Cyclic Modulation Spectrum Research Articles

Intelligent Fault Detection for Rotating Machinery Using Cyclic Morphological Modulation Spectrum and Hierarchical Teager Permutation Entropy

Intelligent fault detection of rotating machines is essentially a pattern classification issue. At the same time, effectively obtaining fault features from the measured signals is a key step to timely diagnose the health status of rotating machinery and evaluate the results of fault classification. To accurately obtain effective fault information to enhance fault accuracy, this paper proposes a novel fault detection scheme based on cyclic morphological modulation spectrum (CMMS) and hierarchical Teager permutation entropy (HTPE). In this scheme, firstly, CMMS was developed to analyze the measured signal to obtain a series of CMMS slices with different frequency bands, which solved the deficiencies of manual empirical selection of frequency band bandwidth in the traditional cyclic modulation spectrum (CMS). Subsequently, by integrating Teager energy operator into hierarchical permutation entropy (HPE), an improved feature selection method named hierarchical Teager permutation entropy (HTPE) is presented to obtain fault information of different frequency band slices, which can improve the fault feature extraction capability of HPE. Finally, the acquired HTPE-based vectors are integrated into the extreme learning machine (ELM) classifier to achieve fault classification of rotating machinery under different working conditions. The proposed scheme is validated by experimental cases including cylindrical roller bearings and planetary gearboxes. The analysis results indicate that the proposed scheme not only can effectively obtain the fault features, but also accurately realize the classification and recognition of the fault mode. In addition, the proposed scheme can achieve higher detection accuracy than some existing algorithms.

Holo-hilbert square spectral analysis: A new fault diagnosis tool for rotating machinery health management

The Holo-Hilbert spectral analysis (HHSA) is an emerging analysis tool for rotating machinery fault diagnosis. It has an excellent performance in reflecting the cross-scale coupling relationship in the nonlinear and non-stationary vibration signals, by combining the two-layer empirical mode decomposition structure. However, the random phase shift involved in the envelope signal analysis limits the wide application of HHSA and may lead to an inaccurate instantaneous frequency estimation. To solve the aforementioned problem and acquire an accurate modulation relationship between amplitude-modulated (AM) and frequency-modulated (FM) characteristics, a novel non-stationary signal analysis method, namely Holo-Hilbert square spectral analysis (HHSSA), is put forward in this paper by combing the square envelope analysis. The proposed HHSSA is integrated to compute a new quantity called the AM-marginal spectrum to reveal the separation efficiency of fault characteristics. Moreover, the relationships between HHSSA with the commonly used second-order cyclostationary methods (Cyclic Modulation Spectrum and Fast Spectral Correlation) are further investigated. Furthermore, the analysis of simulation, experimental data, and industrial data are conducted to demonstrate the high resolution and superior modulation relationship recognition capability of HHSSA. The analysis results further prove that the proposed HHSSA-based AM-marginal spectrum has the best fault identification performance and robustness compared with traditional methods.

Early warning of damaged wind turbine blades using spatial–temporal spectral analysis of acoustic emission signals

A spatial–temporal processing framework is proposed to forecast the wind turbine blade damage in the early stage. The sparse Bayesian learning beamforming (SBL) is applied to data received by a microphone array for enhancement of weak signals and suppressing interference of environmental noise. Then short-time Fourier transform (STFT) is utilized to create a time–frequency spectrum and analyze the nonstationarity of acoustic emission signals. The period of radiation energy change and the cyclic modulation spectrum (CMS) are respectively calculated from the time–frequency spectrum. Blade fault detection is performed based on whether or not the presence of the periodicity or cyclostationary signatures in acoustic emission signals. Numerical simulations have shown that the natural frequencies of acoustic emission signals tend to decrease when there is a hole on the blade surface. The experimental results have verified the effectiveness and robustness of the proposed blade damage detection method.

