Kumar Abhinav
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Kumar Abhinav has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Sanjoy Kumar Parida
Department of Electrical Engineering, Indian Institute of Technology, Patna, India
3 shared publications - Abhineet Prakash
School of Electrical and Computer Sciences, Indian Institute of Technology, Bhubaneswar, India
2 shared publications Piyush Rai
Department of Electrical Engineering, Indian Institute of Technology, Patna, Patna, India
2 shared publications- Priyesh Saini
Department of Electrical Engineering, Indian Institute of Technology, Patna, India
2 shared publications - Ravi Shankar
Department of Electrical Engineering, National Institute of Technology, Patna 800005, India
1 shared publication - Sariki Murali
Tata Consultancy Services
1 shared publication
Latest Publications
- 24 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industry Applications
- Vol. 62 (3) , 5205-5220
Autoencoder-Enhanced Ensemble Neural Network for Forced Oscillation Source Location in Renewable-Rich Power Systems
Conventional energy-based methods struggle to determine the origin of Forced Oscillations (FOs) in renewable-rich systems because of their dependency on Dissipating Energy Flow (DEF) calculations. This paper proposes a novel Autoencoder Enhanced Ensemble Neural Network (AEENN) deep learning model for accurately identifying FO sources in power systems having high wind energy penetration without relying on traditional DEF calculations. The methodology incorporates a two-stage framework that enhances the scalability of the proposed method for larger and more practical power systems. It leverages Wavelet Synchrosqueezing Transform (WSST) and Multivariate Empirical Mode Decomposition (MEMD) to extract relevant harmonic components while filtering out noise. A reduced-order WSST matrix is constructed by retaining only FO-related frequency components based on an energy-based threshold. Moreover, an autoencoder effectively learns nonlinear features from PMU data in wind turbine generator (WTG)-dominated grids, enabling robust feature extraction for multivariate time-series classification. To enhance reliability, a $k$-fold ensemble classifier is employed, where multiple models trained on different data folds are combined to improve accuracy and generalisation. The efficacy of the proposed technique is examined on the IEEE 11-bus system and India's Eastern Regional Grid (ERG) model, demonstrating superior real-world applicability.
- 24 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industry Applications
- Vol. 62 (3) , 5205-5220
Autoencoder-Enhanced Ensemble Neural Network for Forced Oscillation Source Location in Renewable-Rich Power Systems
Conventional energy-based methods struggle to determine the origin of Forced Oscillations (FOs) in renewable-rich systems because of their dependency on Dissipating Energy Flow (DEF) calculations. This paper proposes a novel Autoencoder Enhanced Ensemble Neural Network (AEENN) deep learning model for accurately identifying FO sources in power systems having high wind energy penetration without relying on traditional DEF calculations. The methodology incorporates a two-stage framework that enhances the scalability of the proposed method for larger and more practical power systems. It leverages Wavelet Synchrosqueezing Transform (WSST) and Multivariate Empirical Mode Decomposition (MEMD) to extract relevant harmonic components while filtering out noise. A reduced-order WSST matrix is constructed by retaining only FO-related frequency components based on an energy-based threshold. Moreover, an autoencoder effectively learns nonlinear features from PMU data in wind turbine generator (WTG)-dominated grids, enabling robust feature extraction for multivariate time-series classification. To enhance reliability, a $k$-fold ensemble classifier is employed, where multiple models trained on different data folds are combined to improve accuracy and generalisation. The efficacy of the proposed technique is examined on the IEEE 11-bus system and India's Eastern Regional Grid (ERG) model, demonstrating superior real-world applicability.
- 14 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industry Applications
- Vol. 60 (6) , 8213-8227
Improved LSTM-Based Load Forecasting Embedded 3DOF (FOPI)-FOPD Controller for Proactive Frequency Regulation in Power System
This paper presents the design and analysis of a modified volleyball premier league-optimized 3-DOF (FOPI)-FOPD controller for a proactive LFC scheme. The proposed controller incorporates forecasted load demand as one of its inputs. This unique configuration empowers the controller to proactively eliminate the disturbances. To validate the controller performance, a truncated model of the DPS has been developed which is subjected to different load profiles and are assumed to be known. It was observed that the predicted disturbance to the system (change in load demand) is of utmost importance for this proposed configuration. Thus, improved LSTM, incorporating a multi-approach feature selection and DWT has been developed, which is used to forecast day-ahead load demand. This forecasted demand is given as an input to the controller. The dynamic performance of proposed controller is evaluated by subjecting the modelled DPS with proposed controller to standard test signals. The obtained simulation results are also validated through OPAL-RT real-time simulator. The results demonstrate the superior performance and robustness of the proposed LFC approach in maintaining the frequency of the power system within acceptable limits.
- 14 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industry Applications
- Vol. 60 (6) , 8213-8227
Improved LSTM-Based Load Forecasting Embedded 3DOF (FOPI)-FOPD Controller for Proactive Frequency Regulation in Power System
This paper presents the design and analysis of a modified volleyball premier league-optimized 3-DOF (FOPI)-FOPD controller for a proactive LFC scheme. The proposed controller incorporates forecasted load demand as one of its inputs. This unique configuration empowers the controller to proactively eliminate the disturbances. To validate the controller performance, a truncated model of the DPS has been developed which is subjected to different load profiles and are assumed to be known. It was observed that the predicted disturbance to the system (change in load demand) is of utmost importance for this proposed configuration. Thus, improved LSTM, incorporating a multi-approach feature selection and DWT has been developed, which is used to forecast day-ahead load demand. This forecasted demand is given as an input to the controller. The dynamic performance of proposed controller is evaluated by subjecting the modelled DPS with proposed controller to standard test signals. The obtained simulation results are also validated through OPAL-RT real-time simulator. The results demonstrate the superior performance and robustness of the proposed LFC approach in maintaining the frequency of the power system within acceptable limits.
- 19 September 2023
- Published by Taylor & Francis in Electric Power Components and Systems
- Vol. 53 (14) , 1-16
Distributed Generation Based WADC for Inter-Area Oscillations Considering Unknown Inputs
- 19 September 2023
- Published by Taylor & Francis in Electric Power Components and Systems
- Vol. 53 (14) , 1-16
Distributed Generation Based WADC for Inter-Area Oscillations Considering Unknown Inputs