With the constant evolution of e-manufacturing technologies, there is a clear trend for e-maintenance that involves the integrating of ICT (Information and communication technologies) within the maintenance strategy. This leads to highly sophisticated and complex machinery, which increases the demand for expertise. Unfortunately, a company could always lose the expertise due to experts’ retirement, change of occupation or death. This motivates us in this work to develop an e-maintenance model that enables organizations to exploit expert’s knowledge in the process of machine fault diagnosis. This paper focuses on the building of a Knowledge-Based System (KBS) in order to capture the experts’ knowledge to be permanently kept and cannot be disparaged due to lack of practice. An optimal AI-based tool is proposed that aims at accurate values to retrieve information from KBS, which describes the alarms to diagnose the failure of the machine. An accurate analysis is carried out that yields insight into the impact of KBS on the ability of fault diagnosis. The results illustrate the high-performance of the proposed approach in handling the KBS's data associated.