This PhD thesis concerns the development, implementation and experimental validation of methods for monitoring, diagnostics and quality control of gear systems, based on advanced and innovative techniques of experimental vibration analysis and signal processing, with particular focus on planetary gearboxes. The analysis of the vibrations in this type of gearboxes is complicated by the movement of the planets, that causes the continuous variation of the distance - and consequently of the transfer functions - between the sources of vibration (gear meshing and bearings) and the fixed transducer mounted outside the annulus. For diagnostic purposes, it is necessary to extract from the overall vibration signal the contribution of each gear by means of a synchronous averaging procedure (Planetary TSA), which, in the case of planetary gear trains, has to eliminate the effect of the variation of the transfer functions. This has been obtained by means of a complex algorithm, already proposed in the literature, but requiring further efforts for developing and implementing with specific measures to make it applicable and efficient in an industrial environment. As condition indicators (CI’s) for monitoring of planetary gear trains, numerous TSA statistical parameters have been studied, based on the results of experimental campaigns conducted on a 3 stage planetary gear unit of nominal power of 3 kW; these tests may be summarized as follows: a) tests on bench, under varying load conditions, in the presence of artificial localized defects on planet teeth; b) accelerated wear test on back-to-back test bench, arriving at life end (700 h); c) testing analogous to b), interrupted at 250 h, when the value of the CI’s has indicated the development of pitting; d) tests on bench with different loads, by mounting sun gears in good conditions or worn in the previous tests b) and c). On the basis of these tests, it has been evaluated the effectiveness of metrics proposed in literature; in addition, some original metrics have been developed, more effective and proper for the specific application in industrial environments. In particular, the results show that three original metrics are effective in identifying the onset of pitting and its subsequent development, even with varying load and speed: the variance of the residual signal RV, the corresponding cumulative variance, CRV (average of the previous one at the current time) and an improvement of the already known NA4, NA4mod, defined as the ratio between the forth statistical moment of the residual signal and the current run time average of the squared variance of the regular signal. In addition, a software dedicated to the vibration analysis for diagnostic purposes of planetary gearboxes has been developed; it implements the studied techniques and is suitable for use in industrial environments; it is worth noting that there are no commercial products with similar functions. This software includes modules for the acquisition of vibratory signals, filtering, time and angular resampling, TSA, identification of the phase of the planets, synchronous average for planetary gear units (Planetary TSA), extraction of residual, difference and regular signals, Fourier transform, PSD, Envelope, amplitude and phase demodulation, calculation of statistical parameters and specific metrics for gear diagnostics. The algorithms, initially implemented in Matlab, have required special measures to increase efficiency in the case of large amounts of data. In the final version, the software has been implemented in the ANSI C language with a reduction of computation time by about 50%, greater stability, better memory management and a reduction of costs for the end user. The diagnostic efficiency of the implemented algorithms has been verified by the application to numerous experimental data, generally obtaining satisfactory results.

Questa Tesi di Dottorato di ricerca concerne lo sviluppo,l’implementazione e la validazione sperimentale di metodologie per il monitoraggio,la diagnostica ed il controllo qualità di rotismi,basate su tecniche avanzate ed innovative di analisi sperimentale delle vibrazioni e di analisi del segnale,con particolare riferimento ai rotismi epicicloidali.L’analisi delle vibrazioni in questo tipo di rotismi è resa complessa dal movimento dei satelliti che fa variare continuamente la distanza –e conseguentemente le funzioni di trasferimento- tra le sorgenti di vibrazioni (ingranamenti e cuscinetti) ed il traduttore fisso all’esterno della corona.Ai fini diagnostici,è necessario estrarre dal segnale vibratorio complessivo il contributo di ciascuna ruota dentata mediante una procedura di media sincrona (Planetary TSA),che, nel caso dei rotismi epicicloidali,deve depurare l’effetto della variazione delle funzioni di trasferimento.Ciò è stato effettuato mediante un complesso algoritmo,già proposto in letteratura,che è stato necessario sviluppare ulteriormente ed implementare con accorgimenti specifici per renderlo applicabile ed efficiente in ambiente industriale.Come indicatori di condizione (CI) per il monitoraggio dei rotismi epicicloidali si sono studiati numerosi parametri statistici della TSA,sulla base dei risultati di campagne sperimentali condotte su un riduttore epicicloidale a 3 stadi di potenza nominale 3 kW che possono riassumersi in: a) prove su banco,al variare del carico,in presenza di difetti localizzati artificiali sulla dentatura dei satelliti; b) prova di usura accelerata su banco prova in configurazione back-to-back portata a fine vita (700 h); c) prova analoga alla b),interrotta a 250 h,quando il valore degli CI ha segnalato lo svilupparsi del pitting; d) prove su banco,al variare del carico,montando sia solari nuovi sia usurati nelle precedenti prove b) e c).Sulla base di tali prove sperimentali,è stata valutata l’efficacia delle metriche proposte in letteratura e ne sono state sviluppate di originali,maggiormente efficaci e rispondenti all’applicazione specifica in ambienti industriali.In particolare,tre metriche originali sono risultate efficaci nell’individuare l’insorgere del pitting ed il suo successivo sviluppo,anche al variare del carico e della velocità: la varianza del segnale residuo RV,la corrispondente varianza cumulativa CRV (media al tempo corrente della precedente) e una variazione migliorativa del già noto NA4,NA4mod, definito come rapporto tra il momento del quart’ordine del segnale residuo e la media corrente del quadrato della varianza del segnale regolare. Inoltre,è stato sviluppato un software dedicato all’analisi delle vibrazioni a fini diagnostici di rotismi epicicloidali,che implementa le tecniche sviluppate ed è atto ad essere impiegato in ambiente industriale; si rileva che non esistono prodotti commerciali con analoghe funzioni.Il software contiene moduli per acquisizione dei segnali vibratori,filtraggio,ricampionamento in base tempo e in base angolo,TSA,individuazione della fase dei satelliti,media sincrona per gli epicicloidali (Planetary TSA),estrazione dei segnali residui,differenza e regolari,trasformata di Fourier,PSD,Envelope,demodulazioni di ampiezza e fase,calcolo di parametri statistici e metriche specifiche per gli ingranaggi.Gli algoritmi,inizialmente implementati in Matlab,hanno richiesto accorgimenti particolari per incrementarne l’efficienza,a fronte della necessità di trattare grosse quantità di dati.Nella versione finale il software è stato implementato in linguaggio ANSI C con una riduzione dei tempi di calcolo di circa il 50%,maggiore stabilità,migliore gestione della memoria e una riduzione dei costi per l’utilizzatore finale.L’efficacia diagnostica degli algoritmi implementati è stata verificata mediante l’applicazione a numerosi dati sperimentali,ottenendo risultati complessivamente soddisfacenti.

