Control Predictiu no-lineal i gestió de la producció en temps real

The project has two main aims: Predictive control of non-linear processes and production scheduling of mixed continuous/batch factory sections. We try to approach the first problem using several techniques: neural networks, robust formulations and physical models, and with the idea of obtaining a controller which incorporates in addition management of constraints, set point optimization and fault detection and adaptation. Besides continuous processes, the controller is aimed to batch units driven by economic targets. The production scheduling of a section will be studied with two approaches: one based in constrained predictive control with a mix of continuous and integervariables and another based on timed Petri nets. The project tires also to develop software tools for design and control: a Matlab Toolbox and a real time computer control, that can be accessed from the web. It also includes experiments in simulated systems and pilot plants.