An Artificial Immune System for Fault Detection

The oil well instrumentation generates a set of process variables, which must analyzed by the experts in order to determine the well state. That implicates a highly cognition task where the information generated is very important for maintenance tasks, production control, etc. In other way, the natural energy of an oil field can not be enough to lift the fluids. In these case is necessary to use another procedure to lift the oil, for example gas. That is an interesting case to be modeled by an artificial intelligence technique. Particularly, in this paper we propose an Artificial Immune System for fault detection in gas lift oil well. Our novel approach inspired by the Immune System allows the application of a pattern recognition model to perform fault detection. A significant feature of our approach is its ability to dynamically learning the fluid patterns of the ‘self’ and predicting new patterns of the ‘non-self’