BITS Faculty Publications
Permanent URI for this communityhttp://localhost:4000/handle/123456789/1867
Browse
65 results
Search Results
Item Machine learning-based prediction of axial load and strain capacities for circular FRP-concrete-steel double-skin tubular columns(Taylor & Francis, 2025-12) Singh, Shamsher Bahadur; Barai, Sudhir KumarFiber-reinforced polymer (FRP)-concrete-steel double-skin tubular columns (hybrid DSTCs) are innovative composite columns offering high strength, ductility, corrosion resistance, and lightweight due to hollow cross-section. Despite extensive experimental, numerical, and analytical studies, accurately predicting the behavior using traditional methods remains challenging. Experimental and numerical methods are costly and time-consuming, while analytical approaches are conservative and may not effectively capture complex and nonlinear relationships. This study compares five machine learning (ML) models with two existing empirical equations for predicting the axial load and axial strain of circular hybrid DSTCs. An extensive dataset of 249 specimens from the literature was used to train and test ML models. Five ML models, namely, multiple linear regression (MLR), decision tree (DT), random forest (RF), K-nearest neighbors (KNNs), and extreme gradient boosting (XGBoost), were trained using eight input parameters. Results indicate that the XGBoost model achieved the highest accuracy in predicting both axial load and strain capacities, with R2 values of 0.87 and 0.96, respectively. Among the empirical equations, Louk Fanggi and Ozbakkaloglu’s equation performed better than traditional ML models such as MLR and DT for axial load prediction, achieving an R2 value of 0.785, compared to 0.72 for MLR and 0.74 for DT. Feature importance analysis further identified the significant influence of geometric parameters on axial load prediction and material properties on axial strain prediction. Additionally, a user-friendly web application is developed, allowing users to easily predict the axial load and strain of circular hybrid DSTCs, demonstrating ML’s efficiency as a data-driven alternative to empirical approaches.Item FLEMOflash – flood loss estimation models for companies and households affected by flash floods(2025-04) Guntu, RavikumarIn light of the increasing losses from flash floods intensified by climate change, there is a critical need for improved loss models. Loss assessments predominantly focus on fluvial flood processes, leaving a significant gap in understanding the key drivers of flash floods and the effect of preparedness on losses. To address these gaps, we introduce FLEMOflash—a novel multivariate probabilistic Flood Loss Estimation Model compilation for flash floods. The models are developed for companies and households based on survey data collected after flash flood events in 2002, 2016, and 2021 in Germany. FLEMOflash employs a data-driven feature selection approach, combining machine learning techniques (Elastic Net, Random Forest, XGBoost) to identify key drivers influencing flash flood losses and Bayesian networks to model probabilistic loss estimates, including uncertainty. Model-based findings show that in extreme hazard scenarios, high preparedness can reduce building losses by up to 47 % for large companies. Households who knew exactly what to do during high water depth were able to reduce their building losses by 77 % and contents losses by 55 %. Thus, FLEMOflash can support risk communication and management by providing reliable estimation of flash flood losses along with the loss differential considering the level of risk preparedness.Item Deciphering the drivers of direct and indirect damages to companies from an unprecedented flood event: A data-driven, multivariate probabilistic approach(Copernicus Publications, 2026) Guntu, RavikumarFloods are among the most destructive natural hazards, causing extensive damage to companies through direct impacts on assets and prolonged business interruptions. The July 2021 flood in Germany caused unprecedented damage, particularly in North Rhine-Westphalia and Rhineland-Palatinate, affecting companies of all sizes. While the drivers of company damages from riverine flooding are well documented, the drivers of both direct and indirect damages during an extreme flash flood event have not yet been examined. This study addresses this gap using survey data from 431 companies affected by the July 2021 flood. Results show that 62 % of companies incurred direct damages exceeding EUR 100 000. Machine learning models and Bayesian network analyses identify water depth and flow velocity as the primary drivers of both direct damage and business interruption. However, company characteristics (e.g., size premise, number of employees) and preparedness also play critical roles. Companies that implemented precautionary measures experienced significantly shorter business interruption durations – up to 58 % for water depths below 1 m and 44 % for depths above 2 m. These findings offer important insights for policy development and risk-informed decision-making. Incorporation of behavioural indicators into flood risk management strategies and improving early warning systems could significantly enhance business preparedness.Item Estimation of