Automation of Detection and diagnosis of malfunctions of static and dynamic systems using semiqualitative knowledge, DPI2003-07146-C02-01, 2003-2006.
This project is funded by the Spanish Ministry of Science and Education and ERFD/FEDER, and aims at performing research in computer science with focus on: making advances in fault detection and model-based diagnosis for semiqualitative systems, solving complex problems at a higher efficiency and accuracy. We this project we intend to provide concrete results for the research community with the development of software facilities for fault detection and diagnosis of this systems.
Manager: Rafael M. Gasca
Researchers: Pedro Abad, Rafael Ceballos, María Teresa Gómez, Luis González, Antonio Márquez, Sergio Pozo, Fernando de la Rosa, Antonio J. Suárez, Miguel Toro Bonilla, Carmelo del Valle, Pablo Neira
Project summary: Detection and diagnosis of devices or processes malfunctions is nowadays a very active research topic. Faults can originate undesirable stops and damages in the systems, thus increasing the costs and decreasing the production. Furthermore, faults can have an important negative impact over the environment which must avoided if possible. Therefore, the development of mechanisms to detect and diagnose faults as soon as possible is necessary in order to keep the systems in the desired levels of security, production and reliability, it helps the companies to survive in a market where the competitiveness is always increasing.
In the last twenty years, fault detection and isolation has been based mainly in the use of models. A model of the system is used to detect the faults and to isolate the subsystems or the components that possibly originated them. Usually the models used for fault detection and diagnosis are quantitative or qualitative, but very little research has been made in the field of semiqualitative models. These models combine uantitative and qualitative knowledge so they are very useful when there is uncertainty or incomplete knowledge. The usual engineering models contain semiqualitative functions, constraints or data and structural changes depending on the inputs. This project proposes the application of these models to fault detection and diagnosis by means of techniques from Artificial Intelligence and Systems' Engineering. Moreover, different approaches to optimally correct the detected fault and allow the system to work according to the specifications determined by the engineers will be considered
On the other side, very often a model of complex systems is not available and only experimental data can be used. There is little research about this topic. This project proposes the automatisation of fault detection and diagnosis in these systems by means of rules obtained using learning techniques.. Finally, more and more the systems of this global world are distributed spatially and semantically across the networks. Fault detection and diagnosis of faults in these systems is also important. This is a complex task due to the special characteristics of these systems. This task will be tackled in this project using intelligent software agents.