Sistema experto para la definición de umbrales adaptivos en procedimientos de mantenimiento en transmisión eléctrica

Abstract

In the work the design of a connectionist expert system is realized, which uses tools of the artificial intelligence that allow the decision making for the establishment of adaptive thresholds as classifiers of the type of maintenance to be used. For the selection of a maintenance in electrical systems in general, parameters and conditions are taken into account that facilitate an adequate selection of the maintenance to be applied, for that reason the incidences of a failure, are limited in selection ranges of a maintenance for each incidence ; From 0 - 0.49 preventive maintenance is considered, taking into account the technology of the material to be used, the experience acquired by the personnel in charge of maintenance and considering the presence of a passive hot spot in all the electrical installations in full operation; Of 0.5 -0.69 is considered a predictive maintenance, with presence of an active hot spot, in this maintenance an additional range of 0.7 - 0.79 is considered considering a proactive maintenance due to the possibility of some failure, Of 0.8 - 1.0 is considered a corrective maintenance, for the equipment change. It builds the system starting from the acquisition of the data for the construction of the database, the machine inference expert system based on the operators' expertise and the human machine interface in a comfortable, friendly and reliable manner. The results are displayed on the computer screen, and the connection system database is available for other applications.
 Index Terms— artificial intelligence, expert system, failures, maintenance, classifiers.