|M.Sc Student||Vertzberger Eran|
|Subject||Analysis and Comparison of Dead-Time Compensators|
|Department||Department of Mechanical Engineering||Supervisor||Professor Emeritus Zalman Palmor|
|Full Thesis text - in Hebrew|
Time delays (dead times - DTs) appear frequently in control of industrial and engineering systems. Presence of delays in control loops
complicates the systems analysis and as well reduces the achievable performance. One way to improve the closed loop performance is to include a predictor in the controller. In this fashion, the delays are pulled out of the characteristic polynomial and a closed loop system of a finite order is obtained. That kind of a control scheme is often called a Dead Time Compensator (DTC). The first DTC was proposed by Otto J. M. Smith more than half a century ago and is referred to as the Smith predictor (SP). Since then numerous publications studied the properties of SP and proposed various modifications and predictors. Some of the proposed DTC were continuous, some discrete and incorporate a variety of different structures. The multitude of DTCs and variety of structures make the analysis of the special properties of each DTC and relations between the various DTC’s difficult. As a result, comparisons of DTCs in the literature are often incomplete as some properties are not taken in consideration in a proper way. This work aims to generalize DTCs as a way to simplify the analysis and relations between different structures.
DTCs adequate for unstable systems can be divided into two types. One includes completion-based predictors and the second truncation-based predictors. In this work a generalized scheme is developed for all Single Input-Single Output (SISO) DTC’s incorporating a truncation-based predictor. All members of that group are shown to be formulated as particular cases of the unified scheme using simple block manipulations. Via the unification a balanced and comprehensive comparison among DTC’s of that group can be performed, considering tracking, regulation and robustness properties uniformly.
The unification can also be used as a tool for analyzing properties of DTC’s. Via the unification, the unique properties of a certain DTC are pointed out, equivalence between different structures is revealed and some DTC’s are shown to be more sensitive to model uncertainties.
Although DTC’s incorporating a completion-based predictor are not included in the unification, the mathematical relationship to the generalized scheme is developed. Every member of the unification is shown to have an equivalent completion-based structure. In some cases the truncation-based DTC has better tracking properties than the completion-based counterpart. The equivalence between modern structures and the completion-based DTCs, introduced more than thirty years ago, underlines the contribution of bringing the different structures to a common ground through the unification.