Search

This paper introduces a method based on short periodograms, combining, envelope analysis, and spectral kurtosis (SK). In order to highlight an underlying impulsive spectral component of a vibration signal, to do so, the proposed method takes into account the well-known welch’s method of periodograms applied to a non-overlapping filtered signal segment using the SK. An SK short-time approach is considered, due to, a high variable speed even in angle domain has an underlying non-stationarity caused by the fact that the transfer function is angle variant. In the end, it is obtained a noiseless envelope spectrum, averaging the filtered periodograms of each envelope of the non-overlapped signal segments. To prove the effectiveness of the proposed method, it is tested on a highly non-stationary vibration signal measured from an aircraft engine under a run-up test. As a result, we can identify a bearing failure embedded highly non-stationary noise. Besides as the proposed method makes use of short periodograms is fast, reliable, and entirely non-parametric, making it readily applicable to highlight any underlying impulsive behaviour in a vibration signal.

Vibration and acoustic analysis actively support the nondestructive and noninvasive fault diagnostics of rotating machines at early stages. Nonetheless, the acoustic signal is less used because of its vulnerability to external interferences, hindering an efficient and robust analysis for condition monitoring (CM). This paper presents a novel methodology to characterize different failure signatures from rotating machines using either acoustic or vibration signals. Firstly, the signal is decomposed into several narrow-band spectral components applying different filter bank methods such as empirical mode decomposition, wavelet packet transform, and Fourier-based filtering. Secondly, a feature set is built using a proposed similarity measure termed cumulative spectral density index and used to estimate the mutual statistical dependence between each bandwidth-limited component and the raw signal. Finally, a classification scheme is carried out to distinguish the different types of faults. The methodology is tested in two laboratory experiments, including turbine blade degradation and rolling element bearing faults. The robustness of our approach is validated contaminating the signal with several levels of additive white Gaussian noise, obtaining high-performance outcomes that make the usage of vibration, acoustic, and vibroacoustic measurements in different applications comparable. As a result, the proposed fault detection based on filter bank similarity features is a promising methodology to implement in CM of rotating machinery, even using measurements with low signal-to-noise ratio.

Mostly, a vibration signal from rotating machines comprises stationary and non-stationary components, whose description should be as accurate as possible to infer the internal and external forces that affect the system behavior. Moreover, either component can provide diverse but relevant information about the machine health. Thus, bearing faults foster the non-stationary component that is characterized by time-varying statistical moments, periodically changing through the time (or cyclostationary signals). Therefore, the problem to detect a bearing fault signal is usually addressed to separate the deterministic and the stochastic components of the vibration signal to make clear the damage characteristics. To this end, we present the novel order tracking (OT) method that decomposes the non-stationary vibration signal into narrow-band spectral components, aiming to enhance the cyclostationary characteristics. Moreover, a similarity measure is computed between the envelopes of the raw signal and each component, allowing to quantify the cyclic behavior of signal components. Since the proposed method acts as a narrow-band filter, a comparison with the spectral kurtosis (SK) is performed using a rolling element bearing dataset that includes an inner race, outer race, and rolling element defects. Specifically, the Case Western Reserve University data is carried out aiming to improve the diagnosis of bearing failures that are categorized in a recent work using the benchmark methods. As a result, the proposed blind extraction method allows capturing the cyclostationary behavior hidden in the signal and improves the identification of the bearing faults when the signal is noisy.

Order tracking (OT) is a technique for vibration analysis under non-stationary operating conditions, which considers the speed or load is time-varying in order to obtain machine health indicators. Nonetheless, information about the rotational speed implies a priori knowledge about the machine. Therefore, in this paper is proposed a blind instantaneous rotational speed extraction methodology that extracts a set of instantaneous frequencies (IF), from a filter bank applied to the original signal, using the Hilbert transform. The signal is transformed into order domain using velocity synchronous discrete Fourier transform (VSDFT), taking each estimated IF as a possible reference shaft speed. Then, the kurtosis is used as a non-stationarity criterion aiming to select a suitable instantaneous reference shaft speed. The methodology is tested in a simulated signal and a test rig which exhibit non-stationary behavior. As a result the closest component to the rotational speed is extracted without priori knowledge of the machine.

Time series analysis implies extracting relevant features from real-world applications to improve pattern recognition tasks. In that sense, representation methods based on time series decomposition and similarity measures are combined to select representative features with physical interpretability. In this work, we introduce two similarity measures based on the cross-power spectral density to select representative intrinsic mode functions (IMF) that characterize the time series. The IMFs are obtained by Ensemble Empirical Mode Decomposition because it deals with non-stationary dynamics present into time series. The proposed similarity measures are an extension of the correlation coefficient and are validate using vibration signals acquired in a test rig under three different machine states (undamaged, unbalance and misalignment). Results show that the proposed measures improve the interpretability in terms of association between an IMF and a fault state, preserving a high classification rate.