soil water characteristic curve using machine-learning algorithms and its application in embankment response(ASCE, 2025-01) Showkat, RakshandaThe parameters of the soil water characteristic curve (SWCC) play a pivotal role in the examination of unsaturated soil behavior. This study employs three machine learning models—random forest (RF), extreme gradient boosting (XGBoost), and multiexpression programming (MEP)—to predict the SWCC using key soil properties. Among them, the RF model demonstrated the most robust performance in SWCC prediction. The Shapley Additive Explanation (SHAP) analysis further reveals that suction is the most influential factor affecting SWCC predictions, with other input parameters also contributing significantly. Additionally, the MEP model offers a straightforward expression for SWCC estimation and, thus, proved practical for predicting embankment responses and exhibited superior accuracy over traditional methods, such as the Arya and Paris model (ACAP). For a precise assessment of the hydromechanical response of the embankment subjected to infiltration, an increase in pore pressure is observed when employing the MEP model compared to the ACAP model for fine-grained soils. The findings emphasize the potential of RF and MEP in enhancing SWCC prediction and their practical implications for soil engineering applications.Item Optimizing the upcycling of microplastics to a carbon-based adsorbent for water treatment: An integrated experimental and computational approach(Elsevier, 2026-05) Goonetilleke, AshanthaMicroplastics (MPs) are an escalating environmental hazard because they persist in aquatic ecosystems and resist removal by conventional water treatment technologies. A novel data-driven strategy that upcycles MPs into engineered carbonaceous adsorbents via hydrothermal carbonization (HTC) is presented. By systematically varying three synthesis variables – feedstock loading, acid type and acid concentration – a range of carbonaceous materials (CMs) was produced and evaluated for their ability to adsorb Reactive Orange 84 dye. An integrated full factorial design of experiments encompassing both, material synthesis variables (acid type, acid concentration, and polyester (PES) material concentration) and the application variables (CM dose) was implemented. Subsequent statistical analysis and PCA identified acid type, acid concentration, CM dose, and PES concentration as dominant factors controlling adsorption capacity (q_e) and removal percentage. To refine the optimization, several machine learning (ML) models – linear regression, support vector machines, ensemble methods, and neural networks – were trained on the experimental dataset. Acid treated CMs consistently outperformed those synthesized under neutral conditions, with optimal performance observed at moderate acid concentrations. The key innovation in this study lies in the integrated experimental-computational framework that models the entire process (synthesis −> application), coupling rigorous statistical screening with advanced ML prediction. This delivers actionable guidance for the rational design of acid-modified carbonaceous adsorbents and advances the upcycling of MPs for water treatment applications.Item Revisiting El-sayed synthesis: bayesian optimization for revealing new insights during the growth of gold nanorods(ACS, 2024-02) Rao, AnishIn diverse fields, machine learning (ML) has sparked transformative changes, primarily driven by the wealth of big data. However, an alternative approach seeks to mine insights from “precious data”, offering the possibility to reveal missed knowledge and escape potential knowledge traps. In this context, Bayesian optimization (BO) protocols have emerged as crucial tools for optimizing the synthesis and discovery of a broad spectrum of compounds including nanoparticles. In our work, we aimed to go beyond the commonly explored experimental conditions and showcase a workflow capable of unearthing fresh insights, even in well-studied research domains. The growth of AuNRs is a nonequilibrium process that remains poorly understood despite the presence of well-established seeded growth protocols. Traditional research aimed at understanding the mechanism of AuNR growth has primarily relied on altering one reaction condition at a time. While these studies are undeniably valuable, they often fail to capture the synergies between different reaction conditions, thus constraining the depth of insights they can offer. In the present study, we exploit BO, to identify diverse experimental conditions yielding AuNRs with similar spectroscopic characteristics. Notably, we identify viable and accelerated synthesis conditions involving elevated temperatures (36–40 °C) as well as high ascorbic acid concentrations. More importantly, we note that ascorbic acid and temperature can modulate each other’s undesirable influences on the growth of AuNRs. Finally, by harnessing the power of interpretable ML algorithms, complemented by our deep chemical understanding, we revisited the established hierarchical relationships among reaction conditions that impact the El-Sayed-based growth of AuNRs.Item pLMMoRF: A web server that accurately predicts membrane-interacting molecular recognition features by employing a protein