Fuzzy systems engineering. Theory and practice. (Q2566472)

This book contains some articles on fuzzy system engineering. It consists of two main parts: The first three chapters are theoretical developments of fuzzy logic. The remaining five chapters are applications of fuzzy logic controllers. Chapter one is an introduction to fuzzy set theory and fuzzy logic. Fuzzy propositions, rules, inferences and approximate reasoning are defined. Fuzzy controller is introduces, and applied to the inverted pendulum problem. Chapter two studies symbolic data manipulation. This is done via introducing degrees and scales. Information is converted into sets of ordered degrees attached to variables using generalized linguistic modifier (GLM). Moreover, weights are allocated to each degree to allow data combination. The combination operator is called symbolic weighted median (SWM). These tools (GLM and SWM) can be applied to practical problems, e.g. colorimetry. Chapter three discusses adaptation of fuzzy inference system using neural learning. Such systems should be fast learning, adaptable, achieving global error rate and computationally in-expensive. Generally there are three main categories for neuro-fuzzy models: Cooperative, concurrent and integrated. Here, three types of cooperative neuro-fuzzy models are presented. Also some Mamdani and Takagi-Sugano type integrated neuro-fuzzy models are introduced. Global optimization procedures prevent the system from being trapped in local minima but they are computationally expensive. However, since evolutionary algorithm uses a population of independent solutions, the situation may improve if parallel algorithms are used. Takagi-Sugeno type fuzzy systems are high performers but their learning procedures are complicated. Mamdani type fuzzy system can be modeled using faster heuristics but with less performance. In Chapter four a model based on fuzzy logic for air navigation in interception flight has been developed. Mamdani rule has been used. The method is compared with some traditional proportional navigation methods with promising results. Moreover the fuzzy fully controlled navigation in the proposed model does not impose any analytic relations between the positions of the aircrafts. In Chapter five the performance of a hybrid soft computing and hard computing techniques, to predict the average monthly foreign exchange (forex) rates, are compared. The soft computing models considered are a neuro-fuzzy model implementing a multi-output Takagi-Sugeno system. The hard computing models contains multivariate adaptive regression spline (MARS), classification regression trees (CART) and a hybrid MARS-CART model. The hybrid (fuzzy) model predicted the forex rates more accurately than all the other techniques when applied individually. Chapter six contains stability and sensitivity analysis of fuzzy control system, with Mamdani fuzyy type, dedicated to servo-system control. Four stability methods are used namely state-space, Popov hyperstability, circle criterion and harmonic balance method. Sensitivity analysis is done assuming that fuzzy control systems, in certain cases, are equivalent to linear ones. Some case studies are given and the results validate the analytical results. In Chapter seven two different approaches to the automatic design of fuzzy inference system are presented. The first uses a neuro-fuzzy learning process to adapt the fuzzy system parameters. The second uses genetic algorithm to select the optimal membership functions and rules for the fuzzy system. Both approaches are applied to mobile robots and robotic communications in multi-robot teams. In Chapter eight two modeling techniques for building structures of the Tennessee Eastman chemical reactor process were used. They are least square estimation (LSE) and a fuzzy logic based Takagi-Sugeno method. The second method outperformed the first in the cases of reactor pressure, cooling and temperature. It has a better variance accounted for in the above cases. The book is an interesting addition in a fastly growing field with many interesting applications. However it suffers from the general problem of books on hot topics, that it may become outdated quickly.