Real-Time Analytics and Proactive Control for Smart Manufacturing Systems
The objective of this research is to build a novel analytical framework for real-time monitoring and control in smart manufacturing systems. The rapid development of Cyber-Physical Systems (CPS) has provided unprecedented opportunities for the sensing and control in advanced manufacturing systems. For instance, process sensing data and production operation data are made vastly available to the decision makers in real time through industrial communication networks, which could be potentially turned into actionable insights timely. The goal of this research is to establish an automated and intelligent monitoring and control scheme that translates sensing, operational, and performance related information into smart and timely decisions by connecting, modeling and optimizing the manufacturing system at both the unit and system levels.
Funded by NSF CMMI-1922739 and Intel.
Exploring Discrete Event Dynamics to Model and Control Intelligent Manufacturing Systems
Rapid advances in computer-controlled processes, high-performance computing, and Internet-of-Things (IoT) lay the groundwork for significantly improving manufacturing productivity. Despite the opportunity, modeling methodologies and real-time control methods continue to face two major challenges, namely the lack of predictive models for the dynamic evolution of manufacturing systems, and the lack of real-time optimization and control algorithms to generate effective on-line production control. This research will provide new methods for targeted on-demand simulation, integrated with novel control methodology, to support factory level decision-making based on instantaneous machine status. Specifically, a novel simulation and control architecture for intelligent manufacturing systems will be developed to (1) construct approximate and high-fidelity simulations that can continuously receive information from the real system and generate conditional statements; and (2) define a class of dynamic performance specifications to be controlled and optimized.
Funded by NSF CMMI-1829238.
Real-time Control for Additive Manufacturing
Large scale additive manufacturing has been used to produce vehicles and other large scale products, resulting in reduced design to manufacture times (50%), decreased embodied energy (37%),and lowered carbon dioxide emissions (52%), when compared to traditional manufacturing methods. However, there are areas that can be further optimized to save even more energy and reduce printing cost. The polymer additive manufacturing process involves the layering of molten polymer extrudate to build complex 3-dimensional objects, as such it is inherently dependent on the time-temperature history of each layer to maintain geometric tolerances and mechanical integrity. Our preliminary study shows that regression based layer time control model using thermal images could result in up to 30% built time reduction for simple geometries. This project will invetigate the use of high performance computing to couple the data-driven model with thermal simulation for better predicting layer temperature profiles, improving throughput of large scale additive manufacturing and reducing its energy cost.
Funded by DOE, in collaboration with LM Industries and ORNL.
Online and Dynamic Scheduling for MRI Fleet Management
Since the development of MRI in the 1970s, it has been used in diagnostic medicine with sustained demand increase. As for 2018, the number of MRI scanners in the US is more than 39 per million population, which leads to the second place among the Organization for Economic Co-operation and Development member countries. Despite the increase investment in MR resources, there is a still significant gap from the patients demand. This is largely due to the ineffective scheduling procedure and the nature of uncertainties in MRI jobs, including patients’ arrival time, scanning time, preparation time, etc. Therefore, to deal with these issues and improve quality of care, we are developing a novel scheduling decision framework to be able to translate patients’ clinical data into operational information and utilize them for better MRI workflow coordination and decision making on the fly.
Funded by Mayo Clinic.
Liver Disease Progression Modeling with Serial Multiparametric MRI/MRE
Cirrhosis has become the major liver-related clinical endpoint in non-alcoholic steatohepatitis (NASH). Understanding the temporal changes of Nonalcoholic Fatty Liver Disease (NAFLD) Activity Score (NAS) progression and regression and their relationship with intervention (dietary, clinical) is vital to diagnosis and treatment of NASH. However, progression to cirrhosis is less predictable in NASH than in other chronic liver diseases due to the slow development of NASH in humans, one being the requirement for multiple biopsies during the longitudinal follow-up. The advent of novel and highly sensitive imaging methods — magnetic resonance elastography (MRE) — to assess the dynamics of liver fibrosis in NASH will improve detection, stratification, and follow-up of patients with progressive NASH. Using the serial MRE information, we use novel machine learning algorithms for estimating and predicting the disease progression in the NASH in both preclinical and clinical settings. This non-invasive imaging tool together with the proposed algorithm will promote the clinical development of antifibrotic drugs, by permitting the design of lean proof-of-concept studies and enabling the development of personalized antifibrotic therapy for patients with rapid fibrosis progression or advanced NASH disease.
In collaboration with Mayo Clinic.
Harnessing Interdependency for Resilience: Creating an “Energy Sponge” with Cloud Electric Vehicle Sharing
Operating multiple interdependent infrastructure systems are often criticized as posing threats to the resilience of our modern society. However, with a smart system design, it is possible to create a reciprocal, interdependent cross-system interface to cushion local disturbances and decouple inter-system operations. This “sponge service” enhances system resilience by reserving backup resources for all the associated components, and by converting system-specific exclusive resources into system-indifferent, sharable resources. Motivated by the rapidly increasing adoption of electric vehicles (EV) in recent years, this research aims to materialize the sponge service concept based on an EV-sharing cloud. Several aspects of a potential interface between transportation and power systems will be studied by arranging EV fleets in a way that provides interchangeable mobility and energy in a “smart” way. This research will examine hypotheses, generate knowledge, and create a system design and management methodology for such novel energy sponge service, which can dynamically store and transmit energy across transportation and power systems to stabilize operations in both systems.
Funded by NSF CNS-1638213, in collaboration with UWisc, USF, and SMU.