Synthesis and Application of Adaptive Predictive Control Algorithms for a Transportation System

More than ten years ago the early approaches of predictive control methods, for example, the non-parametric model based DMC-algorithm [1] had been proposed in the literature. Following the main idea of predictive control, several control algorithms have been developed since then. Examples are the EHAC- [2], EPSAC- [3], OCS- [4], GPC- [5], MURHAC- [6] and MUSMAR-algorithms [7], which by using a previewed reference trajectory, enabled faster control than conventional controllers. The combination of all these algorithms with adaptive parameter estimation techniques leads to the main advantage, a capability to deal with processes with variable or unknown dead-time. Within the pool of predictive control algorithms, the GPC-algorithm, which may be seen as a combination of certain other predictive algorithms, has gained a great deal of attention in recent years [8-12]. As the GPC-algorithm seems to be favoured in use, there are certain disadvantages too. One disadvantage of the GPC-algorithm is its extreme sensitivity to additional noise signals. This disadvantage in working with the real-time application of a shaker conveyor for transportation of plastic beads is shown and discussed, and two modifications of the basic algorithm are proposed in this contribution. Both the new algorithms, the MGPC-(Modified GPC) and the DGPC-(Direct GPC) algorithm proposed here, show good performance in reducing the computational burden of the GPC-algorithm and achieving a better disturbance rejection property.