Technology & Research News

TECHNOLOGY AND RESEARCH NEWS
Advanced Control of Energy Systems
Power electronics is devoted to electrical energy conditioning, management and storage. Power converters are the tools to do this job. They process electrical energy in a way that would be 100 per cent efficient if the circuit components were ideal. Switched power electronic circuits are highly nonlinear systems whose performance can be improved by advanced, nonlinear control techniques. Control design for power converters or, more in general, for energy management systems, as well as its implementation, is the aim of the research group ACE (http://www.aces.upc.es) at the Institute of Industrial and Control Engineering, Universitat Politècnica de Catalunya. ACES mission entails contributing to the progress of scientific knowledge, the training of specialized personnel and the diffusion of technological advances in the field of modeling and control of complex systems, and its application to problems related to the generation, conditioning, management and storage of electrical energy.
Power converters belong to the wide class of Port controlled Hamiltonian systems (PCHS), which generalize the Hamiltonian formalism of classical mechanics to physical systems connected in a power-preserving way. A PCHS model encodes the detailed energy transfer and storage in the system, and is thus suitable for control schemes based on, and easily interpretable in terms of, the physics of the system. PCHS are passive in a natural way, and several methods to stabilize them at a desired fixed point have been devised.
In order to use the regulation techniques developed for PCHS, a method to reduce a signal generation or tracking problem to a regulation one is, in general, necessary. One powerful way to do this is averaging, and in particular Generalized State Space Averaging. In this method, the state and control variables are expanded in a Fourier-like series with time-dependent coefficients; for periodic behavior, the coefficients will evolve to constants. In many practical applications, physical consideration of the task to solve indicates which coefficients to keep, and one obtains a finite-dimensional reduced system to which standard techniques can be applied.
Switched power converters can also be modeled as variable structure systems because of the abrupt topological changes that the circuit, commanded by a discontinuous control action, undergoes. They therefore constitute a natural field of application of Sliding Mode Control (SMC) techniques. It is worth noting that SMC theory presumes an infinite switching frequency when the system operates in sliding mode, and that actual switches cannot commute at infinite frequency. At any rate, higher switching frequencies become harmful in some of the applications. In power electronics, for instance, the higher the switching frequency, the higher the losses in the converter. Consequently, actual sliding mode controls operate at high, finite, possibly variable frequency, yielding a chattering around the sliding surface. Appropriate SMC implementations in switching systems must be considered at this stage. The switching frequency is required to be stable and synchronous for this type of systems, resulting in very demanding requirements for non-standard implementation strategies. These problems have been tackled through fixed and variable bandwidth hysteresis comparators, by the addition of an external synchronous signal, by the use of equivalent control as duty cycle, and by our own method, Zero Average Dynamics.
In the case of specifications given by periodic signals, linear controller designs are commonly based on the Internal Model Principle (IMP). This principle states that if a certain signal must be tracked or rejected without steady-state error, the generator must be inside either the control loop, the controller, or the plant itself. In practice, many real systems have to handle tracking and rejecting of periodic signals. Thus, the open-loop transfer function must contain the signals to be tracked or rejected. A well known technique which uses the IMP concept to address this problem is repetitive control. This technique has been extensively used in electronic rectifiers, PWM inverters, and current harmonics active filters. Also, instead of repetitive controllers, a bank of resonators can be used to contribute with a high gain to track the desired reference.
Enric Fossas, Institute of Industrial and Control Engineering, Universitat Politècnica de Catalunya, Spain (Email: enric.fossas@upc.edu)

This article was recommended by Mario di Bernardo.

Quality Issues for Research in Circuits and Systems

Have you ever read or reviewed a paper, where you missed essential data in order to understand or reproduce the results? Were you ever frustrated that the authors of the paper you were reading did not test and evaluate their designs on more than one case or only reported about the cases where it worked well?

As the EXCOM and BOG of CAS Society we feel that we have the responsibility to work on good practices and guidelines for guaranteeing maximal quality and value of the publications. So I would like to discuss with you some initiatives on this issue and ask for your cooperation on it.

Let us first discuss initiatives that can improve the quality and value of publications. First of all modern web technology provides us a channel for more information from the author to the readers and the reviewers than the paper itself. This can typically be additional data, signals, measurements, test cases, tables, or the computer code, and the experimental configurations. Often this information is rather boring and not so important to be included in the main text or appendix and hence for page limitations it is often eliminated. However this information is often essential to reproduce the findings of the paper. Such additional information allows for complete reproducibility of the findings in papers describing and studying new theory, algorithms and computer programs. This is usually called "reproducible research". In fact a number of research groups [1,2] at Stanford and EPFL Lausanne are already systematically implementing it in their work. Our sister society the IEEE Signal Processing Society has also taken a number of actions and initiatives for promoting the idea and studying the implementation with a special session at ICASSP07 [3]. However for experimental setups, and hardware implementations of designs this additional information is not sufficient to verify it completely. Here a special session with hardware implementations like that at ISCAS07 is very valuable. For such physical and experimental setups one cannot talk about reproducibility but about a weaker form of reusability. There are additionally other mechanisms for increasing the value for the readers and those are common targets for the design of algorithms, hardware, or systems. One can distinguish here benchmark design problems, or challenges, and competitions. The competitions distinguish from the benchmarks in the sense that competitions have a common deadline for the submissions and this allows to keep some information or results unknown to the participants until the deadline. When the submissions are then revealed at the workshop or conference they can be compared among each other and also with the optimal solution if it exists. Of course one should in advance agree also on the ways the quality of the submitted designs will be measured. Both benchmark problems and competitions should be based on a commonly accepted target that is considered meaningful and valuable by the scientific community. Typically such benchmarks or competitions can be set up by our Technical Committees, and are certainly encouraged for ISCAS 2008. CEDA has already made programming challenges [4].

Now let us discuss the actions that we can take on these mechanisms. First of all let us mention here that it is certainly not the intention to make the reproducible research mandatory or force all research papers to deal with benchmark problems or competitions. There are several reasons for that. One is the fact that we need a common agreement about the value of these targets or mechanisms, and that confidence takes time to grow in our scientific community. Moreover the creativity in setting up new design problems should be stimulated and the design problems can only be set up when a problem is considered to be widely accepted. Third several industrial researchers may have intellectual property problems to publish the additional material for reproducible research. It is certainly not the intention to refrain industrial researchers from publishing in the CAS Society journals rather on the contrary.

As concrete actions for our Society members I propose:

- reflect and discuss in your research environment these instruments and let us know your ideas, thoughts and concerns.

- make proposals for special sessions at ISCAS on the ideas of reproducible research, benchmark designs and competitions

- make proposals for life demo sessions at ISCAS

It would be nice that I could have your comments by October 1, 2007, so that we can discuss it at our November BOG meeting, and have interesting activities already for the participants of ISCAS 2008.

References:

[1] http://lcavwww.epfl.ch/reproducible_research/

[2] http://sepwww.stanford.edu/research/redoc/

[3] J. Kovacevic, How to Encourage and Publish Reproducible Research, Proc. IEEE International Conference on Acoustics, Speech, and Signal Processing, vol. 4, pp. 1273-1276, 2007.

[4] http://www.iwls.org/challenge/

Joos Vandewalle, VP Technical Activities, IEEE CAS Society (Email: Joos.Vandewalle@esat.kuleuven.be)