FIGURE 3 - uploaded by Zhenyu Chen
Content may be subject to copyright.
RNN and LSTM comparison chart

RNN and LSTM comparison chart

Source publication
Article
Full-text available
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 data of wind turbines have increased dramatically. This problem has caused a need for efficiency and accuracy in monitor...

Contexts in source publication

Context 1
... (x), g(x) represents the number of active faces, and l is the loss function at the time of training. Figure 3 is a comparison of RNN and LSTM. The basic principle of the forward propagation process of the cyclic neural network is similar to that of the general neural network, except that the input of the hidden layer includes not only the input of the current moment but also the state of the network at the previous moment. ...
Context 2
... (x), g(x) represents the number of active faces, and l is the loss function at the time of training. Figure 3 is a comparison of RNN and LSTM. The basic principle of the forward propagation process of the cyclic neural network is similar to that of the general neural network, except that the input of the hidden layer includes not only the input of the current moment but also the state of the network at the previous moment. ...

Similar publications

Article
Full-text available
In order to assess the level of power reserves during down-regulation, the available power of a wind turbine needs to be estimated. The current practice in available power estimation is heavily dependent on the pre-defined performance parameters of the turbine and the curtailment strategy followed. This paper proposes a single-input model-free appr...

Citations

... Zhang Fan et al. [12] analyzed the relationship between SCADA data input/output parameters from a physical point of view, and proposed a calculation formula for the abnormal degree index of the operating state of wind turbines. Fu et al. [13], Zhao et al. [14], and Bangalore et al. [15] used CNN-LSTM, DAE, and NARX-ANN to predict and reconstruct SCADA target parameters, the prediction and reconstruction errors were used as indicators of wind turbine operation status, and the abnormal operation of the wind turbine was successfully monitored. Rezamand et al. [16] proposed a fault detection method based on a mixture of RPCA and wavelet probability distribution functions to achieve early fault monitoring of wind turbine blades. ...
Article
Full-text available
The main transmission system of wind turbines is a multi-component coupling system, and its operational state is complex and varied. These lead to frequent false alarms and missed alarms in existing monitoring systems. To accurately obtain the operational state of the main transmission system and detect its abnormal operation, an early fault warning method for the main transmission system based on SCADA and CMS data is proposed. Firstly, the SCADA and CMS feature parameters relevant to the operating status of the main transmission system are selected by two different methods separately, and the correlation mechanism between the feature parameters and the operating characteristics of the main transmission system is further analyzed. Secondly, the Long Short-Term Memory (LSTM) network-based prediction model of the main transmission system operating parameters is established, in which SCADA and CMS feature parameters are fused as the input feature vectors. Then, the predicted residuals of the state evaluation parameters are used as the operational state evaluation index. The early fault warning model is established by Analytic Hierarchy Process (AHP) and Kernel Density Estimation (KDE). Finally, a case study is used to verify the correct performance of the proposed method. The results show that this method can realize early warning functions 73 h earlier than the existing SCADA system. The method can provide a theoretical basis for the safe operation and condition-based maintenance of wind turbines.
... One of the first, and most likely the most influential, application of normality models to wind turbine monitoring is presented in [5]. Neural networks are a popular choice due to their ability to model complex relationships in data; some examples are provided in [17][18][19]. Other authors have compared the performance of neural networks with standard machine learning algorithms, and regression models [9,20] or alternative neural networks such as Extreme Learning Machines [4]. ...
Article
Full-text available
In this paper, the time variation of signals from several SCADA systems of geographically closed turbines are analysed and compared. When operating correctly, they show a clear pattern of joint variation. However, the presence of a failure in one of the turbines causes the signals from the faulty turbine to decouple from the pattern. From this information, SCADA data is used to determine, firstly, how to derive reference signals describing this pattern and, secondly, to compare the evolution of different turbines with respect to this joint variation. This makes it possible to determine whether the behaviour of the assembly is correct, because they maintain the well-functioning patterns, or whether they are decoupled. The presented strategy is very effective and can provide important support for decision making in turbine maintenance and, in the near future, to improve the classification of signals for training supervised normality models. In addition to being a very effective system, it is a low computational cost strategy, which can add great value to the SCADA data systems present in wind farms.
... Recently maintenance management in WT has seen enormous significance and improved reliability and accuracy. Some of the artificial intelligence (AI) methods, such as Artificial Neural Network (ANN) (Bangalore, 2016), Long Short-Term Memory (LSTM) (Feng et al., 2019), Convolutional Neural Network (CNN) (Fu et al., 2019), are being used for the condition monitoring of WT. ...
... In Turnbull et al. (2018), it shows 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. Fu et al. (2019) introduced deep learning for condition monitoring of WT adaptive elastic network, CNN and LSTM are combined to do feature extraction, dimension reduction and classification. This algorithm solves the issue of gradient explosion and overfitting, reducing the prediction error. ...
