Hybrid PV–wind–battery microgrid power management using osprey optimized convolutional neural network for EV charging applications Sivaraman P R, Devaraj V Engineering Research Express, 2026 The rapid growth of electric vehicle (EV) charging demand, combined with the variability of renewable energy sources, increases the complexity of maintaining power balance and voltage stability in microgrids. This work develops an enhanced power management strategy for a hybrid PV–wind–battery-based EV charging infrastructure. The overall microgrid power coordination is handled using a Convolutional Neural Network whose parameters are tuned through the Osprey Optimization Algorithm (OOA-CNN). The controller generates inverter reference currents using inputs such as PV and wind power, storage state-of-charge (SOC), EV SOC, and DC-link voltage. Simulation studies demonstrate that under constant irradiance and wind profiles, the proposed method achieves a peak grid export of 30.11 kW and storage battery discharge of 8.19 kW at 80% SOC, while sustaining a DC-link voltage of approximately 613.5 V. In variable operating conditions involving fluctuating irradiance and wind speed, the OOA-CNN maintains DC-link stability within ±0.5% of the nominal 612 V and provides a grid exchange of 28.09 kW at 0.4 s, along with effective storage charging of –14.67 kW when the battery SOC is 30%. This method attains a lower RMSE (0.0946) compared with GA-CNN (0.1294), confirming improved prediction accuracy and control precision. A 750 W laboratory prototype further validates the feasibility of the proposed scheme by demonstrating stable EV charging and consistent power sharing under dynamic environmental conditions.
Intelligent Speed Regulation of Electric Vehicles on Unpaved Road via LSTM-Driven RNN Models P R Sivaraman, P Sivakumar, Shivaani Balakrishnan, Velantinaa C J 7th International Conference on Energy Power and Environment Icepe 2025, 2025 This paper presents an adaptive speed control system for BLDC motors using PID and LSTM-based optimization. The system integrates real-time Hall sensor feedback to regulate motor speed by computing the speed error and adjusting the inverter switching pattern. The LSTM model optimizes PID gains dynamically, ensuring stable operation despite torque variations and reference speed changes. Simulation results demonstrate improved speed tracking, reduced overshoot, and enhanced stability, making this approach efficient for intelligent motor control applications.
A Deep Learning-Driven Framework for Sustainable and Intelligent Energy Management in Smart Cities Zokir Mamadiyarov, P R Sivaraman, N M G Kumar, Pavitar Parkash Singh 2025 2nd International Conference on New Frontiers in Communication Automation Management and Security Iccams 2025, 2025 To minimise their negative effects on the environment, cutting-edge technologies are needed to optimise energy consumption in today's fast-growing metropolitan areas. With the integration of Long Short-Term Memory (LSTM) networks, Convolutional Neural Networks (CNNS) & Reinforcement Learning (RL), this study offers a deep learning-based energy optimisation framework for Iot-enabled smart cities that will improve sustainability, encourage the use of renewable energy sources & increase efficiency. The suggested hybrid model improves system performance for better use of renewable energy sources, lower carbon emissions & more energy efficiency as compared to conventional methods. By utilisation of cloud-based analytics, the system allows adaptive learning and real-time decision-making by deploying lightweight algorithms on edge devices. The results show that the current methods are shoddy, and they provide a scalable route for environmental city planning. The integration of digital twin technologies with large-scale Iot deployment is an area that needs more investigation for improved predictive abilities in future work.
Data Science and ML Algorithms to Investigate Different Testing Scenarios for Various Anomalies in Driven Electric Motor Arpit Jain, P. R. Sivaraman, Abhishek Choubey, Guna Sekhar Sajja AI and Machine Learning for Mechanical and Electrical Engineering, 2025 Flexible manufacturing systems are a need in today’s complicated and unpredictable market, which is driving innovation in manufacturing technology. So, new ideas for production systems need integration both horizontally and vertically, both inside and across value networks, as well as within and outside of a manufacturing facility. A large quantity of data providing information on production, procedures, and equipment on the shop floor may be acquired via the integration of different settings. Processing and analysing this data and information may disclose industrial system insights and guide strategic decision-making. Applying predictive maintenance (PdM) is one of the possibilities that arise in this setting. Nevertheless, PdM’s industrial applications are still in the early stages. The purpose of this work is to provide a procedure for reducing the time and effort needed to define the key characteristics of electric motor-driven rotating machines’ instrumentation setup. Famous algorithms in the fields of machine learning and data science are scouted for this purpose. To identify several forms of abnormalities, such as normal, imbalanced, overloaded, uncoupled, and misaligned, various testing scenarios were envisioned for the studies. These algorithms can categorize deviations, and k-nearest neighbour and multi-layer perceptron have the highest accuracy ratings.
