Self-Powered Sensing and Data Science for Smart Manufacturing
Project Lead: Penn State University
Partners: Texas A&M University, University of Texas – Rio Grande Valley
Member % Cost Share: 50.04%
CESMII % Cost Share: 49.96%
Duration: 18 Months
Problem Statement
This proposal addresses the fundamental problem of establishing the foundations of self-powered sensing and data science in smart manufacturing. Specifically, we are interested in developing algorithms for learning and inferencing from sensor data from IIoT; developing energy-harvesting sensors from vibration sensing; identifying context and decision making in manufacturing with use cases from specific industrial settings and processes and designing, developing and implementing data science algorithms ; studying the trade-off between edge, cloud, and hybrid computing and designing, developing and implementing opportunistic computing algorithms.
Project Goal
Enhance the SM capabilities of CESMII SM platform
Technical Approach
- Formal specification of a sensor wrapper to identify sensors and actuators needed to collect information and respond to actions for machines including legacy machines.
- Development of novel self-powered sensors, cloud and edge-based communication and computation architecture, sensor-based models and algorithms for real-time control and scheduling.
- Development of a test-bed with these advanced real-time sensing using energy-harvesting sensors, analytics and optimization capabilities to demonstrate and enhanced process-wide energy efficiency (by 15%).
Deliverables/Outcomes/SM Marketplace
- Delivered Dashboard for Machine Health Monitoring and Production Scheduling (Powered by Reinforcement Machine Learning)
- Delivered Production Equipment Data Classification Tool Using Distributed Random Forest with Edge-Cloud Partitioning
- Delivered Manufacturing Productivity Dashboard, Enabling Operators to Simulate and Visualize Productivity Status and Machine Health, based on M. Hoffman’s Simantha Simulator
- Delivered Smart Surface Grinding Application powered by explainable AI (XAI) vibration sensing
- Sensor information models
- Real-time control and scheduling algorithms
- Dynamic machine service and task scheduling dashboard
Potential Impact
The main impact of the proposed effort is in reducing the costs of adopting SM into small and medium enterprises. Over 70% whose manufacturing equipment in the machine shops of these enterprises lack interfaces to connect to advanced web services, although many of these are beginning to collect data about energy usage at fine granularities. This project would establish a framework of self-powered IIoT sensing and analytics for SM. This will substantially enhance energy productivity to extract valuable data, as well as reduce maintenance costs and equipment downtime. Also, integration of advanced sensing of machines and manufacturing processes, and leveraging of the generated big data, essential for SM, will be realized.
Benefits
The technology developed in this proposal will improve the global competitiveness of American companies. It is essential to unlock huge opportunities for energy reduction especially to accelerate the translation of growing SM tools and platforms. Also, we will build the foundation for machine resource sharing and composition via marketplace, and process optimization. With rigorous theoretical approaches, formal (verifiable) specification and scalable optimization methods, product delivery will be much enhanced and energy efficiency will be improved. The project will also provide CESMII with key testbed infrastructure components that provide test and evaluation capabilities for additional extensions of the SM Platform reducing deployment costs.
