E-News May 2013 Edition

A Major Change in Training Course Provision, by Professor Mike J Grimble

When the ACTC started about two decades ago it was supported by the Department of Trade and Industry and one of the main aims was to provide a range of generic training courses at central locations that members could attend. The focus was on control engineering practice and the vast majority of training events covered general control design techniques that were applicable across a range of industries. In recent years the requirements for training have become very much more specific and related to particular industrial sectors and even particular processes. The consequence is that most of the training is now provided at particular company premises using courses that have been tailored to their requirements.

One of the results of this change in requirements is that there is no longer a standard membership fee within the ACTC and the costs are now determined by a company’s requirements for the year ahead. The ability to provide tailored courses is aided by the maturity the ACTC has achieved. The course material has been built up over many years and it is therefore not so onerous to provide specialized and targeted material.

Another change that has occurred in recent years is that the ACTC and the parent company ISC Ltd., have been providing courses in other engineering areas such as electrical engineering. These courses for companies like Rolls Royce often utilise specialist skills brought in from other Universities and companies. This enables the ACTC and ISC to accommodate a much greater range of training course requirements.

It is not a secret that there is a significant shortage of engineers in certain discipline areas. With the rebalancing of the economy one expects both apprenticeships and professional engineer training to increase significantly over the next decade but this is a very slow process. The only option for most companies that require a boost in engineering numbers is to undertake retraining activities and this is where the ACTC and ISC can help. Further details may be obtained from Dr Meghan McGookin.

Mike Grimble


Scheduled Training Courses, Glasgow, 2013

Early Bird Discount!!! We are offering up to 25% discount on our training courses if registration is received 4 weeks before the scheduled date. Registration and prices can be found at ‘Online Registration’ under the following scheduled training courses:

9th to 11th July 2013 - Control Fundamentals:This three-day course introduces the basic control engineering concepts in terms of Linear Systems Models, Classical Control Theory and Practical Aspects in Control. The course presents analysis of a linear control system in terms of stability and performance. In addition, Classical Control Method such as PID controller (including tuning methods), lead-lag and root locus are also covered. Significant hands-on examples are used to reinforce the lectures..

20th and 21st Aug 2013 - Optimisation Methods:This two day course provides an introduction to optimisation methods for use in practical applications. It covers both unconstrained and constrained optimisation methods and more advanced concepts such as dynamic optimisation and genetic algorithms. The lectures are reinforced by Hands-On simulation and design examples that provide practical experience of the concepts covered.

10th to 12th Sept 2013 - Robust and Reliable Control Systems Design: This course illustrates the basic concepts of robust and reliable control system and problems associated with multivariable systems.

26th to 27th Nov 2013 - Modelling, Identification and Parameter Estimation Methods:The course is aimed at engineers who are involved in system modeling and model based control/simulation. Basic System ID methods such as least square algorithm and Kalman Filter estimation are discussed in this course to provide a good background understanding. Real life issues such as implementing system ID and model validation can be problematic and this will be addressed and discussed in the course. Furthermore, two popular System ID techniques namely parameter estimation for grey-box models and nonlinear system modelling are illustrated as well. Lastly, the application of artificial neural network to identify and/or approximate a static and dynamic model is demonstrated too.

28th Nov 2013 - Predictive Control for Linear and Nonlinear Systems:The course provides overview of Predictive Control techniques and then gives guidelines for analysis and tuning of predictive controllers.


IET Control and Automation Conference, Birmingham

Professor Mike J Grimble has been invited to give a keynote talk at the IET Control and Automation Conference which is being held on the 4th and 5th of June in the Conference Aston Lakeside Centre in Birmingham.

The talk will be based on the benefits and problems arising from the use of Advanced Control Methods in Industrial Applications. This presentation will consider the benefits or problems that arise in the use of advanced control methods in industrial applications. It will highlight the techniques which are the most reliable and those that are more suspect. The needs of industry which sometimes require more advanced controls and the opportunities such solutions can provide will be discussed. At the same time, the question of whether advanced control methods really provide the benefits will be considered and some of the more classical control solutions will be endorsed. The aim is to help engineers to determine which methods are really useful and those which are just fun for academics.