A Fractional and Stochastic Model for Axonal Microtubule Bundles Under Dynamic Loading P M Promince, P A Gowri Sankar, R Madhanraj, P. R. Sivaraman, G Karthikeyan, K. Karthikkumar Proceedings of 3rd International Conference on Intelligent Cyber Physical Systems and Internet of Things Icoici 2025, 2025 Traumatic axonal injury is a key feature of diffuse brain trauma, often caused by structural failure in axonal microtubule bundles under mechanical stress. Existing models typically assume simple viscoelastic behavior and deterministic rupture thresholds, limiting their ability to represent biological variability and time-dependent damage. This study introduces an improved computational model that incorporates fractional-order viscoelastic damping and probabilistic rupture based on both strain magnitude and rate. A two-dimensional network formulation is used to simulate load redistribution and spatially heterogeneous failure, capturing rupture cascades across microtubule bundles. The enhanced model shows significantly greater energy dissipation and more realistic damage progression compared to the original model. It predicts nonlinear and rate-sensitive rupture patterns consistent with experimental findings. Simulation results across various loading conditions demonstrate the model’s ability to reproduce key features of microstructural damage, such as delayed failure onset, progressive degradation, and sensitivity to dynamic strain inputs. This work advances the modeling of axonal injury and offers a framework for more accurate prediction of brain tissue response under traumatic loading.
Analysis of Contemporary Trends in Industrial Stability Across Various Countries Through Text Mining A. Alhazemi, P. R. Sivaraman, Abhishek Sharma, Ankitha Sharma, M. Clement Joe Anand Text Mining and Sentiment Analysis in Climate Change and Environmental Sustainability, 2024 Sustainability in business is more important than ever before due to the current spike in environmental concerns. Throughout the globe, people are looking for businesses to operate in a way that doesn't affect the environment too much while fostering a harmonious relationship between the company, the environment, and society. Companies often disclose their activities through environmental and social responsibility (ESR) reports. This study seeks to comprehend and evaluate current patterns in CSR reports submitted by Fortune 500 corporations via the use of text-mining techniques. It looks at sustainability reports from different nations and different sectors and contrasts their emphasis on economic, social, and governmental sustainability components. According to the study's findings, sustainability reports differ in their emphasis depending on many criteria, including the company's size, industry, duration on the Fortune 500 list, and country of origin. As a result, it's useful for learning why the organisation is so concerned with certain aspects of corporate sustainability.
Stability Enhancement for PV and Wind Penetrated Power System Using Machine Learning Prediction Sivaraman P R, Sivakumar P, Amirthavarshini A, Ashika P G, Hemanth Kumar R, Jagan Balaji T Proceedings of the 3rd IEEE International Conference on Power Electronics Intelligent Control and Energy Systems Icpeices 2024, 2024 In modern power systems, the integration of distributed generation (DG) has introduced both opportunities and challenges. It has also gained significant attention due to its potential benefits in terms of renewable energy utilization, improved system reliability and reduced transmission losses. As PV and wind are intermittent variables of the power system, stability problems arise which can be recovered using conventional PID controllers through the advanced tuning methods which incorporate the AI based tuning parameter. The process of tuning of the parameter is approached through proper artificial intelligence based trained data. This approach will enhance the power angle stability of the existing system.
Design and development of mobility system for double amputees T S Saravanan, R Sagayaraj, P R Sivaraman, D Sivamani, R Jaiganesh, P Ragupathy Intelligent and Soft Computing Systems for Green Energy, 2023 Mobility refers to a human being's ability to move his or her body in an environment and to manipulate objects. Collectively, these activities enable the individual to pursue life activities of their choosing. An individual's ability to perform any mobility task can be compromised by impaired body functions or structures. The proposed system mainly focuses on the “transfemoral” type of above-knee amputation (AKA) in which both knees are amputated above the knee joint. Prosthetic legs are used by shaping the limbs. But not everyone is capable of using prosthetic legs as they have their own concerns. The proposed system aims to develop a mobility system for such persons with system clearance nearly six inches from ground level, at standstill, so that the amputees can use their hands to get onto the system and perform a transition without the help of the caretaker. Horizontal and vertical transitions are included in the system. The horizontal transition is realized using wheels controlled by an electric motor at constant speed. At standstill condition the amputees can elevate themselves using vertical transition. This aids them in performing necessary activities at a considerable height above ground level. Vertical movement is implemented using scissor lift technique actuated by a permanent magnet DC (PMDC) electric motor. Vertical transition height is restricted to two feet in order to ensure safety.
Exploration and applications of electronic balance for high power discharge lamps at high frequency through power factor modification International Journal of Scientific and Technology Research, 2020
Performance enhancement of elementary additional series positive output super lift converter fed PMBLDC drive International Journal of Scientific and Technology Research, 2020
