Sustainable Cr (VI) removal using Ragi Husk: A data-driven metaheuristic optimization approach Lakshmana Rao Kalabarige, D. Krishna, Upendra Kumar Potnuru, M. Raviraja Holla Journal of Engineering Research Kuwait, 2026 Chromium, widely used in industries such as chemicals, textiles, and metal finishing, poses serious environmental and health hazards when discharged in its hexavalent form (Cr(VI)). This investigation highlights the use of Ragi Husk powder as a sustainable, resource-efficient and cost-effective bio-adsorbent for extracting Cr(VI) from industrial wastewater. Traditional optimization methods, like Box–Behnken Design (BBD), often fail to capture nonlinearities in adsorption processes. To overcome this, six machine learning (ML) models—three tree-based ( M 1 :Decision Tree Regressor (DTR), M 2 :Random Forest Regressor (RFR), M 3 :Extra Trees Regressor (ETR)) and three boosting-based ( M 4 :AdaBoost Regressor (ADBR), M 5 :Gradient Boosting Regressor (GBR), M 6 :LightGBR (LGBR))—were employed to predict Cr(VI) removal efficiency. These models integrated with Nelder–Mead Optimization (NMO), forming hybrid frameworks ( NMO − M 1 to NMO − M 6 ) to identify optimal process parameters. Among them, LGBR and ETR exhibited superior prediction score ( R 2 − Score ) of 0.99%. The NMO-LGBR ( NMO − M 6 ) optimization approach achieved a maximum Cr(VI) removal efficiency of 82.26%, which is 1%–4% higher than experimental results. The comparative analysis with existing literature revealed that the proposed models improved Cr(VI) removal efficiency by up to 4% over BBD and 1%–2% over ANN models. Experimental validation showed close alignment with model predictions, confirming the robustness of the proposed framework. Additionally, SEM analysis before and after Cr(VI) adsorption revealed that Ragi Husk retained its structural integrity, affirming its reusability and effectiveness. This work introduces a novel, scalable, data-driven approach that significantly enhances Cr(VI) removal efficiency while reducing experimental costs and time, demonstrating strong potential for eco-friendly wastewater treatment applications. • Machine learning–based framework for Cr(VI) removal using Ragi Husk bio-adsorbent. • LGBR and ETR models showed superior predictive performance among six ML algorithms. • Hybrid ML–Nelder–Mead optimization achieved 82.26% Cr(VI) removal efficiency. • ML–NMO method reduced experimentation time with R² > 0.99 accuracy. • SEM confirmed Ragi Husk integrity, supporting its use in wastewater treatment.
DBLN-Based IC-DSTATCOM for Power-Line Conditioning in the Power Distribution System Mrutyunjaya Mangaraj, Ramana Pilla, Lakshmana Rao Kalabarige, Upendra Kumar Potnuru, M. Raviraja Holla IEEE Access, 2026 With the growing demand for clean and sustainable electricity, modern power distribution systems must ensure not only efficiency but also high-power quality. This study proposes and investigates an advanced three-phase three-wire Inductively Coupled Distributed Static Compensator integrated with a Deep Belief Learning Network control strategy for enhancing current-based power-line conditioning. The proposed system incorporates an Inductive Filtering Transformer alongside a conventional Directly Coupled DSTATCOM, acting as an impedance-matching interface among the grid, compensator, and load. A mathematically grounded DBLN framework is developed to generate optimized switching signals for both DC- and IC-DSTATCOM configurations, ensuring dynamic adaptability and precise control. Extensive simulation studies in MATLAB/Simulink validate the effectiveness of the system under various static and dynamic loading scenarios. Key performance improvements include significant reduction in Total Harmonic Distortion, improved power factor, effective voltage regulation, and reduced compensator size. The system’s performance is further validated through experimental implementation, demonstrating compliance with international standards such as IEEE 519-2017 and IEC 61000-3-12. These findings underscore the potential of the DBLN-based IC-DSTATCOM as a robust and scalable solution for smart grid applications.
