Nien-Che Yang

Dr. Nien-Che Yang

Department of Electrical Engineering, National Taiwan University of Science and Technology, Taipei 10607, Taiwan
orcid

Research Focus & Metrics

Academic Career Length

14Years (2013-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
24 April 2019 in IEEE Sensors Journal
Verified
Journal Article
Closed access
MGMou-Fa Guo
Mou-Fa Guo
Nien-Che YangNien-Che Yang
Nien-Che Yang
WCWei-Fan Chen
Wei-Fan Chen
Fault classification is important for the fault cause analysis and faster power supply restoration. A deep-learning-based fault classification method in small current grounding power distribution systems is presented in this paper. The current and voltage signals are sampled at a substation when a fault occurred. The time-frequency energy matrix is constructed via applying Hilbert-Huang transform (HHT) band-pass filter to those sampled fault signals. Regarding the time-frequency energy matrix as the pixel matrix of digital image, a method for image similarity recognition based on convolution neural network (CNN) is used for fault classification. The presented method can extract the features of fault signals and accurately classify ten types of short-circuit faults, simultaneously. Two simulation models are established in the PSCAD/EMTDC and physical system environment, respectively. The performance of the presented method is studied in the MATLAB environment. Various kinds of fault conditions and factors including asynchronous sampling, different network structures, distribution generators access, and so on are considered to verify the adaptability of the presented method. The results of investigation show that the presented method has the characteristics of high accuracy and adaptability in fault classification of power distribution systems.

11 April 2022 in IEEE Access
Verified
Journal Article
Open access
Nien-Che YangNien-Che Yang
Nien-Che Yang
HIHarun Ismail
Harun Ismail
Photovoltaic (PV) systems are vulnerable to failures due to undesired conditions. Faults can occur unpredictably and remain challenging to recognize. In this study, inverter fault, voltage sag, partial shading, and open circuit were considered as the PV faults. However, the faults become eight target classes since each fault considers two different conditions: intermediate power point tracking and maximum power point tracking. Machine learning techniques have recently become the most interesting methods for solving PV failures. This paper proposes a random forest and modified independent component analysis (RF-MICA) to detect the occurrence of PV faults. The MICA was developed for dimensionality reduction for enhanced performance, whereas previous studies only focused on principal component analysis. Two strategies are introduced to address an imbalanced dataset: the synthetic minority oversampling technique as scenario 1 and random undersampling as scenario 2 for oversampling and undersampling methods, respectively. These techniques are essential for investigation because PV fault data can become unbalanced, and they have not been fully addressed in previous studies. The results indicated that the proposed model performs better than other algorithms, with high accuracy and low computational time. RF-MICA yielded accuracy rates of 99.88% and 99.43% for scenarios 1 and 2, respectively.

1 March 2016 in IET Generation, Transmission & Distribution
Verified
Journal Article
Open access
PDF file
Nien-Che YangNien-Che Yang
Nien-Che Yang
In this study, a three‐phase power flow solution method using graph theory, injection current, and sparse matrix techniques for large‐scale unbalanced distribution networks is proposed. In the bus frame of reference, a direct iterative method is adopted. To integrate the electric characteristics of transformers and step voltage regulators into the proposed method, the existing component models are modified by equivalent injected currents. To validate the performance and effectiveness of the proposed method, four three‐phase IEEE test systems and random test systems are used for comparison purposes. As the size of the network increases, the proposed direct Z BUS method drastically shows its superiority over the other methods. The results reveal that the proposed method has good potential for improving the computational efficiency of optimal planning and design as well as real‐time power dispatch applications in large‐scale distribution systems.

1 April 2015 in IET Generation, Transmission & Distribution
Verified
Journal Article
Open access
PDF file
Nien-Che YangNien-Che Yang
Nien-Che Yang
MLMinh‐Duy Le
Minh‐Duy Le
In this study, a novel method is developed to optimise the design of passive power filters (PPFs). The proposed method is based on the famous bat algorithm (BA) and is expected to make the PPF design easier and more effective than conventional trial‐and‐error approaches. Furthermore, the effects of inertia weight variations on the performance of the proposed method are examined. To optimise performance, the most suitable inertia weight, with the best effects, will be selected. For solving multi‐objective optimisation problems, a so‐called external archive is also integrated into the proposed method. A case study of PPF design is also presented to demonstrate the accuracy, efficiency and superiority of the proposed method.

23 June 2023 in IEEE Access
Verified
Journal Article
Open access
Nien-Che YangNien-Che Yang
Nien-Che Yang
JYJen-Ming Yang
Jen-Ming Yang
Traditional fault classification methods typically rely on manual feature extraction and the application of machine-learning algorithms. However, these approaches encounter difficulties when extracting features and handling large-scale datasets. This study proposes a data preprocessing method for accurately detecting various types of short-circuit faults in power systems, which can lead to more effective power repair and maintenance processes. The proposed method involves converting the measured voltage and current signals into time and frequency domains using the short-time Fourier transform (STFT) to produce a time-frequency energy map. A convolutional neural network (CNN) is subsequently trained and tested to classify the short-circuit faults. However, overfitting may occur during the CNN training process owing to the large volume of data with similar features. To address this issue, this study proposes a data reduction method based on the fast dynamic time warping (Fast-DTW) algorithm, which compares waveform features and eliminates highly similar data regarded as redundant data from the dataset. The simulation results show that the proposed method can improve the model training performance and its adaptability to different power system topologies, as tested in two simulation environments: power systems computer-aided design (PSCAD)/electromagnetic transients, including DC (EMTDC), and the real-time digital simulator (RTDS). The STFT transformation is implemented in MATLAB. The simulation results demonstrate that the proposed method reduces redundant data by 40.2%, while decreasing the model training time. Consequently, the overall accuracy of the fault classification reaches 99.37%, confirming the effectiveness of the proposed method for fault classification.

17 January 2022 in Mathematics
Verified
Journal Article
Open access
PDF file
Nien-Che YangNien-Che Yang
Nien-Che Yang
HIHarun Ismail
Harun Ismail
A photovoltaic (PV) system is one of the renewable energy resources that can help in meeting the ever-increasing energy demand. However, installation of PV systems is prone to faults that can occur unpredictably and remain challenging to detect. Major PV faults that can occur are line-line and open circuits faults, and if they are not addressed appropriately and timely, they may lead to serious problems in the PV system. To solve this problem, this study proposes a voting-based ensemble learning algorithm with linear regression, decision tree, and support vector machine (EL-VLR-DT-SVM) for PV fault detection and diagnosis. The data acquisition is performed for different weather conditions to trigger the nonlinear nature of the PV system characteristics. The voltage-current characteristics are used as input data. The dataset is studied for a deeper understanding, and pre-processing before feeding it to the EL-VLR-DT-SVM. In the pre-processing step, data are normalized to obtain more feature space, making it easy for the proposed algorithm to discriminate between healthy and faulty conditions. To verify the proposed method, it is compared with other algorithms in terms of accuracy, precision, recall, and F-1 score. The results show that the proposed EL-VLR-DT-SVM algorithm outperforms the other algorithms.