Serhii Mykhalkiv | Ukrainian State University of Railway Transport (original) (raw)
Papers by Serhii Mykhalkiv
Naukovyi Visnyk Natsionalnoho Hirnychoho Universytetu, ISSN 2071-2227, E-ISSN 2223-2362, 2024
Purpose. To develop a hybrid method based on ICEEMDAN and OMEDA for high-fidelity detection of th... more Purpose. To develop a hybrid method based on ICEEMDAN and OMEDA for high-fidelity detection of the diagnostic features of the technical condition of roller bearing.
Methodology. Due to digital signal processing the search was made of the informative components on the broadband spectra, envelope spectra, squared envelope spectra. The usage of the methods of mathematical statistics for the selection of informative IMFs as a result of iterative calculate procedures. The introduction of blind deconvolution method with a proper objective function for the enhancement of impulse fault features.
Findings. The use of traditional spectral methods provided an incomplete list of diagnostic features of bearing roller damage. The empirical mode decomposition (EMD) method and its minor versions provide the series of intrinsic mode functions (IMFs). On the broadband vibration spectra of the selected low-frequency IMFs, diagnostic features were not detected in a sufficient number. Further research focused on the development of a hybrid method that takes into account the demodulation procedure of the high-frequency components of those IMFs that were selected according to the complex evaluation index. The Fast Kurtogram technique determined an informative frequency band of 7.2–7.5 kHz for demodulation and construction of the squared envelope spectra, where harmonics of the roller spin frequency and some side bands with a width equal to the separator rotation frequency appeared. The use of OMEDA caused the reduction of residual noise on the squared envelope spectra and bilateral sidebands to appear around all harmonics of the roller spin frequency.
Originality. For the first time, the whole list of the diagnostic features of roller faults of the axle-box bearing was identified using the developed hybrid method, which involves steps of noise reduction for enhancing a physical meaning of a proper frequency components, associated with faults. The usage of the optimal minimum entropy deconvolution adjusted method helped to eliminate noise and provide all diagnostic features of the technical condition on the squared envelope spectrum.
Practical value. The results of the study help to control the quality of repair of the axle-box rolling bearings on the test rig at the workshop of the freight railway car depot.
Keywords: car, vibration, diagnostics, bearing, deconvolution, decomposition, spectrum
2021 IEEE 2nd KhPI Week on Advanced Technology (KhPIWeek), 2021
The article presents the results of the study of the energy characteristics of three-phase voltag... more The article presents the results of the study of the energy characteristics of three-phase voltage source inverters with modified pulse width modulation with the addition of a third harmonic to the modulating signal with the implementation of adaptive change of the carrier sawtooth signal. Computer simulation in MATLAB / Simulink software was performed for the described modulation algorithm. The results of studies of the proposed modulation algorithm showed the improvement of the harmonic composition of the input current, output voltage and output current, increase the output power and reduce the power loss of a two-level three-phase inverter.
Eastern-European Journal of Enterprise Technologies, 2021
An analysis of the method for ensuring the sinusoidality of the output voltage in power generatio... more An analysis of the method for ensuring the sinusoidality of the output voltage in power generation systems with self-commutated voltage inverters under the requirements of the international standard IEEE-519 is presented. In a number of programs, especially low-power generation systems, a low-cost solution is needed to provide the sinusoidal waveform of the output voltage with the total harmonic distortion of 5 %. This solution is to use two-level voltage inverters with an output sine LC filter. However, the feature of the sine filter with the frequency converter is that the PWM frequency affects the spectrum of higher harmonics of the output voltage. In addition, there is the starting current of the filter capacitor, which can disable the power switches of the voltage inverter. The developed method for calculating the values of the LC filter with the two-level voltage inverter in the PWM mode is presented meeting the requirements of the international standard IEEE-519, taking into ...
