Model Reduction and Robust Control
Model Reduction and Robust Control
People
- Jemonde Taylor
- Sam Chang
Description
Engineers commonly use reduced models for controller design. Without guarantees about the interaction of the neglected dynamics with the feedback control law, testing and intuition must be relied on for confidence in the performance of the overall system. This research seeks to develop a framework wherein the validity of reduced models and restrictions resulting from reduction can be rigorously
treated. Focusing on Lagrangian systems, we study the robustness of feedback control laws designed for reduced models when applied to the actual systems (from which the reduced models were derived). We hope to complement existing work on robust nonlinear control and Lagrangian systems.
An application for this research arises in the design of heavy trucks for safe performance. Advanced Vehicle Control Systems (AVCS) - in the form of either driver assistance systems or full automation - hold considerable promise for increasing heavy vehicle safety. Unfortunately, the complexity of truck dynamics makes it difficult to determine exactly what constitutes a safe truck and, even more so, to realize the benefits of implementing advanced control systems on such vehicles. The initial goal of the project is to establish a framework of performance measures that successfully cover the range of often competing safety demands (such as yaw stability, rollover avoidance, and stopping distance) faced
by heavy trucks. These performance measures will provide a basis for evaluating the safety benefits of active systems designed using reduced models. Being able to quantify the uncertainties introduced by model reduction will lead to greater confidence in the controller designs.
Results
Our first task was to build a library of references containing information on heavy truck mechanical properties. These resources will provide realistic parameters for simulation and model evaluation.
We have compiled a list of common heavy truck safety performance measures along with test maneuvers used by industry and researchers to evaluate these measures.
For future model reduction work, we have developed a baseline model of a tractor semi-trailer combination using ADAMS dynamic analysis software (below right, truck beginning to roll over). The model accurately represents the PATH experimental vehicle, a Freightliner FLD120 tractor with 45-foot semi-trailer (below left, photo taken at test site in Crow's Landing, CA).
Reports and Publications
Sponsors
Work in Progress
We are currently validating the ADAMS heavy truck model by comparing simulation results to actual test data from the PATH vehicle.
» 140 reads