Condition Monitoring of Wind Turbine Main Bearing Based on Multivariate Time Series Forecasting

Abstract

Condition monitoring and overheating warnings of the main bearing of large-scale wind turbines (WT) plays an important role in enhancing their dependability and reducing operating and maintenance (O&M) costs. The temperature parameter of the main bearing is the key indicator to characterize the normal or abnormal operating condition. Therefore, forecasting the trend of temperature change is critical for overheating warnings. To achieve forecasting with high accuracy, this paper proposes a novel model for the WT main bearing, named stacked long-short-term memory with multi-layer perceptron (SLSTM-MLP) by utilizing supervisory control and data acquisition (SCADA) data. The model is mainly composed of multiple LSTM cells and a multi-layer perceptron regression layer. By combining condition parameters into a characteristic matrix, SLSTM can mine nonlinear, non-stationary dynamic feature relationships between temperature and its related variables. To evaluate the performance of the SLSTM-MLP model, experimental analysis was carried out from three aspects: different sample capacity sizes, different sampling time segments, and different sampling frequencies. Furthermore, the model’s capability of online fault detection was also carried out by simulation. The results of comparative studies and online fault simulation tests show that the proposed SLSTM-MLP has better performance for temperature forecasting of the main bearing of large-scale WTs.

A Review of Predictive Techniques Used to Support Decision Making for Maintenance Operations of Wind Turbines

“…Yet, there is a physical reason why the temperature of this component is quite responsive to incoming faults: the main bearing is a large component, which rotates relatively slowly, and it is, therefore, reasonable that it releases much heat in a way that can be easily captured (together with its anomalies) by data-driven algorithms. In fact, in [35], it is shown that by monitoring the temperature of the main bearing, it is possible to diagnose a stator fault. Finally, it is worth noting that a few studies [36] approach the diagnosis of bearing faults through the analysis of tower sound and vibration, without knowing the transfer function between the bearings and the tower.…”

Review of Wind Power Bearing Wear Analysis and Intelligent Lubrication Method Research

“…The health condition of the wind turbine plays an indispensable role in its normal operation, and if the bearings are in various adverse conditions, such as wear and tear, straining, micro-erosion, and other problems, it will cause damage to other parts of the wind turbine [16,17]. To improve the performance of wind power bearings, it is important to regularly monitor their status to detect any potential problems and take appropriate maintenance measures, which can help reduce the occurrence of failures and maximize the service life of the bearings [18,19]. Improving the operational efficiency and performance of wind turbines reduces the cost of wind power production and helps reduce dependence on non-essential energy sources [20,21].…”

“…Improving the operational efficiency and performance of wind turbines reduces the cost of wind power production and helps reduce dependence on non-essential energy sources [20,21]. Therefore, monitoring wind power bearings and maintaining their health status is essential to ensure regular operation and sustainable development of wind power systems [18]. To write the review more clearly, a PRISMA flowchart was used for the writing process of this paper (Figure 2).…”

Wind Turbine Gearbox Failure Detection Through Cumulative Sum of Multivariate Time Series Data

“…IEA (2020) predicted that over 2023-25, average annual wind energy additions could range from maintenance records. Chapter 12 of Ding (2019) explains the two major schools of thought of fault diagnosis and anomaly detection: a statistical learning-based approach (Orozco et al, 2018;Vidal et al, 2018;Ahmed et al, 2019;Moghaddass and Sheng, 2019;Ahmed et al, 2021b;Ahmed et al, 2021;Xiao et al, 2022), including control chart approaches (Hsu et al, 2020;Riaz et al, 2020), and physical model-based approach . There are naturally approaches combining the two schools of thought (Yampikulsakul et al, 2014;Hsu et al, 2020;Yucesan and Viana, 2021).…”

“…Supervised learning needs appropriately labeled data to train a predictive model, which, once a future input is given, predicts whether the future instance is a fault/failure event. Least-squares support vector regression (LS-SVR) (Yampikulsakul et al, 2014), support vector machine or regression (Vidal et al, 2018;Natili et al, 2021), random forest (Hsu et al, 2020;Pang et al, 2021), XG-Boost and long short-term memory (LSTM) networks (Desai et al, 2020;Xiao et al, 2022) are examples of this category. Labeling the training data can be challenging because the fault tags are often added manually.…”