Collection of scientific works of the State University of Infrastructure and Technologies series "Transport Systems and Technologies", 2022
The paper deals with the properties of the Fast Kurtogram technique for the vibrodiagnostics of r... more The paper deals with the properties of the Fast Kurtogram technique for the vibrodiagnostics of rolling bearings of electric motor. Taking into account the disadvantages of the classic tools of vibrodiagnostics in the time domain, the proper frequency band selection procedure was suggested for the further demodulation and envelope spectrum extraction. Fast Kurtogram represents the spectral kurtosis value of the signal on the (f, f) plane. The frequency and the frequency resolution are used as the key functions to determine the magnitude of the spectral kurtosis. The best combination makes the kurtosis maximum. During the experimental research the vibration of the rolling bearing of electric motor of electric locomotive ChS7 series was acquired. The broadband spectrum in the frequency range 0 – 9 kHz was extracted and with the help of empiric approach the four frequency bands with resonance excitations were selected. None of four envelope spectra did not have any informative features among numerous random components. After the selection of a proper center frequency and the frequency band by the Fast Kurtogram, the extracted envelope spectrum has shown the series of harmonics related to the outer race faults.
NAUKOVYI VISNYK Natsionalnoho Hirnychoho Universytetu, 2020
Purpose. The purpose is to create discrete stochastic ARIMA models for forecasting the remaining... more Purpose. The purpose is to create discrete stochastic ARIMA models for forecasting the remaining life of pads of modernized
brake rigging (BR) devices of bogies of industrial railway cars.
Methodology. Accounting of statistical data on the wear of pads of typical and modernized BR devices obtained in the relevant
studies. On the basis of analytical designs of BR, changes in the junction of the vertical lever with the spacer are proposed. Akaike
and Bayesian information criteria are used for selecting the optimal integrated autoregression and moving average model within
the Box-Jenkins methodology for forecasting the remaining mileage of pads.
Findings. The ARIMA model was identified, evaluated, and checked for adequacy according to the Akaike and Bayesian information
criteria. It is established that abnormal wear of the top of the pads of typical BR devices due to a number of design and
operational reasons occurs when the mileage is about 3.5 times less than the forecasted life before the abnormal wear of the top of
the pads of modernized BR devices. The forecasted remaining life of the top of the pad of the modernized BR is 3.3 thousand km
shorter than that for the bottom of the same pad.
Originality. For the first time, the remaining life of the pads of the modernized BR devices of industrial freight cars was forecasted
using discrete stochastic ARIMA models, which require only the availability of discrete values that are recorded during the
relevant experimental measurements.
Practical value. The results of the study were verified on experimental rolling stock with modernized devices in the brake systems
of bogies. They can be used in the design, upgrade and operation of the brake systems of both the rolling stock which is currently
in operation and the new generation of bogies of freight cars.
Keywords: ARIMA model, freight car, wear, forecasting, brake pad, brake rigging
Collection of scientific works of the Ukrainian State University of Railway Transport, 2019
The article deals with the forecasting of the technical state of the traction gearboxes of electr... more The article deals with the forecasting of the technical state of the traction gearboxes of electric trains with discrete stochastic models. During every 3rd level maintenance, the vibration signals of the previously selected fourteen traction gearboxes were recorded, and the box counting dimension was estimated.
For the forecasting of the technical state, among the broadcast deterministic methods of exponential smoothing and trend analysis, the most effective ARIMA model was selected. This model in certain applications has higher precision than GRNN and BPNN neural networks. The implementation of the ARIMA model requires accomplishment of the identification, evaluation, fitting and practical application stages for the forecasting model. The identification was performed due to the one-step differentiation with a further stationarity check according to the autocorrelation function and partial autocorrelation function.
