Computer Aided Control Algorithm Design

Abstract

Conventionally analog controllers and digital control algorithms of PID-type are designed and tuned by trial and error supported by rules of thumb and sometimes by simulation studies. For processes with large settling time or for multivariable processes with strong interactions this procedure is generally quite time consuming and does not result in the best possible control. Better control in a shorter time can be achieved by the computer aided design of digital control systems. Based on the design methods for feedback and feedforward control algorithms treated in this book, programs can be developed which provide for interactive computer aided design. A necessary pre-condition is, however, the knowledge of suitable mathematical process models and possibly of signal models. Process models may be obtained either by theoretical modeling or by process identification, as in section 3.7.4. Theoretical modeling must be used if the process is not available, for example before its construction. However, there are some natural limitations on accuracy of theoretical modeling. There are, for example, the limited accuracy of available process data and parameters, the simplifying assumptions made during the model derivation, or imprecisely known actuator, valve or sensor models. Particularly in the field of industrial processes (chemical, energy and basic industries) some physical or chemical laws are either unknown or are difficult to formulate with a reasonable number of equations. Therefore process models can often be obtained much more rapidly and with greater accuracy by measuring the dynamics of an existing process, i.e. by applying identification me thods. This can be performed off-line or, if a computer is already connected to the process, on-line. As parametric process models are very suitable for the design of parametric control algorithms the identification methods described in chapter 23 may be used. There are program packages for process identification which contain perturbation signal generation, process signal filtering, parameter estimation methods, model order search and model verification [3.13], [29.1], [29.2], [29.3], [23.16], [29.4].