A new method of plotting and navigating self organising maps for improved condition monitoring and prognosis

This paper proposes a powerful new method for plotting self organising maps which enables more effective convergence analysis for training algorithms, and facilitates clear identification and distinction of major and minor class or feature boundaries. A simple example is presented to show how these boundaries enable direct interactive and automatic classification, without an output classification layer, and facilitate adaptive trajectory analysis for temporal condition monitoring with inherent time to failure prediction capabilities.