Pneumatic cylinders are used in many industrial applications to position loads using a rectilinear motion. Although pneumatic cylinders are easy to set up and to maintain, they are limited to a narrow range of applications because their motion trajectory is difficult to control. Conventional linear control methods can not compensate for both the nonlinear flow of compressed air and the internal friction present in the cylinders. Multilayer neural networks (MNN) are nonlinear mappings which can be used to compensate for the nonlinear nature of these dynamic systems. A model of a pneumatic cylinder was developed to provide training data for a MNN. The MNN was designed to cancel the cylinder dynamics and was implemented as a feedforward controller in conjunction with a PID feedback controller. The MNN was trained over a range of constant velocity trajectories. The resultant controller allows the model to track the constant velocity training trajectories as well as trajectories for which the MNN was not trained. An adaptive MNN controller was developed to provide adequate control in cases where the cylinder dynamics are continually changing.
展开▼