The evaluation of the ARIMA model with the different orders of the autoregression component and moving average component was done. For the model fitting, the Akaike’s information criterion and Bayes information criterion were calculated, and the autoregression component of the ARIMA model having the minimum values of these criteria was selected. The forecasting of the fitted ARIMA models with an 80 % confidence interval was done for the period since the 71st 3rd level maintenance up to the next 2nd level current repair. After disassembling 14 traction gearboxes during the 2nd level current repair, the insufficient amount of a lubricant in the first gearbox, wear of a roller bearing of the front cover in the second gearbox, a crack of a bearing ring of the front cover in the third gearbox, a tooth break of a gear in the fourth traction gearbox, destruction of a bearing in the fifth gearbox and a tooth crack of its gear were detected. It was established that the ARIMA model is difficult to upgrade for the evaluation of other data, which requires the implementation of measures for a resimulation. Despite high complexity, the absence of the automatic process for the calculation and the necessity of performing several iterative procedures, the calculated minimum value precision of the ARIMA model forecasting is equal to 91.4 %
Key words: accuracy, ARIMA model, electric train, forecasting, traction gearbox
Collection of scientific works of the Ukrainian State University of Railway Transport, 2019
The paper deals with the property of discrete wavelet transform based on the multiresolution anal... more The paper deals with the property of discrete wavelet transform based on the multiresolution analysis feature to identify different types of the gear and bearing vibration. The direct analysis of the vibration time series by the use of conventional statistical measures, such as mean, root mean square, standard deviation, is not always useful due to the complexity of the signal. It was proposed to choose the best mother wavelet which is able to identify the transients in vibration signal according to the calculated minimum value of Shannon entropy, which quantifies the level of uncertainty of a given vibration time series. The main idea is that when a bearing is healthy, it will produce low amplitude random vibration with a uniform-like probability mass function and as the fault occurs and progresses some probability mass function component will be prevalent with a higher probability of occurrence. The chosen Daubechies wavelet of the 4-th order has decomposed the acquired vibration signal of the traction gearbox of electric train into approximated and detailed coefficients on four decomposition scales with further reconstruction of the signals on the appropriate scales according to the above-mentioned coefficients. The autocorrelation was applied for the detection of deterministic and random components in the reconstructed signals through evaluation of the impulse periodicities of the reconstructed signals according to the detailed coefficients at all scales and has taken the sinusoidal shape for the reconstructed signals according to the approximated coefficients. It was established that a deterministic vibration component dominates and there are no bearing damage features in the reconstructed signals according to the approximated coefficients due two strong gearmesh harmonics. The presence of impulse periodicity on the reconstructed signals according to the detailed coefficients at the second decomposition scale is possible to monitor due to the correlogram, which can be explained by the periodic contact of the damaged element with other elements during their rotation in bearing. The kurtosis is applied as a reliable tool for the frequency band selection where the bearing vibration has the strongest
excitation.
Кeywords: bearing, impulse, spectrum, traction gearbox, vibration, wavelet.
Technical Sciences and Technologies, 2019
Urgency of the research. It is established that introduction of effective vibrodiagnostics techni... more Urgency of the research. It is established that introduction of effective vibrodiagnostics techniques, which are capable of the early identification of the traction gearbox faults of electric trains, is a relevant task.
Target setting. The vibrodiagnostics of mechanical units of electric trains is known to be a trustworthy technique that is able to extract impulsive components with a periodic repetition in accordance with the revolution of the faulty parts of gears or bearings. The main problem is development of effective methods for the noise elimination and identification of the technical condition features of gears and bearings.
Actual scientific researches and issues analysis. Recently the calculation of the kurtosis index has taken a leading role in the extraction of weak periodic impulses. However, unclear recommendations on increasing the accuracy of these results needed improvement, which led to emergence of the spectral kurtosis method based on the filters for the reconstruction of random signals with a high level of additive stationary noise.
Uninvestigated parts of general matters defining. The identification of a bearing vibration component and the influence of a gear component in high frequency band on it were not clearly understood.
The research objective. The objective of the paper is identification of the informative frequency band of excited bearing vibration by means of the spectral kurtosis method.
The statement of basic materials. The paper selects the best window length of the Short-Time Fourier Transform by means of the spectral kurtosis method, which allows to identify the highest level of the spectral kurtosis and to find the proper frequency band.
Conclusions. By means of Wiener filter, the broadband structures of the traction gearboxes vibration of electric trains were identified, and the frequency band with excited resonance component of the bearing vibration was registered.
Keywords: bearing, frequency, gear, electric train, spectral kurtosis, vibration.
Collected scientific works of Ukrainian State University of Railway Transport, 2018
The paper deals with the time-synchronous averaging technique of vibration signals for extraction... more The paper deals with the time-synchronous averaging technique of vibration signals for extraction of the features of the technical state of electric train traction gearboxes. Simulation of the vibration of the faultless gear mesh, and gearwheel with a cracked tooth and strong deterministic component was carried out. The rotation frequencies of gearwheel and pinion, and the mesh frequency are clear visible on the vibration power spectrum. Taking into account the poor visibility of the sidebands around mesh frequency on the power spectra, the time-synchronous averaging technique enhances the sidebands, the width of which corresponds to the gearwheel rotation frequency and influences weakening of the sidebands of the faultless pinion. During the experimental research the traction gearbox vibration of electric train was acquired on the test bench in a depot, and the calculated power spectrum has not identified sidebands and mesh
frequency. After the time-synchronous average within 18 revolutions for the pinion and gearwheel
frequency, the vibration power spectra were acquired. The time-synchronous average technique
eliminates random and non-synchronous components while keeping the frequency components that
are synchronous to the rotating frequency of the pinion and gearwheel of the traction gearbox. On
the vibration power spectra in the vicinity of the mesh frequency the strong sidebands are visible,
and they do not coincide with the previously calculated rotation frequencies, which means the
absence of gearwheel faults. At the vibration power spectrum for the pinion there are six sidebands
in the vicinity of the meshing frequency that coincide with the previously calculated pinion rotation
frequencies, which means the presence of localized or distributed faults for the pinion.
