Abstract
A new method of leakage detection in pneumatic or mechanic components was studied on the basis of the estimate of the leakage by the model knowledge of the component under test.
Important results were achieved like a drastic reduction of the time necessary to the measurement, a good repeatability, excellent disturbance rejection thanks, moreover, to the knowledge of the statistical characteristics both of the measured signal and of the process under test .
Furthermore, the knowledge of process parameters by estimate is a straightforward and reliable methodology which permits a good adaptability to the changes of the process under test by a fit of the process model.
Also The new methodology is suitable to extract the model parameters by a test on a good component which represents a master device. Moreover, a genetic algorithm, can optimize the model parameters to obtain the best estimate response, and the fastest convergence.
A ratio at least of 5:1 in terms of time response was achieved respect to the standard methods.