Residual Life Prediction of Rotating Machines Using Acoustic Noise Signals

Intelligent Condition Based Monitoring Using Acoustic Signals for Air Compressors

“…Taking this intuition ahead, 0.025% of the samples are eliminated from either extremes of the statistical mean while finding maximum and minimum values. Further, normalization is done using (13), which is similar to the max-min normalization, and scales the values from to 1. In (13), , and respectively refer to the lower, and upper values of the scale to which normalization needs to be done.…”

“…Experimentation in [47] shows that the proposed method gives better performance than the regular max-min normalization method. (13) A simple way of implementing this normalization method would be to sort the sampled signal values in ascending order of their values, and then exclude the top 0.025% and bottom 0.025% of sample values while finding maxima and minima values. Sorting can be computationally quite expensive with a best complexity of .…”

“…In [5]- [12], induction motor faults such as damage in bearings, rotors, and stator bars have been diagnosed by analyzing motor current and vibration data. For rotating machines, the most commonly used measurements for fault recognition are acoustic, and vibration measurements [13]- [16]. James et al [79] developed a fault diagnosis system for high pressure air compressors based on the signatures of different parameters, specifically pressure, temperature, voltage, and water flow rate.…”

PDF and Breakdown Time Prediction for Unobservable Wear Using Enhanced Particle Filters in Precognitive Maintenance

“…where the normalizing constant p(y k |y 1:k−1 ) is p(y k |y 1:k−1 ) = p(x k |y 1:k−1 ) p(y k |x k ) dx k (5) and the likelihood function p(y k |x k ) is defined in (2). A relationship of the prior prediction and posterior estimation between noisy observations and system states is set up via (3) and (4).…”

“…The internal machinery wear is usually unobservable, which has to be indirectly monitored under the electrical and electronic instruments by other physical quantities, such as force, acoustic emission, and vibration [1], [2]. After signal processing, the corresponding industrial degradation can be predicted based on the logged sensor signals in practice to evaluate remaining useful life (RUL) and schedule maintenance [3]- [5]. In consideration of the existing uncertainties, including environment, model, and measurement uncertainties [6], [7], the probability density function (pdf) prediction will provide more statistical information for engineers to analyze its prediction reliability and confidence in industry [8]- [10].…”