RESUMEN
The shafting systems of hydropower units work as the core component for the conversion of water energy to electric energy and have been running for a long time in the hostile hydraulic-mechanical-electrical-coupled environment-their vibration faults are frequent. How to quickly and accurately identify vibration faults to improve the reliability of the unit is a key issue. This study proposes a novel shafting vibration fault identification framework, which is divided into three coordinated stages: nonlinear modeling, signal denoising, and holographic identification. A nonlinear dynamical model of bending-torsion coupling vibration induced by multiple excitation vibration sources of the shafting system is established in the first stage. The multi-stage signal denoising method combines Savitzky-Golay (SG) smoothing filtering, singular value decomposition (SVD), and variational mode decomposition (VMD). SG-SVD-VMD is used for the guide bearing the vibration signals in the second stage. Further, the holospectrum theory is innovatively introduced to obtain the holospectra of the simulated and measured signals, and the shafting vibration faults of the real unit are identified by comparing the holospectrum of the measured signal with the simulated signal. These results show that the shafting nonlinear model can effectively reflect the vibration characteristics of the coupled vibration source and reveal the influence and fault characteristics of each external excitation on the shafting vibration. The shafting vibration faults of operating units can be identified by analyzing the holospectra of the shafting simulation signals and measuring the noise reduction signals. Thus, this framework can guide the safe and stable operation of hydropower units.
RESUMEN
Accurate degradation tendency prediction (DTP) is vital for the secure operation of a pumped storage unit (PSU). However, the existing techniques and methodologies for DTP still face challenges, such as a lack of appropriate degradation indicators, insufficient accuracy, and poor capability to track the data fluctuation. In this paper, a hybrid model is proposed for the degradation tendency prediction of a PSU, which combines the integrated degradation index (IDI) construction and convolutional neural network-long short-term memory (CNN-LSTM). Firstly, the health model of a PSU is constructed with Gaussian process regression (GPR) and the condition parameters of active power, working head, and guide vane opening. Subsequently, for comprehensively quantifying the degradation level of PSU, an IDI is developed using entropy weight (EW) theory. Finally, combining the local feature extraction of the CNN with the time series representation of LSTM, the CNN-LSTM model is constructed to realize DTP. To validate the effectiveness of the proposed model, the monitoring data collected from a PSU in China is taken as case studies. The root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) obtained by the proposed model are 1.1588, 0.8994, 0.0918, and 0.9713, which can meet the engineering application requirements. The experimental results show that the proposed model outperforms other comparison models.
RESUMEN
Rolling bearings are a vital and widely used component in modern industry, relating to the production efficiency and remaining life of a device. An effective and robust fault diagnosis method for rolling bearings can reduce the downtime caused by unexpected failures. Thus, a novel fault diagnosis method for rolling bearings by fine-sorted dispersion entropy and mutation sine cosine algorithm and particle swarm optimization (SCA-PSO) optimized support vector machine (SVM) is presented to diagnose a fault of various sizes, locations and motor loads. Vibration signals collected from different types of faults are firstly decomposed by variational mode decomposition (VMD) into sets of intrinsic mode functions (IMFs), where the decomposing mode number K is determined by the central frequency observation method, thus, to weaken the non-stationarity of original signals. Later, the improved fine-sorted dispersion entropy (FSDE) is proposed to enhance the perception for relationship information between neighboring elements and then employed to construct the feature vectors of different fault samples. Afterward, a hybrid optimization strategy combining advantages of mutation operator, sine cosine algorithm and particle swarm optimization (MSCAPSO) is proposed to optimize the SVM model. The optimal SVM model is subsequently applied to realize the pattern recognition for different fault samples. The superiority of the proposed method is assessed through multiple contrastive experiments. Result analysis indicates that the proposed method achieves better precision and stability over some relevant methods, whereupon it is promising in the field of fault diagnosis for rolling bearings.
RESUMEN
This paper proposes a distributed model predictive control based load frequency control (MPC-LFC) scheme to improve control performances in the frequency regulation of power system. In order to reduce the computational burden in the rolling optimization with a sufficiently large prediction horizon, the orthonormal Laguerre functions are utilized to approximate the predicted control trajectory. The closed-loop stability of the proposed MPC scheme is achieved by adding a terminal equality constraint to the online quadratic optimization and taking the cost function as the Lyapunov function. Furthermore, the treatments of some typical constraints in load frequency control have been studied based on the specific Laguerre-based formulations. Simulations have been conducted in two different interconnected power systems to validate the effectiveness of the proposed distributed MPC-LFC as well as its superiority over the comparative methods.