Dataset for optimized design parameters of three-phase induction motors with validation through machine learning Upendra Kumar Potnuru, Srinivasa Kishore Teegala, Lakshmana Rao Kalabarige, Vidyabharati Ippili, M Raviraja Holla Data in Brief, 2025 Three-phase induction motors continue to dominate industrial and commercial sectors due to their high efficiency, robustness, and low maintenance requirements. This data article presents a curated dataset of optimized design parameters for three-phase induction motors covering output ratings from 0.5 kW to 100 kW. Motor parameters (stator/rotor dimensions, winding details, air-gap flux, copper/core losses, torque, slip, efficiency, power factor, currents, flux per pole, temperature rise, etc.) were computed by a Python based computational framework implementing standard electromechanical design equations. The original 200 design instances were scientifically expanded to 6000 to represent viable design alternatives. To demonstrate dataset reliability and practical utility, descriptive statistics and tree-based regressors (Decision Tree, Random Forest, Extra Trees) were applied on held out test sets and evaluated with MAE, RMSE, and R². The Extra Trees model yielded the lowest errors (e.g., MAE ≈ 7.31 W and RMSE ≈ 11.62 W for full-load losses; MAE ≈ 0.0073% and RMSE ≈ 0.0232% for efficiency) with R² ≳ 0.9996 and residuals concentrated near zero (≈0.0073-9.1536). These results confirm the internal consistency of the physics-driven dataset and its suitability for simulation, preliminary design studies, controller tuning, and predictive maintenance. However, the current dataset does not incorporate nonlinear magnetic effects, thermal constraints, or experimental validation, which will be addressed in future extended versions.
Optimizing dielectric properties of corannulene nanomaterial for enhanced performance of next-generation electric vehicle batteries: A Machine learning and Nelder-Mead optimization approach Upendra Kumar Potnuru, Lakshmana Rao Kalabarige, Manohar Mishra, Thirumala Rao Gurugubelli, Salman S Alharthi, Mohan Rao Tamtam, Ravindranadh Koutavarapu Results in Physics, 2025 • Nelder-Mead optimization algorithm, coupled with a stacking model. • Gradient Boosting Regression sequentially builds weak learners to improve predictive performance. • The integrated machine learning and optimization approach represents a significant step. • Stacking as the preferred choice for predictive modeling on the given dataset. Enhancing the performance of next-generation electric vehicle (EV) batteries relies on advancements in energy storage technologies. This study explores the potential of corannulene, a nanomaterial renowned for its exceptional electronic properties, to optimize battery performance by fine-tuning its dielectric properties. Traditionally, such optimization methods have been limited to conventional approaches. However, this research adopts a novel machine learning methodology utilizing an experimental dataset encompassing five key features: “frequency,” “real part,” “imaginary part,” “dielectric strength,” and “dielectric loss.” These features are employed to train various machine learning models, including baseline, ensemble-tree, and boosting techniques, with the aim of predicting optimal values for dielectric loss (to minimize negative impact) and dielectric strength (to maximize efficiency). Subsequently, a Nelder-Mead optimization algorithm, coupled with a stacking model, is employed to determine the optimal range for these features, thereby enhancing battery performance. Promising results are obtained, with Ensemble Tree Regression (ETR) and Stacking models achieving remarkable R 2 scores of 0.9973 and 0.9995 for predicting dielectric loss and dielectric strength, respectively. The Nelder-Mead optimization, guided by these machine learning models, effectively recommends optimal ranges for corannulene nanomaterial properties. Notably, ETR and Stacking based optimization outperform other models. This integrated machine learning and optimization approach represents a significant step toward designing not only more efficient but also more sustainable EV batteries, thereby accelerating the automotive industry’s transition to a greener future.
Hashed identity based secure key and data exchange in wireless sensor networks using IEEE 802.15.4 standard International Journal of Applied Engineering Research, 2015
A sturdy compression based cryptography algorithm using self-key (ASCCA) International Journal of Engineering and Technology, 2015