Novel Cyclo-Nonstationary Indicators for Monitoring of Rotating Machinery Operating Under Speed and Load Varying Conditions

Abstract Condition monitoring arises as a valuable industrial process in order to assess the health of rotating machinery, providing early and accurate warning of potential failures and allowing for the planning and effective realization of preventative maintenance actions. Nowadays, machinery (gas turbines, wind turbines, etc.) manufacturers adopt new business models, providing not only the equipment itself but also additionally taking on responsibilities of condition monitoring, by embedding sensors and health monitoring systems within each unit and prompting maintenance actions when necessary. Among others, rolling element bearings are one of the most critical components in rotating machinery. In complex machines, the failure indications of an early bearing damage are weak compared to other sources of excitations (e.g., gears, shafts, rotors). Vibration analysis is most widely used and various methods have been proposed, including analysis in the time and frequency domain. In a number of applications, changes in the operating conditions (speed/load) influence the vibration sources and change the frequency and amplitude characteristics of the vibroacoustic signature, making them nonstationary. Under changing environments, where speed and load vary, the assumption of quasi-stationary is not appropriate and as a result, a number of time-frequency and time-order representations have been introduced, such as the Short Time Fourier Transform and the Wavelets. Recently, an emerging interest has been focused on modeling rotating machinery signals as cyclostationary, which is a particular class of nonstationary stochastic processes. The classical cyclostationary tools, such as the Cyclic Spectral Correlation Density (CSCD) and the Cyclic Modulation Spectrum (CMS), can be used in order to extract interesting information about the cyclic behavior of cyclostationary signals, only under the assumption that the speed of machinery is constant or nearly constant. Global diagnostic indicators have been proposed as a measure of cyclostationarity under steady operating conditions. In order to overcome this limitation, a generalization of both SCD and CMS functions has been proposed displaying cyclic Order versus Frequency as well as diagnostic indicators of cyclo-nonstationarity in order to cover the speed-varying operating conditions. The scope of this paper is to propose a novel approach for the analysis of cyclo-nonstationary signals based on the generalization of indicators of cyclo-nonstationarity in order to cover the simultaneous and independently varying speed and load operating conditions. The effectiveness of the approach is evaluated on simulated and real signals captured on a dedicated test rig.

Improved cyclostationary analysis method based on TKEO and its application on the faults diagnosis of induction motors

Cyclostationary analysis has been strongly recognized as an effective demodulation tool in identifying fault features of rotating machinery based on vibration signature analysis. This study improves two current mature cyclostationary approaches, cyclic modulation spectrum (CMS) and fast spectral correlation (Fast-SC), combined with the novel frequency-domain application of Teager Kaiser energy operator (TKEO). They can enhance fault feature identification with the lower computational burden. Firstly, the raw vibration signal is transformed into the time–frequency domain through the short-time Fourier transform (STFT) to realize the conversion of the multi-carrier signal to a multiple signal-carrier signal. Secondly, the TKEO is utilized to enhance the fault feature by taking full advantage of demodulating the mono-component. Finally, the spectral coherence and enhanced envelope spectrum (EES) are calculated to effectively exhibit fault features. The superiority of the proposed methods is successfully validated by the simulation study and diagnosing the broken rotor bar (BRB) and bearing outer race faults of induction motors (IMs) under various operating conditions.

An informative frequency band identification framework for gearbox fault diagnosis under time-varying operating conditions

The application of informative frequency band identification methods makes it possible to enhance weak damage components in the vibration signals acquired from rotating machines. Some rotating machines (e.g. wind turbines) operate inherently under time-varying operating conditions, however, very few frequency band identification methods have been developed with varying operating conditions in mind. Therefore, in this work, a systematic framework for obtaining consistent feature planes under time-varying operating conditions is proposed. This framework utilises the angle-frequency instantaneous power spectrum and the order-frequency cyclic modulation spectrum to construct feature planes. The kurtogram, the sparsogram, the infogram, the ICS2gram and the log-cycligram are obtained on numerical and experimental datasets acquired under time-varying operating conditions using this framework. In addition to this, we also implement the Informative Frequency Band Identification method using targeted cyclic orders, abbreviated to IFBIαgram, in this framework and compare the performance of this method against the other frequency band identification methods. Ultimately, we found that the feature used in the construction of the IFBIαgram is very well-suited for gear and bearing fault diagnosis under time-varying operating conditions.

Cyclic Modulation Spectrum Based Bearing Defect Recognition

A novel practical technique for bearing defect detection and recognition based on cyclic modulation spectrum (CMS) is presented. Cyclic modulation spectrum stands as a much faster method to detect periodic modulation feature and has a lower complexity. CMS has a very fast calculation speed, but it can provide similar results with spectral correlation density. This advantage enables the cyclic modulation spectrum to extract useful information from the noisy environment in rotating machine. The validity of this technique is exhibited on bearing defect vibration data from an industrial gearbox. The results show the advantage of cyclic modulation spectrum. CMS can correctly pick up the transient features from interference signals and can accurately recognize the rolling element bearing defect.

