Research Projects

The proposed project addresses a critical problem in modern power systems: the need for real-time event detection and intelligent automation in the presence of massive, diverse, and high-speed data generated across a wide-area smart grid. The Indian power grid, operating at high voltages (1200kV AC, 800kV DC) and integrating increasing levels of renewable energy, requires robust monitoring and control solutions to ensure resilience and reliability. Phasor Measurement Units (PMUs), complying with IEEE C37.118, are key enablers of Wide Area Monitoring Systems (WAMS), but their high cost and sparse deployment limit coverage in distribution networks. Moreover, existing PQM instruments and legacy devices (relays, fault recorders) often lack the ability to measure all power quality (PQ) parameters effectively. Several AI/ML approaches (FFT, STFT, EMD, PCA, DBSCAN, DNN, CNN, SVM, etc.) have been applied for event detection and classification. However, these methods often suffer from poor time-frequency resolution, scalability constraints, convergence inefficiencies, and a need for full-bus data. Most critically, they are rarely implemented in real-time hardware environments, limiting their operational readiness. A significant challenge that remains inadequately addressed is the lack of seamless integration among smart grid subsystems—including AMI, PMU, SCADA, and WAMS. These operate as isolated units with fragmented data pipelines, hindering coordinated analytics and real-time control. Furthermore, the volume, velocity, and variety of data within the smart grid demand a robust and fault-tolerant architecture capable of handling failure scenarios and inter-system inconsistencies without compromising grid stability. This project proposes a big data-enabled and feedback-controlled model that integrates machine learning with real-time grid monitoring to address these challenges. The specific objectives are: 1. Real-time sensing for fast fault response using advanced big data analytics on wide-area monitoring data. 2. Development of a neuromorphic data mining model based on Spiking Neural Networks (SNNs) for intelligent detection and classification of dynamic grid events from large-scale, post-disturbance records. 3. Hardware implementation using GPS-assisted FPGA platforms to validate real-time applicability using synchronized WAMS/PMU data. 4. Subsystem integration framework to ensure seamless data flow and coordinated decision making across AMI, SCADA, PMU, and WAMS. 5. The novelty of this proposal lies in: ● Its brain-inspired SNN approach for sparse, energy-efficient, and high-speed event detection with a feedback loop that dynamically tunes event detection thresholds. ● Hardware realization, ensuring real-world applicability besides complying with IEEE C37.118 standards and IEC 61000-4-30 Class A. ● A comprehensive integration framework for disparate smart grid subsystems. The proposed model will enable intelligent, and resilient operation of the smart grid, contributing significantly to the realization of Smart Grid Industry 4.0.