Variable-structure controls are normally understood to be controls that have sliding modes and robustness as their main objective. In addition to sliding-mode controls, there are also variable-structure controls, which were developed for the purpose of intentionally precluding sliding modes and achieving high regulation rates and short settling times. Two types of such controls may be distinguished, variable-structure controls that switch between different parameters and a systematic further development of them called "soft variable-structure controls" that continuously vary controllers' parameters or structures and achieve nearly time-optimal control performance. This paper surveys soft variable-structure controls, compares them to other controls, taking a submarine dive-control as an example, and presents an outlook on their auspicious further development.

Unlike static fuzzy systems, recurrent fuzzy systems allow representing knowledge-based dynamic processes that can be stated in the form of "if ... , then ..." rules, making it possible to model systems that can only be described qualitatively. Further possible applications exist in the case where the dynamics of a system are quantitatively known, but only in certain mesh points. The interpolating character of the fuzzy system between the dynamics of the mesh points yields a complete dynamic model. Based on discrete-time recurrent fuzzy systems this article presents first steps towards the theory of continuous-time recurrent fuzzy systems and provides criteria for the investigation of the dynamics of this class of systems.

Continuous-time recurrent fuzzy systems allow modeling continuous-time processes whose dynamic behavior can be stated in linguistic if-then rules. If the dynamics are known in certain mesh points, the interpolating character of the fuzzy system between the mesh points yields a complete dynamical model of the process. In this contribution, continuous-time recurrent fuzzy systems are employed to model the electrical behavior of a solid oxide fuel cell, which both can be described qualitatively and is known quantitatively from measurements. Due to the transparency of the model an easy estimation of its parameters is possible. Additionally, the model parameters are optimized numerically to enhance the model performance.

No abstract

No abstract