Deep learning classification for improved bicoherence feature based on cyclic modulation and cross-correlation.

This paper aims to present an improved bicoherence spectrum (IBS) combined with cyclic modulation spectrum (CMS) and cross-correlation that is suitable for classification of hydrophone signals involving deep learning (DL). First, the proposed feature utilizes the all-phase fast Fourier transform to modify the spectrum leakage caused by CMS; this can be used to detect line spectra with low signal-to-noise ratios (SNRs). Second, the cross-correlation and bispectrum are both exploited to suppress non-periodic line spectra interference from CMS. Based on numerous numerical simulations and experimental verification, compared with CMS and conventional bispectrum, the prominent characteristics of IBS include: detecting higher-precision periodic harmonics without single-line interference, superior robustness under low SNR, and greatly reducing the data redundancy. In addition, to test the performance of IBS for DL application, three deep belief network (DBN)-based classifiers-DBN-softmax, DBN-support vector machine, and DBN-random forest-are introduced and employed for five experimental scenarios (including ships and underwater source). The results indicate that benefiting from DBN pre-training, the IBS classification accuracy of DBN-based models is generally higher than 80%.

An Improved Cyclic Modulation Spectral Analysis Based on the CWT and Its Application on Broken Rotor Bar Fault Diagnosis for Induction Motors

Induction motors (IMs) are widely used in many manufacturing processes and industrial applications. The harsh work environment, long-time enduring, and overloads mean that it is subjected to broken rotor bar (BRB) faults. The vibration signal of IMs with BRB faults consists of the reliable modulation information used for fault diagnosis. Cyclostationary analysis has been found to be effective in identifying and extracting fault feature. The estimators of cyclic modulation spectrum (CMS) and fast spectral correlation (FSC) based on the short-time fourier transform (STFT) have higher cyclic frequency resolution, which has proven efficient in demodulating second order cyclostationary (CS2) signals. However, these two estimators have limitations of processing the maximum cyclic frequency αmax that is smaller than Fs/2 (Fs is the sampling frequency) according to Nyquist’s Theorem. In addition, they have lower carrier frequency resolution due to the fixed window size used in STFT. In order to resolve the initial shortcomings of the CMS and FSC methods, in this paper, we extended the analysis of CMS algorithm based on the continuous wavelet transform (CWT), which enlarged the maximum cyclic frequency range to Fs/2 and provides higher carrier frequency resolution because the CWT has the advantage of multi-resolution analysis. The reliability and applicability of the proposed method for fault components localization were validated by CS2 simulation signals. Compared to CMS and FSC methods, the proposed approach shows better performance by analyzing vibration signals between healthy motor and faulty motor with one BRB fault under 0%, 20%, 40%, and 80% load conditions.

Fault Identification of Broken Rotor Bars in Induction Motors Using an Improved Cyclic Modulation Spectral Analysis

Induction motors (IMs) play an essential role in the field of various industrial applications. Long-time service and tough working situations make IMs become prone to a broken rotor bar (BRB) that is one of the major causes of IMs faults. Hence, the continuous condition monitoring of BRB faults demands a computationally efficient and accurate signal diagnosis technique. The advantage of high reliability and wide applicability in condition monitoring and fault diagnosis based on vibration signature analysis results in an improved cyclic modulation spectrum (CMS), which is one of the cyclic spectral analysis algorithms. CMS is proposed in this paper for the detection and identification of BRB faults in IMs at a steady-state operation based on a vibration signature analysis. The application of CMS is based on the short-time Fourier transform (STFT) and the improved CMS approach is attributed to the optimization of STFT. The optimal window is selected to improve the accuracy for identifying the BRB fault types and severities. The appropriate window length and step size are optimized based on the selected window function to receive a better calculation benefit through simulation and experimental analysis. Compared to other estimators, the improved CMS method provides better fault detectability results by analyzing vertical vibration signatures of a healthy motor, and damaged motors with 1 BRB and 2 BRBs under 0%, 20%, 40%, 60%, and 80% load conditions. Both synthetic and experimental investigations demonstrate the proposed methodology can significantly reduce computational costs and identify the BRB fault types and severities effectively.