DIAGNOSTICA VIBRAZIONALE DI RIDUTTORI EPICICLOIDALI IN AMBIENTE INDUSTRIALE: PROCEDURE, IMPLEMENTAZIONE E VALIDAZIONE

BERLATO, Francesco
2017

Abstract

This PhD thesis concerns the development, implementation and experimental validation of methods for monitoring, diagnostics and quality control of gear systems, based on advanced and innovative techniques of experimental vibration analysis and signal processing, with particular focus on planetary gearboxes. The analysis of the vibrations in this type of gearboxes is complicated by the movement of the planets, that causes the continuous variation of the distance - and consequently of the transfer functions - between the sources of vibration (gear meshing and bearings) and the fixed transducer mounted outside the annulus. For diagnostic purposes, it is necessary to extract from the overall vibration signal the contribution of each gear by means of a synchronous averaging procedure (Planetary TSA), which, in the case of planetary gear trains, has to eliminate the effect of the variation of the transfer functions. This has been obtained by means of a complex algorithm, already proposed in the literature, but requiring further efforts for developing and implementing with specific measures to make it applicable and efficient in an industrial environment. As condition indicators (CI’s) for monitoring of planetary gear trains, numerous TSA statistical parameters have been studied, based on the results of experimental campaigns conducted on a 3 stage planetary gear unit of nominal power of 3 kW; these tests may be summarized as follows: a) tests on bench, under varying load conditions, in the presence of artificial localized defects on planet teeth; b) accelerated wear test on back-to-back test bench, arriving at life end (700 h); c) testing analogous to b), interrupted at 250 h, when the value of the CI’s has indicated the development of pitting; d) tests on bench with different loads, by mounting sun gears in good conditions or worn in the previous tests b) and c). On the basis of these tests, it has been evaluated the effectiveness of metrics proposed in literature; in addition, some original metrics have been developed, more effective and proper for the specific application in industrial environments. In particular, the results show that three original metrics are effective in identifying the onset of pitting and its subsequent development, even with varying load and speed: the variance of the residual signal RV, the corresponding cumulative variance, CRV (average of the previous one at the current time) and an improvement of the already known NA4, NA4mod, defined as the ratio between the forth statistical moment of the residual signal and the current run time average of the squared variance of the regular signal. In addition, a software dedicated to the vibration analysis for diagnostic purposes of planetary gearboxes has been developed; it implements the studied techniques and is suitable for use in industrial environments; it is worth noting that there are no commercial products with similar functions. This software includes modules for the acquisition of vibratory signals, filtering, time and angular resampling, TSA, identification of the phase of the planets, synchronous average for planetary gear units (Planetary TSA), extraction of residual, difference and regular signals, Fourier transform, PSD, Envelope, amplitude and phase demodulation, calculation of statistical parameters and specific metrics for gear diagnostics. The algorithms, initially implemented in Matlab, have required special measures to increase efficiency in the case of large amounts of data. In the final version, the software has been implemented in the ANSI C language with a reduction of computation time by about 50%, greater stability, better memory management and a reduction of costs for the end user. The diagnostic efficiency of the implemented algorithms has been verified by the application to numerous experimental data, generally obtaining satisfactory results.
DALPIAZ, Giorgio
File in questo prodotto:
File Dimensione Formato  
Tesi_Berlato_Unife_ScienzeIngegneria_CicloXXIX.pdf

accesso aperto

Descrizione: Tesi
Tipologia: Tesi di dottorato
Dimensione 20.02 MB
Formato Adobe PDF
20.02 MB Adobe PDF Visualizza/Apri

I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11392/2487841
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact