Predictive Informatics Research Lab

Welcome

Advanced and Intelligent Manufacturing Systems (AIMS):

My research lab is focused on developing physics-informed AI/ML models for analyzing and assessing the performance and health condition of engineering assets within intelligent manufacturing systems and complex engineering environments, providing predictive and adaptive decision-support tools for optimal operations, maintenance, and control. Key focus areas include:

  • AI-enabled digital manufacturing and cyber manufacturing systems
  • Diagnostics and prognostics for advanced manufacturing processes and systems
  • Data-driven optimal design for new materials fabrication processes, including 2D materials synthesis, nanomaterials synthesis
  • Sensor data analytics and multi-modal data fusion for manufacturing process monitoring and control
  • Roll-to-roll process for flexible electronics printing
  • High-precision manufacturing processes, such as cold spray, wire-arc additive mfg. and gravure printing
  • Manufacturing system design and optimization in sectors such as automotive assembly line, and semiconductor fabrication

News

  • Prof. Jin is awarded a $2M NSF grant to develop Integrative Manufacturing and Production Engineering Curricula That Leverage Data Science in collaboration with colleagues Professors Kamarthi (PI), Issacs, Moghaddam and Jona.
  • Congratulations!  Ph.D student Anqi He received the 2019 John and Katharine Cipolla MS/PhD Award.
  • Congratulations to Ph.D student Mengkai Xu’s paper on “Resiliency of Mutualistic Supplier-Manufacturer Networks” published in Scientific Reports [link]
  • Dr. Jin was invited to present her roll-to-roll manufacturing research at the MIE department seminar in UMasss Amherest. [link]
  • Dr. Xiaoning Jin (PI) is leading a $550K NSF grant on Precision Alignment of Roll-to-Roll Printing of Flexible Paper Electronics, in collaboration with Hongli Zhu (Co-PI) and UMass Amherst.
  • Congratulations to PhD student Anqi He  for winning the ASME Manufacturing Science and Engineering Conference Best Paper Award (1st Place), Erie, PA 2019. “Failure Detection and Remaining Life Estimation for Ion Mill Etching Process Through Deep-Learning Based Multimodal Data Fusion”
  • Dr. Jin’s paper on “Virtual sensing and virtual metrology for spatial error monitoring of roll-to-roll manufacturing systems” is published in CIRP Annals [link]
  • Congratulations to PhD student Xiaomeng Peng for receiving the 2018 John and Katharine Cipolla PhD Award!
  • Dr. Jin is co-organizing a technical symposium on Advances in Data Analytics and Engineering Modeling for Intelligent & Resilient Manufacturing Systems at 2019 ASME International Manufacturing Science and Engineering Conference (MSEC), June 10-14, 2019, Pennsylvania State University. Papers are welcome to submit at https://www.asmeconferences.org/MSEC2019/login.cfm
  • The Multimodal Data Fusion Workshop (MMDF 2018) Report is available now! [See the news][Archives]
  • Xiaoning “Sarah” Jin is the guest editor of the Special Issue of IEEE Sensors Letters. [See details]
  • Anqi He received CASE2018 Student Travel Grant to present his paper in Munich, Germany 2018
  • Mengkai Xu’s new article on “A failure-dependency modeling and state discretization approach for
    condition-based maintenance optimization of multi-component systems” has been published in Journal of Manufacturing Systems. [pdf]
  • Anqi He’s first paper on “NARNET-based Prognostics Modeling for Deteriorating Systems under Dynamic Operating Conditions”  has been accepted by IEEE Conference on Automation Science and Engineering. Congrats!
  • Dr. Jin is awarded a $750k project from Digital Manufacturing Design Innovation Institute (DMDII) on Predictive Factory research.

Research Highlights

NSF_R2R reports
pytsc_logo

PyTSC: Traffic Signal Control Environment for Multi-Agent Reinforcement Learning

Rohit Bokade, Xiaoning Jin

We are pleased to announce that the MARL test platform for the Traffic Signal Control (TSC) environment is now available for research purposes, including testing and benchmarking.

fleet learning1
fleet 3

Fleet Learning: Active learning-assisted semi-supervised learning for fault detection and
diagnostics with imbalanced dataset