language model(Elsevier, 2025-09) Basu, SushmitaInteractions between proteins and lipids are crucial for numerous cellular processes. Some of the lipid interacting segments in protein sequences are intrinsically disordered regions (IDRs), which may gain secondary structures upon binding. We collected experimentally annotated lipid-interacting IDRs, named membrane molecular recognition features (MemMoRFs). We used this dataset to develop and test an accurate and relatively fast sequence-based MemMoRF predictor, pLMMoRF, thereby supporting tedious and costly experimental identification of MemMoRFs. Our predictor utilizes a protein language model (pLM) which we processed to generate inputs to a deep convolutional neural network. We considered various pLMs (ESM-2, ProstT5, ProtT5 and Ankh) and applied feature selection to reduce their outputs, creating a more compact neural network model. pLMMoRF leverages the Ankh-based model, selected for its higher accuracy compared to our other models. Tests on low similarity test datasets demonstrate that pLMMoRF is more accurate than the sole current predictor of MemMoRFs, CoMemMoRFPred. Moreover, pLMMoRF has a relatively small computational footprint because of the compact network size and use of dedicated GPU nodes. This allowed us to make MemMoRF predictions for the human proteome. We analyzed these predictions and made them publicly available, facilitating an improved understanding of functions of membrane-coupled proteins. Our work underscores the importance of selecting key embedding features to enhance predictive performance and reduce computational footprint of sequence-based predictors of protein functions. The web server for the pLMMoRF predictor and the predictions for human proteinsItem Machine learning-based predictive life cycle assessment approach during product design(Elsevier, 2025) Sangwan, Kuldip SinghMost of the past efforts for the environmental impact assessment has been carried out based on the manufactured product. Environmental improvement strategies intertwined with the product design will be more effective as 70-80% of the environmental impacts are fixed during the design phase. However, the major challenge to carry out the life cycle assessment (LCA) during design phase – predictive LCA – is the data scarcity during the design phase. Therefore, this paper proposes data augmentation using deep learning techniques to overcome this challenge for predictive LCA. This paper proposes a four-phase predictive LCA methodology consisting of (i) identification of design requirements and environmental aspects, (ii) database building based upon product descriptors & environmental performance data, (iii) deep learning assisted data preprocessing, and (iv) machine learning based predictive LCA models – random forest, support vector machine, and neural network. It was found that random forest gives the better prediction based on model evaluation metrics of mean absolute error (MAE), mean square error (MSE), root mean squared error (RMSE), and R-square (R2). This research equips environmentalists, companies, researchers, and businesses with a predictive environmental conscious approach to their decision-making processes early in the design phase, thereby, fostering a sustainable approach right from the inception of a product’s design.Item Implementation of battery degradation on lithiumion batteries using PYNQ-FPGA(IEEE, 2024) Srinivasan, P.Predicting the remaining usable life (RUL) of a lithium-ion battery properly is vital for appropriate maintenance and overall health evaluation, which is particularly pertinent in the burgeoning electric vehicle industry, where optimising battery performance is essential. Determining the rate of battery deterioration is a complex task because of the wide variety of internal and external elements that could affect it. Our study addresses this challenge by using datasets on battery ageing sourced from NASA's Prognostic Center of Excellence (PCoE) to introduce a data-driven approach for State of Health (SOH) estimation. In our pursuit of RUL prediction, we have devised a machine-learning model employing the ADAM optimiser for optimisation. Consequently, our proposed model utilises software programming on PYNQ FPGA to discern battery degradation. The findings of these innovative approaches are thoroughly analysed and assessed, showcasing the effectiveness of our approach in navigating the complexities associated with predicting battery RUL.Item A data analytic-based logistics modelling framework for E-commerce enterprise(Taylor & Francis, 2022-01) Verma, AbhishekData-driven approaches have noteworthy significance in managing and improving logistics in E-commerce enterprises. This study focuses on the development of an integrated framework to analyse the Brazilian E-Commerce enterprise public dataset. From the analysis, it is found that sellers of Ibitinga city of SP state had the most count of late deliveries where 42 sellers are under-performing in terms of estimated delivery time. Locations of customers and sellers were spotted on a map to get a geographical representation. The proposed framework may help E-Commerce enterprise owners and retail merchants to make better decisions related to sales and E-Commerce enterprise logistics.