Active and semi-active control of suspension systems for commercial vehicles based on preview

The performance of a vehicle axle/wheel suspension system is primarily determined by the comfort of the occupants, the required working space, the handling properties and, for commercial vehicles in particular, the dynamic tire forces and the load on the chassis components and on the cargo. General design considerations, leading to our research objectives, are given in Sec.1. The investigation described in Sec.2 focuses first on the development of an active control at the rear wheels of the tractor of a tractor-semitrailer combination. An active suspension system is not very likely to be applied in practice however, so the results of this research are only considered as a basis for comparison with more feasible solutions. A more practical approach, using semi-active control,, is considered in Sec.3. There we make use of fast switching shock absorbers to change the damping of the suspension at the rear wheels of a tractor semitrailer. All control strategies considered are based on knowledge of the road surface between the front and the rear wheels. This so-called preview information is obtained indirectly by reconstruction from on-line measurements of accelerations and suspension deflections. Though studies of this kind usually only consider stochastic road inputs, we are more interested in incidental inputs (e.g. traffic humps and railway crossings), which turn out to be more severe with respect to the performance measures and the peak values of the controlled quantities. The improvements obtained with active suspension system simulations are promising and justify further investigations in the semi-active field. Here the control problem gets more complicated due to the non-linear characteristics of the switching shock absorbers of a discrete on/off type, to the non-linear way the actuator inputs enter the system equations, as well as to the non-linear behaviour of the rest of the system. Also,the control objectives cannot suitably be formulated in a simple mathematical form and there are hard constraints on certain state variables such as the upper/lower bounds of the suspension deflections. For this purpose we developed numerical control methods - similar to Model Predictive Control - which determine the best control input sequence over a certain horizon given the current state of the system and given the external disturbance (the road input) over the preview interval. This numerical optimisation problem must be solved on-line and requires computing power which is not easily available at reasonable costs for a practical application. Various calculation schemes have been considered and worked out. The various schemes were implemented in a discrete time controller using dSpace hardware and a rugged computer system which can be mounted in a test vehicle; see Sec.4. However, the computational burden of these schemes turned out to be too high to allow detailed on-line calculations. Because of this problem, parallel to the digital approach a circuit in analogue electronics was developed to perform the simulation and optimisation, which can speed up the calculations substantially - thus broadening the concept's potentials while decreasing its flexibility -, at the same time keeping costs low. The test vehicle (a DAF tractor semitrailer) is supplied with newly designed controllable shock absorbers and with the digital controller (the analogue controller not yet being available). The implemented control strategy as well as the control parameters can be adjusted during the test runs on the proving grounds. Some first experimental results are discussed in Sec.5. Finally, we draw some conclusions in Sec.6.