Morphological signal processing and computational intelligence for engineering system prognostics

Today's complex industrial and engineering systems - especially with the appearance of large-scale embedded and/or real-time systems - confront researchers and engineers with completely new challenges. Measurement and signal processing systems are involved in almost all kinds of activities in that field where control problems, system identification problems, industrial technologies, etc., are to be solved, i.e., when signals, parameters, or attributes must be measured, monitored, approximated, or determined somehow. In a large number of cases, traditional information processing tools and equipment fail to handle these problems. Not only is the handling of previously unseen spatial and temporal complexity questionable but such problems have also to be addressed such as the interaction and communication of subsystems based on entirely different modeling and information expression methods, the handling of abrupt changes within the environment and/or the processing system, the possible temporal shortage of computational power and/or loss of some data due to the former. Signal processing should even in these cases provide outputs of acceptable quality to continue the operation of the complete system, producing data for qualitative evaluations and supporting decisions. It means the introduction of new ideas for specifying, designing, implementing, and operating sophisticated signal processing systems. Intelligent - artificial intelligence, soft computing, anytime, etc. - methods are serious candidates for handling many theoretical and practical problems, providing a better description, and, in many cases, are the best if not the only alternatives for emphasizing significant aspects of system behavior. These techniques, however, are relatively new methods and up until now, not widely used in the field of signal processing because some of the critical questions related to design and verification are not answered properly and because uncertainty is maintained quite differently than in classical metrology. After the initiation of the 1999 IEEE International Workshop on Intelligent Signal Processing, WISP'99, which was the first event to start linking scientific communities working in the fields of intelligent systems and signal processing and hoping that it will attract more and more scientists and engineers in these hot topics, this special issue continues this pioneering work by offering a selection of nine papers fitting into the profile of the journal from the numerous high quality ones presented at WISP'99. These excellent papers deal with different aspects of advanced computational intelligence in signal processing, including the application of neural networks, fuzzy techniques, genetic and anytime algorithms in modeling, signal processing, noise cancellation, identification, and pattern recognition, multisensorial information fusion and intelligent classification in image processing, exact and nonexact complexity reduction, and nonclassical and mixed data and uncertainty representation and handling. As an editor of this special issue, I would like to express my thanks to all of the contributors and my belief in that the excellent research results it contains provide the basis for further strengthening and spreading of advanced computational intelligence in signal processing opening wide possibilities for new theoretical and practical achievements.

Download Full-text