Agile

The wide availability of control systems and sensing elements, along with networking components to facilitate communication, and the development of software tools that ease the programming and implementation of control logics in a distributed network of such elements, has led to the acceptance and extensive deployment of such control infrastructures. Existing state-of-the-art SCADA systems provide the control designer with an abundance of tools so that the control logic can easily be implemented, hiding many of the intricacies of communication with the decentralized components and the programming of the individual subsystems. In effect, such advances in the “standard SCADA toolset” have considerably simplified the process of system monitoring, control and data acquisition. In the past, positive feedback from the deployment of SCADA systems has led initially to their acceptance and gradually to their ubiquitous presence – for all serious engineering applications. Especially for larger-scale applications, optimism and the positive results obtained from the use of such systems initially overshadowed some of their problems: supervisory control approaches employed often had poor performance and reliability, and the resilience to atypical scenarios and out-of-design situations was, at best, poor. The pressing need for improved efficiencies and performance is reinvigorating the discussion on improved control strategies.

Within AGILE we take the view that reconfigurability, (near-) optimality and performance can only be achieved through the continuous adaptation of the control strategies employed. For smaller systems it might be argued that with careful planning, a sufficient number of control scenarios can be a priori prescribed in the design phase and used as and when the need arises. Unfortunately, this is not the case in larger-scale systems whereby the number of fault-scenarios requires that combinatorially many controllers be a priori designed – making this impractical. This observation marks the need for a departure from existing approaches and precludes the utilization of a “fixed-set” of control strategies, often encountered in existing SCADA systems. Therefore, the proposed AGILE system is a meta-system capable of “automagically” creating, deploying and fine-tuning SCADA/DCS controllers. There are three essential elements to the AGILE system: the ConvCD approach “convexifies” the problem by creating local convex approximations (for appropriately defined partitions of the state space) that can be selected to be arbitrarily close to the optimal LSCS. ConvCD by incorporating performance requirements and constraints (T2.2) yields meaningfully accurate results in a computationally-efficient manner, while the possibility of using predictive models (T2.3), make the AGILE system respond proactively. The Hybrid AdaptST/MMACM, the second ingredient of the AGILE system, is the actual enabler of self-tuning, self-reconfiguration and fault-tolerance capabilities when the actual system is changing. The hybrid approach, as described in Section 1.2.2, allows for the efficient response avoiding poor transient performance and/or loss of stabilizability, while mitigating, or at least bypassing, the curse of dimensionality. Finally, the third ingredient, critical for the successful deployment of the AGILE system in existing open-architecture SCADA systems and DCSs infrastructures, is the development of open-source interfaces to make AGILE straightforwardly interfaceable and implementable. The AGILE system being intrinsically self-tuneable and able to rapidly and efficiently optimize LSCS performance when short- medium- and long-time variations affect the large-scale system; by providing efficient, rapid and safe fault-recovery and LSCS re-configuration; and, last, but not least, by being scalable and modular, AGILE will be a next generation automation product superior in terms of functionality, accuracy, dynamic range, autonomy, reliability and resilience.

The two AGILE test cases serve to illustrate the benefits of a next-generation advanced control system but also help to fine-tune salient points in the theoretical and practical implementations of the AGILE system. For both test cases, it is expected that the AGILE benefits will quickly be realized as the ability to have near-optimal performance and, perhaps more importantly, resilience to faults, will yield tangible direct and indirect impacts. The first test case, urban traffic control in the city of Chania, is an archetypal example of a large-scale system, and will strenuously test the ability of the AGILE system to optimize traffic conditions in real-time. This case is particularly egregious, as faults and other incidents occur frequently, so that fault-recovery and re-configurability are essential for good performance. Directly expected impacts from application of the AGILE system in this particular case, are lower-traffic times and consequently lower fuel consumption and emissions, and overall better QoS for that particular application. The second test case, control of EPBs, is also especially interesting both from a theoretical and a practical viewpoint. In the control of EPBs the goal is not only to reduce energy consumption at the building-level but also to utilize most effectively energy obtained from renewable energy sources, something referred to as generation-consumption matching. The stochasticity of the weather – and consequently, of energy generation from renewable energy sources, – along with the unpredictable behavior of users pose a very challenging problem in that the controller actions should be extremely proactive if (nearly-) optimal performance is required. Existing building energy management systems (BEMS) are severely limited and simplistic in their algorithms and control methodologies, and use of the AGILE system will lead to improved operation and efficiencies. From the practical viewpoint, improved operational performance will lead to reduced energy intensity for EU economies and enhance competitiveness, along with the obvious environmental benefits. More to this, at the EU-level there is an effort to move from the building- to the neighborhood-level in devising control algorithms to improve energy efficiencies. At present there are significant efforts to generate this infrastructure (e.g. the FP7 program InTube) and when this technology becomes available, AGIILE-type control systems are going to be essential for the proper orchestration of the various independent subsystems leading to effective Neighborhood Energy Management Systems (NEMS) for the environment.

These two totally distinct cases will strengthen the confidence that AGILE is truly applicable in a variety of application areas, involving extremely complex dynamics and stochasticity and attest to the fact that with minimal effort the AGILE system can be adapted and efficiently utilized in other large-scale applications, including, but not limited to: transportation systems, the energy-management and environmental sector, the manufacturing sector and beyond. Once the AGILE system is developed the deployment costs are expected to be minimal: there are no costs involved with the installation of new hardware or modification to existing installations, it is expected that the AGILE payback period will be markedly small – in most cases, less than six months. In addition by avoiding the need for operators and engineers to fine-tune and calibrate the system the operational costs will also be reduced. Improved performance through the availability of next generation automation products will indirectly lead to strengthened competitiveness of the industry supplying monitoring and control systems.

Achieving the, rather ambitious, goals set in the AGILE project, and before the aforementioned benefits can be reaped, significant advances are required beyond the state of the art in a number of areas. For this reason the AGILE consortium comprises members that are world leaders in areas relevant to AGILE project. The three academic partners have demonstrated excellence in the area of control and optimization (TUC, CUT, PSU); the industrial partners ASE and SICE have significant experience on the deployment of SCADA/DCS systems; SIE, FIBP, and TCD have extensive experience in the respective test case areas. The AGILE consortium comprises members from 4 EU-member countries, 1 partner country (ASE, Israel) and PSU from the USA, making for a truly international partnership that transcends strict EC levels. The fruitful interactions and cross-fertilization of ideas that led to this research project proposal will continue more vigorously upon funding and commencement of the proposed work. An “extrovert approach” will be taken in communication of the findings and sharing of the developed tools to policy-makers, researchers, students, end-users. A comprehensive dissemination plan along with the diverse research group and effective communication strategies will lead to reinforced European inter-disciplinary excellence in control and systems engineering.

Full Title: rApidly-deployable, self-tuninG, self-reconfIgurabLE, nearly-optimal control design for large-scale nonlinear systems
Funded by: EU-FP7, ICT-2009.3.5
Start-End Date: 01/09/2012 – 30/11/2013