For more information on please visit http://conferences.theiet.org/control/venue/index.cfm


BP Upstream Process Control Training, Houston - 22nd - 26th April 2013

ISC has just delivered the eleventh Upstream Process Control course for BP, this time in Houston, Texas. Almost 200 of BP engineers have now gone through this course, which continues to be very well received. The delegates mostly value the application orientated computer-based exercises that were specifically built by ISC for this course. These use simulations of typical offshore processes (separators, compressors, gas turbines) and thanks to the use of LabVIEW they are highly intuitive to use meaning the trainees can focus on the learning exercise, rather than getting bogged down in an unfamiliar application. Also, as these examples are pre-compiled they do not need an expensive simulation application to run, and can be used anytime, not only during the course.

If you are interested in hearing more about this course, or maybe even having a course with bespoke elements please contact Meghan McGookin or Andy Clegg.


Marine Control System Design – Software and Hardware Development

Since winning two awards from National Instruments UK & Ireland in November 2012 (http://uk.ni.com/gsdawards/2012) for its work on the Turbine Access System (TAS) project for Houlder Ltd., ISC has continued to work on projects involving both software and hardware control system development and commissioning. ISC’s work has involved simulation to determine the system configuration and components needed in order to achieve a required performance. These simulations are often also used to develop an emulator (i.e. a system simulation executed in real time) to accurately mimic the real system/plant. Such an emulator allows the developed control system software to be tested early, which in turn de-risks and shortens the subsequent commissioning stages with the real system. Furthermore, the proposed control system can be reviewed, refined and optimised until the desired performance has been met. This provides the engineers (and even the ultimate clients) with a good level of confidence before the factory tests, and can be used to demonstrate the user interface to the end users.

If projects of this type are of interest, please contact Meghan McGookin or Andy Clegg for further information.


Advanced Control in Automotive Applications

There is a very strong trend towards the use of model-based control methods in automotive systems, replacing PID and other classical control methods. There are various drivers for this push towards more advanced methods, including the larger number of actuators that are now available and a growing number of measurements, making it more difficult to solve the resulting multivariable interactive problems. There is also a need to develop new engines and get them into production faster, both to satisfy regulatory requirements and to give customers what they need such as low emissions and good fuel economy.

From research studies, the indications are that predictive and other model-based control will become as successful in automotive applications as it is in the process industries. Since automotive systems often have more irregular dynamics and non-linearities, there are greater challenges for algorithm development. However, recent results appear very promising. As in other industrial sectors, new methods of modelling systems such as linear-parameter-varying and/or state-dependent models are finding application and there is considerable interest in hybrid control techniques.

The great benefit of model-based control methods is that proforma design procedures can be developed so that the time to produce new control solutions can be much reduced. In addition, new engineers can pick up and use such methods with much less effort, so that projects can be shared between groups and engineers used more flexibly. The main problem with most advanced control techniques is in the educational process of enabling calibration engineers to tune such controllers, when their intuitive experience has been on more classical methods such as PID control. Nevertheless, the demands of the sector are such that there will have to be a gradual move towards such methods since they are essential to meet the above-mentioned regulatory and customer demands. Even when existing control loops work well there is a possibility that in future they will be replaced, so that a common unified approach may be used across the company’s product range.


Just Published!! - Paper by Professor Mike J Grimble & Pawel Majecki (ISC Ltd.)
Title: NGMV Control Using Unstable State-Dependent Multivariable Models.

