Time-domain Averaging Research Articles

Fault Diagnosis of Marine Diesel Engine Based on Multi-scale Time Domain Decomposition and Convolutional Neural Network

Abstract Marine diesel engines work in an environment with multiple excitation sources. Effective feature extraction and fault diagnosis of diesel engine vibration signals have become a hot research topic. Time-domain synchronous averaging (TSA) can effectively handle vibration signals. However, the key phase signal required for TSA is difficult to obtain. During signal processing, it can result in the loss of information on fault features. In addition, frequency multiplication signal waveforms are mixed. To address this problem, a multi-scale time-domain averaging decomposition (MTAD) method is proposed and combined with signal-to-image conversion and a convolutional neural network (CNN), to perform fault diagnosis on a marine diesel engine. Firstly, the vibration signals are decomposed by MTAD. The MTAD method does not require the acquisition of the key phase signal and can effectively overcome signal aliasing. Secondly, the decomposed signal components are converted into 2-D images by signal-to-image conversion. Finally, the 2-D images are input into the CNN for adaptive feature extraction and fault diagnosis. Through experiments, it is verified that the proposed method has certain noise immunity and superiority in marine diesel engine fault diagnosis.

Time–frequency analysis of gustatory event related potentials (gERP) in taste disorders

In taste disorders, the key to a correct diagnosis and an adequate treatment is an objective assessment. Compared to psychophysical tests, EEG-derived gustatory event-related potentials (gERP) could be used as a less biased measure. However, the responses identified using conventional time-domain averaging show a low signal-to-noise ratio. This study included 44 patients with dysgeusia and 59 healthy participants, who underwent a comprehensive clinical examination of gustatory function. gERPs were recorded in response to stimulation with two concentrations of salty solutions, which were applied with a high precision gustometer. Group differences were examined using gERP analyzed in the canonical time domain and with Time–Frequency Analyses (TFA). Dysgeusic patients showed significantly lower scores for gustatory chemical and electrical stimuli. gERPs failed to show significant differences in amplitudes or latencies between groups. However, TFA showed that gustatory activations were characterized by a stronger power in controls than in patients in the low frequencies (0.1–4 Hz), and a higher desynchronization in the alpha-band (8–12 Hz). Hence, gERPs reflect the altered taste sensation in patients with dysgeusia. TFA appears to enhance the signal-to-noise ratio commonly present when using conventional time-domain averaging, and might be of assistance for the diagnosis of dysgeusia.

Adaptive time-domain sliding average method for spall size estimation of roller bearing with skidding

The size of the bearing spall fault can be estimated by the time interval between the roller's entry and exit points in the bearing vibration signal. However, the entry vibration response is typically low amplitude and masked in environmental noise. The time-domain average (TDA) method is effective for periodic impulse enhancement. However, bearing skidding is unavoidable during its running, and the cage's speed fluctuations caused by skidding will lead to non-periodicity of the impulses. The performance of TDA seriously deteriorated with such skidding because of the time-varying period. This paper presents the adaptive time-domain sliding average (ATDSA) method to address the bearing skidding problem. The similarity feature between each impulse segment is utilized, and a time delay matrix is constructed to estimate the skidding rate and mitigate the skidding effect. The frequency spectrum of the reconstructed signal shows clear sidebands with BCF interval. And the obtained averaged waveform has high resolution in both the low-frequency entry and high-frequency exit events. Simulation and experiment results show that it can effectively suppress the skidding effect and estimate the spall size correctly under different speeds.

Signal Processing in Optical Frequency Domain Reflectometry Systems Based on Self-Sweeping Fiber Laser with Continuous-Wave Intensity Dynamics

We report on the development of an optical frequency domain reflectometry (OFDR) system based on a continuous-wave Er-doped self-sweeping fiber laser. In this work, we investigate the influence of the input data processing procedure in an OFDR system on the resulting reflectograms and noise level. In particular, several types of signal averaging (in time and frequency domain) and Fourier analysis are applied. We demonstrate that the averaging in the frequency domain can be applied to evaluate absolute values of the local scattering amplitudes related to the Rayleigh light scattering (RLS), which is associated with the interference of scattering signals on microscopic inhomogeneities in optical fibers. We found that the RLS signal remains unchanged in the case of signal averaging in time domain, while the noise floor level decreases by 30 dB with an increasing number of points from 1 to ~450. At the same time, it becomes possible to detect the spectral composition of the scattering at each point of the fiber using windowed Fourier transform. As a result, the sensitivity of the developed system allows us to measure the RLS signal at a level of about 20 dB above the noise floor. The described analysis methods can be useful in the development of distributed sensors based on Rayleigh OFDR systems.

