Research objectives and content
The cryogenic system of the LHC (Large Hadron Collider) accelerator will require the simultaneous control of several thousand of variables with strong non-linearities and many degrees of freedom. To guarantee smooth operation of the machine, reliability, fault tolerance and robustness is essential.
Prototype work has been based so far on conventional or modified PID-type loops, implemented on PLC's The main objective of the work is to investigate the potential of intelligent modelling (based on non-linear predictive models) and new modern control techniques, e.g. adaptive, predictive, non-linear parametric predictive control, fuzzy or neural type, for LHC applications, and to experiment them quasi full-scale on the prototype cryomagnet string, as well as on the cryogenic plants. Research areas of particular interest include neuro-fuzzy techniques: studies in artificial neural network based control (modelling non-linear dynamic systems), fuzzy models for non-linear system identification (grey-box modelling) and the use of first principles models for controllers 'based on models' (MBC) as Parametric Predictive Control. Therefore one of the main objectives is to explore new research avenues in the fields of non-linear and adaptive control that exploit the new theoretical developments while emphasizing their connection with engineering practice. To attain this objective the following broad but specific problems are addressed: robust performance in adaptive systems; nonlinear control for performance improvement; incorporation of plant prior information (intelligent modeling) and structure in the control design.
Training content (objective, benefit and expected impact)
The main project training goal will be the research and technical development of intelligent modelling and new modern control techniques which involves many potential application fields as engineering, computer science, physics, mathematics, chemistry, neurophysiology, psychology and others now strongly voiced. As practical benefit is concerned mentioning the optimum operation of the cryogenics systems of the superconducting magnets of LHC accelerator and a deeper knowledge of these complex processes.
As far as control engineering is concerned the benefits that can be gained by using non-linear model approximation techniques having into account that at present very little work is available for understanding the potential of non-linear MBPC (Model Based Predictive Control)
The impact foreseen will be the contribution, with new results, to the validation of these new control engineering techniques applied to strong non-linear processes, i.e., cryogenics, superconductivity, thermodynamic process, ... which are one of the best frameworks for these test benches. Links with industry / industrial relevance (22)
As direct beneficiary of the technology developed in the project, the companies in charge of the final implementation of the control systems in the LHC accelerator will get all the experience accumulated in the project development. Each country will get the mandatory technology return according with the agreement between the CERN member states in the fields mainly of cryogenics, superconductivity magnets, industrial software, and control engineering.
The project will also be close followed by several Spanish international companies highly interested in the final results, as GTD, INISEL Space, Mondragon-Corporaci6n-Cooperativa which are currently working in industrial software and control engineering in several organisations in Europe (CERN, ESA, EMBL, ...), and those which are linked with the universities involved in the project: University of Valladolid (Spain) and the Norwegian Institute of Technology in Trondheim (Norway).