Browsing by Author "Singh, Karanpreet"

Now showing 1 - 4 of 4
  • Thumbnail Image
    Accelerating Structural Design and Optimization using Machine Learning
    Singh, Karanpreet (Virginia Tech, 2020-01-13)
    Machine learning techniques promise to greatly accelerate structural design and optimization. In this thesis, deep learning and active learning techniques are applied to different non-convex structural optimization problems. Finite Element Analysis (FEA) based standard optimization methods for aircraft panels with bio-inspired curvilinear stiffeners are computationally expensive. The main reason for employing many of these standard optimization methods is the ease of their integration with FEA. However, each optimization requires multiple computationally expensive FEA evaluations, making their use impractical at times. To accelerate optimization, the use of Deep Neural Networks (DNNs) is proposed to approximate the FEA buckling response. The results show that DNNs obtained an accuracy of 95% for evaluating the buckling load. The DNN accelerated the optimization by a factor of nearly 200. The presented work demonstrates the potential of DNN-based machine learning algorithms for accelerating the optimization of bio-inspired curvilinearly stiffened panels. But, the approach could have disadvantages for being only specific to similar structural design problems, and requiring large datasets for DNNs training. An adaptive machine learning technique called active learning is used in this thesis to accelerate the evolutionary optimization of complex structures. The active learner helps the Genetic Algorithms (GA) by predicting if the possible design is going to satisfy the required constraints or not. The approach does not need a trained surrogate model prior to the optimization. The active learner adaptively improve its own accuracy during the optimization for saving the required number of FEA evaluations. The results show that the approach has the potential to reduce the total required FEA evaluations by more than 50%. Lastly, the machine learning is used to make recommendations for modeling choices while analyzing a structure using FEA. The decisions about the selection of appropriate modeling techniques are usually based on an analyst's judgement based upon their knowledge and intuition from past experience. The machine learning-based approach provides recommendations within seconds, thus, saving significant computational resources for making accurate design choices.
  • Thumbnail Image
    ALMO: Active Learning-Based Multi-Objective Optimization for Accelerating Constrained Evolutionary Algorithms
    Singh, Karanpreet; Kapania, Rakesh K. (MDPI, 2024-10-31)
    In multi-objective optimization, standard evolutionary algorithms, such as NSGA-II, are computationally expensive, particularly when handling complex constraints. Constraint evaluations, often the bottleneck, require substantial resources. Pre-trained surrogate models have been used to improve computational efficiency, but they often rely heavily on the model’s accuracy and require large datasets. In this study, we use active learning to accelerate multi-objective optimization. Active learning is a machine learning approach that selects the most informative data points to reduce the computational cost of labeling data. It is employed in this study to reduce the number of constraint evaluations during optimization by dynamically querying new data points only when the model is uncertain. Incorporating machine learning into this framework allows the optimization process to focus on critical areas of the search space adaptively, leveraging predictive models to guide the algorithm. This reduces computational overhead and marks a significant advancement in using machine learning to enhance the efficiency and scalability of multi-objective optimization tasks. This method is applied to six challenging benchmark problems and demonstrates more than a 50% reduction in constraint evaluations, with varying savings across different problems. This adaptive approach significantly enhances the computational efficiency of multi-objective optimization without requiring pre-trained models.
  • Thumbnail Image
    Force-exerting perpendicular lateral protrusions in fibroblastic cell contraction
    Padhi, Abinash; Singh, Karanpreet; Franco-Barraza, Janusz; Marston, Daniel J.; Cukierman, Edna; Hahn, Klaus M.; Kapania, Rakesh K.; Nain, Amrinder S. (2020-07-21)
    Aligned extracellular matrix fibers enable fibroblasts to undergo myofibroblastic activation and achieve elongated shapes. Activated fibroblasts are able to contract, perpetuating the alignment of these fibers. This poorly understood feedback process is critical in chronic fibrosis conditions, including cancer. Here, using fiber networks that serve as force sensors, we identify "3D perpendicular lateral protrusions" (3D-PLPs) that evolve from lateral cell extensions named twines. Twines originate from stratification of cyclic-actin waves traversing the cell and swing freely in 3D to engage neighboring fibers. Once engaged, a lamellum forms and extends multiple secondary twines, which fill in to form a sheet-like PLP, in a force-entailing process that transitions focal adhesions to activated (i.e., pathological) 3D-adhesions. The specific morphology of PLPs enables cells to increase contractility and force on parallel fibers. Controlling geometry of extracellular networks confirms that anisotropic fibrous environments support 3D-PLP formation and function, suggesting an explanation for cancer-associated desmoplastic expansion. Padhi et al. employ nanofibers with controlled structure and alignment as an extra-cellular matrix model, on which they study the exertion of forces from adherent fibroblasts. Identifying force exerting 3D perpendicular lateral protrusions, authors describe a mechanism which leads to the contraction of parallel, neighbouring fibers, and the forces needed to move and align the neighbouring fibers. These findings have relevance in understanding cancer-associated desmoplastic expansion.
  • Thumbnail Image
    Lightweight Chassis Design of Hybrid Trucks Considering Multiple Road Conditions and Constraints
    De, Shuvodeep; Singh, Karanpreet; Seo, Junhyeon; Kapania, Rakesh K.; Ostergaard, Erik; Angelini, Nicholas; Aguero, Raymond (MDPI, 2020-12-28)
    The paper describes a fully automated process to generate a shell-based finite element model of a large hybrid truck chassis to perform mass optimization considering multiple load cases and multiple constraints. A truck chassis consists of different parts that could be optimized using shape and size optimization. The cross members are represented by beams, and other components of the truck (batteries, engine, fuel tanks, etc.) are represented by appropriate point masses and are attached to the rail using multiple point constraints to create a mathematical model. Medium-fidelity finite element models are developed for front and rear suspensions and they are attached to the chassis using multiple point constraints, hence creating the finite element model of the complete truck. In the optimization problem, a set of five load conditions, each of which corresponds to a road event, is considered, and constraints are imposed on maximum allowable von Mises stress and the first vertical bending frequency. The structure is optimized by implementing the particle swarm optimization algorithm using parallel processing. A mass reduction of about 13.25% with respect to the baseline model is achieved.