Disassembling of the traction gearbox shows a crack of the rolling bearing inner race and faultless
state of the gear mesh. It was established that the sensitivity of the time-synchronous average
technique makes it possible to identify not only localized faults extraction but also distributed faults
such as eccentricity and misalignment of the pinion shaft through the modulation of the sidebands
on vibration power spectra of the gearwheel.
International Journa l of Engineering &Technology (IJET), 2018
Purpose: To improve the performance of vibration spectral methods in identification of bearing el... more Purpose: To improve the performance of vibration spectral methods in identification of bearing element faults of freight car axle-boxes. Approach: An algorithm for simulating the expected vibration signal of outer race bearing was implemented. The autoregressive filter and minimum empirical deconvolution method was applied to identify the ball pass outer-race fault frequency and its harmonics on the envelope spectra and squared envelope spectra which were extracted in the proper frequency band. Results: The simulated vibration signal of a faulty bearing shows suitability of the autoregressive filter and minimum empirical deconvo-lution method, envelope and squared envelope spectra for outer race fault identification. There were observed a lower amount of features and their impulse sharpness in outer race faults in the bearing test rig than on the spectra in the wheelset test rig. Conclusions: The deterministic components are removed in the residual signal after using the AR filter and the impulse and noise components that decrease the kurtosis value remain in it. The MED technique additionally enhances the magnitude of increased BPFO components after using the AR filter and, together with it, provides satisfied performance and increases the efficiency of vibration diagnostics.
Collection of scientific works of the Ukrainian State University of Railway Transport, Dec 15, 2014
Taking into consideration the necessity of applying monitoring means to diagnose the actual techn... more Taking into consideration the necessity of applying monitoring means to diagnose the actual technical condition of driving axle bearings to reduce the number of their failures, some new devices have been developed which take account of a sudden nature of failures of some bearings elements. These devises have been designed to detect and analyse certain bearing vibration characteristics during the motion of the locomotive. The built-in device for monitoring the vibration of the driving axle bearings is attached to the surface of the tractive motor bearing assembly by means of a pin. The pin is screwed in a pre-drilled hole and the part is attached to it. Attaching by means of pins is one of the ways that don't distort the operation characteristics, the resonance frequency and the dynamic range of vibroaccelerometers. To increase the trustworthiness of the results obtained, the monitoring signals are analyzed only in the mode of the locomotive stationary speed. Objective: develop built-in means to monitor vibration characteristics of driving axle bearings to increase the efficiency of planned preventive system of locomotive maintenance and repairs.
Keywords: bearing, diagnosis, failure, locomotive, operation, vibration
Journal of Mechanical Engineering National Technical University of Ukraine "Kyiv Politechnic Institute" , Dec 29, 2016
Goal: increase of efficiency for diagnostics of rolling bearing faults using an autoregressive mo... more Goal: increase of efficiency for diagnostics of rolling bearing faults using an autoregressive model to calculate AR coefficients and further application of pre-whitening AR filter and envelope spectra to extract weak fault features. Method of doing research: diagnostics of rolling bearing faults involves signal acquisition technique and application of the AR model for better analysis of short duration signal properties and impulses. The Akaike information criterion is used to ensure optimum adaptation of AR coefficients to bearing faults. The AR coefficients are defined with Yule-Walker equitation. The advantages of pre-whitening AR filter are presented in comparison of the low efficiency of the power spectral density of the parametric model. The experimental study of vibration characteristics of the auxiliary electromotor body of electric locomotive determines the frequency band 5,5 — 7 kHz of the rolling bearing vibration, and this frequency band can be used for further extraction of the envelope spectra. Value: the research shows a capability of the pre-whitening AR model to gain valuable information not only about different faults concerning outer and inner races and bearing rollers, but also about the technical condition of the cage, the features of which are displayed on the envelope spectra directly after the AR filter
Naukovyi Visnyk Natsionalnoho Hirnychoho Universytetu, ISSN 2071-2227, E-ISSN 2223-2362, 2024
Purpose. To develop a hybrid method based on ICEEMDAN and OMEDA for high-fidelity detection of th... more Purpose. To develop a hybrid method based on ICEEMDAN and OMEDA for high-fidelity detection of the diagnostic features of the technical condition of roller bearing.