Xiaomeng Peng, Xiaoning Jin, Shiming Duan, and Chaitanya Sankavaram  

Data-driven Fault Detection and Diagnostics (FDD) methods often assume that sufficient labeled samples are class-balanced and faulty classes in testing are precedent or seen previously during model training. When monitoring a large fleet of assets at scale, these assumptions may be violated:
(I) only a limited number of samples can be manually labeled due to constraints of time and/or cost; (II) most of the samples collected in the engineering systems are under normal conditions, leading to a highly imbalanced class distribution and a biased prediction model. This work presents a robust and cost-effective FDD framework that integrates active learning and semi-supervised learning methods to detect both known and unknown failure modes iteratively. This framework allows to strategically select the samples to be annotated from a fully unlabeled dataset, while labeling cost is minimal.  We tested the framework and algorithms in three synthetic datasets and one real-world dataset of vehicle air intake systems, and demonstrated the superior performance compared to the state-of-the-art methods for fleet-level FDD.

2D materials
cvd 2

Accelerating Optimal Synthesis of 2D Materials:  A Constrained Bayesian Optimization Guided Brachistochrone Approach  

Yujia Wang, Guoyan Li, Xiaoning Jin, Swastik Kar

We present a machine learning (ML) guided approach for the accelerated optimization of chemical vapor deposition (CVD) synthesis of 2D materials towards the highest quality, starting from low-quality or unsuccessful synthesis conditions. Using 26 sets of synthesis conditions as our initial training dataset, we could systematically progress towards optoelectronic-grade monolayer MoS2 flakes with A-exciton linewidth (σA) as narrow as 38 meV after only an additional 35 trials (reflecting only 15% of the full factorial design dataset for training purposes). This translates to an 85% reduction in wasteful “trial-and-error” experiments. This remarkable efficiency, without any domain knowledge intervention, was accomplished by formulating a constrained sequencing optimization problem solved via a combination of constraint learning and Bayesian Optimization.  We provide a clear visualization of “sweet spots” for a CVD reactor to an experimentalist. Our method is scalable to a higher number of synthesis parameters and target metrics and is transferrable to other materials and types of reactors.

mfg resilience
SC

Resiliency of Mutualistic Supplier-Manufacturer Networks

Mengkai Xu, Srinivasan Radhakrishnan, Sagar Kamarthi, and Xiaoning Jin*

Current Supplier-Manufacturer (SM) networks are highly complex and susceptible to local and global disruptions, due to connectivity and interdependency among suppliers and manufacturers. Resiliency of supply chains is critical for organizations to remain operational in the face of disruptive events. In this work we investigate resiliency of SM networks using the quantitative methods employed to study mutualistic ecological systems. We create a bipartite representation and generate a multidimensional nonlinear model that captures the dynamics of a SM network and predicts the point of collapse. We extensively validate the model using real-world global automotive SM networks. The current work offers a means for designing resilient supply chains that can remain robust to local and global perturbations. An interactive visualization tool of the SM network and its resilience analysis has been developed by Capstone Project team. See https://rainbowfalcons.herokuapp.com/.

Picture2

NARNET-Based Prognostic Modeling

Anqi He, Xiaoning Jin

This paper presents a new prognostics modeling method based on a nonlinear autoregressive neural network (NARNET) for computing the remaining useful life (RUL) of a deteriorating system under dynamic operating conditions. We particularly investigate how the degradation process is affected by the unit-specific operating conditions. The operating conditions are forecasted by a NARNET model based on the unit’s historical operating conditions. We show that the prognostics model integrating the operating condition forecast provides more accurate and efficient RUL prediction.

Picture1

Stochastically-dependent Multi-component Degradation and Failure

Mengkai Xu, Xiaoning Jin, Sagar Kamarthi, Md. Noor-E-Alam

Unexpected component failures in a mechanical system always cause loss of performance and functionality of the entire system. Condition based maintenance decisions for a multi-component mechanical system are challenging because the interdependence of individual components’ degradation is not fully understood and lack of physical models.  An extended proportional hazard model (PHM) is developed to characterize the failure dependence and estimate the influence of degradation state of one component on the hazard rate of another. An optimization model is developed to determine the optimal hazard-based threshold for a two-component repairable system.

Graduate Students

No funded graduate student positions are available at the moment, although applications are always welcome.

Post-docs

One post-doctoral position is currently available, please contact Prof. Jin for details.

Visiting  Students/Scholars

Interested students/scholars from other institutions are welcome to contact Prof. Jin by email for the position of visiting students.

NSF_logomanufacturingusa_logo

mxd