A Robust Extraction Approach of Auditory Brainstem Response Using Adaptive Kalman Filtering Method.

The Auditory brainstem response (ABR) can provide valuable information on the function of the auditory pathway. However, the ABR signal has a very small amplitude, and it is easily submerged in different background noises with large amplitude. Conventional ABR extraction methods such as time-domain averaging (TDA) and Kalman filter (KF) were greatly affected by noise intensity, and the result relies on the empirical settings of parameters. ABR extraction method that can automatically adjust parameters to adapt different background noises was needed. An adaptive Kalman filtering (AKF) based ABR signal extraction method was proposed, in which two recursive rules were introduced to constantly update the parameters according to the real-time noise properties. It was used for ABR extraction from recordings in noises with different orders of larger magnitude. The AKF method demonstrated the best performance in obtaining reliable ABR waveform morphologies in the presence of large EMG noises compared with traditional methods of TDA or KF. It could extract satisfactory ABR signal with fewer trials of acoustic stimulus repetition, even from noise 10000 times larger than ABR signal. The AKF results also showed smaller absolute errors and higher correlation coefficients with the target ABR signal when different types (gum chewing, mouth opening and milk drinking) or levels of noises were introduced. The proposed AKF method is a great candidate to increase the robustness of current ABR measurements. It could provide reduced testing time and relaxed recording conditions for ABR and other evoked potentials extraction.

Retrospective comparison of motor and somatosensory MEG mapping—Considerations for better clinical applications

While magnetoencephalography (MEG) has proven to be a valuable and reliable tool for presurgical functional mapping of eloquent cortices for at least two decades, widespread use of this technique by clinicians has remained elusive. This modest application may be attributable, at least in part, to misunderstandings regarding the success rate of such mapping procedures, as well as the primary sources contributing to mapping failures. To address this, we conducted a retrospective comparison of sensorimotor functional mapping success rates in 141 patients with epilepsy and 75 tumor patients from the Center for MEG in Omaha, NE. Neurosurgical candidates either completed motor mapping (i.e., finger tapping paradigm), somatosensory mapping (i.e., peripheral stimulation paradigm), or both motor and somatosensory protocols during MEG. All MEG data underwent subsequent time-domain averaging and source localization of left and right primary motor (M1) and somatosensory (S1) cortices was conducted using a single equivalent dipole model. Successful mapping was determined based on dipole goodness of fit metrics ∼ 95%, as well as an accurate and conceivable spatial correspondence to precentral and postcentral gyri for M1 and S1, respectively. Our results suggest that mapping M1 in epilepsy and tumor patients was on average 94.5% successful, when patients only completed motor mapping protocols. In contrast, mapping S1 was successful 45–100% of the time in these patient groups when they only completed somatosensory mapping paradigms. Importantly, Z-tests for independent proportions revealed that the percentage of successful S1 mappings significantly increased to ∼ 94% in epilepsy patients who completed both motor/somatosensory mapping protocols during MEG. Together, these data suggest that ordering more comprehensive mapping procedures (e.g., both motor and somatosensory protocols for a collective sensorimotor network) may substantially increase the accuracy of presurgical functional mapping by providing more extensive data from which to base interpretations. Moreover, clinicians and magnetoencephalographers should be considerate of the major contributors to mapping failures (i.e., low SNR, excessive motion and magnetic artifacts) in order to further increase the percentage of cases achieving successful mapping of eloquent cortices.

