Sudipto Roy

Sudipto Roy

Business Administration and Management, University of the Cumberlands, Williamsburg, Kentucky, USA
+4
orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2021
Verified
Journal Article
Open access
PDF file
Sudipto RoySudipto Roy
Sudipto Roy
MHMesbaul Hasan
Mesbaul Hasan
This study investigated the effectiveness of a machine learning–based process-mining framework for anomaly detection and quality assurance in high-throughput manufacturing environments. The research aimed to determine whether integrating process-mining metrics with data-driven machine learning techniques could outperform the traditional Statistical Process Control (SPC) system in predicting product non-conformance and operational inefficiencies. A quantitative research design was employed using a dataset of 142,368 production instances collected from Manufacturing Execution System (MES), Supervisory Control and Data Acquisition (SCADA), and Quality Management System (QMS) logs across three automated production lines. Variables included cycle-time variation, waiting time, machine utilization, rework frequency, conformance fitness score, and sensor-derived telemetry aggregates. A binary logistic regression model was developed to estimate the probability of product non-conformance. The model achieved a statistically significant fit (χ² = 482.76, p < 0.001) and explained 46.2% of the variance in defect occurrence (Nagelkerke R² = 0.462). Key predictors included Conformance Fitness Score (β = –0.72, p < 0.001) and Rework Frequency (β = 0.58, p < 0.01), indicating that lower process conformance and higher rework activity substantially increased the likelihood of defects. Comparative performance analysis showed that the machine learning–based model achieved higher predictive accuracy (91.3%) than the SPC baseline (84.9%), along with improved AUROC (0.93 vs. 0.82) and AUPRC (0.48 vs. 0.41) scores. These findings demonstrated that the proposed model provided superior anomaly detection capability, reduced false-alarm rates, and enhanced predictive precision. Overall, the integration of machine learning and process-mining analytics significantly improved operational reliability and quality assurance performance. The study concluded that adopting such intelligent process-monitoring systems can strengthen defect prevention, streamline production decision-making, and support real-time process optimization in smart manufacturing environments. The results contributed to advancing data-driven quality engineering and reinforced the role of AI-powered analytics in modern industrial process control.

1 September 2023
Verified
Journal Article
Open access
PDF file
Sudipto RoySudipto Roy
Sudipto Roy
Manufacturing enterprises increasingly operate under the dual imperative of enhancing operational efficiency while simultaneously reducing energy consumption. However, conventional improvement approaches often remain siloed—addressing either productivity or sustainability—thereby constraining opportunities for reproducible and system-wide optimization. This study aims to develop and empirically evaluate an AI-enhanced, multi-objective optimization framework that aligns lean manufacturing principles with energy-conscious operations. Using a quantitative, cross-sectional, case-based design, enterprise cloud data pipelines were established to integrate Manufacturing Execution System (MES) and Supervisory Control and Data Acquisition (SCADA) historians with sub-metered energy logs across multiple discrete-part production lines in three industrial plants. The primary variables included Lean Practice Intensity (measured through a five-point Likert instrument), Overall Equipment Effectiveness (OEE) and its components, and Energy Intensity (kWh per good unit). Additional operational moderators such as peak kW, demand variability, product mix, machine age, and shift patterns were also captured to account for contextual heterogeneity. The analysis pipeline consisted of scale reliability testing, descriptive and inferential statistics, correlation analysis, and hierarchical multiple regression incorporating plant- and shift-level controls. Furthermore, mediation analysis tested the role of efficiency as an intervening variable, while moderation analysis assessed the conditional effects of demand variability. To complement empirical inference, a surrogate-assisted evolutionary optimization algorithm was employed to generate Pareto-optimal fronts and knee solutions, benchmarked against baseline production schedules. Key results demonstrate that Lean Practice Intensity exhibits a positive association with OEE and a negative association with energy intensity, confirming efficiency as a significant mediator with a residual direct effect of lean practices on energy outcomes. The influence of lean practices was found to diminish under high demand variability, indicating a boundary condition for implementation. The optimization results further revealed that knee-point policies typically improved OEE by approximately 2–5 percentage points, reduced energy intensity by 6–15%, and curtailed peak demand by around 6%, all without compromising delivery performance or schedule adherence. From a managerial and systems integration perspective, the findings highlight the importance of viewing lean routines as actionable energy levers rather than ancillary practices. The study recommends institutionalizing variability-aware and tariff-aware production planning, and adopting a governed digital stack that links key performance indicators (KPIs), digital-twin telemetry, and AI-driven optimization engines.

