Real time control of an active pendulum arm suspension system for a downscaled XT28 forwarder model
Abstract: A forwarder named XT28 has been developed in joint collaboration between eXtractor AB, Skogforsk AB, KTH, Linköpings Unversity, Sveaskog, and Bosch Rexroth. The forwarder is equipped with six hydraulic active suspension pendulum arms. This allows for a lower impact on the soils in the forest but also increases the productivity and operation of the forwarder. Attempts have been done to find a controller design in order to control the pendulum arms effectively. Even though the pendulum arms have successfully been controlled using combinations of P- and PID controllers, these controllers have shown to not be effective enough for end use. Therefore there is still a need for a more appropriate controller. This is what this thesis work set out to solve using a real-life downscaled model of the XT28 forwarder. During this thesis work the pendulum arms of the downscaled XT28 forwarder model were redesigned in order to fit passive springs and accommodate load cell force sensors to the pendulum arms. This was done to reduce the stiffness of the chassis and allow for new controller design approaches. Four controllers were implemented during the project based on simplified vertical full-vehicle models, including two Offset-free Model Predictive controllers and two Linear Quadratic Controllers with integral action, with both load cell force- and piston rod position control of the linear actuators in the pendulum arms. The controllers were tested in real-time using a self-built test track with obstacles. A quantitative approach was used to minimize the roll- and pitch angle changes, and their derivatives, of the vehicle body and was compared between the implemented controllers when the forwarder drove through the test track in real time. Furthermore, the effect on controller performance by implementing springs and load cells in the system was also investigated. The results showed that it was possible to successfully implement the proposed controllers using simplified vertical full-vehicle models. The best-performing controllers, which were shown to be the Model Predictive controllers with load cell force- and piston rod position control performed the best with both the springs and load cells implemented. Further, the load cells showed to have a vital part in the success of all of the controllers.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)