Search

Abstract: This paper offers a new hybrid position estimation method for the sensorless control of piezoelectric actuators. Often mathematical models which map easy to measure signals to displacement are used in this role. The proposed method is aimed at enhancing the accuracy of commonly accepted black box models. Three easy to measure signals are available to be used to estimate displacement. One, the induced voltage, is not suitable for piezoelectric stacks. Two others are more generally usable: the piezoelectric voltage and the sensing voltage. This paper proposes, then theoretically and experimentally verifies a hybrid algorithm that uses the two latter signals to produce estimates of displacement with improved accuracy. [Copyright &y& Elsevier]

In this article, different approaches of the use of black box modelling techniques for piezoelectric actuators are particularly addressed, regardless of the employed technique/algorithm. A modelling approach in this paper refers to two matters: the first, the role of black box techniques in modelling (i.e. if physics-based techniques are also involved in modelling; if so, how and to what extent). From this aspect, the spectrum of approaches ranges from those merged with/inspired by classical phenomenological models to an approach based on purely system identification-based techniques. The second aspect of modelling approaches, in this article, is the input variables to the model. Current and previous values of input voltage, previous values of the output (displacement), derivatives and extremum values of the system's input/output have been used as the inputs to the model so far. Both aforementioned aspects of modelling approaches are addressed appropriately in this article, and various modelling approaches in the literature are categorized and presented in a uniform and comparable manner, so that readers can easily identify research trends in this area and the gaps in the literature. One of the identified unanswered questions in the literature is whether the extremum values of the system's input/output should/should not be used as an input to black box models of piezoelectric actuators. There are works in the literature which have/have not used the aforementioned input variables, but there is no published investigation to evidently answer the proposed question. This article, in the last section, answers this question by reporting and discussing an experimental study. [ABSTRACT FROM AUTHOR]

In this paper, a Neuro-Predictive (NP) controller was first designed and implemented on a highly non-linear system, a model helicopter in a constrained situation. It was observed that the closed loop system with the NP controller had a significant overshoot and a long settling time in comparison to the same system with an fuzzy controller. In order to improve the system's performance, a Sugeno-type fuzzy compensator, having only two rules, is added to the control loop to adjust control input. The newly designed Neuro-Predictive control with the Fuzzy Compensator (NPFC) improves the system's performance. Furthermore, it is shown that the NPFC controlled system is robust to disturbance and system parameter changes. [ABSTRACT FROM AUTHOR]