Improved time domain synchronous averaging based on the moving interpolation and kurtosis criterion searching

The fault diagnosis of rotating machinery involves revealing the possible faults in advance to reduce dispensable breakdowns; the difficulty of this lies in identifying the periodic features of rotating machines contaminated by background noise. Time-domain synchronous averaging (TSA) has been studied to eliminate random noise in signals. However, TSA sometimes cannot extract useful signals more accurately because of the accumulative phase error caused by the discrete sampling process and the difficulties in obtaining accurate prior information. Hence, the moving interpolation and Kurtosis searching criterion are used for more accurate extraction of harmonics and transient impacts. Also, an improved compensation algorithm based on moving interpolation is proposed to overcome the amplitude attenuation caused by cumulative phase error for low signal to noise ratio (SNR) signal. To determine some parameters in the algorithm such as the number of periods and the time delay of windows which depend on a priori information relevant to the fault period, a searching method for the prior information in vibration signals including transient impacts and harmonics with Kurtosis and minimum mean square error criterion is proposed to optimize the algorithm in the process of feature extraction. Finally, the improved TSA (ITSA) is applied to extracting the fault features in a real factory, and the performance of fault feature extraction in low SNR signal conditions with the ITSA has been enhanced effectively.

Research on the rotor/stator interaction noise through vane unsteady pressure test based on polyvinylidene fluoride sensor array

The rotor/stator interaction noise is the primary noise source of modern turbofan engines. Vane unsteady pressure test is essential to understand the generation mechanism and propagation characteristics of the noise, as well as examine the passive/active noise control technologies. However, limited by the Vane’s thickness as well as the requirement for high spatial and temporal resolutions of the sensor, traditional transducers still have considerable limitations and difficulties carrying out the measurement. In this paper, a polyvinylidene fluoride (PVDF) piezoelectric-film sensor array was developed for the vane unsteady pressure test. The dynamic characteristics of the sensor were obtained by the shock tube test. The calibration results further confirm the effectiveness of the self-developed PVDF sensor and also show the advantage in frequency response over traditional pressure transducer. Time histories of the vane fluctuating pressures were obtained by the PVDF sensor. The blade passing frequency (BPF) harmonic component and the broadband component of the unsteady pressure were analyzed individually by time-domain averaging and spectral analysis. The distributions of BPF content and broadband content on both sides of the vane at different fan tip speeds and operating conditions were investigated in detail. The results indicate that the operating conditions have little effect on both of them. The broadband component yielded by the rotor-wake turbulence seems to be quasi-isotropic. The average broadband level increases linearly with the fan tip speed. The developed PVDF sensor shows potential in noise evaluation and noise source diagnosis, which can properly guide further noise reduction technology development.

Filter Realization of the Time-Domain Average Denoising Method for a Mechanical Signal

Time-domain averaging (TDA) is an effective signal processing technique in fault diagnosis that can extract the periodic components of interest from signals mixed with noise interference while suppressing other irrelevant periodic signals. However, there are two obvious shortcomings to TDA: first, the acquisition of keyphasor signals is often restricted by the application environment and conditions. Even if the signal is obtained by TDA, owing to the existence of periodic truncation errors, satisfactory results cannot be obtained. Second, due to the velocity fluctuation, the actual mechanical signal is easy to produce a large error in TDA stacking. To solve the above challenges, first, based on the disadvantage of using traditional resampling to solve the TDA synchronization problem, this paper proposes a new method of subsection resampling, which improves the analysis effect of the traditional TDA. Second, to further expand the range of the practical applications, according to the amplitude-frequency map of TDA, a method for realizing TDA function by using the FIR multiband filter is proposed. This approach effectively avoids the requirement of traditional methods to collect the keyphasor signals and broadens the application in practical engineering. Finally, the improved TDA method is compared with the filter implementation technology, and their respective application conditions are given.

Efficient and de-noising calculation for extracting impulse response function in time-domain

ABSTRACT Impulse response function (IRF) and frequency response function (FRF) are the bases and preconditions for acquiring the parameters of a dynamic system and accurately obtaining the dynamic response of the system both in time and frequency domains. In the test of dynamic characteristics of a structural system, for some cases, it is not suitable to use the average power spectrum and cross-spectrum methods to obtain the FRF of the system. This paper studies the methods of extracting the IRF in the time domain, where the measured input and output signals contain white Gaussian noise. According to Duhamel’s integral and the time-domain averaging technique, an efficient average recursive (AR) algorithm with anti-noise performance is proposed to obtain the impulse response function. In the AR method, it is assumed that the input and output data of the system are composed of white Gaussian noise with a mean value of zero. The proposed AR algorithm avoids the inversion of a larger matrix and improves the efficiency of calculation. Simulation and experimental results show that it has apparent advantages in computing efficiency and de-noising.

