This thesis deals with an application of discriminant analysis to equipment diagnosis and preventive maintenance policy. Assuming that the equipment profile vector follows a multivariate normal distribution, the reliability of the equipment which can be easily computed via a computer program is obtained. Then, using the data obtained concomitantly from the program, we formulate two important maintenance problems, one of which is maximizing the reliability of the equipment under maintenance budget constraint and the other is minimizing the maintenance cost under reliability requirement. For the two problems, we give simple rules which guarantees optimal solutions and a numerical example is presented to demonstrate these processes.
