Editorial Self tuning control As a form of adaptive control, self-tuning control has been of major interest for a number of years now, with the total of operative industrial controllers constantly growing. In fact early developments in the field can easily be traced back to the beginning of the 1970s, and in the following decade many theoretical developments and variations were witnessed. The topic really grew up in the 1980s, and indeed the IEE organised four very well attended workshops on self-tuning and adaptive control in 1981, 1983, 1985 and 1987. These workshops resulted in the publication of a book with the workshop title, with a further text, more concerned with implementation issues, appearing three years ago. The last few years have seen quite a number of excit- ing steps forward in what is now the fairly well estab- lished field of self-tuning control. In common with other computer-based controllers, as technological improve- ments allow for greater processing speed andflexibilityof operation, so the range of controllers realistically con- sidered has broadened considerably. This allows for the application of self-tuning control to systems of varying complexity and with a wide diversity of desired control objectives. Owing to the general software base present, so variability and expansion of the controllers is possible, with guarantees of performance improvement and oper- ational success. Further, the need for operator interven- tion is much reduced, leading to higher reliability with fewer enforced maintenance or upgrade costs. The aim of this special issue is to present a number of recent developments in the field of self-tuning control, with as broad a scope as possible. This means that a range of theoretical and applications oriented papers concentrate more on plant identification/estimation issues, while others are more concerned with the type of control employed. It can be readily witnessed however that certain trends are exhibited, and perhaps most prominent amongst these is the employment of a predic- tive control approach, which is something that has only been realised over the last few years. Nevertheless, although they exist within a self-tuning doctrine, the con- trollers discussed cover aspects such as multivariable systems, nonlinear systems, servo and optimal control, state-space design and parameter tracking. Shieh, Zhao and Sunkel put forward a hybrid state- space controller with dual-rate sampling, this being based on a pole-assignment philosophy. Essentially a fast rate discrete time model of a continuous time system is obtained, and this is subsequently converted into a con- tinuous time model for controller design, the resultant controller is then implemented as a slow-rate digital con- troller. In the paper by Chotai, Young and Behzadi, a PIP control scheme is applied to operate a nonlinear heated bar control system. This also employs pole assign- ment via state variable feedback and makes use of a priori physical information about the system in order to improve parameter tracking properties of the incorpo- rated recursive identifier. The paper by Shook, Mohtadi and Shah contains a description of a multistep ahead-costing function for parameter estimation, this being of use with a long-range predictive controller. A parameter estimator of this type makes use of a range of prediction errors, rather than a single step arrangement. Predictive control is also con- sidered at length in the paper by Zhu, Warwick and Douce, in which a Hammerstein model is used for the control of nonlinear systems. The method employed makes use of a simple one-step root-solving procedure as a crucial part of the overall tuning algorithm. Robinson and Clarke, in their paper, look at closed- loop robustness properties of generalised predictive con- trollers in the presence of modelling errors. This leads to direct controller commissioning procedures and guide- lines for the selection of user specified prefilters. Fern- andez del Busto, Wellstead and Zarrop describe a transient specification approach to parameter tracking in recursive estimation and self-tuning control. In their paper, a specific class of algorithms is formed by con- sidering recursive estimators as dynamical systems. The operation and performance of several self-tuning controllers is investigated by Tham, Vagi, Morris and Wood with online application to a binary distillation column, a direct comparison with multiloop PI/PID control being paramount. Results from this work show that multirate-multivariable self-tuning control can provide improved regulatory performance when com- pared with multiloop PI/PID schemes. A further multi- variable application, this time for the control of a boiler-turbine unit for power generation, is presented by Rossiter, Kouvaritakis and Dunnett. This work employs a generalised predictive control scheme which clearly exhibits improved performance when compared with conventional PID methods. Hang, Lim and Ho put forward a generalised minimum variance self-tuning controller which is based on overshoot and settling-time performance through z-plane pole restriction. A dead-band algorithm is employed, and this ensures that systems stability robust- ness is enhanced, providing superior results to straight- forward pole placement methods. Finally, Deng and Huang present a multivariable self-tuning controller which achieves pole assignment with static and dynamic decoupling. This approach avoids the need for solutions to a Diophantine equation, and provides simpler implementation features, again in comparison with straightforward pole placement. It is difficult to foresee in which ways the field of self- tuning control will progress over the next few years from its present position, although it is felt that many of the ideas put forward in this special issue will have consider- able influence, not only on the basic designs and algo- rithms obtained, but also on the way in which such controllers are implemented. It is particularly felt that the fruits of predictive forms of self-tuning control have not yet been fully realised, and this is seen as one area in particular where considerable profitable developments can be made. It is hoped that this special issue will be useful in pointing the way for the future of the subject and will provide a stimulus for the consideration of further new, and possibly distinctly diferent, design pro- cedures. K. WARWICK IEE PROCEEDINGS-D, Vol. 138, No. 1, JANUARY 1991