http://dspace.dtu.ac.in:8080/jspui/handle/123456789/80 2026-07-22T20:38:53Z http://dspace.dtu.ac.in:8080/jspui/handle/repository/22982 Title: ANALYSIS, DESIGN AND DEVELOPMENT OF ENERGY STORAGE SYSTEM FOR ELECTRIC VEHICLES Authors: KUMAR, DEEPAK; Rizwan, M. (SUPERVISOR); Panwar, Amrish K. (CO-SUPERVISOR) Abstract: The rapid adoption of electric vehicles requires the development of energy storage systems with high energy density, enhanced safety, and long cycle life. However, the performance of lithium-ion batteries in electric vehicle applications is significantly influenced by operating and environmental conditions, including material degradation, temperature variations, charge/discharge rates, dynamic drive profiles, and thermal stresses. These factors introduce nonlinear electrochemical behaviour that accelerates battery aging and hinders the accurate estimation of critical battery states, such as state of charge, state of energy, and state of health. To address these challenges, this thesis aims to develop advanced artificial intelligence driven frameworks combining advanced cathode material development, AI-based state estimation, and Green AI–based control strategies. to address the challenges of nonlinear battery behaviour, ageing, and safety. The proposed approach significantly improves the accuracy, reliability and sustainability of lithium-ion battery systems in electric vehicle applications. A Ni-rich layered NMC811 cathode was synthesised through a solid-state reaction route and characterized through thermogravimetric analysis, Fourier transform infrared spectroscopy, X-ray diffraction, scanning electron microscopy and energy dispersive x-ray spectroscopy analysis. These results confirmed a hexagonal α-NaFeO₂ phase (R-3m space group), crystallite size ~22 nm. Thermal and structural analyses confirmed phase formation, while scanning electron microscopy and energy dispersive x-ray spectroscopy revealed homogeneous morphology and elemental composition, validating its suitability for high-energy-density EV batteries. To address battery monitoring challenges, advanced AI-based algorithms were developed for accurate estimation of state of charge and state of energy. A novel filtering technique was proposed to reduce redundant data by maintaining original critical patterns. The method achieved up to 80% reduction in dataset size, and 79-80% memory consumption, 59-66% computational efficiency, and 61-82% energy consumption without losing critical lithium-ion battery dynamic information. The filter technique with convolutional neural network and bidirectional long short-term memory model achieved significantly higher V estimation accuracy compared to traditional models for battery states estimation under different dynamic drive cycles, improving RMSE, MSE, and MAE by up to 99.98%, 99.97%, and 99.96%, respectively. However, the filter technique with gated recurrent unit model improved state of health estimation accuracy by over 93% in RMSE and maintained R² values above 0.999 across multiple aging datasets. These findings validate the proposed energy efficient AI frameworks for battery health monitoring. This research introduces a Green AI framework for accurate, scalable, and efficient battery state estimation and health monitoring under dynamic conditions. Further, the impact of the environmental and operational influences, such as temperature variations and dynamic load profiles, was systematically investigated. Condition-aware AI driven control strategies were developed for adaptive charging behaviour, thermal management, and degradation mitigation. The proposed models were tested across −10 °C to 40 °C, confirmed by the convolutional neural network, which achieved superior accuracy i.e. RMSE of 0.0043 and 67% faster computation, while long short-term memory performed best at 25–40 °C and showed higher error at −10 °C due to electrochemical limitations. CNN based SOH and SOC estimation frameworks outperform long short-term memory and gated recurrent unit in terms of accuracy, flexibility, and efficiency under diverse thermal conditions. This interdisciplinary research contributes a unified framework that integrates materials engineering, Green AI-driven modelling, condition-aware architecture and sustainable computational techniques. The proposed methodology significantly improves battery states prediction accuracy, computational efficiency, and operational reliability. The outcomes of this research deliver strong foundation for the development of intelligent, energy-efficient, condition-aware BMS for the next generation of electric mobility and energy storage applications. Furthermore, the incorporation of Green AI principles ensures reduction in computational and energy demands, improved resource efficiency. This development supports sustainable and energy-efficient energy storage technologies for greener EVs aligned with global decarbonization goals. 