Force Estimation and Event Localization (FEEL) of Impacts Using Structural Vibrations

Abstract

Force applied and location it occurred are of growing interest for smart buildings, particularly in the area of human activity. The ability to localize human activity through the vibrations caused by the activity on the structure has the potential to be a low-cost, privacy-respecting solution to areas of interest such as building occupancy estimations, customer flow through retail stores, and human fall detection. This paper introduces the Force Estimation and Event Localization (FEEL) Algorithm, which estimates the force of an impact that caused a structural vibration and additionally locates the location the impact occurred. The main feature of FEEL is that it does not require time synchronization like other time-of-flight techniques. FEEL was validated using the human-induced vibration benchmark data set totaling 3,500 impact events of seven different types at five different locations, and an additional 75 force hammer impacts. FEEL displayed 96.4% location accuracy and a force magnitude estimate accuracy of −2.0%±1.3% in a 99% confidence interval in the experiments.