E-News March 2019 Edition

Advanced Control for Applications
Mike Grimble

There has been a shift in academic research on advanced control systems design towards the solution of challenging application problems. The main interest from the academic community has been in the way mathematical tools could help to analyse and synthesise advanced control solutions. There has been a lot of work on robust control systems design but other areas of design have been relatively neglected. The solution of optimal control problems was often assumed to be the end of the matter but the route to satisfying an industrial specification was mysterious.

Many theoretical tools now seem to have a more important role in real application problems but the design procedures still need development. There is a continuing interest in new ideas to address the challenging problems that have arisen in applications. New problems are often driven by technological developments in areas like robotics, bioengineering, smart grids and like autonomous vehicle controls. Control problems in applications are often more difficult than in days gone by because of increasing demands on product quality or system performance.

There are some advances in control systems theory that have been in response to application needs. For example, the coordinated control of squadrons of autonomous flying vehicles is a problem that has involved and stimulated interest in the use of multi-agent systems theory. Agent-based systems provides a way of designing and implementing software systems. A multi-agent system provides a coupled network of software agents for complex problems. They are valuable for complex distributed control problems such as modern power systems where there are difficult problems due to the use of far more renewable power sources.

One of the biggest problems in implementing an advanced control system on a real application is to tune the controller so that adequate performance is achieved in a reasonable timescale and with a solution that is robust. There has been a large increase in the number of sensors and actuators available in most applications. The use of multivariable control techniques is therefore more important than in the past. Applications that require fast processing speeds are also providing new challenges. The next generation of advanced controls will therefore need to be more reliable, easier to design and numerically efficient to implement. There is a very bright future for modern advanced control methods and in fact a new journal published by Wiley should stimulate an interest in such developments (Journal of Advanced Control for Applications: Engineering and Industrial Systems).

Mike Grimble

Top

Assessment of Company Research and Development Services in Advanced Control

ISC has been developing a service to help companies assess the state of their research and development teams working on research and design issues in advanced control. This was initially at customers request but then the need for a more general service was recognised. In addition to covering an investigation of the type of tools and design techniques a company uses the type of training that should be undertaken is also now considered in cooperation with the ACTC training engineers. The aim is to provide a comprehensive report usually at the end of a 2 or 3 months period which summarises the most efficient way forward and the type of tools that might be employed. This should be a help to busy managers who because of the pressure of work find it hard to pause and assess the strengths and weaknesses of their design and commissioning teams.

ISC Limited has had three decades of experience in working across industrial sectors and finding the advanced control methods that are most effective. Many problems arise due to a poor understanding of the systems involved and investment in modelling and simulation is often therefore very beneficial, and this is even before different more advanced control techniques are considered. The ACTC Training Programme has been active for about two decades and it has built up a huge library of information and training course material. This expertise can be called upon when assessing company needs and for the provision of technology transfer workshop material. Further information on the provision of such services can be obtained from Dr Meghan McGookin

Top

New Course: Artificial Intelligence for Control

A new course has recently been developed on Artificial Intelligence and Machine Learning for Control and Signal Processing Applications. The ACTC courses can be presented at company premises and this course is just one day in duration but depending upon the application area could be extended to 2 days by including appropriate hands-on examples. The use of hands-on simulations is always considered one of the most effective ways of providing practical experience of the design methods for attendees. There seems to be considerable interest in the use of big data which provides new opportunities but also challenges into how best to exploit this information. In some industries like automotive and robotics there is a strong interest in the potential of Artificial Intelligence (AI) systems, to exploit machine learning for control and tuning, and for condition monitoring and fault detection.

The course includes a historical introduction to the topic and a state of the art overview of recent developments. It may be of interest to engineers involved with systems modelling and system identification. It covers topics such as neural networks to exploit big data and support vector machines for plant model identification. The course also covers the use of fuzzy logic and fuzzy control methods, and reviews the opportunities in industrial applications for the future. Further details of the course or any questions can be addressed to Dr Meghan McGookin

Top

Marine Control Systems

In years gone by there was not a particular interest in developing marine control systems that were energy efficient. However, it is now often a priority of ship owners to reduce energy costs and to limit harmful emissions. Improvements can be gained both at device level and through total system control. For example, an integrated energy management system can attempt to optimise the total system whilst control of the rudder and fins can be considered more of a local optimisation problem. Advanced controls can be used in both cases but the greatest improvements will be through the changes to upper level integrated control systems.

Optimal control methods and optimisation clearly play an important role since they enable criteria to be established and controllers to be tuned to minimise costs. Even classically controlled systems which already exist can benefit from the use of benchmarking analysis methods which enables controllers to be re-tuned to optimise behaviour. In fact just the very focus on energy minimisation will probably lead to improvements. The future will belong to companies that are friendly to the environment and at the same time save them money.

