SUNConferences, 27th Annual Conference of the SA Institute for Industrial Engineering

Font Size: 
Lungile Nyanga, Andre Francois Van der Merwe, Nelson Masanganise, Stephen Matope, Mncedisi Trinity Dewa

Last modified: 2016-10-13


As manufacturing companies thrive to achieve global competitiveness, effective plant monitoring and short decision making time has proven to be a big contribution for them to achieve their goal. To cut operational costs and offer a competitive edge, the companies need not only to capture equipment failure but to predict the failure occurrence. Relevant decisions also have to be made in the shortest possible time in the case of a failure. The paper discusses an analytical decision support tool for monitoring a generator thrust bearing of a hydro power station. The system consists of a distributed Multi-Agent System (MAS) comprising of diagnosis, decision alarm, database and distribution agents. The diagnosis agent is responsible for obtaining the parameter readings from the sensors using a Siemens S7-1200 PLC and Kepware OPC server. The decision alarm agent uses a Java Expert System Shell (JESS) rule base to monitor the parameters obtained from the diagnosis agent and make decisions on equipment availability using the Maintenance Free Operating Period (MFOP) concept. The desired action and equipment performance trends are then conveyed to the user by the distribution agent. The decision support tool is developed using the Java Agent Development framework (JADE). The proposed system showed an increase in machine availability and reduction in the maintenance scheduling time and parts inventory.


Multi-Agent System, JESS, JADE