• • AI-Based Modeling Development: Introduced a groundbreaking AI-based modeling approach designed to accurately estimate high-fidelity heat transfer calculations and foresee thermal distortion in metal additive manufacturing, specifically focusing on the multi-laser powder bed fusion (ML-PBF) process.
  • In-depth Analysis of ML-PBF Process: Conducted extensive research to understand the effects of start position and printing orientation on deformation and stress distribution in parts produced using the ML-PBF process. Successfully identified strategies that resulted in a 53% reduction in deformation by comparing the best and worst printing cases.
  • Neural Network Framework: Spearheaded the creation of a low-fidelity modeling framework, utilizing a feedforward neural network to swiftly predict thermal displacements with remarkable accuracy. Demonstrated the model's efficacy by achieving a strong correlation between high-fidelity and neural network-predicted outputs, paving the way for cost-effective and efficient preliminary evaluations in design processes.

  
  

Technologies used

• • Metamodeling Approach for Composite Manufacturing: Pioneered a computationally efficient and physically precise metamodeling technique to scrutinize the spring-in angle deformation during composite manufacturing processes, addressing the inherent uncertainties arising from the heterogeneous thermo-mechanical attributes of composite materials.
  • Gaussian Processes (GP) Optimization: Delved deep into machine learning techniques, surrogate models, and Gaussian processes (GP) to propagate uncertainties effectively. Recognized the criticality of model parameter tuning and hyperparameter optimization for enhancing predictive modeling performance. Developed a systematic approach, moving beyond traditional empirical methods, to optimize these elements, significantly reducing computational cost and enhancing model accuracy.
  • Nature-Inspired Hyperparameter Selection: Innovated a nature-inspired methodology, leveraging an improved firefly algorithm (iFA) that incorporated environmental factors, to guide the GP in the optimal selection and tuning of hyperparameters for uncertainty analysis in composite manufacturing. The newly proposed method outperformed contemporary deterministic/metaheuristic algorithms, as corroborated by multiple nonparametric comparison tests.
    

Technologies used

• • Open Innovation Platform Development: Spearheaded the design and development of an intelligent, robust, modular, and responsive system.
  • Security and User Role Design: Enhanced system security by incorporating distinct roles with specialized activities and privileges, ensuring a streamlined user experience.
  • Global Collaboration: Established a platform that serves as a nexus for around 10,000 innovators and industry partners globally, fostering engagement, innovation, and mutual collaboration.
    

Technologies used

• • AI-based Modeling of EHD Inkjet Printing Systems: Pioneered the development of an Artificial Intelligence (AI) system tailored for the intricate modeling of non-linear Electrohydrodynamic (EHD) inkjet printing processes. The devised methodologies elucidated the intricate cause-effect dynamics inherent to the system.
  • Feed-forward Neural Network Training Enhancement: Innovated a unique learning mechanism designed to train feed-forward neural networks (FFNN) with an emphasis on heightened precision for predicting complex real-world systems. When juxtaposed with prevailing state-of-the-art algorithms, the proposed method showcased superior performance, marking it as a vanguard technique for training neural network-based models.
  • Quantifying Uniformity in EHD Inkjet Printing: Introduced a groundbreaking offline procedure dedicated to quantifying the uniformity grade of the deposited features within the EHD inkjet printing framework. Leveraging AI-based techniques, the system's stability was significantly enhanced, leading to a more uniform deposition of features.
  • Refinement of the Firefly Algorithm: Proposed an enhanced version of the nature-inspired firefly algorithm (FA) to counteract its inherent slow convergence rate. Eschewing the traditional constant initial brightness coefficient, a novel rule was introduced to update firefly brightness based on generational selection probability, striking an optimal balance between exploration and exploitation. The efficacy of this improved FA was substantiated through its application to mathematical benchmark functions and real-world engineering challenges.
    
    

Technologies used

• • EHD Inkjet System Development: Led the design and establishment of a foundational Electrohydrodynamic (EHD) inkjet experimental setup by seamlessly integrating diverse hardware components.
    
  • GUI Design for Hardware Synchronization: Crafted a sophisticated Graphical User Interface (GUI) to facilitate seamless communication and synchronization among critical hardware components, such as the positioning system, ink supply system, and electrical system.
    
  • Taylor Cone Realization: Successfully achieved a stable Taylor cone from the developed system, visualized using a high-speed camera.
    
  • Advanced Material Patterning: Leveraged the indigenously developed system to produce miniaturized patterns of functional materials on a variety of substrates.

Technologies used

• • Development of GUI for Braille Embosser: Played a pivotal role in crafting a sophisticated Graphical User Interface for the high-speed Interpoint Braille Embosser. This interface seamlessly facilitated the conversion of traditional text documents (such as Word, PDF) into their corresponding Braille language format. Additionally, integrated quality testing modules within the GUI ensured the precision of printed features.
  • Wireless Communication Enhancement: Delved into the intricacies of wireless communication systems, implementing a myriad of error control coding schemes. This endeavor culminated in the proposition of a novel hybrid ARQ scheme. Impressively, this newly introduced system demonstrated superior performance compared to the conventional ARQ using CRC, particularly in the arena of energy consumption during single-hop data transmissions in the CDMA Wireless Sensor Network.
    

Technologies used