SUMMARY
The fault diagnosis process in Computer Numerical Control (CNC) hydraulic machines for steel processing relies on skills, experiences, and maintenance technicians' understanding of the machine. The problem is many junior maintenance technicians are inexperienced and unskilled. This paper proposes a conceptual framework for a fault diagnosis system for the CNC hydraulic machine to help a maintenance technician in a fault diagnosis process. The framework uses association rule mining to discover hidden association patterns between fault symptoms and causes from historical machine fault data. The framework has consisted of data standardization, knowledge acquisition, and a model of the fault diagnosis system. The data standardization aims to make the data ready to be mined by assigning a fault tag for each record of historical fault data. The tagged repair records are used to produce symptoms–cause associative knowledge. The produced knowledge is refined by corrective actions acquired from expert knowledge. The knowledge is then stored in the fault knowledge database in the form of IF-THEN rules. The reasoning machine is developed to map the fault symptoms as IF and the causes as THEN. Production operators can fill in the fault symptoms by choosing the standardized fault symptom tag. When a maintenance technician reviews a fault report, the system, through a reasoning machine, will access the appropriate IF-THEN rules based on the fault symptoms that the production operator has filled in. The system concludes the fault cause and recommends suitable corrective action.