As the scope of automation in the manufacturing process is increasing every day, maintaining a fault-free production line becomes a regular standard to be followed. Hence early detection and diagnosis of fault situation play a very important role. Hence comes the need to build an intelligent fault diagnosis system. The thesis work uses two heterogeneous data sources as the knowledge source to extract information about fault events namely Failure Mode and Effects Analysis (FMEA) and digital logbook. The intelligent system uses Semantic Web Technologies (SWT) to represent the knowledge contained in the two sources. FMEA describes the possible fault scenarios which can occur in a manufacturing domain. The knowledge is prepared by domain experts to give the working personnel an idea of possible faults and their preventive measure. On the other hand, a digital logbook contains information about the real-time fault situation and the entries are maintained by workers. It often contains data in natural language form. The primary goal is to build a unified knowledge source from the semi-structured and unstructured data. This is achieved by building an ontology that focuses on the semantic relationship between the data elements. Next, to support automated reasoning, rules have been defined and reasoning task is performed via a rule engine in the ontology editor tool on the knowledge base instances. This method helps to establish an inference that can express the implicit knowledge present in the knowledge base. With the help of automated reasoning, the working personnel can be informed about the possible cause and remedy for the failure. An automatic translation from conceptual model to ontology model and automatic ingestion of data points in the knowledge base can accelerate the usability of the knowledge base. In addition, the knowledge base uses priori information determining the severity of the failure to detect the probable cause and countermeasures. The results achieved from different experiments prove the validity of the proposed solution. In addition to reasoning, several querying options can fetch the relevant fault cause and its controlling measure too. Apart from the contributions mentioned earlier, quantifying knowledge of digital logbook was a challenging task, which is overcome by applying heuristics to determine the priori information for the fault scenario.