The automation of industrial processes is a hallmark of modern society. It has improved production, efficiency and safety in a variety of industries from manufacturing to transport. The basis of engineering control systems for many tasks is mathematical models and algorithms. However, many systems display nonlinear or non-stationary behaviour which complicates this modelling process. Examples of this nonlinear or non-stationary behaviour can be found in aircrafts. The variables acting on the plane which a model or algorithm is attempting to automate are constantly changing over time. Air speed, direction, temperature, etc., are all dynamic variables and the algorithms must adapt. Current mathematical theories and principles employed to deal with these systems have done a good job increasing economic performance and helping many industries meet environmental regulations. However, demands are increasing and methods in the linear time-invariant (LTI) framework, the current standard, are being stretched to the limit. To address this Dr Roland Toth and colleagues in the APROCS project have been working with a new framework for addressing the challenges of nonlinear behaviour. Members of APROCS have been working on the linear parameter-varying (LPV) framework, which according to Toth is a better way forward. It improves on the LTI framework by operating under the principle that modelling these complicated systems requires varying linear models. Impressively, this new framework extends the functionality of the LTI methods but maintains one of the best features of LTI - high levels of simplicity and reliability. According to Toth, LPV has three important key features: guaranteed performance; easiness of design; and computational efficiency. 'We have already shown the powerful capabilities of the method on laboratory examples achieving high-accuracy servo control of electro-mechanic motion applications with guaranteed stability and performance specs,' says Toth. With the positive attributes of the LPV framework established, the team is now moving to make the method available to the widest possible set of users.
This work is licensed under a Creative Commons Attribution 4.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/