Prognostics using morphological signal processing and computational intelligence

A procedure is presented for monitoring and prognostics of machine conditions using computational intelligence (CI) techniques. The machine vibration signals are processed using morphological operations to extract an entropy based feature characterizing the signal shape-size complexity for assessment of machine conditions. An evolutionary average entropy of the system is introduced as the dasiamonitoring indexpsila for prognostics of the system condition. The progression of the dasiamonitoring indexpsila is predicted using CI techniques, namely, recursive neural network (RNN), adaptive neuro-fuzzy inference system (ANFIS) and support vector regression (SVR). The proposed prediction procedures have been evaluated through benchmark datasets. The prognostic effectiveness of the CI techniques has been illustrated through vibration dataset of a helicopter drivetrain system gearbox. The performances of ANFIS and SVR have been found to be better than RNN for the dataset used. The results are helpful in understanding the relationship of machine conditions, the corresponding indicating feature, the level of damage/degradation and their progression.