2026-04-01T00:00:00Z http://dspace.dtu.ac.in:8080/jspui/handle/repository/22760 Title: DEVELOPMENT OF AI-BASED MICROGRID RENEWABLE GENERATION FORECASTING Authors: POONAM; Sreejeth, Mini (SUPERVISOR); Tripathi, M. M (CO-SUPERVISOR) Abstract: The growing integration of renewable energy resources into modern microgrids has intensified the need for accurate, scalable, and uncertainty-aware forecasting frameworks capable of operating under dynamic meteorological conditions. Wind and solar generation exhibit strong intermittency, non-stationarity, and complex spatial dependencies, making conventional statistical and deep learning models insufficient for real-time operational decision-making. This thesis develops a comprehensive suite of artificial intelligence–based methodologies that address multi-horizon deterministic forecasting, ensemble-based prediction refinement, uncertainty quantification, and cross-source spatio-temporal modeling tailored for microgrid applications. The research begins with a systematic review of existing renewable forecasting paradigms, emphasizing the evolution from classical physical and statistical models to modern machine learning, deep learning, ensemble learning, and probabilistic approaches. Critical research gaps are identified in multi-scale temporal modeling, hybridization strategies, uncertainty calibration, and the treatment of spatial correlations in hybrid microgrid settings. Motivated by these insights, the first methodological contribution introduces a hybrid deep learning pipeline that integrates PCA- and EMD-based feature engineering with an attention-augmented LSTM model and XGBoost residual refinement. This framework significantly improves short-term wind power forecasting accuracy across multiple real-world datasets from Tamil Nadu, India. Building upon these findings, the thesis proposes the Confined Attention-enabled LSTM (CAELSTM) architecture, which explicitly separates localized short-term dynamics from long-term periodic components via a dual-branch temporal modeling strategy. The confined-attention mechanism restricts the receptive field to the most informative temporal windows, while a parallel periodic extraction module captures diurnal and multi-scale variations. Extensive experiments demonstrate superior accuracy and robustness compared to classical LSTM variants and state-of-the-art hybrid models. To enhance v multivariate forecasting reliability, the research advances an ensemble-based framework combining CEEMDAN decomposition, Stacked GRUs, and a hybrid bagging–boosting strategy. This design effectively mitigates noise, reduces variance, and strengthens generalization under diverse operating regimes, providing a highly stable forecasting alternative for microgrid supervisory control. Recognizing the limitations of deterministic forecasting in operational environments, the thesis develops a Physics-Informed Adaptive Conformal Prediction (PI-ACP) approach for uncertainty quantification. The proposed method integrates regime-aware nonconformity scoring, adaptive calibration via exponentially weighted residuals, and physics-based generation constraints to ensure reliable, distribution-free predictive intervals that remain consistent under temporal drift and wind ramp events. An extended multivariate formulation enables joint multi-horizon interval forecasting for improved operational risk assessment. The final contribution presents a unified spatio-temporal transformer architecture designed for hybrid solar–wind microgrids. The model employs dual-spatial attention to capture geographical and meteorological couplings across distributed nodes, and dual-phase temporal attention to learn both short-term fluctuations and long-term seasonal trends. Coupled with a PI-ACP layer, the framework delivers physically consistent probabilistic forecasts suitable for real-time microgrid deployment. Validation on a six-node hybrid microgrid in the Khavda–Bhuj region demonstrates significant advancements in accuracy, generalization, and probabilistic reliability relative to deterministic, quantile-based, and diffusion-based baselines. Overall, the thesis establishes an integrated forecasting ecosystem combining hybrid deep learning, attention mechanisms, ensemble learning, adaptive probabilistic calibration, and spatio-temporal transformers. The developed methodologies collectively advance renewable generation forecasting for intelligent microgrid operations, supporting improved stability, planning, and decision-making in renewable-rich power systems. 2025-11-01T00:00:00Z http://dspace.dtu.ac.in:8080/jspui/handle/repository/22689 Title: DESIGN AND ANALYSIS OF SYNCHRONIZATION TECHNIQUES FOR CONTROL OF POWER ELECTRONIC CONVERTERS Authors: DEVI, OINAM LOTIKA; Singh, Alka (SUPERVISOR) Abstract: The increasing integration of renewable energy sources into electrical power systems has necessitated advanced synchronization techniques for power electronics converters. This thesis investigates synchronization methods, with particular emphasis on Phase-Locked Loop (PLL) technologies, to address challenges in grid-integrated systems under various disturbances. The primary objective is to develop and validate enhanced synchronization algorithms that overcome limitations of conventional PLLs in applications including photovoltaic systems, doubly fed induction generator (DFIG) wind energy converters, and electric vehicles. A comprehensive methodology combining mathematical modelling, stability analysis, extensive MATLAB Simulink simulations, and experimental validation using OPAL- RT real-time simulator was employed. Multiple PLL architectures were evaluated under abnormal grid conditions including voltage sags/swells, frequency jumps, DC offsets, and harmonic distortions. The research examined single-phase and three-phase PLLs, including