Visible to the public EAGER: Collaborative: Predictive Maintenance of HVAC Systems using Audio SensingConflict Detection Enabled

Project Details
Lead PI:Shahriar Nirjon
Performance Period:03/01/16 - 02/28/17
Institution(s):University of North Carolina at Chapel Hill
Sponsor(s):National Science Foundation
Award Number:1619967
302 Reads. Placed 555 out of 803 NSF CPS Projects based on total reads on all related artifacts.
Abstract: Acoustic sensing-based preventive maintenance approach focuses on mapping auditory information, captured from mechanical systems in buildings, to their health status and probability of impending failures. An important application of this methodology is reducing energy waste in commercial heating, ventilating, and air-conditioning (HVAC) systems, which accounts for over 42% of the total U.S. commercial building energy usage. The outcome of this project is a robust acoustic sensing technology that has a high accuracy in predicting actual failures of HVAC systems. This research will be integrated with new user interfaces that will allow building managers to virtually navigate the equipment and appliances in large buildings (or collections of buildings), and to quickly identify potential failures. This EArly-Concept Grants for Exploratory Research (EAGER) project addresses the following technology gaps as it translates from research discovery toward commercial applications: (a) ensuring privacy, and (b) minimizing false positives in predicting equipment failure. This project develops acoustic signal acquisition and processing techniques that preserve the privacy of everyone and everything that is susceptible to privacy violations due to continuous acoustic monitoring. The proposed collaborative research enables buildings to be retrofitted with a low-cost, acoustic sensing solution to monitor its HVAC systems to predict their impending failures. A major goal of this project is to reduce false positives when making these predictions that are primarily caused by inadequate modeling of sounds from a faulty component, inadequate modeling of different types of faults, and errors in sound source recognition. Furthermore, this project creates a foundation for the next generation of intelligent systems that autonomously monitor equipment and predict failure. The project engages University of Florida and University of North Carolina at Chapel Hill to augment research capability in conducting visualization-based dynamic assessment of HVAC systems, and building low-cost, embedded device-based centralized HVAC monitoring systems. With a cloud-connected network of embedded audio monitoring devices deployed in the University of Florida campus buildings for running acoustic processing and classification tasks, this novel and transformative technology is aimed at identifying and solving challenges in large-scale, commercial-grade deployment of such systems in real world scenarios. This project will engage an industrial partner to develop privacy-preserving algorithms, build test environments, and guide commercialization aspects of this technology.