Dynamic model of a cylindrical roller on a rough surface, with applications to wind turbine gearbox planetary bearings

Wind turbines of larger power ratings have become increasingly prevalent in recent years, improving the viability of wind energy as a sustainable energy source. However, these large wind turbines have been subjected to higher rates of failure of the wind turbine gearbox, resulting in larger downtime of operation and an increase in cost due to repairs. These failures most frequently initiate in the gearbox’s bearings, especially in the planetary bearings of the planetary stage and high-speed bearings. Currently, most of the research on the detection of planetary bearing faults only addresses the case of localised faults in the outer bearing race, while fewer research considering the detection of distributed bearing faults. The research that does consider distributed bearing faults relies on techniques – such as machine learning for the identification of faulty bearings – that do not account much for the underlying physics of the bearing. In this paper, a model is developed to simulate and analyse the dynamic interaction of a planetary bearing in the presence of surface roughness, which can be used to represent a distributed fault. The model presented uses random vibration theory for simulating the response of the planet bearing induced by distributed faults. The input of the model considers statistical expressions of the roughness geometry using multiple parallel tracks. Numerical simulation of the random vibration of the model is performed using 16 tracks, and the power spectral density of the radial deflection of the roller and the roller–race contact force is determined. The results of the simulation with the multi-track model show that a single-track model significantly overestimates the power spectral densities, and also suggests the stiffness of the bearing race is too high to have an effect on the roller dynamics for a planet bearing.

The application of the cyclic coherence for distributed planet fault detection in planetary gears

The vibration-based condition monitoring of planetary gears is a highly active and challenging field of research. Many signal processing techniques have been proposed, with the aim of promoting the fault component in the signal and, consequently, highlighting the fault signature (ie the damage symptom). Most of these techniques consider the fault contribution as being deterministic, which is true in the case of an advanced and localised fault. Such techniques may fail in other scenarios in which, for instance, the fault is of a distributed nature and its vibrational component is weak. In such a case, the fault component is likely to be randomised and turns cyclostationary. The present paper suggests the presence of an additional cyclostationary component in planetary gear vibrations. The presence of this component is explained by the presence of load fluctuations at the meshing points and the random micro-irregularity in the stiffness of gear components, as well as the presence of distributed faults. A simplified signal model is proposed to explain the vibration signal structure of healthy and faulty planetary gears (with a planet fault in the case of the faulty gear). Also, a cyclostationary-based condition monitoring approach is proposed, based on the cyclic coherence. The proposed approach is validated on real-world vibration signals acquired from a planetary gear benchmark.

Application of Cyclostationary Indicators for the Diagnostics of Distributed Faults in Ball Bearings

This paper deals with the detection of distributed faults in ball bearings. In literature most of the authors focus their attention on the detection of incipient localized defects. In that case classical techniques (i.e. statistical parameters, envelope analysis) are robust in recognizing the presence of the fault and its characteristic frequency. In this paper the authors focalize their attention on bearings affected by distributed faults, due to the progressive growing of surface wear or to low-quality manufacturing process. These faults can not be detected by classical techniques; in fact, in this case the signal does not contain impulses at the fault characteristic frequency, but more complex components with strong non-stationary contents. Distributed faults are here detected by means of advanced tools directly derived from the theory of cyclostationarity. In particular three metrics — namely Integrated Cyclic Coherence (ICC), Integrated Cyclic Modulation Coherence (ICMC) and Indicator of Second-Order Cyclostationarity (ICS2x) — have been calculated in order to condense the information given by the cyclostationary analysis and to help the analyst in detecting the fault in a fast fault diagnosis procedure. These indicators are applied on actual signals captured on a test rig where a degreased bearing running under radial load developed accelerated wear. The results indicated that all the three cyclostationary indicators are able to detect both the appearance of a localized fault and its development in a distributed fault, whilst the usual approach fails as the fault grows.

Gearbox Systems Dynamic Modelling for Diagnostic Fault Detection

The paper deals with mathematical modelling and computer simulation of a gearbox driving system with a double stage gearbox. Mathematical modelling and computer simulations are used for supporting diagnostic inference. Vibration is thought of as a signal of gear condition. It is stressed that vibration generated by gears is influenced by many factors. These factors are divided into four groups: design, production technology, operational, condition change. The condition change of a gearbox is given by gear faults that are divided into single faults such as a tooth crack or breakage or distributed faults as pitting, scuffing, and erosion. The faults are modelled in the case of a crack as a change of tooth stiffness in the case of distributed faults they are given multi-parameter functions. Simulated signals undergo signal analysis by spectrum, cepstrum, time-frequency spectrogram. It has been shown by computer simulation that single and distributed faults are identified by cepstrum. For explicit fault identification time-frequency spectrogram has to be additionally used. The computer simulation results are confirmed by analysis of measured vibration signals received from a gearbox wall/housing. The aim of mathematical modelling and computer simulation, besides finding the relationship between gear condition and vibration signal is in the future to give vibration signals for neural network training.

Differential Diagnosis of Gear and Bearing Faults

This paper deals with the vibration-based diagnosis of rolling element bearings in the presence of strong interfering gear signals, such as is typical of helicopter gearboxes. The key idea consists in recognizing gear signals as purely periodic, whereas bearing signals experience some randomness and are close to cyclostationary, i.e. with a periodic bivariate autocorrelation function. This assertion is demonstrated by introducing a comprehensive model for the vibration generating process of bearing faults: distinctions are made between localized and distributed faults, between cyclostationary and pseudo-cyclostationary processes, and between additive and multiplicative interactions with gear signals. Finally, an original diagnostic procedure is proposed and its performance illustrated using simulated, experimental and actual cases.