Improvements in the Theory of Identification of Burst-Shaped Events for Fault Diagnosis

Szappanos, G.; Por, G.

doi:10.13182/nse98-a2005

Cited by 6 publications

(6 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The same is true for H 1 : if the growing value of lambda function exceeds the +4.6 value, then we can state with 95 % of confidence level that the stochastic noise belongs to a process with probability density function characterized by H 1 parameters. More, in-depth information about these steps can be found in references [20] and [21].…”

Section: Time-domain Methodsmentioning

confidence: 99%

Managing measured vibration data for malfunction detection of an assembled mechanical coupling

Manhertz

Gardonyi

Por

2014

Int J Adv Manuf Technol

View full text Add to dashboard Cite

Vibration data recorded by accelerometer fixed on a mechanical coupling were treated using different data processing techniques as estimating general root mean square, partial root means square, mean, variance, partial variance, kurtosis, full Fourier spectra, and partial spectra, as well as using sequential probability ratio test (SPRT). The numerical results of these methods measured on an electromotive drive train-coupling were collected into a common data set, and a search method of deviation from the normal behavior was elaborated. This selects the bad couplings if at least three of the presented qualification parameters crosses either theoretically or experimentally settled limits. It was shown that the confidence level of such classification of good and bad species was better than 95 % compared with the results of thoroughly tested ones by conventional methods.

show abstract

Section: Time-domain Methodsmentioning

confidence: 99%

Managing measured vibration data for malfunction detection of an assembled mechanical coupling

Manhertz

Gardonyi

Por

2014

Int J Adv Manuf Technol

View full text Add to dashboard Cite

show abstract

“…where y is the time signal a i are the AR coefficients and v i is a random white noise. One can obtain the Yule-Walker equations [6,7] featuring the i autocorrelation function and solve it by an iterative process, the so called Durbin's recursive formula:…”

Section: Autoregressive Modeling-filteringmentioning

confidence: 99%

Advanced acoustic detection system for monitoring nuclear power plant components

Por¹

2016

Insight

View full text Add to dashboard Cite

An acoustic event detection system is presented, which was installed at all four Units of Paks Nuclear Power Plant in 2002, and since that it was modified several times to follow technical developments. It monitors the whole reactor coolant system including its components. The original aim of the system was to monitor and detect loose parts. Fortunately loose part occurs rather rarely, but several events have been registered since the beginning of its operation. Event detection is based on Sequential Probability Ratio Test (SPRT), which is the fastest method in the sense it requires the minimum number of samples to find the event. It is demonstrated, that SPRT can detect events buried in the background noise. Monitoring system is fully automated, collecting all abnormal occurrences, i.e. deviation from normal operation, deviation from usual background noises, etc. We also demonstrate the typical spectra of the operation of different components. During the 13 years of operation flow induced vibrations of motor operated valves were monitored continuously, movement of the control rod driving mechanism were registered. System automatically records the start of the main coolant pumps detecting small size loose parts named debris. Sensitivity is demonstrated with automated registration of the water flowing from the control rod bank during outage of the reactor.

show abstract

“…We carried out measurement sampling signals of accelerometers in frequency range from 1kHz up to 50 kHz. The lower frequencies may have very high industrial noise, while the frequency range above audible 16 kHz is still containing weak components from the knocks of the loose parts and the background noise is surprisingly weak in that range [2,4].…”

Section: Test Sectionmentioning

confidence: 99%

“…The next step is the Sequential Probability Ratio Test, which serves for event detection in the filtered signal. The theoretical formula of SPRT [2]:…”

Section: Sequential Probability Ratio Testmentioning

confidence: 99%

See 1 more Smart Citation

Identification of the Place and Materials of Knocking Objects in Flow Induced Vibration

et al. 2011

View full text Add to dashboard Cite

Flow induced vibration can be found and identified by acoustic methods. Using acoustic sensors, the first task, the event detection has been solved using the sequential probability ratio test after autoregressive filtering the measured signal. The LABView program and actual test of the event recognition technique are presented. The signals were recorded in a 100 m long test loop having artificially placed flow induced vibrating objects hitting the wall. Time delay between the fronts of detected events has been used to localize the actual place of the acoustic source. The recognition of the material of the knocking object is based traditionally on the spectrum estimation. However, this is rather time consuming task by naked eyes. We are proposing to introduce the skeleton method for event identification.

show abstract

Improvements in the Theory of Identification of Burst-Shaped Events for Fault Diagnosis

Cited by 6 publications

References 3 publications

Managing measured vibration data for malfunction detection of an assembled mechanical coupling

Managing measured vibration data for malfunction detection of an assembled mechanical coupling

Advanced acoustic detection system for monitoring nuclear power plant components

Identification of the Place and Materials of Knocking Objects in Flow Induced Vibration

Contact Info

Product

Resources

About