Methodology. Due to digital signal processing the search was made of the informative components on the broadband spectra, envelope spectra, squared envelope spectra. The usage of the methods of mathematical statistics for the selection of informative IMFs as a result of iterative calculate procedures. The introduction of blind deconvolution method with a proper objective function for the enhancement of impulse fault features.
Findings. The use of traditional spectral methods provided an incomplete list of diagnostic features of bearing roller damage. The empirical mode decomposition (EMD) method and its minor versions provide the series of intrinsic mode functions (IMFs). On the broadband vibration spectra of the selected low-frequency IMFs, diagnostic features were not detected in a sufficient number. Further research focused on the development of a hybrid method that takes into account the demodulation procedure of the high-frequency components of those IMFs that were selected according to the complex evaluation index. The Fast Kurtogram technique determined an informative frequency band of 7.2–7.5 kHz for demodulation and construction of the squared envelope spectra, where harmonics of the roller spin frequency and some side bands with a width equal to the separator rotation frequency appeared. The use of OMEDA caused the reduction of residual noise on the squared envelope spectra and bilateral sidebands to appear around all harmonics of the roller spin frequency.
Originality. For the first time, the whole list of the diagnostic features of roller faults of the axle-box bearing was identified using the developed hybrid method, which involves steps of noise reduction for enhancing a physical meaning of a proper frequency components, associated with faults. The usage of the optimal minimum entropy deconvolution adjusted method helped to eliminate noise and provide all diagnostic features of the technical condition on the squared envelope spectrum.
Practical value. The results of the study help to control the quality of repair of the axle-box rolling bearings on the test rig at the workshop of the freight railway car depot.
Keywords: car, vibration, diagnostics, bearing, deconvolution, decomposition, spectrum
2021 IEEE 2nd KhPI Week on Advanced Technology (KhPIWeek), 2021
The article presents the results of the study of the energy characteristics of three-phase voltag... more The article presents the results of the study of the energy characteristics of three-phase voltage source inverters with modified pulse width modulation with the addition of a third harmonic to the modulating signal with the implementation of adaptive change of the carrier sawtooth signal. Computer simulation in MATLAB / Simulink software was performed for the described modulation algorithm. The results of studies of the proposed modulation algorithm showed the improvement of the harmonic composition of the input current, output voltage and output current, increase the output power and reduce the power loss of a two-level three-phase inverter.
Eastern-European Journal of Enterprise Technologies, 2021
An analysis of the method for ensuring the sinusoidality of the output voltage in power generatio... more An analysis of the method for ensuring the sinusoidality of the output voltage in power generation systems with self-commutated voltage inverters under the requirements of the international standard IEEE-519 is presented. In a number of programs, especially low-power generation systems, a low-cost solution is needed to provide the sinusoidal waveform of the output voltage with the total harmonic distortion of 5 %. This solution is to use two-level voltage inverters with an output sine LC filter. However, the feature of the sine filter with the frequency converter is that the PWM frequency affects the spectrum of higher harmonics of the output voltage. In addition, there is the starting current of the filter capacitor, which can disable the power switches of the voltage inverter. The developed method for calculating the values of the LC filter with the two-level voltage inverter in the PWM mode is presented meeting the requirements of the international standard IEEE-519, taking into ...
Collection of scientific works of the State University of Infrastructure and Technologies series "Transport Systems and Technologies", 2022
The paper deals with the properties of the Fast Kurtogram technique for the vibrodiagnostics of r... more The paper deals with the properties of the Fast Kurtogram technique for the vibrodiagnostics of rolling bearings of electric motor. Taking into account the disadvantages of the classic tools of vibrodiagnostics in the time domain, the proper frequency band selection procedure was suggested for the further demodulation and envelope spectrum extraction. Fast Kurtogram represents the spectral kurtosis value of the signal on the (f, f) plane. The frequency and the frequency resolution are used as the key functions to determine the magnitude of the spectral kurtosis. The best combination makes the kurtosis maximum. During the experimental research the vibration of the rolling bearing of electric motor of electric locomotive ChS7 series was acquired. The broadband spectrum in the frequency range 0 – 9 kHz was extracted and with the help of empiric approach the four frequency bands with resonance excitations were selected. None of four envelope spectra did not have any informative features among numerous random components. After the selection of a proper center frequency and the frequency band by the Fast Kurtogram, the extracted envelope spectrum has shown the series of harmonics related to the outer race faults.
