Real-Time Optimal Control of Seat-Belt Systems

This work presents a real-time implementable, computationally light algorithm for semi-active seat-belt systems. It consist of a linear feedback loop on the belt force that is cascaded by an algebraic reference governor. The governor is based on the explicit solution to a widely used optimization problem regarding optimal control of seat-belts. The algorithm is applied to a prototype semi-active hydraulic seat belt actuator and is demonstrated on an experimental setup simulating frontal collisions. In comparison to an uncontrolled experiment, the controlled one managed to reduce the injury criterion with 15%, without increasing the occupant travel. When twice the travel was allowed, the criterion was reduced by 51%, showing the effect of variable settings to occupant/vehicle dimensions. However, the injury criteria were still 2.0 - 2.5 times the optimal injury criterion, as calculated with perfect future knowledge of the crash event. (C) 2017, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.