Research on Wind Turbines Fault Diagnosis Technology Based on CMS Data Feature Extraction

As a rich clean and environmentally friendly renewable resources, wind energy has emerged as a strategic choice for countries around the world. Because the wind turbines often operate in severe working conditions such as variable load and large temperature difference they are prone to failures and possible shutdowns. The shutdowns however seriously affect the economic benefits of the wind turbines. Initiative maintenance has become a worldwide recognized scientific method for planning and determining preventive maintenance work, the implementation of this strategy relies on real-time condition monitoring and fault signal identification methods. The condition monitoring of wind turbine can help master the health state and power generation performance of wind turbine, so as to timely formulate maintenance strategies and adopt technical modification measures to improve power generation performance, reduce the down time of wind turbine, avoid the occurrence of major faults, save maintenance cost and improve power generation capacity. Therefore, a condition monitoring system is built on a wind turbine of Zhangjiakou, and a systematic signal analysis method is proposed, time-domain synchronous averaging technology, based on variable period, impulse signal feature extraction technology based on Teager and signal decomposition technology based on CEEMD. The proposed method realizes the signal analysis and feature extraction of non-stationary nonlinear, weak signal and frequency aliasing signals, and successfully diagnose the gearbox secondary meshing failure during the long-term monitoring. This confirms that the monitoring system methods and signal analysis technology proposed in this paper can effectively realize the condition monitoring and fault diagnosis of wind turbines.

Low-Complexity Time-Domain Averaging-Based DAC Clock Tone Leakage Compensation for Short-Reach PAM4 Transmission

High sampling-rate digital-to-analog converters (DAC) are usually applied in optical fiber communication systems for the signal generation with advanced modulation formats such as four-level pulse amplitude modulation (PAM4). There exist various imperfections for high sampling-rate DACs. Clock tone leakage (CTL) is one of the imperfections and may seriously degrade the bit error rate (BER) performance, especially for single-carrier transmission systems. In this paper, we study a low-complexity time-domain averaging (TDA) based CTL compensation (TDA-CTLC) method for short-reach PAM4 transmission. By numerical simulation, we investigate the impacts of chromatic dispersion, CTL interference-to-signal power ratio (ISPR), and vertical resolution of the analog-to-digital converter on the BER performance. The results exhibit that the TDA-CTLC enabled PAM4 transmission system in the presence of a wide range of ISPR has a similar BER performance with the one without CTL interference. Moreover, we experimentally evaluate the compensation performance of the TDA-CTLC in a directly-modulated laser-based short-reach PAM4 transmission system. Up to −4 dB ISPR can be compensated with a slight BER performance degradation due to nonlinear distortions.

Low Complexity Two-Stage FOE Using Modified Zoom-FFT for Coherent Optical M-QAM Systems

We propose a novel low complexity two-stage frequency offset estimation (FOE) scheme based on modified Zoom-FFT for square 16/64-quadrature amplitude modulation (QAM) coherent optical systems. The first stage roughly estimates the frequency offset through a small size FFT. After frequency shifting, the modified Zoom-FFT can select and reconstruct the spectrum of interest by time domain average. Since the average provides a good estimate of the amplitudes of the frequency components, we can accurately find the peak after the second-stage FFT. We conduct simulations and compare the normalized frequency variance of the proposed scheme with different algorithms, e.g., the single-stage FFT and the Chirp-Z transform (CZT). The simulation results show that the proposed FOE scheme has the similar normalized frequency variance as other algorithms of $4.0\times 10^{-9}$ and $1.6\times 10^{-9}$ for 16QAM and 64QAM, respectively. The numerical simulations and experimental results demonstrate that the proposed scheme can achieve comparable performance, while the complexity is reduced by 86.6% (81.2%) and 60.9% (61.2%) for 16QAM (64QAM) compared with single-stage FFT and FFT+CZT, respectively.

