Control and filtering of Markovian jumping parameters stochastic systems

Abstract

Main objectives the motivation for this project is to provide special infrastructure and support to interactions among well established research groups. These groups present analogous research activities and substantial research production in the proposed field. It is important to indicate the occurrence of previous scientific cooperation among the main researchers in this project. The groups reside at UNICAMP, Faculdade de Engenharia Elétrica e de Computação, at Escola Politécnica da USP, Departamento de Engenharia Elétrica and at Laboratório Nacional de Computação Científica LNCC-CNPq. The project also anticipates interactions with researches at USP-ICMC, Departamento de Computação e Estatística, and at Universidade Federal do Espírito Santo, Departamento de Engenharia Elétrica. The project comes as recognition to the synergy of combined research efforts, it is motivated by the special support needed to magnify these efforts and to strengthen the goals attained by the groups involved. Special objectives the main line of research is devoted to the study of Markovian stochastic systems with parameter jumps, seeking stability and the solution of the control and filtering problems. These systems have been receiving broad attention in the last decades regarding both, theoretical and applications aspects. There are many situations of practical interest stressing the importance of the study of dynamic systems subject to abrupt variation in their structures. Very frequently, systems are intrinsically vulnerable to failures in their components, subject to uncertain period of repairs, sudden changes in the environmental or in connections among subsystems, sudden changes of operating points in non-linear industrial plants, etc. One especially important class of stochastic systems that suits the purpose of modeling many of the situations described above are the Markovian processes subject to parameter jumps. The Markovian structure is important from the analytical viewpoint, but it also directly adequate to many interesting phenomena in Engineering, Networks, Operations Research, Biology, etc. In particular, in the System and Control area, the class of Markovian jump linear systems (MJLS) is considered very important and representative of the Hybrid Systems class. The MJLS are seen by many researchers as a viable alternative to attain Fault Tolerant behavior in Control Systems, a theme that is high relevant in the Control Theory field (within the so-called safety-critical and high-integrity systems in the specialized literature)... (AU)