Fault Diagnosis of Rotating Machinery Based on Wavelet Domain Denoising and Metric Distance

Abstract

In the monitoring process of petrochemical equipment rotating machinery, the collected large data easily lead to valuable data loss in the pre-processing process and affecting the accuracy of the fault diagnosis. This paper proposes a method for the fault diagnosis of the rotating machinery based on the wavelet-domain denoising and metric distance. The wavelet-domain denoising uses wavelet coefficients of signal and noise that have different properties on different scales and process noisy signal wavelet coefficients. Metric distance is to compare two independent statistical samples with each other after denoising to determine whether they belong to the same sample. First, the denoising of the vibration timedomain signal is based on the wavelet-domain denoising method. Then, the tested fault samples are compared with the known fault samples by metric distance. Finally, the fault types are identified according to the metric distance. Verification of the algorithm performance and the simulation experiment of petrochemical large units show that the method is not only simple and effective but also has better faults recognition. It can guide the faults diagnosis of large petrochemical units and other large units rotating machinery. INDEX TERMS Fault diagnosis, wavelet domain denoising, metric distance, rotating machinery.

Pressure Vessel Diagnosis by Eliminating Undesired Signal Sources and Incorporating GA-Based Fault Feature Evaluation

“…For example, as input samples x and z belong to a same category, their within-category RBF value is expected as (11). On the other hand, if x and z belong to different categories, their between-category RBF value is expected as (12).…”

“…From (11), (12), (13) and (14), it can be concluded that a subset of fault signatures providing both large value of ( ), This means that the algorithm explores an optimal feature vector which yields the best feature space of categories to categorize by the classifier in the following stage.…”

“…Another alternative approach, currently outperforming former methods, is a wavelet threshold-based de-noising method. However, this method still faces difficulties in choosing a proper threshold, and sometimes is incapable of resolving overlapping frequencies [11][12][13]. Empirical mode decomposition (EMD), a cutting-edge method with compelling noise-reducing capabilities, is another potential option, especially for vibration signal noise.…”

A Bi-Level Nested Sparse Optimization for Adaptive Mechanical Fault Feature Detection

“…Thus, the golden section search method is adopted to find the optimal solution f θ * (y). Substituting the iterative schemes of algorithm 2 into the second-stage optimization (10), then, a Bi-level nested sparse optimization framework is obtained and shown in algorithm 3. The mapping SureSDS shown in algorithm 2 denotes the sparse solver with SURE-based MSE estimation, and the mapping GSS is the update criterion in golden section search method.…”

“…A primary issue in the FDI procedure is to detect feature information from noisy measurements and many advanced signal processing methods have been developed [7], such as spectrum analysis [8], time-frequency analysis [9], wavelet transform (WT) [10], spectrum kurtosis(SK) [11], adaptive mode decomposition [12], cyclostationary descriptors [13], and deep learning [14], [15], etc. In recent years, sparsity representation based fault diagnosis (SRFD) techniques have been one of the hottest topics in the signal processing society and aroused extensive interests.…”