Unsupervised Classification of Hydrophone Signals With an Improved Mel-Frequency Cepstral Coefficient Based on Measured Data Analysis

Recently, passive detection technology has developed the ability to detect surface ships based on the noise emissions recorded by hydrophones, making it possible in some cases to classify surface ships. One of the most concerning issues with ships and underwater targets is the current lack of reliable features for unsupervised classification. To solve this problem, this paper proposes an improved Mel-frequency cepstral coefficients (IMFCC) feature for unsupervised clustering of marine targets. As the feature extraction of hydrophone signals rely on preprocessing, the IMFCC adds cyclic modulation spectrum (CMS) and cross-correlation bispectrum (CCB) in the preprocessing module before traditional the Mel-frequency cepstral coefficient process, and the principal component analysis (PCA) is added as the backend processing module. There are four contributions as follows: First, for IMFCC extraction, it combines the advantages of the CMS, CCB, MFCC, and PCA. Second, the two common unsupervised clusters, Gaussian mixture model (GMM) and fuzzy C-means are used to evaluate the CMS, CCB, and several MFCCs—MFCC-vector quantization (MFCC-VQ), MFCC-Gaussian mixture model (MFCC-GMM), TEO-MFCC (Teager Energy Operator based MFCC), and IMFCC. Third, the performances of traditional MFCC, Teager-energy operator (TEO)-MFCC, IMFCC, MFCC-VQ, and MFCC-GMM are discussed under different dimensions, signal-to-noise ratios, distances, and depths. Finally, the experimental results prove that the IMFCC method has a strong anti-interference ability and is robust and has a high-success rate for clustering multitarget or different depth targets, which enables the IMFCC to be a reliable feature for unsupervised classification.

Compressive Sensing for Detecting Ships With Second-Order Cyclostationary Signatures

Amplitude modulation of the broadband propeller noise as a result of the cavitation yields a second-order cyclostationary ship-radiated noise. The spectrum of the modulating signal, consisting of the so-called propeller (or cavitation) tonals, enables the detection and the classification of the submarines or surface ships. However, data acquisition for this purpose causes vast data sizes due to high sampling rates and multiple sensor deployment. To mitigate the negative effects of this acquisition process such as on energy efficiency, hardware complexity, and storage capacity, we propose a scheme for compressive sensing of propeller tonals. We show that the spectral correlation function of cyclostationary propeller noise is sparse and derive a linear relationship between the compressive and Nyquist-rate cyclic modulation spectra, i.e., the approximation of spectral correlation function that allows utilizing matrix representations required in compressive sensing. It also enables use of the cyclic modulation coherence, i.e., the normalized cyclic modulation spectrum, to demonstrate the effect of compressive sensing in terms of statistical detection. We compare the recovery and detection performance results of the sparse approximation algorithms based on the so-called iterative hard thresholding and compressive sampling matching pursuit. Results show that compression is achievable without affecting the detection performance negatively. The main challenges are the weak modulation, low signal-to-noise ratio, and nonstationarity of the ambient noise, all of which reduce the sparsity level, hence causing degraded recovery and detection performance.

A faster algorithm for the calculation of the fast spectral correlation

Abstract One of the main aims of second order cyclostationary (CS2) analysis is the estimation of the full spectral correlation, allowing the identification of different CS2 components in a signal and their characterisation in terms of both spectral frequency f and cyclic frequency α . Unfortunately, traditional estimators of the full spectral correlation (e.g. averaged cyclic periodogram) are highly computationally expensive and hence their application has been quite limited. On the other hand, fast envelope-based CS2 indicators (e.g. cyclic modulation spectrum, CMS) are bound by a cyclic-spectral form of the uncertainty principle, which limits the extent of the cyclic frequency axis α max at approximately the value chosen for the spectral frequency axis resolution Δ f . A recent work has however introduced a ground-breaking approach resulting in a fast algorithm for the calculation of the spectral correlation. This approach is based on the calculation of a series of CMS-like quantities, each scanning a different cyclic-frequency band, given a certain spectral frequency resolution. The superposition of all these quantities allows covering a larger α -band breaking the constraint between maximum cyclic frequency α max and spectral frequency axis resolution Δ f , at a limited computational cost. In this paper a new algorithm for the calculation of the same fast spectral correlation is introduced, resulting in a further computational efficiency gain, and a simplification of the computational procedure.

