Nowadays, roads are becoming more and more congested, resulting in increasing economic losses due to delays. One way to solve this problem is to persuade people use public transp ortation more frequently. To achieve this, public transportation has to be improved. One way to improve public transportation is to construct a new kind of vehicle that combines the advantages of both commuter busses and railroad vehicles. Such a vehicle could be an all-wheel steered multiple-articulated vehicle. In the city of Eindhoven in The Netherlands, a new kind of transportation system will be operational in the year 2003, which is based on such vehicles. To achieve a track following behaviour similar to railroad vehicles, these vehicles have to be equipped with a lateral guidance system for steering them along a prede¯ne d path. This thesis deals with the design of such a guidance system. To achieve good tracking performance, the guidance system has to be modelbased. Therefore, a dynamic vehicle model has been derived. This model describes the nonlinear planar dynamics of an all-wheel steered n-carriage multiple-articulated vehicle. To validate this model, its frequency responses have been compared with the frequency responses of a 125 degrees of freedom multi-body model. This comparison shows good performance between both models. In order to show that a dynamic vehicle model is required, a comparison has been made between the dynamic vehicle model and a model describing only the kinematics of the vehicle. The position of the vehicle with respect to the path to be followed is crucial for proper control and therefore a measurement method based on the utilization of rotation symmetric bar magnets is presented. These magnets are buried in the road. By utilizing the rotation symmetry, the position to the magnet can be determined independently of the measurement height and the strength of the magnet. One requirement is that two ¯eld components are measured. It is shown that the sensitivity of this method to slant of the magnet and/or vehicle can be reduced by using a second dual-axes ¯eld sensor instead of one. Validation measurements show that the distance to the magnet can be determined with about 2 cm accuracy with 90 slant of the magnet. The permanent magnets yield position information exclusively and only at discrete instances. For controller design, knowledge of the complete state of the vehicle is desirable. An extended Kalman ¯lter has been designed to obtain continuous estimates of this state. To keep the in°uence of varying vehicle parameters small, accelerometers are used as input of the Kalman ¯lter. The accelerometer o®sets, road banking angle and vehicle roll angle are estimated online, to reduce the e®ect of these parameters. The position information obtained from the permanent magnets is used to apply corrections to state predictions that are based on the accelerometer outputs. The discrete and asynchronous character of this position information has been dealt with by implementing the Kalman ¯lter in a multi-rate fashion. Articulation angle sensors, wheel encoders and rate gyros are added to the Kalman ¯lter to improve the performance and to obtain redundan cy of sensors. The vehicles that will be used in the public transportation system in Eindhoven have, apart from all-wheel steering, also independent electrical drives on each of the wheels, except the two wheels at the front. This independent drive can in principle also be used for steering the vehicle, by using the drives on one axle in a di®erential way. A singular value analysis shows that steering with normal steering angles is much more in°uential than using these di®erential torques. It has also been analyzed that both at low and high speed all-wheel steering is bene¯cial to reduce o®-tracking of the rear axles and to improve the yaw dynamics of the vehicle. Two di®erent controllers have been designed for steering the vehicle, based on the outputs of the Kalman ¯lter, along the path to be followed. The ¯rst of these controllers is a feedback linearizing controller. This controller can be considered to consist of two control loops. The inner loop linearizes the planar vehicle dynamics, under the assumption that the steering system dynamics can be neglected. The outer control loop is used to counteract parameter uncertainty and disturbances. For this outer loop, a PID controller has been used. The second controller is a so-called backsteppin g controller. With this controller, also the steering actuator dynamics are taken into account. To simulate the behavior of the lateral guidance system, a more complex vehicle model has been used. This model describes besides the planar vehicle dynamics also the dynamics of the susp ension system. A nonlinear tire model has been used in this model. Simulations with this 3D simulation model show good tracking performance for both the feedback linearizing controller and the backstepping controller. The backstepping controller shows improved tracking performance compared to the feedback linearizing controller. However, this goes at the cost of increased high frequent behavior or the lateral acceleration.
Lateral guidance of all-wheel steered multiple-articulated vehicles
2001-01-01
Bruin, de , D 2001 , ' Lateral guidance of all-wheel steered multiple-articulated vehicles ' , Doctor of Philosophy , Electrical Engineering , Eindhoven . https://doi.org/10.6100/IR544960
Hochschulschrift
Elektronische Ressource
Englisch
DDC: | 629 |
Backstepping Control for Lateral Guidance of All-Wheel Steered Multiple Articulated Vehicles
British Library Conference Proceedings | 2000
|Automatic guidance control of an articulated all-wheel-steered vehicle
Taylor & Francis Verlag | 2014
|Automatic guidance control of an articulated all-wheel-steered vehicle
Tema Archiv | 2014
|