examining any deviation of the baseline model versus newly monitored data. Given that the CM data is mostly time series, the long-short term memory (LSTM) neural network can be employed for baseline CM data modeling. The LSTM is free from the choice of the number of lagged input time series and can also store both long-term and short-term time series dependency information. However, we found that an LSTM with its hyperparameters selected whilst minimizing validation mean squared error (VAMSE) does not differentiate the faulty and healthy states well. There is still room for detectability improvement. In this paper, we propose a physics-informed hyperparameters selection strategy for the LSTM identification and subsequently the fault detection of gearboxes. The key idea of the proposed strategy is to select hyperparameters based on maximizing the discrepancy between healthy and physics-informed faulty states, as opposed to minimizing VAMSE. Case studies have been conducted to detect the gear tooth crack and tooth wear using laboratory test rigs. Results have shown that the proposed physics-informed hyperparameters selection strategy returns an LSTM that can better detect these faults than the LSTM returned from minimizing VAMSE.
Planetary gearboxes significantly differ from fixed-axis gearboxes and exhibit unique behaviors, which invalidate fault diagnosis methods working well for fixed-axis gearboxes. Much work has been ...done for condition monitoring and fault diagnosis of fixed-axis gearboxes, while studies on planetary gearboxes are not that many. However, we still notice that a number of publications on condition monitoring and fault diagnosis of planetary gearboxes have appeared in academic journals, conference proceedings and technical reports. This paper aims to review and summarize these publications and provide comprehensive references for researchers interested in this topic. The structures of a planetary gearbox as well as a fixed-axis one are briefly introduced and contrasted. The unique behaviors and fault characteristics of planetary gearboxes are identified and analyzed. Investigations on condition monitoring and fault diagnosis of planetary gearboxes are summarized based on the adopted methodologies. Finally, open problems are discussed and potential research topics are pointed out.
•The HSWT method can effectively improve the TF energy concentration and obtain a high accuracy reconstruction for signals with fast varying IF.•The performance of the HSWT method has been ...theoretically proved.•The effectiveness of the HSWT method has been verified by a simulated signal and experimental signals.
The synchrosqueezing transform (SST) is a powerful tool for time-frequency analysis of signals with slowly varying instantaneous frequency (IF). However, the SST and its extensions provide poor time-frequency resolution for signals with wide frequency range and fast varying IF. In this paper, a new SST method called high-order synchrosqueezing wavelet transform is proposed to achieve a highly energy-concentrated time-frequency representation (TFR) for nonstationary signals with wide frequency range and fast varying IF. This method uses high-order group delay and chirp rate operators to obtain the accurate estimation of instantaneous frequency. The proposed method can effectively improve the energy concentration of the TFR and remain invertible simultaneously. The numerical simulations investigate the performance and noise robustness of the proposed method when analyzing a typical amplitude-modulated and frequency-modulated (AM-FM) multicomponent signal. Finally, the application of planetary gearbox fault diagnosis in the variable operating condition verifies the effectiveness of the proposed method.
The synchrosqueezing transform can effectively improve the readability of time–frequency representation of mono-component and constant frequency signals. However, for multi-component and time-variant ...frequency signals, it still suffers from time–frequency blurs. In order to address this issue, the synchrosqueezing transform is improved using iterative generalized demodulation. Firstly, the complex nonstationary signal is decomposed into mono-components of constant frequency by iterative generalized demodulation. Then, the instantaneous frequency of each mono-component is accurately estimated via the synchrosqueezing transform, by exploiting its merit of enhanced time–frequency resolution. Finally, the time–frequency representation of the original signal is obtained by superposing the time–frequency representations of all the mono-components with restored instantaneous frequency. This proposed method generalizes the synchrosqueezing transform to multi-component and time-variant frequency signals, and it has fine time–frequency resolution and is free of cross-term interferences. The proposed method was validated using both numerically simulated and lab experimental vibration signals of planetary gearboxes under nonstationary conditions. The time-variant planetary gearbox characteristic frequencies were effectively identified, and the gear faults were correctly diagnosed.
•The proposed method has fine time–frequency resolution and is free from cross term interferences.•It is suited to analyze signals of multi-component and time-varying frequency.•It is evaluated using numerical simulated and lab experimental planetary gearbox signals.•It is effective in diagnosing both the local and distributed gear faults.
Display omitted
•A CNN based feature learning and fault diagnosis method for gearboxes is proposed.•The performance of feature learning of CNN with various data types is tested.•The selection of key ...parameters of CNN is discussed.•Feature learning with CNN provides better results than manual feature extraction.•CNN is more suitable to learn features from vibration signal in frequency domain.
Feature extraction plays a vital role in intelligent fault diagnosis of mechanical system. Nevertheless, traditional feature extraction methods suffer from three problems, which are (1) the requirements of domain expertise and prior knowledge, (2) the sensitive to the changes of mechanical system and (3) the limitations of mining new features. It is attractive and meaningful to investigate an automatic feature extraction method, which can adaptively learn features from raw data and discover new fault-sensitive features. Deep learning has been widely used in image analysis and speech recognition with great success. The key advantage of this method lies into the ability of mining representative information and sensitive features from raw data. However, the application of deep learning in feature leaning for mechanical diagnosis is still few, and limited studies have been carried out to compare the effectiveness of feature leaning with various data types. This paper will focus on developing a convolutional neural network (CNN) to learn features directly from frequency data of vibration signals and testing the different performance of feature learning from raw data, frequency spectrum and combined time-frequency data. Manual features from time domain, frequency domain and wavelet domain as well as three common intelligent methods are used as comparisons. The effectiveness of the proposed method is validated through PHM 2009 gearbox challenge data and a planetary gearbox test rig. The results demonstrate that the proposed method is able to learn features adaptively from frequency data and achieve higher diagnosis accuracy than other comparative methods.