Time Frequency Domain Information Research Articles

Deep learning-based remaining useful life estimation of bearings using multi-scale feature extraction

Accurate evaluation of machine degradation during long-time operation is of great importance. With the rapid development of modern industries, physical model is becoming less capable of describing sophisticated systems, and data-driven approaches have been widely developed. This paper proposes a novel intelligent remaining useful life (RUL) prediction method based on deep learning. The time-frequency domain information is explored for prognostics, and multi-scale feature extraction is implemented using convolutional neural networks. Experiments on a popular rolling bearing dataset prepared from the PRONOSTIA platform are carried out to show the effectiveness of the proposed method, and its superiority is demonstrated by the comparisons with other approaches. In general, high accuracy on the RUL prediction is achieved, and the proposed method is promising for industrial applications.

Research on signal processing of segment-grout defect in tunnel based on impact-echo method

In shield tunneling construction, grout of precast concrete segments is widely applied to various geological conditions. However, the connection performance between the grout and segments after the grout cannot be determined by destructive means, such as coring. The impact-echo method is used extensively in nondestructive testing of civil engineering structures. Nevertheless, there are inadequate researches on determining defects in segment-grout structure. In this paper, the segment-grout structure was simulated experimentally and artificial defects were introduced in the grout. The Short-time Fourier Transform (STFT) and Wavelet Transform (WT) of signals obtained by the impact-echo method were carried out to analyze the time-frequency domain features of the impact-echo signals with and without defects in the grout. On this basis, defects in the grout layer were determined. Finally, energies of wavelet packet were calculated and characteristic relative energy indexes of wavelet packet at different positions were detained. It’s found that the values at positions without defects are relatively low. It’s demonstrated that the signal analysis of impact-echo method has to combine time-frequency domain information and energy distribution for detection of defects in the grout layer. Wavelet transform and energy analysis methods are effective means to detect defects in the grout layer.

The Application of Heterogeneous Information Fusion in Misalignment Fault Diagnosis of Wind Turbines

The misalignment of the drive system is one of the important factors causing damage to gears and bearings on the high-speed output end of the gearbox in doubly-fed wind turbines. How to use the obtained information to determine the types of the faults accurately has always been a challenging problem for researchers. Under the restriction that only one kind of signal is used in the current wind turbine fault diagnosis, a new method based on heterogeneous information fusion is presented in this paper. The collected vibration signal, temperature signal, and stator current signal are used as original sources. Their time domain, frequency domain and time-frequency domain information are extracted as fault features. Taking into account the correlation between the features, t-distributed Stochastic Neighbor Embedding (t-SNE) is used to reduce the dimensionality of the original combinations. Then, the fusion features are put into the Least Square Support Vector Machine (LSSVM), which is optimized by artificial bee colony (ABC) algorithm. The simulation tests show that this method has higher diagnostic accuracy than other methods.

Dynamic behaviors of flexible fluid containers with fuel consumption

This study focuses on the dynamic characteristics of the containers with fuel consumption. With fuel consumption, the fuel container system vibrates with decreasing mass, which is a typical variable mass system. In order to investigate the influences of the time-varying liquid on vibration of the containers, this fluid–structure coupling system with time-varying mass is modeled combining finite element method, used to model the structure, and virtual mass method representing the fluid. The dynamic responses and time–frequency analysis of the time-varying mass system are presented based on the numerical simulations. From the results, it is shown that decrease of the mass results in increase of the vibration frequencies of the system and induces an additional negative damping. The additional negative damping is proportional to the rate of mass change and causes the vibration to decay slowly. The results also suggest that the time–frequency analysis possess appealing advantages in processing nonstationary signals and thus are able to provide reliable time–frequency domain information for time-varying mass systems.

Inverse approach to chronotaxic systems for single-variable time series.

Following the development of a new class of self-sustained oscillators with a time-varying but stable frequency, the inverse approach to these systems is now formulated. We show how observed data arranged in a single-variable time series can be used to recognize such systems. This approach makes use of time-frequency domain information using the wavelet transform as well as the recently developed method of Bayesian-based inference. In addition, a set of methods, named phase fluctuation analysis, is introduced to detect the defining properties of the new class of systems by directly analyzing the statistics of the observed perturbations.We apply these methods to numerical examples but also elaborate further on the cardiac system.

Leakage current characterization for estimating the conditions of ceramic and polymeric insulating surfaces

Leakage current produced on contaminated ceramic and polymeric insulating surfaces was analyzed using a wavelet transform technique providing time-frequency domain information. Salt-fog tests with and without the deposition of non-soluble contamination were conducted in a laboratory, during which the leakage current was separated into sinusoidal, local arc, or their transition components per halfwave of power frequency. The wavelet transform allowed the detection of the zero-cross point of the base current at 50 Hz without the need for voltage information. The source information regarding the distortion level (I/sub 150 Hz//I/sub 50 Hz/) of the current half-wave separated by the zero-cross point and a time lag to onset in current attributable to arcs are used in order to obtain the above three components. The time-integrated values (cumulative charges) of these components were found to correlate with hydrophobicity and contamination level. It was shown that time variations of the cumulative charges and of their component ratios were useful for estimating the conditions of ceramic and polymeric insulating surfaces.