Monitoring the operational condition of a diesel engine by evaluating the parameters of its thermodynamic operation cycle

The great challenge in the field of combustion monitoring and fault diagnosis in diesel engines is understanding the dynamic interactions that reveal the trajectory between the causes located in engine items and their effects or measurable symptoms to establish the rules that can generate diagnoses. In this context, the fault tree analysis (FTA) technique was applied to transform the engine and its subsystems into a structured logical diagram, in which were arranged the various combinations of failures in the engine component items that possibly may lead to undesirable events. With the set of diagnostic rules generated by the FTA serving as a knowledge base, inferences were made by an artificial neural network created to provide the operational condition of the engine submitted to tests. Despite the value errors and knowing that a perfect classification model must have values close to the expected values, it was very easy to distinguish the simulated operating conditions, which makes the diagnosis very consistent when compared with the expected results.