Phase-based time domain averaging (PTDA) for fault detection of a gearbox in an industrial robot using vibration signals

This paper proposes a fault detection method that uses vibration signals in the gearboxes of industrial robots. The vibration signals from gearboxes consist of both deterministic signals and residual signals; fault-related signals usually exist in the residual signals. Previously, time domain averaging (TDA) has been studied to derive the deterministic signals. However, the performance of TDA method is limited when the signals are poorly synchronized. Therefore, we propose a new phase-based time domain averaging (PTDA) method. The proposed PTDA method can estimate deterministic signals that are more synchronized by considering the phase angle of the vibration signals. Then, the residual signals can be calculated by subtracting the estimated deterministic signals from the measured vibration signals using the PTDA method. We use two health features, root-mean-square (RMS) and power spectrum entropy, to quantify the fault severity in the residual signals. To demonstrate the proposed method, we use vibration signals measured from a six-degree-of-freedom (6-DOF) industrial robot test-bed under 1) a simple one-joint rotating motion, 2) a complicated arc welding motion, and 3) a spot welding motion. The results show that the proposed PTDA method can improve the performance of fault detection for gearboxes in industrial robots.

LED Nonlinearity Estimation and Compensation in VLC Systems Using Probabilistic Bayesian Learning

In this paper, we propose and evaluate a novel light-emitting diode (LED) nonlinearity estimation and compensation scheme using probabilistic Bayesian learning (PBL) for spectral-efficient visible light communication (VLC) systems. The nonlinear power-current curve of the LED transmitter can be accurately estimated by exploiting PBL regression and hence the adverse effect of LED nonlinearity can be efficiently compensated. Simulation results show that, in a 80-Mbit/s orthogonal frequency division multiplexing (OFDM)-based nonlinear VLC system, comparable bit-error rate (BER) performance can be achieved by the conventional time domain averaging (TDA)-based LED nonlinearity mitigation scheme with totally 20 training symbols (TSs) and the proposed PBL-based scheme with only a single TS. Therefore, compared with the conventional TDA scheme, the proposed PBL-based scheme can substantially reduce the required training overhead and hence greatly improve the overall spectral efficiency of bandlimited VLC systems. It is also shown that the PBL-based LED nonlinearity estimation and compensation scheme is computational efficient for the implementation in practical VLC systems.

Online digital offset mismatch compensation for high-speed time-interleaved ADC in real-time optical OFDM receiver.

The time-interleaved analog-to-digital converter (TI-ADC) is a good option to realize high-speed data conversion for both single-carrier and multi-carrier optical communication systems. Offset mismatch is one of its drawbacks, which causes distortion in the sampled data and degrades the bit error rate (BER) performance of communication systems. In this article, a low-complexity online digital offset mismatch compensation (OMC) scheme based on time-domain averaging is proposed and implemented with a commercial off-the-shelf field programmable gate array (FPGA) chip for high-speed optical OFDM communications. The proposed OMC scheme is experimentally demonstrated in real-time direct-detection optical OFDM receiver with a 5 GS/s TI-ADC. The experimental results show that the BER performance can be improved by more than an order of magnitude, by using the proposed OMC scheme, for both 16- and 64-QAM modulation formats. The receiver sensitivity can be improved by more than 5-dB in terms of the received optical power at the BER of 1e-3. What's more, the real-time measured BER performance is similar to that of by using offline DSP approaches and has good stability during the measurement period. Numerical simulations are performed under additive white Gaussian noise (AWGN) channel to fully verify the performance of the OMC scheme. It exhibits that the error floor in BER performance can be eliminated by using the OMC scheme and a negligible signal-to-noise ratio (SNR) penalty can be achieved compared with that of offset mismatch-free case.

Cross-Correlation-Based Identification of Rotating Sources

Noise-source identification using microphone arrays has become well-established in wind-tunnel testing. When the sources are moving, the accepted approach is first to “de-Dopplerize” the microphone signals, producing estimated source signals for subsequent analysis. A straightforward time-domain average of these signals is vulnerable to flow-noise contamination, and so an alternative approach is necessary for wind-tunnel applications. This method is equivalent to the basic, “diagonal-removal,” cross-spectral technique employed with fixed sources, but, unlike the fixed-source case, cannot be generalized to account simultaneously for multiple sources. The alternative presented here, which works with the time-domain cross-correlations between microphones, is not subject to this limitation. It is illustrated using simulated data from an example rotating-source configuration, and the results are compared against a de-Dopplerization-based analysis. For three equal-strength sources, the current method eliminates the contamination evident in de-Dopplerization estimates. Equally, it is able to locate and accurately characterize weak secondary sources that are masked by primary-source contributions in de-Dopplerization beamforming maps. Hence it extends the option of higher-fidelity wind-tunnel beamforming, currently only available for fixed sources, to rotating-source configurations.