NAUKOVYI VISNYK Natsionalnoho Hirnychoho Universytetu, 2020
Purpose. The purpose is to create discrete stochastic ARIMA models for forecasting the remaining... more Purpose. The purpose is to create discrete stochastic ARIMA models for forecasting the remaining life of pads of modernized
brake rigging (BR) devices of bogies of industrial railway cars.
Methodology. Accounting of statistical data on the wear of pads of typical and modernized BR devices obtained in the relevant
studies. On the basis of analytical designs of BR, changes in the junction of the vertical lever with the spacer are proposed. Akaike
and Bayesian information criteria are used for selecting the optimal integrated autoregression and moving average model within
the Box-Jenkins methodology for forecasting the remaining mileage of pads.
Findings. The ARIMA model was identified, evaluated, and checked for adequacy according to the Akaike and Bayesian information
criteria. It is established that abnormal wear of the top of the pads of typical BR devices due to a number of design and
operational reasons occurs when the mileage is about 3.5 times less than the forecasted life before the abnormal wear of the top of
the pads of modernized BR devices. The forecasted remaining life of the top of the pad of the modernized BR is 3.3 thousand km
shorter than that for the bottom of the same pad.
Originality. For the first time, the remaining life of the pads of the modernized BR devices of industrial freight cars was forecasted
using discrete stochastic ARIMA models, which require only the availability of discrete values that are recorded during the
relevant experimental measurements.
Practical value. The results of the study were verified on experimental rolling stock with modernized devices in the brake systems
of bogies. They can be used in the design, upgrade and operation of the brake systems of both the rolling stock which is currently
in operation and the new generation of bogies of freight cars.
Keywords: ARIMA model, freight car, wear, forecasting, brake pad, brake rigging
Collection of scientific works of the Ukrainian State University of Railway Transport, 2019
The article deals with the forecasting of the technical state of the traction gearboxes of electr... more The article deals with the forecasting of the technical state of the traction gearboxes of electric trains with discrete stochastic models. During every 3rd level maintenance, the vibration signals of the previously selected fourteen traction gearboxes were recorded, and the box counting dimension was estimated.
For the forecasting of the technical state, among the broadcast deterministic methods of exponential smoothing and trend analysis, the most effective ARIMA model was selected. This model in certain applications has higher precision than GRNN and BPNN neural networks. The implementation of the ARIMA model requires accomplishment of the identification, evaluation, fitting and practical application stages for the forecasting model. The identification was performed due to the one-step differentiation with a further stationarity check according to the autocorrelation function and partial autocorrelation function.
