故障診断に対する2乗距離法の適用

抄録

In recent years, research and development in the field of ship automation and labor-saving have progressed in countries all over the world, and some practical systems have already been developed. But the diagnosis of the condition of diesel engines is very difficult on account of their complex mechanisms, and many of the practical systems for diesel engines are monitoring systems, such as“DATA LOGGER”. With existing engine condition monitoring systems, it is very difficult to monitor the cylinder pressure of a diesel engine continuously. The method of monitoring the cylinder pressure still largely depends on the ability of skilled engine operators to judge the problems in the components around the combustion chamber. In view of this, we have studied the application of the multivariate analysis method to the diagnosis of the components around the combustion chamber by using various characteristics derived from the cylinder pressure. The current engine condition and each failure can be expressed by vectors having several characteristics. It follows naturally that the engine failure can be discriminated by calculating the statistical distances between the vectors of the current engine condition and the vectors of each failure, and then by selecting the engine condition vector which is most similar to the failure vector.