Condition monitoring of rotating machinery under varying operating conditions based on Cyclo-Non-Stationary Indicators and a multi-order probabilistic approach for Instantaneous Angular Speed tracking

During the last decades, an emerging interest has been reported on modelling rotating machinery signals as cyclostationary. Several tools, such as the Spectral Correlation Density (SCD) and the Cyclic Modulation Spectrum (CMS) have been proposed, assuming constant or almost constant rotating speed. To overcome this limitation, generalizations of SCD and CMS have been proposed that display cyclic Order versus Frequency. On the other hand the estimation of the instantaneous angular speed of the rotating machine is a crucial issue that has to be tackled before any analysis or diagnosis of the machine operating under non-stationary conditions. This operation is generally carried out thanks to a tachometer, or an angle encoder, but in some cases this information can be missing for different reasons. The goal of this paper is to combine a novel approach for the analysis of cyclo-non-stationary signals based on the generalization of the indicators of cyclostationarity in order to cover the speed varying conditions, introducing a new speed-dependent angle averaging operator, with a new speed estimation method, that extracts from a vibration signal the most probable instantaneous rotation frequency without focusing on one specific order based on the a priori knowledge of the kinematics of the machine. The effectiveness of the ensemble method is evaluated on an acceleration signal captured at the wind turbine gearbox operating under speed varying conditions.

On-line updating of cyclostationary tools for fault detection in rotating machines - the filter bank approach

The theory of cyclostationarity provides many valuable tools for fault detection in rotating machines. Cyclostationarity indicators and bi-frequency maps help to quantify fault-related modulations of vibration signals. High quality and robustness of these tools are achieved at a relatively high computational cost. Thus, industrial application of cyclostationarity is limited to off-line analysis in the post-processing stage. In this paper an on-line updating of cyclostationary tools is investigated alternatively to the moving block approach. As an exemplary cyclostationary tool the cyclic modulation spectrum is involved. The filter bank approach is proposed in order to carry out the sample-by-sample updating. We compare properties of the newly proposed method with the one based on sliding discrete Fourier transform.

Specifics of DEMON acoustic signatures for large and small boats

Marine vessel propellers produce noise by the formation and shedding of cavitation bubbles. This process creates both narrow-band tones and broad-band amplitude modulated noise. The Detection of Envelope Modulation on Noise (DEMON) is an algorithm to determine the frequencies that modulate this noise. Results of DEMON processing depend on the selection of a ship noise frequency band to analyze. It is well known that the best passband to use may vary dramatically between vessels. Despite this, there has been no systematic investigation how the DEMON spectra depend of the carrier noise frequencies, and the modulation indices of vessel noise have not been investigated. We use a modification of the Cyclic Modulation Spectrum (CMS) to determine the modulation index of cavitation noise across the entire spectrum of carrier frequencies. We investigated how speed and vessel size affect the modulation index and carrier frequency of vessel noise. Several phenomena in the distribution of modulation indices for large and small boats were observed. These can be used for vessel classification. For small boats, the DEMON spectra have a different set of frequency peaks at various carrier frequencies. This is explained by the engine exhaust which produces amplitude modulated noise much like a propeller.

Fast computation of the spectral correlation

Although the Spectral Correlation is one of the most versatile spectral tools to analyze cyclostationary signals (i.e. signals comprising hidden periodicities or repetitive patterns), its use in condition monitoring has so far been hindered by its high computational cost. The Cyclic Modulation Spectrum (the Fourier transform of the spectrogram) stands as a much faster alternative, yet it suffers from the uncertainty principle and is thus limited to detect relatively slow periodic modulations. This paper fixes the situation by proposing a new fast estimator of the spectral correlation, the Fast Spectral Correlation, based on the short-time Fourier transform (STFT). It proceeds from the property that, for a cyclostationary signal, the STFT evidences periodic flows of energy in and across its frequency bins. The Fourier transform of the interactions of the STFT coefficients then returns a quantity which scans the Spectral Correlation along its cyclic frequency axis. The gain in computational cost as compared to the conventional estimator is like the ratio of the signal length to the STFT window length and can therefore be considerable. The validity of the proposed estimator is demonstrated on non trivial vibration signals (very weak bearing signatures and speed varying cases) and its computational advantage is used to compute a new quantity, the Enhanced Envelope Spectrum.

