Introduction: The properties of gold nanoparticles (AuNPs) are governed by their shape, structure, and intrinsic characteristics. The applications of gold nanoparticles in medical diagnosis and photovoltaics rely on precisely controlled features. However, achieving this precision is expensive, time-consuming, and labor-intensive due to the need for multiple reagents and highly dependent experimental conditions. Methods: We propose an artificial neural network (ANN)–based feature optimization approach for predicting nanoparticle features to facilitate nanoparticle synthesis. First, computationally less expensive machine learning (ML) models such as random forest and decision tree were used to rank input features to reduce computational time and complexity. Second, forward sequential feature selection (SFS) was applied as a greedy procedure that iteratively identifies the best new feature to add to the selected feature set. Results: We collected a large dataset, including reagent concentrations, temperature, SPR peak, and pH, alongside nanoparticle outcomes. The ANN optimization model was used to design a nanoparticle synthesis experiment that provides the best parameters for synthesizing precise nanoparticles, thereby tailoring particle size for various applications. Discussion: We introduce a novel approach that uses ML methods to determine the best feature ordering and applies an ANN-based SFS strategy to predict the optimized size of AuNPs as the desired output. The proposed ANN-SFS model reduces the need for multiple laboratory procedures typically required for optimizing ANN models in nanoparticle synthesis. Conclusion: This ANN approach improves nanoparticle synthesis compared to traditional optimization methods and supports the advancement of nanomaterial development. The proposed ANN model achieves 90.61% accuracy and a minimum mean square error (MMSE) of 9.4% in the predicted outcomes.
This work presents a bi-inspired optimization method, in particular the Flower Pollination Algorithm (FPA), for handling the Economic Dispatch (ED) problem. The algorithm is motivated by the biological process of pollination, especially biotic pollination, in which insects and birds help pollen move from one bloom to another so that blossoming plants can reproduce and survive best. The suggested method uses models of these natural processes to quickly find the best solutions. We examine the effectiveness of FPA on three conventional test systems with 3, 6, and 20 producing units, taking into consideration transmission losses. In addition, a comparison is made between Grey Wolf Optimization (GWO) and Ant Lion Optimization (ALO) methods. The simulation findings show that FPA works better than GWO and ALO on several evaluation criteria, especially when it comes to lower fuel costs and shorter computation times. This shows that it has a lot of promise as an effective way to solve ED problems.
Although Automatic Speech Recognition (ASR) in Bengali has seen significant progress, processing long-duration audio and performing robust speaker diarization remain critical research gaps. To address the severe scarcity of joint ASR and diarization resources for this language, we introduce Lipi-Ghor-882, a comprehensive 882-hour multi-speaker Bengali dataset. In this paper, detailing our submission to the DL Sprint 4.0 competition, we systematically evaluate various architectures and approaches for long-form Bengali speech. For ASR, we demonstrate that raw data scaling is ineffective; instead, targeted fine-tuning utilizing perfectly aligned annotations paired with synthetic acoustic degradation (noise and reverberation) emerges as the singular most effective approach. Conversely, for speaker diarization, we observed that global open-source state-of-the-art models (such as Diarizen) performed surprisingly poorly on this complex dataset. Extensive model retraining yielded negligible improvements; instead, strategic, heuristic post-processing of baseline model outputs proved to be the primary driver for increasing accuracy. Ultimately, this work outlines a highly optimized dual pipeline achieving a ∼0.019 Real-Time Factor (RTF), establishing a practical, empirically backed benchmark for low-resource, long-form speech processing.
Electric mobility has emerged as a cornerstone of global decarbonization strategies, with its successful deployment critically dependent on the coordinated integration of vehicle powertrain engineering, advanced battery technologies, charging infrastructure, power grid interaction, and intelligent control systems. This paper presents a comprehensive system-level critical assessment of electric mobility, providing an integrated analytical framework that unifies electric vehicle (EV) powertrains, electrochemical energy storage, grid impacts, artificial intelligence (AI), and sustainability considerations. The study systematically examines EV propulsion architectures, charging technologies, and the operational characteristics of contemporary and emerging battery chemistries, including lithium-ion variants ( Nickel–Manganese–Cobalt, Nickel–Cobalt–Aluminum, and Lithium Iron Phosphate), solid-state batteries, and sodium-ion batteries, with particular emphasis on degradation mechanisms, thermal safety, second-life utilization, and recycling pathways. The impacts of large-scale EV charging on power distribution networks are rigorously analyzed through power quality and voltage stability modeling, highlighting harmonic distortion, feeder loading, and voltage deviation challenges associated with high-power fast-charging infrastructure. Advanced mitigation strategies, including active filtering and AI-based grid impact prediction, are discussed to enhance grid resilience. AI is positioned as a core enabling technology throughout the EV ecosystem, with detailed coverage of data-driven and physics-informed approaches for battery health estimation, remaining useful life prediction, range estimation, smart charging control, traffic-aware routing, and charging queue optimization. Furthermore, emerging quantum-inspired optimization and quantum machine learning paradigms are identified as promising tools for addressing high-dimensional uncertainty in routing, charging scheduling, and battery diagnostics. A life-cycle sustainability perspective is incorporated to evaluate the environmental performance of EVs, emphasizing the influence of electricity generation mix, battery manufacturing emissions, material criticality, and recycling efficiency on overall greenhouse gas reduction potential. By synergizing engineering models, AI-driven intelligence, grid interaction analysis, and life-cycle assessment, this work delivers a unified blueprint for accelerating the transition toward sustainable electric mobility. The presented framework offers clear technical guidance for researchers, policymakers, and industry stakeholders seeking to design resilient, intelligent, and environmentally responsible electric transportation systems.
The swift rise in electrical power demand has resulted in frequent power shortages, highlighting the necessity for effective renewable energy usage. Solar photovoltaic systems represent a viable solution; however, their output can vary considerably based on the sun's position and environmental factors. This paper introduces a smart solar tracking and monitoring system aimed at enhancing power extraction while facilitating real-time performance evaluation. The system features a light-sensing mechanism based on LDRs, along with an amplifier, ADC, microcontroller, motor driver and limit switches to ensure automatic alignment of the solar panel in line with the sun's trajectory. Furthermore, an IoT-based monitoring system is incorporated through an Arduino platform to continuously track and transmit important parameters from a $5-\mathrm{W}$ solar panel. The collected data is released online, allowing for remote monitoring of output power, early detection of panel problems, wiring issues and performance degradation caused by dust accumulation. The proposed technology improves energy harvesting efficiency and offers a scalable solution for smart solar power plant management. The proposed system has been simulated and experimentally verified and its performance is satisfactory.