Article
Full-text available
This paper reviews the recent advancement made in data-driven technologies based on SCADA data for improving wind turbines’ operation and maintenance activities (e.g. condition monitoring, decision support, critical components failure detections) and the challenges associated with them. Machine learning techniques applied to wind turbines’ operation and maintenance (O&M) are reviewed. The data sources, feature engineering and model selection (classification, regression) and validation are all used to categorise these data-driven models. Our findings suggest that (a) most models use 10-minute mean SCADA data, though the use of high-resolution data has shown greater advantages as compared to 10-minute mean value but comes with high computational challenges. (b) Most of SCADA data are confidential and not available in the public domain which slows down technological advancements. (c) These datasets are used for both, the classification and regression of wind turbines but are used in classification extensively. And, (d) most commonly used data-driven models are neural networks, support vector machines, probabilistic models and decision trees and each of these models has its own merits and demerits. We conclude the paper by discussing the potential areas where SCADA data-based data-driven methodologies could be used in future wind energy research.
... As an important part of artificial neural networks, the deep learning algorithm is widely applied. Fu et al. designed a new modeling structure to analyze the gearbox-bearing temperature changes by the convolutional neural network (CNN) and the long shortterm memory (LSTM) [21]. Yang et al. used a new modeling structure to analyze temperature changes during high-speed train operation using the LSTM model, which showed that the forecasting errors were arranged within a reasonable range [22]. ...
Article
Full-text available
Recently, with the increasing scale of the volume of freight transport and the number of passengers, the study of railway vehicle fault diagnosis and condition management is becoming more significant than ever. The axle temperature plays a significant role in the locomotive operating condition assessment that sudden temperature changes may lead to potential accidents. To realize accurate real-time condition monitoring and fault diagnosis, a new multi-data-driven model based on reinforcement learning and deep learning is proposed in this paper. The whole modeling process contains three steps: In step 1, the feature crossing and reinforcement learning methods are applied to select the suitable features that could efficiently shorten the redundancy of the input. In step 2, the stack denoising autoencoder is employed to extract deep fluctuation information in the features after the reinforcement learning. In step 3, the bidirectional gated recurrent unit algorithm is utilized to accomplish the forecasting model and achieve the final results. These parts of the integrated modeling structure contributed to increased forecasting accuracy than single models. By analyzing the forecasting results of three different data series, it could be summarized that: (1) The proposed two-stage feature selection method and feature extraction method could greatly optimize the input for the predictor and form the optimal axle temperature forecasting model. (2) The proposed hybrid model can achieve satisfactory forecasting results which are better than the contrast algorithms proposed by other researchers.
... ElSaid et al. [29] proposed to improve the LSTM cell structure using the ant colony optimization algorithm (ACO) for forecasting engine data and the new model presented an improvement of 1.35%. Fu et al. [30] used CNN to extract features and then used LSTM for gearbox bearing forecasting to achieve bearing high speed-side monitoring and super high temperature warning. Based on the sliding window forecasting methods, there was an error accumulation problem in time series forecasting. ...
Article
Full-text available
Electric energy, as an economical and clean energy, plays a significant role in the development of science and technology and the economy. The motor is the core equipment of the power station; therefore, monitoring the motor vibration and predicting time series of the bearing vibration can effectively avoid hazards such as bearing heating and reduce energy consumption. Time series forecasting methods of motor bearing vibration based on sliding window forecasting, such as CNN, LSTM, etc., have the problem of error accumulation, and the longer the time-series forecasting, the larger the error. In order to solve the problem of error accumulation caused by the conventional methods of time series forecasting of motor bearing vibration, this paper innovatively introduces Informer into time series forecasting of motor bearing vibration. Based on Transformer, Informer introduces ProbSparse self-attention and self-attention distilling, and applies random search to optimize the model parameters to reduce the error accumulation in forecasting, achieve the optimization of time and space complexity and improve the model forecasting. Comparing the forecasting results of Informer and those of other forecasting models in three publicly available datasets, it is verified that Informer has excellent performance in time series forecasting of motor bearing vibration and the forecasting results reach 10−2∼10−6.
... In this case, Wavelet Packet Transform (WPT) preprocesses the signals. Fu uses deep learning to monitor the condition of the gearbox bearing [25]. Li proposes the use of Deep Small-World Neural Network on the basis of unsupervised learning to detect the early failures of wind turbines [26]. ...
Article
Full-text available