CS2 analysis in presence of non-Gaussian background noise – Effect on traditional estimators and resilience of log-envelope indicators

Second-order cyclostationary (CS2) analysis has become popular in the field of machine diagnostics and a series of digital signal processing techniques have been developed to extract CS2 components from the background noise. Among those techniques, squared envelope spectrum (SES) and cyclic modulation spectrum (CMS) have gained popularity thanks to their high computational efficiency and simple implementation. The effectiveness of CMS and SES has been previously quantified based on the hypothesis of Gaussian background noise and has led to statistical tests for the presence of CS2 peaks in squared envelope spectra and cyclic modulation spectra. However a recently established link of CMS with SES and of SES with kurtosis has exposed a potential weakness of those indicators in the case of highly leptokurtic background noise. This case is often present in practice when the machine is subjected to highly impulsive phenomena, either due to harsh operating conditions or to electric noise generated by power electronics and captured by the sensor. This study investigates and quantifies for the first time the effect of leptokurtic noise on the capabilities of SES and CMS, by analysing three progressively harsh situations: high kurtosis, infinite kurtosis and alpha-stable background noise (for which even first and second-order moments are not defined). Then the resilience of a recently proposed family of CS2 indicators, based on the log-envelope, is verified analytically, numerically and experimentally in the case of highly leptokurtic noise.

Order-frequency analysis of machine signals

Latest researches have asserted the eligibility of angle/time cyclostationarity in analyzing machine signals when operating under variable speed conditions. A core descriptor in this framework is the order-frequency spectral correlation (OFSC), basically estimated by the averaged cyclic periodogram (ACP), being able to jointly decode (i) the angle-dependent modulations related to the machine kinematics and (ii) the time-dependent carriers related to the machine dynamics. The present paper comes into this context with the aim of enriching this framework with new tools excerpted from cyclostationarity. In particular, a new estimator of the OFSC based on the cyclic modulation spectrum (CMS) is proposed and compared with the ACP in terms of resolution, statistical performance and computational cost. In addition, two related tools are theoretically addressed and their estimators are derived through the ACP and CMS. Specifically, the optimality of the “order-frequency spectral coherence” (the normalized/whitened form of the OFSC) in revealing cyclic components according to their SNR is demonstrated. Also, the “improved envelope spectrum” is derived from the latter by integrating over the spectral frequency variable, evidencing considerable enhancement over the squared envelope spectrum. The potentiality of the proposed tools and the adequacy of the related estimators are experimentally investigated on simulated and real-world vibration signals.

Combining blind separation and cyclostationary techniques for monitoring distributed wear in gearbox rolling bearings

This work seeks to study the potential effectiveness of the Blind Signal Extraction (BSE) as a pre-processing tool for the detection of distributed faults in rolling bearings. In the literature, most of the authors focus their attention on the detection of incipient localized defects. In that case, classical techniques (i.e. envelope analysis) are robust in recognizing the presence of the fault and its characteristic frequency. However, when the fault grows, the classical approach fails, due to the change of the fault signature. De facto, in this case the signal does not contain impulses at the fault characteristic frequency, but more complex components with strong non-stationary contents. Moreover, signals acquired from complex machines often contain contributions from several different components as well as noise; thus the fault signature can be hidden in the complex system vibration. Therefore, pre-processing tools are needed in order to extract the bearing signature, from the raw system vibration. In this paper the authors focalize their attention on the application of the BSE in order to extract the bearing signature from the raw vibration of mechanical systems. The effectiveness and sensitivity of BSE is here exploited on the basis of both simulated and real signals. Among different procedures for the BSE computation, the Reduced-Rank Cyclic Regression algorithm (RRCR) is used. Firstly a simulated signal including the effect of gear meshing as well as a localized fault in bearings is introduced in order to tune the parameters of the RRCR. Next, two different real cases are considered, a bearing test-rig as an example of simple machine and a gearbox test-rig as an example of complex machine. In both examples, the bearings were degreased in order to accelerate the wear process. The BSE is compared with the usual pre-processing technique for the analysis of cyclostationary signals, i.e. the extraction of the residual signal. The fault detection is carried out by the computation of the Integrated Cyclic Modulation Spectrum on the extracted signals. The results indicate that the extracted signals via BSE clearly highlight the distributed fault signature, in particular both the appearance of the faults as well as their development are detected, whilst noise still hides fault grow in the residual signals.