Wind power scenarios have a significant impact on stochastic optimization problems for power systems in which wind power is a significant component. Generative adversarial networks (GANs) are a powerful class of generative models, and can generate realistic scenarios for renewable power sources without the need for any modeling assumptions. However, the performance of GANs in generating scenarios can be further improved by modifying the way in which the Lipschitz constraint on discriminator network is imposed. Another critical problem of applying deep neural networks is overfitting, a phenomenon especially prone to appear on small training sets. In this paper, we propose an improved GAN for the generation of wind power scenarios. To improve the training speed, we use a gradient penalty term to enforce a Lipschitz constraint based on the output and input of the discriminator network. To improve the scenario quality, we further use a consistency term in the training procedure. Besides, the overfitting problem can be effectively alleviated by the enforced Lipschitz continuity. The proposed method is applied to actual time series data from the NREL wind integration dataset. The experimental results demonstrate that our method outperforms existing methods.
1 March 2021 in IEEE Transactions on Instrumentation and Measurement
To increase the reliability and availability of wind turbines, a new diagnosis method was proposed for multiple open-circuit faults in the back-to-back converters of turbines using permanent magnet synchronous generators (PMSGs). This method is the first to use the instantaneous amplitude (IA) as the principal quantity for formulating diagnostic variables that can distinguish different states of any switch accurately without being affected by speed variations. The method also uses the weighted sliding Hilbert transform (WSHT) to estimate the IA to remove the influence of end effects and carry the information of multiple open-circuit faults. To locate faulty switches, a localization method using signal samples adaptive processing was employed to eliminate the influence of speed changes. Finally, combining these diagnostic variables with identification thresholds makes it possible to detect and locate 21 open-circuit fault combinations for the PMSG-side and grid-side converters. Compared to the conventional fault diagnosis methods, simulation and experimental results showed that the proposed method more efficiently and reliably detects and locates multiple open-circuit faults of PMSG wind power converters.
26 December 2022 in IEEE Transactions on Instrumentation and Measurement
Pig face recognition has a wide range of applications in breeding farms, including precision feeding and disease surveillance. This article proposes a method to guarantee its performance in complex environments such as with dirty faces and in unconstrained outdoor conditions. First, inspired by the shape of the pig face, a trapezoid normalized pixel difference (T-NPD) feature is designed to achieve more accurate detection in unconstrained outdoor conditions. Subsequently, a trimmed mean attention mechanism (TMAM) uses the trimmed mean based squeeze method to assign more precise weights to feature channels, and then fuses it into a ResNet50 backbone network to classify detected pig face images with high accuracy. In addition, the TMAM can be applied in numerous common networks due to its universality. Finally, comprehensive experiments conducted on the publicly available JD pig face dataset indicate that the proposed method has superior performance compared with other methods, with an overall accuracy of 95.06%.
8 October 2017 in The Canadian Journal of Chemical Engineering
This paper deals with the current sensor fault diagnosis and isolation (FDI) problem for a permanent magnet synchronous generator (PMSG) based wind system. An observer based scheme is presented to detect and isolate both additive and multiplicative faults in current sensors, under varying torque and speed. This scheme includes a robust residual generator and a fault estimation based isolator. First, the PMSG system model is reformulated as a linear parameter varying (LPV) model by incorporating the electromechanical dynamics into the current dynamics. Then, polytopic decomposition is introduced for design of an LPV residual generator and fault estimator in the form of linear matrix inequalities (LMIs). The proposed gain-scheduled FDI is capable of online monitoring three-phase currents and isolating multiple sensor faults by comparing the diagnosis variables with the predefined thresholds. Finally, a MATLAB/SIMULINK model of wind conversion system is established to illustrate FDI performance of the proposed method. The results show that multiple sensor faults are isolated simultaneously with varying input torque and mechanical power.
7 March 2024 in IEEE Transactions on Instrumentation and Measurement
Pig face recognition is a crucial tool in preventing and controlling swine fever, as well as optimizing pig farming practices. This study proposes an end-to-end lightweight pig face recognition method called EHV-YOLO, which can accurately identify pig faces from multiple angles and detect faces covered with dirt or external objects. First, to enhance the effectiveness of pig face feature extraction, the study innovatively designs the efficient mobile network (EMobileNet) as the backbone network and constructs a spatial pyramid pooling-fast (SPPF) module to be added to the end of the network. Subsequently, to improve the accuracy of pig face detection and identification, a horizontal vertical attention mechanism (HVAM) is designed and inserted into the EMobileNet. Meanwhile, HVAM is a generic module that can be easily integrated into other network models. Based on the above designs, the EHV-YOLO model has a parameter count of only 0.97M and achieves a real-time detection speed of over 120 frames per second (FPS) with a 99.34% mean average precision (mAP) on the JD pig face dataset, outperforming other existing methods. These results demonstrate the excellent performance of the proposed method and its potential usefulness in practical applications.