This work focuses on the control of the pitch angle of wind turbines. This is not an easy task due to the nonlinearity, the complex dynamics, and the coupling between the variables of these renewable energy systems. This control is even harder for floating offshore wind turbines, as they are subjected to extreme weather conditions and the disturbances of the waves. To solve it, we propose a hybrid system that combines fuzzy logic and deep learning. Deep learning techniques are used to estimate the current wind and to forecast the future wind. Estimation and forecasting are combined to obtain the effective wind which feeds the fuzzy controller. Simulation results show how including the effective wind improves the performance of the intelligent controller for different disturbances. For low and medium wind speeds, an improvement of 21% is obtained respect to the PID controller, and 7% respect to the standard fuzzy controller. In addition, an intensive analysis has been carried out on the influence of the deep learning configuration parameters in the training of the hybrid control system. It is shown how increasing the number of hidden units improves the training. However, increasing the number of cells while keeping the total number of hidden units decelerates the training.
... These building blocks include: convolution layers, pooling layers, and fully connected layers" [7]. Jian Fu et al. [15] proposed a system that uses a CNN-LSTM model to predict faults in the wind turbines' gearbox. Both CNN (Convolutional Neural Network) combined with RNN (Recursive Neural Network) for feature extraction, dimension reduction and classification on the 1-min level data of 90 consecutive days collected by the SCADA system. ...
... Recently initiatives have been taken to build the NBMs by considering both the spatial and temporal features of SCADA data. For example, Fu [17] proposed a NBM method to monitor WT gearboxes based on the convolutional neural network (CNN) and the LSTM, where the CNN was used to extract the spatial features while the LSTM was used for the temporal features. Kong [18] presented a WT condition monitoring method based on the fusion of spatial and temporal features. ...
... By considering the information change of SCADA data on both the temporal and spatial scales, the proposed NBM based on the CNN and the gated recurrent unit (GRU) demonstrated a superior performance for the condition monitoring of WT gearboxes. Generally, existing deep learning based NBM studies, for example, [17] and [18], commonly use the CNN to extract the spatial features while the recurrent neural network (RNN) to extract the temporal features from SCADA data. Although improvement made, there are still two main challenges. ...
... Considering the oil temperature of WT gearboxes often rises quickly in failures [17], it is selected as the target output variable of the proposed STAGN model. Then, the other 22 operation and environment variables in Table 1 are taken as the input variables. ...
Article
Full-text available
Abstract Effective monitoring and early warning of gearbox operating status are of great significance to the operation and maintenance (O&M) of offshore wind turbines (WTs). This study proposes a normal behaviour modelling (NBM) method based on the spatial‐temporal attention module and the gated recurrent unit (GRU), for the condition monitoring of offshore WT gearboxes. The proposed method has a superior performance by extracting the spatial and temporal features from the supervisory control and data acquisition (SCADA) system, and also has the unique advantage of model interpretability. Specifically, in the NBM training stage, the spatial features of offshore wind farm SCADA data are extracted by the spatial attention module firstly. Then, the temporal features of the spatial feature sequences above are extracted and fused by the GRU network. Afterwards, the temporal attention module is applied to strengthen the expression of key time points. In the NBM testing stage, the output residual between the predicted and the measured values is calculated and monitored by the exponential weighted moving average (EWMA) control chart. Finally, the effectiveness and superiority of the proposed NBM method are verified by detailed simulations on the Donghai Bridge offshore wind farm.
... Among them, fault diagnosis methods based on deep learning have been verified to have excellent performance. Common deep learning models, such as convolutional neural networks (CNN) [3,4], deep autoencoders (DAE) [5,6], recurrent neural networks (RNN) [7,8], long short-term memory (LSTM) [9,10], deep belief networks (DBN) [11] etc., have been applied to wind turbine fault diagnosis research because they usually have a strong ability to automatically learn hidden fault features. However, deep learning methods have heavy data dependence, which means that sufficient labeled data is required to support model training. ...
Preprint
Full-text available
In recent years, intelligent fault diagnosis methods based on deep learning have been widely used in the wind power industry. These methods mostly rely on sufficient labeled data to support model training. However, new wind farms often face the challenge of insufficient labeled data, which will hinder the practical application of the deep learning methods. In addition, data distributions of different wind farms are usually different, so the model trained by other wind farm data cannot be directly used for fault diagnosis of new wind farms. In view of the lack of labeled data in new wind farms, a transfer learning method: weighted joint matching adaptive network (wJMAN) is proposed. This method realizes the alignment of the source domain (other existing wind farms) and the target domain (the new wind farm) through sub-domain matching and joint distribution adaptation. Therefore, the information of other wind farms can be more effectively used in the construction of the fault diagnosis model of the new wind farm. In addition, in view of the imbalance between classification loss and adaptive loss in existing domain adaptation methods, an adaptive loss weight method is proposed, which makes the classification loss and adaptive loss more balanced and converges faster in the optimization process. The proposed wJMAN method is tested with the operating data from 4 real wind farms. The results show that wJMAN has higher diagnostic accuracy than common transfer learning methods.
... The results have shown that the proposed method outperformed the four diagnostic methods. Fu et al. [39] proposed a deep learning-based method for gearbox fault diagnosis. The relationship between temperature measurements was extracted by combining a CNN and a LSTM network using the adaptive elastic network. ...
Article
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.