Synchronous Reference Frame (SRF), Second-Order Generalized Integrator (SOGI), Least Mean Square (LMS), Least Mean Fourth (LMF), Modified Synchronous Reference Frame (MSRF), and fractional-order PLLs. Key findings demonstrate that the LMF PLL outperforms SRF PLL under phase shift and frequency change disturbances, while Type III Enhanced PLL exhibits superior frequency response with polluted grid voltage and DC offset. The proposed fractional- order FO-LPFO-PI MSRF-PLL with optimized parameters shows enhanced stability during grid abnormalities. Additionally, the novel H-LMS-SOGI PLL architecture provides exceptional dynamic responses, surpassing conventional structures under all adverse grid conditions tested. Experimental validation of DFIG systems with LMF- PLL confirms satisfactory performance under variable wind speeds and grid disturbances. This research significantly contributes to grid stability and renewable energy integration by providing validated solutions for maintaining reliable power system operation with high penetration of distributed energy resources. The dual validation approach ensures practical applicability, offering a comprehensive framework for xxi designing robust power electronics-based systems capable of handling complex grid disturbances in modern power networks. 2026-02-01T00:00:00Z http://dspace.dtu.ac.in:8080/jspui/handle/repository/22688 Title: DESIGN AND DEVELOPMENT OF SECURITY FRAMEWORK FOR SMART GRID INFRASTRUCTURE Authors: KUMAR, CHANDAN; Chittora, rakash (SUPERVISOR) Abstract: The smart grid is considered the future of electricity networks because it brings in intelligence, flexibility, and reliability to the power system. Unlike traditional power grids, smart grids use advanced information and communication technologies to enable two-way communication between utility companies and consumers. This makes it possible to integrate renewable energy sources, improve energy efficiency, and ensure better management of resources. However, as the smart grid becomes more connected and dependent on digital systems, it faces serious challenges related to cybersecurity. The openness and interconnectedness that make it efficient also make it vulnerable to different types of cyberattacks, data breaches, and unautho- rized access. If these issues are not addressed properly, they can affect the privacy of consumers, disrupt services, or even cause large-scale blackouts. This thesis presents an advanced, multi-layered framework for secure smart grid in- frastructure, integrating the strengths of deep learning, blockchain technology, and immersive collaborative platforms. Addressing the critical gaps in cybersecurity, privacy, and operational scalability, the research rethinks smart grid protection by unifying decentralized trust, intelligent anomaly detection, and efficient large-scale data management. The first major contribution is a robust secure data sharing architecture combin- ing a hybrid deep learning intrusion detection system using Variational Autoen- iv v coder (VAE) and Attention-based Bidirectional LSTM (ABiLSTM) with blockchain- backed audit trails and off-chain Inter Planetary File System (IPFS) storage. Ex- perimental validation on benchmark ToN-IoT and BoT-IoT datasets demonstrates significant performance improvements. The proposed system achieves near-perfect detection in percentage, accuracy (99.99), precision (98.99), recall (99.9), and F1 score (99.91), outperforming classical approaches (Naive Bayes, Decision Tree, Ran- dom Forest) by wide margins. Importantly, the hybrid on/off-chain design substan- tially reduces blockchain overhead, enabling real-time scalability lacking in earlier ledger-centric models. The research extends to IoT-enabled Electric Vehicles (EVs), devising a secure and privacy-preserving framework that leverages Stacked Sparse Denoising Autoencoders (SSDAE) and Attention-based LSTM for anomaly detection and data anonymiza- tion. Coupled with smart contract-driven authentication, this approach achieves highly effective multi-class threat detection and privacy protection, with detection and recall rates surpassing leading prior Long Short Term Memory (LSTM) and Artificial Neural Network (ANN) models while ensuring low-latency communication essential for mobile EV-grid integration. A novel facet of this thesis is the application of metaverse and digital twin technolo- gies, which enable unprecedented real-time, immersive collaboration and situational awareness for grid operators. Through federated and collaborative intrusion detec- tion, multi-operator security, and grid incident response now occur with reduced detection latency and enhanced visibility, an improvement over the isolated or man- ual supervision that dominated earlier solutions. In summary, this thesis offers (1) higher detection accuracy and reliability through advanced deep learning architectures; (2) integrated, scalable security and privacy vi solutions across grid and IoT-vehicle domains; (3) immersive, collaborative security and operational platforms; and (4) efficient, tamper-evident storage and auditing suitable for next-generation smart grid requirements. Collectively, these contribu- tions pave the way for intelligent, secure, and sustainable energy systems addressing technical, economic, and social imperatives in real-world smart grid deployments. 2025-06-01T00:00:00Z