A wavenet-ES integrated fault diagnosis system for cracking furnace

A kind of integrated FDD approach based on wavenet and ES was studied for the practical application in industrial cracking furnace. The FDD system was based on the data from field-site. It was constructed as several independent wavelet sub-networks designed for each kind of process fault, and all sub-networks´ outputs were sent into the decision fusion system (ES) to give an integrated diagnosis result for the cracking furnace process. Deep process knowledge of experts and operators was incorporated in the decision fusion system to support reasonable decision results. Simulation tests were conducted for each kind of faults. The result shows that the system has high confidence, robust, and good practicability.