The evaluation of the ARIMA model with the different orders of the autoregression component and moving average component was done. For the model fitting, the Akaike’s information criterion and Bayes information criterion were calculated, and the autoregression component of the ARIMA model having the minimum values of these criteria was selected. The forecasting of the fitted ARIMA models with an 80 % confidence interval was done for the period since the 71st 3rd level maintenance up to the next 2nd level current repair. After disassembling 14 traction gearboxes during the 2nd level current repair, the insufficient amount of a lubricant in the first gearbox, wear of a roller bearing of the front cover in the second gearbox, a crack of a bearing ring of the front cover in the third gearbox, a tooth break of a gear in the fourth traction gearbox, destruction of a bearing in the fifth gearbox and a tooth crack of its gear were detected. It was established that the ARIMA model is difficult to upgrade for the evaluation of other data, which requires the implementation of measures for a resimulation. Despite high complexity, the absence of the automatic process for the calculation and the necessity of performing several iterative procedures, the calculated minimum value precision of the ARIMA model forecasting is equal to 91.4 %
Key words: accuracy, ARIMA model, electric train, forecasting, traction gearbox
Collection of scientific works of the Ukrainian State University of Railway Transport, 2019
The paper deals with the property of discrete wavelet transform based on the multiresolution anal... more The paper deals with the property of discrete wavelet transform based on the multiresolution analysis feature to identify different types of the gear and bearing vibration. The direct analysis of the vibration time series by the use of conventional statistical measures, such as mean, root mean square, standard deviation, is not always useful due to the complexity of the signal. It was proposed to choose the best mother wavelet which is able to identify the transients in vibration signal according to the calculated minimum value of Shannon entropy, which quantifies the level of uncertainty of a given vibration time series. The main idea is that when a bearing is healthy, it will produce low amplitude random vibration with a uniform-like probability mass function and as the fault occurs and progresses some probability mass function component will be prevalent with a higher probability of occurrence. The chosen Daubechies wavelet of the 4-th order has decomposed the acquired vibration signal of the traction gearbox of electric train into approximated and detailed coefficients on four decomposition scales with further reconstruction of the signals on the appropriate scales according to the above-mentioned coefficients. The autocorrelation was applied for the detection of deterministic and random components in the reconstructed signals through evaluation of the impulse periodicities of the reconstructed signals according to the detailed coefficients at all scales and has taken the sinusoidal shape for the reconstructed signals according to the approximated coefficients. It was established that a deterministic vibration component dominates and there are no bearing damage features in the reconstructed signals according to the approximated coefficients due two strong gearmesh harmonics. The presence of impulse periodicity on the reconstructed signals according to the detailed coefficients at the second decomposition scale is possible to monitor due to the correlogram, which can be explained by the periodic contact of the damaged element with other elements during their rotation in bearing. The kurtosis is applied as a reliable tool for the frequency band selection where the bearing vibration has the strongest
excitation.
Кeywords: bearing, impulse, spectrum, traction gearbox, vibration, wavelet.
Technical Sciences and Technologies, 2019
Urgency of the research. It is established that introduction of effective vibrodiagnostics techni... more Urgency of the research. It is established that introduction of effective vibrodiagnostics techniques, which are capable of the early identification of the traction gearbox faults of electric trains, is a relevant task.
Target setting. The vibrodiagnostics of mechanical units of electric trains is known to be a trustworthy technique that is able to extract impulsive components with a periodic repetition in accordance with the revolution of the faulty parts of gears or bearings. The main problem is development of effective methods for the noise elimination and identification of the technical condition features of gears and bearings.
Actual scientific researches and issues analysis. Recently the calculation of the kurtosis index has taken a leading role in the extraction of weak periodic impulses. However, unclear recommendations on increasing the accuracy of these results needed improvement, which led to emergence of the spectral kurtosis method based on the filters for the reconstruction of random signals with a high level of additive stationary noise.
Uninvestigated parts of general matters defining. The identification of a bearing vibration component and the influence of a gear component in high frequency band on it were not clearly understood.
The research objective. The objective of the paper is identification of the informative frequency band of excited bearing vibration by means of the spectral kurtosis method.
The statement of basic materials. The paper selects the best window length of the Short-Time Fourier Transform by means of the spectral kurtosis method, which allows to identify the highest level of the spectral kurtosis and to find the proper frequency band.
Conclusions. By means of Wiener filter, the broadband structures of the traction gearboxes vibration of electric trains were identified, and the frequency band with excited resonance component of the bearing vibration was registered.
Keywords: bearing, frequency, gear, electric train, spectral kurtosis, vibration.
Collected scientific works of Ukrainian State University of Railway Transport, 2018
The paper deals with the time-synchronous averaging technique of vibration signals for extraction... more The paper deals with the time-synchronous averaging technique of vibration signals for extraction of the features of the technical state of electric train traction gearboxes. Simulation of the vibration of the faultless gear mesh, and gearwheel with a cracked tooth and strong deterministic component was carried out. The rotation frequencies of gearwheel and pinion, and the mesh frequency are clear visible on the vibration power spectrum. Taking into account the poor visibility of the sidebands around mesh frequency on the power spectra, the time-synchronous averaging technique enhances the sidebands, the width of which corresponds to the gearwheel rotation frequency and influences weakening of the sidebands of the faultless pinion. During the experimental research the traction gearbox vibration of electric train was acquired on the test bench in a depot, and the calculated power spectrum has not identified sidebands and mesh
frequency. After the time-synchronous average within 18 revolutions for the pinion and gearwheel
frequency, the vibration power spectra were acquired. The time-synchronous average technique
eliminates random and non-synchronous components while keeping the frequency components that
are synchronous to the rotating frequency of the pinion and gearwheel of the traction gearbox. On
the vibration power spectra in the vicinity of the mesh frequency the strong sidebands are visible,
and they do not coincide with the previously calculated rotation frequencies, which means the
absence of gearwheel faults. At the vibration power spectrum for the pinion there are six sidebands
in the vicinity of the meshing frequency that coincide with the previously calculated pinion rotation
frequencies, which means the presence of localized or distributed faults for the pinion.
