AGILE - RAPIDLY-DEPLOYABLE, SELF-TUNING, SELF-RECONFIGURABLE, NEARLY-OPTIMAL CONTROL DESIGN FOR LARGE-SCALE NONLINEAR SYSTEMS
Computationally efficient and scalable Large Scale Control System (LSCS) designs will be developed within AGILE with the aim of substantially improving large-scale non-linear system performance (and thus productivity). These model-based approaches used in conjunction with convex optimization and self-tuning tools will achieve near-optimal system performance by intelligently avoiding the "curse of dimensionality"
Most LSCDs operate at much less than their optimal level owing to non-linear phenomena, atypical situations, restrictions on performance, etc. For example, in a traffic system the most appropriate action is not always taken when there are sudden changes in traffic flow, when detectors malfunction or when vehicle access is restricted because special events are taking place - all situations that can cause the system to collapse.
AGILE is based on tools and technologies developed recently by AGILE's members, providing a scalable, modular system that is an automatically adjustable, resettable and nearly optimal LSCD design that can be used with large, heterogeneous, complex non-linear systems. The system will contain user-friendly tools for adjusting the restrictions, characteristics and objectives of current SCADA or DCS systems.
Development of the interfaces and the middleware software between the optimized engine and sensors/actuators will be the heart of the SCADA partner role (WP4).
To ease implementation and deployment of the AGILE system in existing open-architecture SCADA/DCS infrastructures, a set of open-source interfacing tools will be developed. The integrated LSCS design system to be developed within AGILE along with the interfaces will be extensively tested and evaluated into two real-life large-scale Test Cases:
- a 20-junction urban traffic network (city of Chania, Greece)
- a large-scale energy-controlled building (Kassel, Germany)
The above test cases will possess a rich variety of design and performance characteristics, extremely complex nonlinear dynamics, highly stochastic effects, uncertainties and modeling errors, as well as reconfiguration and modular design requirements.
As there is no well-established LSCS strategy for controlling Energy Positive Buildings, a range of different strategies will be tested and evaluated. Such strategies will include standard linear optimal control strategies (based on a linearized version of the building dynamics), fuzzy-tuned strategies (which have been proven quite efficient in energy-controlled building applications) as well as approximate dynamic programming controllers that are based on highly nonlinear building models.
AFCON's role in the project
AFCON will be involved in the development, implementation and integration of the SCADA/DCS - (WP4) and have main roles in test case #2 design, implementation and data collection (WP5) and will lead the dissemination and exploitation (WP7).
Project web site: http://www.agile-fp7.eu/default.asp?node=page&id=127&lng=2