摘要：AbstractThis 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.
