Data based tools for sensors continuous monitoring in industry applications

This paper presents a 10 years experience of data driven models for sensor validation applied for petroleum and natural gas industry. Auto-associative kernel regression has been used as the main modeling method. The models achieved were embedded in software called Sentinell, which is used for sensors diagnosis. The software is being used in a natural gas compression station, and it has been evaluated in other industries such as: refineries, offshore petroleum platforms, and thermoelectric power plants. In this work the theoretical background is presented, as well as the performance metrics indexes used to evaluate the models. The developed methodology and the results in the real plants are presented and discussed. The experience of these previous works might open future applications in high reliability automated processes.