Program: UFSC / POSMEC
Date: August, 2017
Advisor: Jonny Carlos da Silva
Abstract: From the advancement of medicine in the development of powerful drugs requiring a controlled dose, it is necessary to develop systems that aim to increase the reliability during the infusion of such drugs. For this, Artificial Intelligence has become a widespread area in the detection of failures, receiving several advances that, when applied, can assist the health professional during the malfunction of the equipment. Motivated by the reduction of failures in an infusion system, this work studies the main faults and possible symptoms that can be diagnosed from the signal of typical components of an infusion system like: pressure sensor, drop counter sensor and air bubble sensor. Thus, an Expert System (SE) is proposed in order to monitor the infusion and interpret the symptoms of failure, making inferences about the monitored signals. Sensor signals and faults are generated by the SE, which are detected through the Limit Check and the Sequential Probability Ratio Test (SPRT). For this work 6 types of failures were studied with 2 different start times. The diagnosis was based on a Fault Tree Analysis (FTA) also presented in this work. For the validation of the work, health professionals tested the SE in order to verify the consistency of the causes generated according to system symptoms.
Key-words: Knowledge-based system, Fault Detection, Infusion Systems.
Reference: PAGATINI, G. Sistema Especialista Protótipo para Diagnóstico de Falhas em Equipamentos de Infusão. 2017. 116 p. Dissertação (Mestrado em Engenharia Mecânica). Universidade Federal de Santa Catarina, Florianópolis.