Application Of Mathematical Morphology To Fault Diagnosis Of Rotary Machines

With the development of science and the aggravation of world economic competition,modern industry puts forward higher requirements for mechanical equipment, not only instable and high production efficiency, but slso in secure and sustained production capacity.The working state of hydraulic pump and rolling bearings has great influence on the wholeequipment, for hydraulic pump is the core of hydraulic system and rolling bearings arevulnerable parts of equipment which are widely used. Thus, it is very necessary to monitorand diagnose the hydraulic pump and rolling bearings of the concerned locations. In axialpiston pump and rolling bearings, the fault signal is usually performed as nonlinear andnon-stationary, showing complex motion characteristics. As a nonliner signal processingmethod, mathematical morphology method has its unique advantage in dealing with thiskind of signals.Firstly, according to the problem in selecting the scale of structure element insingle-scale morphology, a composite evaluation index is used for the optimizationcalculation in this paper, based on the discussion about the factors of morphological filter.The simulated signal and hydraulic pump fault signal are analysed, then the result showsthat the method can extract shock characteristic clearly.Secondly, for the single-scale morphology has too many requirements when selectingthe scale of the structural element, a multi-scale method for signal filtering is used in thispaper. In the condition of high sampling frequency and intensity noise, the existingadaptive multiscale morphological method can not get an ideal result. Therefore thethreshold constrained adaptive multiscale morphological difference (TCAMMA) methodis proposed to solve the problem. The simulated signal and pump fault signal are analysedto verify the validity of this method. The result demonstrates that, the method canovercome the lack of the original adaptive multiscale morphological analysis methods andget an ideal result.Finally, the iterative adaptive multiscale morphological difference (IAMMA) methodis proposed to solve the problem in the existing adaptive multiscale morphological methodin a different way. The simulated signal and roller bearings signal which is collected from the Case Western Reserve University are analysed to verify the validity of this method.The result demonstrates that, the method can get an ideal result similarly.