1 June 2024
Verified
Journal Article
Open access
PDF file
Sudipto RoySudipto Roy
Sudipto Roy
HIHasan Imam
Hasan Imam
This study investigates how data-driven capabilities and sustainable resource-allocation policies jointly influence supply chain resilience (SCRes) under disruption and uncertainty. Adopting a quantitative, cross-sectional, multiple case-study design, the research integrates survey-based measurement, archival operational data, and stochastic simulation–optimization to link organizational capabilities—visibility, collaboration, flexibility, supplier diversification, redundancy, risk orientation, and allocation efficiency—to key resilience outcomes, including service recovery, time-to-recovery (TTR), backorder intensity, cost variance, and emissions. Data were collected from 190 professionals across four international firms in discrete manufacturing, FMCG, healthcare logistics, and electronics sectors, each providing both perceptual and objective data spanning 12–24 months of operations. Hierarchical multiple regression models, supported by mediation and moderation analyses, revealed that collaboration, digital visibility, and allocation efficiency were the strongest predictors of resilience performance, while flexibility and diversification contributed moderate incremental effects. Allocation efficiency partially mediated the impact of visibility and collaboration on outcomes, demonstrating that information and coordination enhance resilience primarily through improved resource allocation. Moreover, capability effects intensified under uncertainty—visibility’s and collaboration’s benefits were significantly amplified by demand volatility and lead-time variability. A composite Resilience Performance Index (RPI) was developed to align statistical and operational metrics, linking survey constructs to observed KPIs. Monte Carlo simulation experiments using empirically calibrated disruption parameters validated these statistical insights: sustainability-aware optimized policies improved mean service levels by 3–6 percentage points, reduced TTR by 15–27%, lowered backorder intensity by up to 24%, and cut emissions intensity by 6–11% compared to status quo. These findings confirm that resilience and sustainability can be jointly enhanced through data-driven allocation strategies. The study contributes an integrated, replicable framework that bridges measurement, inference, and decision experimentation—offering both theoretical clarity on capability mechanisms and practical guidance for managers seeking to design resilient, carbon-conscious supply chains under stochastic conditions.

2 September 2023 in American Journal of Scholarly Research and Innovation
Verified
Journal Article
Open access
PDF file
Sudipto RoySudipto Roy
Sudipto Roy
AOAnanna Roy Orthy
Ananna Roy Orthy
PRPartho Protim Roy
Partho Protim Roy
DIDoreen Power Generations And Systems Ltd. Engineer (Deputy Shift In-Charge)
Doreen Power Generations And Systems Ltd. Engineer (Deputy Shift In-Charge)
BTBachelor of Science in Electrical and Electronic Engineering IUBAT— International University of Business Agriculture and Technology
Bachelor of Science in Electrical and Electronic Engineering IUBAT— International University of Business Agriculture and Technology
Efficient resource allocation plays a crucial role in optimizing supply chain performance within the textile and apparel manufacturing industries. The ability to effectively distribute resources—such as raw materials, labor, equipment, and financial investments—directly impacts production efficiency, cost management, and supply chain resilience. This study systematically reviews and synthesizes findings from 75 peer-reviewed research papers, industry reports, and case studies to examine the impact of resource allocation efficiency on supply chain operations. The study explores how strategic resource distribution enhances production workflow, minimizes operational disruptions, and improves financial performance in a highly competitive and globalized industry. Key themes include the integration of digital technologies, lean manufacturing principles, sustainable resource utilization, and supply chain agility. Findings indicate that companies that implement structured resource allocation strategies, supported by automation and AI-driven forecasting tools, achieve greater operational efficiency, cost savings, and supply chain adaptability. Moreover, the study highlights major challenges faced by the industry, such as geopolitical risks, supply chain disruptions, labor shortages, and the bullwhip effect, which often lead to resource mismanagement and inefficiencies. The results also emphasize the growing importance of sustainability in resource allocation, as environmentally responsible practices contribute to cost reduction and long-term industry viability. The study concludes that firms that prioritize efficient resource allocation through advanced planning, real-time decision-making, and supply chain optimization can achieve a significant competitive advantage in the global textile and apparel sector. These insights provide valuable recommendations for supply chain managers, policymakers, and industry stakeholders aiming to improve resource allocation efficiency and enhance overall supply chain sustainability.

