Searching for solutions to manufacture industries, which every day deal with problems of faults in their process, that generate economics and humans main losses, an algorithm to construct a Petri Nets based model and diagnoser to isolate and fault detection of Discrete Events Systems is presented. This algorithm is developed in a real process of liquids packaging, where we can see that its implementation allows detecting individuals, simultaneous and dependents faults. The process to construct the model and diagnoser is systematic and useful, and it reduces the problems of combinational explosion, which is the main problem present in other investigations. This research has an excellent projection in Intermittent Faults Diagnosis and Hybrids Systems.