Camps Echevarría, L., Llanes Santiago, O. e Silva Neto, A. J., A Bioreactor Fault Diagnosis Based on Metaheuristics, Capítulo 7, pp. 139-163, em Computational Intelligence, Optimization and Inverse Problems with Applications in Engineering, (Eds.), ISBN: 978-3-319-96432-4, Editora Springer, 2018
Fault Diagnosis is a very important issue in the industry. Some essential topics in the industry, e.g. reliability, safety, efficiency, and maintenance, depend on the correct diagnosis of systems. Robustness in relation to external disturbances, which may affect the system, sensible to incipient faults, and a proper diagnosis time are desired characteristics of the diagnosis, in order to prevent propagation of faults. In the particular case of the chemical and biochemical industries, the use of nonlinear bioreactors is common. Therefore, the diagnosis of these systems is of high importance for both industries. This chapter presents the application of three metaheuristics, Ant Colony Optimization with Dispersion (ACO-d), Differential Evolution with Particle Collisions (DEwPC), and Covariance Matrix Adaptation Evolution Strategy (CMA-ES), in the diagnosis of a nonlinear bioreactor through a Fault Detection and Isolation (FDI) inverse problem approach. This technique deals with the solution of an optimization problem, which is solved with the help of these three metaheuristics. The analysis of the quality of the diagnosis is based on the robustness and diagnosis time. Furthermore, the results are compared with other reported ones in the literature. The main contributions of this chapter are, at first, a proposal for collecting information regarding the quality of the diagnosis based on the FDI inverse problem approach and the use of metaheuristics, as well as the organization of this information in tables. Furthermore, it is shown how to improve the stopping criteria of the metaheuristics, when they are applied to FDI inverse problems.
DOI: https://doi.org/10.1007/978-3-319-96433-1_7