Top

Next Major Control Conference

The next major control conference is the American Control Conference to be held in Philadelphia during early July (10th to 12th). This event is probably one of the best to attend for engineers in industry. It contains a nice mixture of real applications work with new and innovative developments in advanced control. It is organised by the American Automatic Control Council and it is proceeded by a number of workshops involving tutorial sessions in new areas of control. It will be held at the Marriott Downtown Philadelphia, Pennsylvania. Both the workshops and the conference can be recommended.

Top

Book Review: Optimal Space Flight Navigation

This text on Optimal Space Flight Navigation is by Doctor Ashish Tewari and is published by Birkhauser. The cost of the hard cover version is £79.99. Although this text is concerned with an application area it provides a nice introduction to the control methods required in the space flight navigation problem. These include optimal control theory and optimisation methods dealing with equality constraints. The application related areas include material on orbital mechanics, two body manoeuvres and flight control in non-spherical gravity fields.

This text is really aimed at students and researchers working in the areas of spacecraft dynamics and control systems. It is well organised and nicely presented. It is certainly a very desirable text to have for anyone working in the control of space flight systems.

Book Nerd.

Top

Book Review: Frequency Domain Techniques for H∞ Control of Distributed Parameter Systems

This text on Frequency Domain Techniques for H∞ Control of Distributed Parameter Systems is published by SIAM and involves a distinguished set of authors including Hitay Özbay, Suat Gümüşsoy, Kenji Kashima and Yutaka Yamamoto. The Control of Distributed Parameter Systems may be thought of as a quite mathematical area and not too relevant to industrial applications. However, there are many applications where the model for the system is a distributed parameter model. It is often the case (for simplicity) we approximate such a model using a lumped parameter model but this is sometimes a poor approximation. If for example the problem is flatness control in a rolling mill it is the tension distribution across the strip which is of interest and not just the tension at various points across the strip. The techniques covered in this text are therefore important, particularly for researchers and those faced with problems where such approximations lead to poor performance.

The theory is of course rather more complex than for traditional robust control problems. Nevertheless, this text is laid out in a very clear and understandable manner. The results presented are mathematical but there is a lot of engineering design material which helps to make the book more accessible. The illustrations and figures are particularly valuable. There is a comprehensive list of references included and the text is relatively short. The cost of the paperback version is $79 and this is certainly recommended for anyone working in the area of distributed parameter systems.

Book Nerd.

Top

Book Review: Uncertainty in Complex Networked Systems

This text is edited by Professor Tamer Basar who is very well known for his contributions to control engineering education. The hard copy version of this book costs £99.99 and the e-book version £79.50. It is published by Birkhauser and the topics chosen pay homage to the work of Professor Tempo. Uncertainties in systems are of course one of the most difficult problems a control engineer has to face.

A good example of the problems faced is covered in the chapter concerning stochastic optimisation for energy storage allocation in smart grids. Power engineers are of course very concerned with the developments in smart grids and this chapter is quite practical despite the title of the text. In fact the whole book is more accessible than the title suggests. It is certainly a research text but it deals with many down to earth engineering problems such as fault diagnosis. It is also good value for the 600 pages hard copy version.

Book Nerd.

Top

Book Review: Linear Stochastic Systems

The author Professor Peter Caines of McGill University in Montreal is one of the leading researchers in stochastic systems. Stochastic control does of course deal with disturbances and measurement noise in a very formal manner. He has produced a very weighty text in more ways than one. Although the topic may sound a little esoteric for some engineers in industry it does underpin many of the advanced control approaches now used. It is one of the most comprehensive texts on the subject and is written in a very clear and precise manner. Although the subject is rather mathematical the book does include a lot of discussion making it far more accessible than many competing texts.

The book has a very comprehensive introduction to stochastic processes and linear systems and it then moves onto the most interesting chapter on estimation theory. The Kalman Filter is becoming ubiquitous in industrial application in the same way it has dominated in the aerospace industry. System identification methods which are related to filtering techniques are covered and again they are written in a way that is very accessible but also covers a lot of the main areas of interest to industry.

The linear quadratic optimal control methods are also covered which are now probably the second most popular advanced control technique (after predictive control). The final chapter is concerned with adaptive control which also has an important role in industry although it is accompanied by certain risks.

The text is published by SIAM and the cost of the soft cover version is £90.50, which is very good value considering the large amount of material covered in the almost 900 pages. This is certainly a text to be recommended for research use and students but is also valuable for engineers grappling with the intricacies of advanced control theory.

Book Nerd.

Top