On real time adaptive and dynamically constrained control allocation for stability control of heavy vehicles
Typ
Examensarbete för masterexamen
Master Thesis
Master Thesis
Program
Publicerad
2008
Författare
Sundström, Peter
Tagesson, Kristoffer
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
This report evaluates the option of using real-time optimization for the coordination of motion actuators in motion stability control of heavy vehicles. The general assumption is that the system is over actuated which means that there are more actuators than controlled motions. This is often the case for today’s heavy vehicles. With over actuation comes the possibility to perform the same type of motion control in several ways. When a specific motion is asked for, some sort of mapping is needed to actuator input. This mapping is known as control allocation. Roll-over accidents are far more common for heavy vehicles, than for passenger cars and therefore roll-over prevention is crucial to lower the rate of accidents. Roll-over occurs when the lateral forces are too high. A controller is therefore proposed to reduce the lateral forces under critical situations. Yaw instability, in the form of over- and understeer, is also critical for vehicles. Therefore a combined controller which accounts for roll as well as yaw stability simultaneously is proposed. The control allocation strategy used is a constrained weighted least square minimization criterion. Two algorithms for solving this criterion were studied, an Active Set method and a Primal-dual Interior Point method. In literature, the algorithms have been proposed suitable for fast real-time applications such as in automotive and aerospace control. The Active Set method has shown better results than the Primal-dual IP method with respect to convergence rate and computational load when the number of actuators is low. When the number of actuators is above 25 the same holds for the Primal-dual IP method. Furthermore the convergence pattern for Primal-dual IP method is smoother than that of the Active Set method, this makes it less sensitive if using a solution generated before convergence. The control system used shows performance similar to current systems in terms of vehicle stability. The proposed system can easily be reconfigured for varying numbers and types of actuators. Real-time performance benchmarking indicates that the system should be realizable in production vehicles in terms of execution time versus required sample rate. System robustness has been shown to be good in general.
Beskrivning
Ämne/nyckelord
Elektroteknik och elektronik , Electrical Engineering, Electronic Engineering, Information Engineering