Disassembling of the traction gearbox shows a crack of the rolling bearing inner race and faultless
state of the gear mesh. It was established that the sensitivity of the time-synchronous average
technique makes it possible to identify not only localized faults extraction but also distributed faults
such as eccentricity and misalignment of the pinion shaft through the modulation of the sidebands
on vibration power spectra of the gearwheel.
International Journa l of Engineering &Technology (IJET), 2018
Purpose: To improve the performance of vibration spectral methods in identification of bearing el... more Purpose: To improve the performance of vibration spectral methods in identification of bearing element faults of freight car axle-boxes. Approach: An algorithm for simulating the expected vibration signal of outer race bearing was implemented. The autoregressive filter and minimum empirical deconvolution method was applied to identify the ball pass outer-race fault frequency and its harmonics on the envelope spectra and squared envelope spectra which were extracted in the proper frequency band. Results: The simulated vibration signal of a faulty bearing shows suitability of the autoregressive filter and minimum empirical deconvo-lution method, envelope and squared envelope spectra for outer race fault identification. There were observed a lower amount of features and their impulse sharpness in outer race faults in the bearing test rig than on the spectra in the wheelset test rig. Conclusions: The deterministic components are removed in the residual signal after using the AR filter and the impulse and noise components that decrease the kurtosis value remain in it. The MED technique additionally enhances the magnitude of increased BPFO components after using the AR filter and, together with it, provides satisfied performance and increases the efficiency of vibration diagnostics.
Collection of scientific works of the Ukrainian State University of Railway Transport, Dec 15, 2014
Taking into consideration the necessity of applying monitoring means to diagnose the actual techn... more Taking into consideration the necessity of applying monitoring means to diagnose the actual technical condition of driving axle bearings to reduce the number of their failures, some new devices have been developed which take account of a sudden nature of failures of some bearings elements. These devises have been designed to detect and analyse certain bearing vibration characteristics during the motion of the locomotive. The built-in device for monitoring the vibration of the driving axle bearings is attached to the surface of the tractive motor bearing assembly by means of a pin. The pin is screwed in a pre-drilled hole and the part is attached to it. Attaching by means of pins is one of the ways that don't distort the operation characteristics, the resonance frequency and the dynamic range of vibroaccelerometers. To increase the trustworthiness of the results obtained, the monitoring signals are analyzed only in the mode of the locomotive stationary speed. Objective: develop built-in means to monitor vibration characteristics of driving axle bearings to increase the efficiency of planned preventive system of locomotive maintenance and repairs.
Keywords: bearing, diagnosis, failure, locomotive, operation, vibration
Journal of Mechanical Engineering National Technical University of Ukraine "Kyiv Politechnic Institute" , Dec 29, 2016
Goal: increase of efficiency for diagnostics of rolling bearing faults using an autoregressive mo... more Goal: increase of efficiency for diagnostics of rolling bearing faults using an autoregressive model to calculate AR coefficients and further application of pre-whitening AR filter and envelope spectra to extract weak fault features. Method of doing research: diagnostics of rolling bearing faults involves signal acquisition technique and application of the AR model for better analysis of short duration signal properties and impulses. The Akaike information criterion is used to ensure optimum adaptation of AR coefficients to bearing faults. The AR coefficients are defined with Yule-Walker equitation. The advantages of pre-whitening AR filter are presented in comparison of the low efficiency of the power spectral density of the parametric model. The experimental study of vibration characteristics of the auxiliary electromotor body of electric locomotive determines the frequency band 5,5 — 7 kHz of the rolling bearing vibration, and this frequency band can be used for further extraction of the envelope spectra. Value: the research shows a capability of the pre-whitening AR model to gain valuable information not only about different faults concerning outer and inner races and bearing rollers, but also about the technical condition of the cage, the features of which are displayed on the envelope spectra directly after the AR filter