Summary: A non-linear generalised minimum variance (NGMV) control law is derived for systems represented by an input– output state dependent non-linear (NL) subsystem that may be open-loop unstable. The solution is obtained using a model for the multivariable discrete-time process that includes a state-dependent (NL and possibly unstable) model that links the output and any ‘unstructured’ NL input subsystem. The input subsystem can involve an operator of a very general NL form, but this has to be assumed to be stable. This is the first NGMV control solution that is suitable for systems containing an unstable NL sub-system which is contained in the state-dependent model. The process is also assumed to include explicit common delays in input or output channels. The generalised minimum variance cost index to be minimised involves both dynamically weighted error and control signal costing terms. It may also include weighted values of the system states for greater generality. The controller derived is simple to implement considering the complexity of the system represented. If the plant is stable the controller structure can be manipulated into an internal model control form. This form of the controller is like an NL version of the Smith Predictor which is valuable for providing confidence in the solution.


Benefits of Auto-Tuning

It is well known that full adaptive control systems based on ideas like self-tuning control are theoretically interesting but not so valuable practically. The main problem is robustness since even laboratory experiments can involve an adaptive controller that works well for a period but suddenly diverges with the system going unstable. This is not a situation that can be allowed in most applications.

An auto-tuner is one that often uses similar techniques to adaptive control but which is utilised just for commissioning or retuning. There is not therefore the intention to leave the auto tuner running for long periods and hence the controller in question operates mostly with fix gains providing a degree of certainty and robustness. If the operator or engineer wishes to retune they can of course press a button or activate the auto tuner but this would normally involve close supervision of what the system does.

The most successful auto tuning methods have been developed by a Swedish Professor and his team at Lund Institute of Technology in Sweden. This is Professor Karl Johan Åström who is still at Lund University. In this approach a relay can be inserted in the system so that the control loop enters continuous oscillations. The magnitude and the period of the oscillations is measured and used to choose a stabilising control gain. In fact, a result from describing function analysis is utilised. The describing function for a relay involves a line along the real axis and the point at which it intersects the open loop frequency response for the system determines this magnitude and period.

A further link that is required is the type of Ziegler Nichols tuning rules used for PID controllers where by a simple test can be undertaken on a system by increasing the gain until sustained oscillations occur. The magnitude and period of these oscillations are then linked by empirical rules to the recommended PID controller gains. The auto tuning procedure provides equivalent information and hence the tuning rules may be applied.

There are of course other auto tuning methods based on matching desirable step characteristics to some chosen reference signal. Fuzzy logic, neural networks and expert systems tools have for example been employed. However, simplicity is often a valuable feature in all engineering systems and it is here that the Åström auto tuning methods stands out.


New Tools for Nonlinear Control Problems

Over the last decade there has been significant developments in the production of new software tools for the control of nonlinear systems. This applies to, for example, both Matlab based design packages and to those utilising National Instruments Labview. Some of the new design approaches are based on relatively new ways of modelling systems using for example, state dependant, input dependant or LPV models.

One of the new features of such approaches is the relative simplicity of the design process and the structure of the algorithm. In the past nonlinear methods were either empirical and had few guarantees or were so theoretically contrived that they did not apply to real problems. Some of the methods like sliding mode control have been around for quite some time but it is only in recent years that a better understanding of the real application problems has been forthcoming. In some cases rather complex methods such as nonlinear predictive control have been simplified considerably and there is now a real potential for such solutions. If you have a control problem where nonlinearities dominate and where the solution has thus far been intractable please feel free to contact us at iscmail@isc-ltd.com for further details of off the shelf solutions available.


Adaptive Predictive Control

The ACTC and ISC have been working with a Spanish supplier of adaptive predictive control solutions over the last few years; this is a topic which is entering training courses and where we are also gaining applications experience. It combines two successful methodologies of predictive control and of course adaptive systems. Further details of such solutions are available from Ken Slaven at Adex.


CDC Conference Maui, Hawaii

The 51st IEEE Conference on Decision and Control was held on the beautiful island of Maui, Hawaii.

