Self-Organizing Maps with Unsupervised Learning for Condition Monitoring of Fluid Power Systems

The goal of this paper is to study a proactive condition monitoring system for fluid power systems where the Self-Organizing Maps (SOM) with unsupervised learning is used to classify and interpret high-dimensional data measurements. If all the damages are not assumed to be known before diagnostics, an ordinary neural network with supervised learning for their detection can not be used. Operation of the proactive condition monitoring system is tested in a test system where two fault types are used. The test system is run in normal and two different fault situations. Measurement results are used for training and testing the SOM. In this paper these measurement results and also the quality of state recognition are shown.