Accuracy Improvement of In-line Near-Infrared Spectroscopic Moisture Monitoring in a Fluidized Bed Drying Process.

An exploratory analysis of a large representative dataset obtained in a fluidized bed drying process of a pharmaceutical powder has revealed a significant correlation of spectral intensity with granulate humidity in the whole studied range of 1091.8–2106.5 nm. This effect was explained by the dependence of powder refractive properties, and hence light penetration depth, on the water content. The phenomenon exhibited a close spectral similarity to the well-known stochastic variation of spectral intensities caused by the process turbulence (the so-called “scatter effect”). Therefore, any traditional scatter-corrective preprocessing incidentally eliminates moisture-correlated variance from the data. To preserve this additional information for a more precise moisture calibration, a time-domain averaging of spectral variables has been suggested. Its application resulted in a distinct improvement of prediction accuracy, as compared to the scatter-corrected data. Further improvement of the model performance was achieved by the application of a dynamic focusing strategy when adjusting the model to a drying process stage. Probe fouling was shown to have a minor effect on prediction accuracy. The study resulted in a considerable reduction of the root-mean-square error of in-line moisture monitoring to 0.1%, which is close to the reference method's reproducibility and significantly better than previously reported results.

Adaptive determination of fundamental frequency for direct time-domain averaging

Time-domain averaging (TDA), as an effective signal processing technique for periodic signal enhancement has been widely used for fault detection of gearboxes and bearings. TDA normally requires a tachometer signal to provide an accurate estimation of fundamental frequency. However, tachometer is unavailable or difficult to use in many applications. In this paper, an adaptive fundamental-frequency determination method without tachometer signal is proposed. Firstly, the effect of fundamental-frequency estimation error on direct TDA is investigated, and it indicates that minor error of estimation can cause severe deterioration in curve of attenuation coefficient. Secondly, harmonic-to-noise ratio (HNR) is proposed to evaluate the result of direct TDA for a given fundamental frequency, and it is utilized to optimize the fundamental frequency from a number of candidates. Finally, average waveform is obtained by implementing direct TDA with the optimized fundamental frequency. Examples of gear faults are given to verify the proposed method. The derived fundamental frequency is accurate, and average waveforms indicate the presence of gear faults successfully.

Influence of Airflow Rate and Stimulus Concentration on Olfactory Event-Related Potentials (OERP) in Humans.

Although the association between odor concentration and olfactory event-related potential (OERP) has been studied, less is known about the influence of airflow on OERP. The aim of this study was to investigate the influence of airflow rate and stimulus concentration on OERP in humans. Electroencephalogram data were collected from young healthy volunteers (n = 17) in separate sessions where 2-phenylethanol (PEA) was delivered in the following conditions: 8 L/min 50% v/v, 8 L/min 30% v/v, 4 L/min 100% v/v, and 4 L/min 60%v/v. Odor concentrations are referred to the %v/v achieved with air dilution and was not measured in the nose. Odor intensity ratings were recorded immediately after stimulus presentation. Data recorded at 5 electrodes (Fz, Cz, Pz, C3, and C4) were pooled and analyzed using both time-domain averaging and single-trial time-frequency domain approaches. Higher airflow rate significantly increased intensity ratings (F = 10.98, P < 0.01), and improved the signal-to-noise-ratio (F = 5.42, P = 0.025). Results from time-frequency analysis showed higher concentration versus lower concentration increased brain oscillations in the slow frequency band (1-3 Hz) at 0-600 ms; while higher airflow rates versus lower airflow rate increased theta-band oscillations (300-600 ms and 5-9 Hz) and decreased delta-band oscillations at 900-1500 ms after stimulus onset. In conclusion, compared to stimulus concentration, airflow rate was associated with improved OERP quality and more pronounced responses. The results suggest that intensity ratings and OERP are strongly related to the steepness of stimulus onset. High airflow rates are suggested for odor delivery in order to record OERP.