The very prestigious Bode lecture was presented by Jessy Grizzle of the University of Michigan on Bipedal Robots. Jessy is well known for his work on automotive systems but in recent years he has been considering robotic control problems very successfully. This presentation showed how nonlinear feedback control methods are providing model-based solutions that enhance the ability of bipedal robots to walk, run, and recover from stumbles.

Kameshwar Poolla, from Berkley, presented a plenary on the control of the power grid with so-called intelligent periphery. Pressing environmental problems, energy supply security issues, and nuclear power safety concerns drive the worldwide interest in renewable energy. Renewable energy sources such as wind and solar exhibit considerable variability, have large fluctuations, and are uncertain. Variability is the most important obstacle to deep integration of renewable generation. The current approach is to absorb this variability in operating reserves.

As with most major conferences it was accompanied by workshops. One was concerned with nonlinear delay systems and was organised by Miroslav Krstic (University of California, San Diego). This workshop covered a range of applications from automotive to network control systems.

The CDC conference has a larger proportion of academic papers than perhaps the American Control Conference and nevertheless, it is a sign of the times that quite a few of the sessions were concerned with applications and there was even a session on emerging control applications. Much of the new work in applied control is in the area of networks and systems. There is both the problem of controlling networks and also the problem of using networks for control. These represent two different communities. One of the more unusual sessions was concerned with the modelling and the control of disease.

There were quite a few sessions and papers on the topic of sliding mode control which is now a mature topic but it is still experiencing strong growth. Some new ideas on adaptive control were also explored and in fact there were areas where very interesting controversies arose, such as on L1 adaptive control.

Since theses are hard times and because Hawaii is so far away I was not expecting a very large attendance at this event. However, there were about 1500 attendees which is larger than most CDC’s. The conference facilities at the Waldorf Astoria Resort hotel were excellent. The golden beach and blue ocean were a great aid for deep thought on control engineering.


Book Review : Hybrid Control and Motion Planning of Dynamical Legged Locomotion

For those interested in Robotics the new Hybrid Control and Motion Planning of Dynamical Legged Locomotion text By Nasser Sadati, Guy A. Dumont, Kaveh Akabri Hamed and William A. Gruver published by the IEEE through the good offices of Wiley represents a very comprehensive look at the type of problem of legged locomotion and the challenges in dynamic walking and running. Legged robots are one of the most difficult control problems even though human beings seem to get on quite easily. Nevertheless, each aspect of the problem from modelling to control is difficult and this book discusses all aspects of the problem.

There is no doubt that in future years despite the ready availability of humans we shall have the help of robots with human like behaviour. There will be many applications where such things are very desirable such as hazardous environments. There now seems to be emerging a methodology of control of these hybrid systems and even software tools. In fact this text has downloadable Matlab code which makes it a good starting point for research studies or industrial applications in this area.

This is an invaluable book for anyone working in the area and is applicable equally to engineers and industry and to undergraduate students. It is well illustrated and has extensive references that are very relevant to the topic. It is to be recommended particular for those starting out in the area. The cost is £66.95 and it may be obtained from John Wiley Publishers or through the IEEE press.

Book Nerd


Book Review : Linear and Nonlinear Inverse Problems with Practical Applications

The text by Jennifer L Mueller and Samuli Siltanen on Linear and Nonlinear Inverse Problems with Practical Applications is of value in a wide range of engineering areas and not just control engineering. It is a rather mathematical text published by the Society for Industrial and Applied Mathematics (SIAM) and as the title suggests it is about the use of inversion methods in engineering.

The use of the inverse of the plant model is not of course really practical but there are many circumstances where the idea is exploited in some way to try to get a system which is easier to control. This is particularly true in feed-forward control.

This particular book considers all aspects of the general problem of the analysis and solution of inverse problems, mainly in engineering, but it also covers some other areas. I am afraid the basic mathematics needed to be able to be able to understand this book is at quite a high level and hence it would be more useful for students and researchers working on such problems. The cost of the text is £84 and may be obtained from SIAM.

Book Nerd