26 June 2022
Verified
Journal Article
Open access
PDF file
PMProject Implementation Assistant Manager
Project Implementation Assistant Manager
SSShaikh Shofiullah
Shaikh Shofiullah
Sudipto RoySudipto Roy
Sudipto Roy
This study examines how multi-objective thermo-economic and supply chain optimization capabilities influence hydrogen energy integration in smart factories operating within Industry 4.0 environments. Using a quantitative, cross-sectional, case-study–based design, the research surveyed 120 respondents across energy, production, maintenance, and supply-chain roles in hydrogen-relevant smart factories. Key constructs measured on Likert’s five-point scale indicate that smart factory digitalization (mean 3.65), thermo-economic optimization capability (mean 3.48), and hydrogen supply chain optimization (mean 3.36) are present at moderate-to-high maturity levels, while hydrogen integration performance (mean 3.72) and overall operational/economic performance (mean 3.68) reflect favorable perceived outcomes. Reliability analyses demonstrate strong internal consistency (Cronbach’s α ranging from 0.84 to 0.89). Correlation coefficients show statistically significant positive relationships across all constructs, with the strongest association between hydrogen integration performance and operational/economic performance (r = 0.71, p < .01). Regression modeling confirms that thermo-economic optimization (β = 0.32, p < .01) and supply chain optimization (β = 0.29, p < .01) significantly predict hydrogen integration performance (Adj. R² = .54), while both capabilities jointly explain substantial variance in operational and economic performance (Adj. R² = .61). A supporting multi-objective optimization model generates Pareto-optimal scenarios illustrating trade-offs among cost, efficiency, and reliability: a cost-focused configuration achieves 46% exergy efficiency with a disruption risk index of 0.42 at +12% CAPEX, whereas a reliability-focused configuration achieves 55.8% efficiency with risk reduced to 0.17, requiring +31% CAPEX. These quantitative findings reinforce that hydrogen integration success in smart factories depends not only on digital infrastructure but on the coordinated development of exergy-aware thermo-economic routines and resilient hydrogen supply chains. The study thus provides a validated theoretical, empirical, and optimization-based foundation for guiding hydrogen-enabled smart manufacturing and industrial decarbonization strategies.

11 May 2025
Verified
Conference Paper
Closed access
IHIsmail Hossian
Ismail Hossian
HAHasanur Zaman Anonto
Hasanur Zaman Anonto
Sudipto RoySudipto Roy
Sudipto Roy
NANur Uddin Mahmud Arif
Nur Uddin Mahmud Arif
SASazid Ashraf
Sazid Ashraf
ASAbu Shufian
Abu Shufian
This work investigates a variety of anomaly detection models in the scope of machinery faults including the 1D CNN, LSTM Autoencoders, RNN architectures. We tested and compared the models in terms of test accuracy, convergence time, and predictive accuracy with the fewest misclassifications in profile anomalies. Of all the models evaluated, the LSTM Autoencoder was the most accurate, with a $99.5 \%$ test accuracy. This performed well in learning temporal patterns and achieved low RMSE, which made it well-suited to anomaly detection and sensor data reconstruction. But the parameter tuning was tedious, and it took a long time to train it. The 1D CNN model achieved a test accuracy of $98.1 \%$, indicating rapid initial learning and high accuracy in anomaly detection, while demonstrating slower convergence and minor misclassifications over time. While the RNN did well on sequential data achieving $97.7 \%$ test accuracy, its convergence speed was low, and this also contributed to the high training loss. The experimental results show that the LSTM Autoencoder has the fastest convergence speed and the highest accuracy among various machine learning and deep learning models used for normal and abnormal fault machinery detection, which can be treated as a more reliable anomaly detection model.