More Like this

1. An Additive Manufacturing Testbed to Evaluate Machine Learning-Based Autonomous Manufacturing

   https://doi.org/10.1115/1.4064321  

   Zhang, Zhi; George, Antony; Ferdous_Alam, Md; Eubel, Chris; Prasad_Vallabh, Chaitanya Krishna; Shtein, Max; Barton, Kira; Hoelzle, David J (March 2024, Journal of Manufacturing Science and Engineering)

   Abstract This paper details the design and operation of a testbed to evaluate the concept of autonomous manufacturing to achieve a desired manufactured part performance specification. This testbed, the autonomous manufacturing system for phononic crystals (AMSPnC), is composed of additive manufacturing, material transport, ultrasonic testing, and cognition subsystems. Critically, the AMSPnC exhibits common manufacturing deficiencies such as process operating window limits, process uncertainty, and probabilistic failure. A case study illustrates the AMSPnC function using a standard supervised learning model trained by printing and testing an array of 48 unique designs that span the allowable design space. Using this model, three separate performance specifications are defined and an optimization algorithm is applied to autonomously select three corresponding design sets to achieve the specified performance. Validation manufacturing and testing confirms that two of the three optimal designs, as defined by an objective function, achieve the desired performance, with the third being outside the design window in which a distinct bandpass is achieved in phononic crystals (PnCs). Furthermore, across all samples, there is a marked difference between the observed bandpass characteristics and predictions from finite elements method computation, highlighting the importance of autonomous manufacturing for complex manufacturing objectives. 

   more » « less
2. Stochastically-Trained Physics-Informed Neural Networks: Application to Thermal Analysis in metal Laser Powder Bed Fusion

   Pierce, Justin; Williams, Glen; Simpson, Timothy; Meisel, Nicholas; McComb, Christopher (September 2021, International Design Engineering Technical Conferences)

   null (Ed.)

   Modern digital manufacturing processes, such as additive manufacturing, are cyber-physical in nature and utilize complex, process-specific simulations for both design and manufacturing. Although computational simulations can be used to optimize these complex processes, they can take hours or days--an unreasonable cost for engineering teams leveraging iterative design processes. Hence, more rapid computational methods are necessary in areas where computation time presents a limiting factor. When existing data from historical examples is plentiful and reliable, supervised machine learning can be used to create surrogate models that can be evaluated orders of magnitude more rapidly than comparable finite element approaches. However, for applications that necessitate computationally- intensive simulations, even generating the training data necessary to train a supervised machine learning model can pose a significant barrier. Unsupervised methods, such as physics- informed neural networks, offer a shortcut in cases where training data is scarce or prohibitive. These novel neural networks are trained without the use of potentially expensive labels. Instead, physical principles are encoded directly into the loss function. This method substantially reduces the time required to develop a training dataset, while still achieving the evaluation speed that is typical of supervised machine learning surrogate models. We propose a new method for stochastically training and testing a convolutional physics-informed neural network using the transient 3D heat equation- to model temperature throughout a solid object over time. We demonstrate this approach by applying it to a transient thermal analysis model of the powder bed fusion manufacturing process. 

   more » « less
3. Multi-Agent Artificial Intelligence to Self-Organize Machine Networks for Resilient Manufacturing

   https://doi.org/10.1115/MSEC2025-155761  

   Lee, Hankang; Yang, Hui (June 2025, American Society of Mechanical Engineers)

   Modern manufacturing is increasingly challenged by larger product varieties, shorter product life cycles, and unexpected production disruptions. Examples of such disruptions include market uncertainty, machine failures, and delivery backlogs. These disturbances are intricately interrelated, exacerbating system complexity and necessitating the adaptive organization (or re-configuration) of machine networks within factory layouts. However, traditional factory layouts are often stationary and lack the flexibility to rearrange or adjust machine networks in response to volatile markets and unexpected disruptions. Also, layout planning typically emphasizes offline design and configuration of machine networks and resources within a facility to optimize process flow and production performance, but tends to overlook the self-organizing arrangement of machines in a dynamic environment. Therefore, to address this gap, this paper presents a novel Self-organizing Machine Network (SOMN) model that optimizes the spatial layout of machine positions and queue configurations, thereby enhancing the manufacturing system’s resilience to unexpected disruptions. First, as opposed to traditional fixed machine positions, we design intelligent machine agents that communicate and autonomously reorganize in real-time to optimize key performance indicators (KPIs). Second, we develop the machine network model in a Digital Twin (DT) environment, facilitating cyber-physical interactions and capturing variations of state-action space in machine agents. Third, multi-agent reinforcement learning (MARL) algorithms empower these networked machine agents to adapt layouts and minimize the impact of disruptions on production performance. We evaluate and validate the proposed SOMN model through computer experiments, benchmarking it against random search and simulated annealing approaches. Experimental results show that the SOMN model significantly improves material handling efficiency, reduces computational overhead, and maintains productivity in different scenarios of manufacturing disruptions. This research holds strong potential for enabling distributed intelligence within self-organizing machine networks for resilient manufacturing. 

   more » « less
4. Superhuman Fairness

   Memarrast, Omid; Vu, Linh; Ziebart, Brian D. (July 2023, International Conference on Machine Learning)

   The fairness of machine learning-based decisions has become an increasingly important focus in the design of supervised machine learning methods. Most fairness approaches optimize a specified trade-off between performance measure(s) (e.g., accuracy, log loss, or AUC) and fairness measure(s) (e.g., demographic parity, equalized odds). This begs the question: are the right performance-fairness trade-offs being specified? We instead recast fair machine learning as an imitation learning task by introducing superhuman fairness, which seeks to simultaneously outperform human decisions on multiple predictive performance and fairness measures. We demonstrate the benefits of this approach given suboptimal decisions. 

   more » « less
5. Grid Search Hyperparameter Tuning in Additive Manufacturing Processes

   Ogunsanya, M.; Isichei, J.; Desai, S. (July 2023, Manufacturing letters)

   In Machine learning (ML) and deep learning (DL), hyperparameter tuning is the process of selecting the combination of optimal hyperparameters that give the best performance. Thus, the behavior of some machine learning (ML) and deep learning (DL) algorithms largely depend on their hyperparameters. While there has been a rapid growth in the application of machine learning (ML) and deep learning (DL) algorithms to Additive manufacturing (AM) techniques, little to no attention has been paid to carefully selecting and optimizing the hyperparameters of these algorithms in order to investigate their influence and achieve the best possible model performance. In this work, we demonstrate the effect of a grid search hyperparameter tuning technique on a Multilayer perceptron (MLP) model using datasets obtained from a Fused Filament Fabrication (FFF) AM process. The FFF dataset was extracted from the MakerBot MethodX 3D printer using internet of things (IoT) sensors. Three (3) hyperparameters were considered – the number of neurons in the hidden layer, learning rate, and the number of epochs. In addition, two different train-to-test ratios were considered to investigate their effects on the AM process data. The dataset consisted of five (5) dominant input parameters which include layer thickness, build orientation, extrusion temperature, building temperature, and print speed and three (3) output parameters: dimension accuracy, porosity, and tensile strength. RMSE, and the computational time, CT, were both selected as the hyperparameter performance metrics. The experimental results reveal the optimal configuration of hyperparameters that contributed to the best performance of the MLP model. 

   more » « less