An implementation of an analog neural network trained to identify signatures from the AC electrical power system on the Space Shuttle Orbiter is described. This demonstration project shows that a small stand alone system in the form of a VME-module can be designed, constructed and tested within days, provided a proper set of training vectors are available.