Limits...
A monotonic degradation assessment index of rolling bearings using fuzzy support vector data description and running time.

Shen Z, He Z, Chen X, Sun C, Liu Z - Sensors (Basel) (2012)

Bottom Line: DSI inherits all advantages of ε⁻ and overcomes its disadvantage.A run-to-failure test is carried out to validate the performance of the proposed method.The results show that DSI reflects the growth of the damages with running time perfectly.

View Article: PubMed Central - PubMed

Affiliation: State Key Laboratory for Manufacturing System Engineering, School of Mechanical Engineering, Xi'an Jiaotong University, Xi'an 710049, China. zjshen.2007@stu.xjtu.edu.cn

ABSTRACT
Performance degradation assessment based on condition monitoring plays an important role in ensuring reliable operation of equipment, reducing production downtime and saving maintenance costs, yet performance degradation has strong fuzziness, and the dynamic information is random and fuzzy, making it a challenge how to assess the fuzzy bearing performance degradation. This study proposes a monotonic degradation assessment index of rolling bearings using fuzzy support vector data description (FSVDD) and running time. FSVDD constructs the fuzzy-monitoring coefficient ε⁻ which is sensitive to the initial defect and stably increases as faults develop. Moreover, the parameter ε⁻ describes the accelerating relationships between the damage development and running time. However, the index ε⁻ with an oscillating trend disagrees with the irreversible damage development. The running time is introduced to form a monotonic index, namely damage severity index (DSI). DSI inherits all advantages of ε⁻ and overcomes its disadvantage. A run-to-failure test is carried out to validate the performance of the proposed method. The results show that DSI reflects the growth of the damages with running time perfectly.

No MeSH data available.


Related in: MedlinePlus

The Time-domain waveforms and Hilbert spectrums of test 1 failure bearing: (a) Time-domain waveform on 4,000 min; (b) Hilbert spectrum on 4,000 min; (c) Time-domain waveform on 9,860 min; (d) Hilbert spectrum on 9,860 min; (e) Time-domain waveform on 10,000 min; (f) Hilbert spectrum on 10,000 min; (g) Time-domain waveform on 12,250 min; (h) Hilbert spectrum on 12,250 min.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3472819&req=5

f9-sensors-12-10109: The Time-domain waveforms and Hilbert spectrums of test 1 failure bearing: (a) Time-domain waveform on 4,000 min; (b) Hilbert spectrum on 4,000 min; (c) Time-domain waveform on 9,860 min; (d) Hilbert spectrum on 9,860 min; (e) Time-domain waveform on 10,000 min; (f) Hilbert spectrum on 10,000 min; (g) Time-domain waveform on 12,250 min; (h) Hilbert spectrum on 12,250 min.

Mentions: The monitoring coefficient ε is a comprehensive index which is sensitive to the initial defect and steadily increases with the damage development. The monitoring coefficient ε is utilized to find the initial defect time. The degradation beginning threshold is defined as ε > 0. The samples with ε > 0 are outside of the sphere acquired by training the normal samples. The two assumptions of Equation (18) are necessary in the processes of the initial defect time determined by the threshold ε > 0. They assure that the initial defect time is free from the influence of the vibration randomness, so the moments of 0 ∼ 220 min of test 1 failure bearing are not the degradation beginning time though εi > 0, because of the bearing in the run-in period. Likewise, the moments of 5,585 min, 7,340 min, 7,830 min are not the initial defect time because they are the exceptional points. Their εi > 0 but their neighbor points are all εi < 0. The similar situations appear in the tests 2 and 3 failure bearings. The degradation beginning times of three bearings are determined as 9,860 min, 10,710 min and 3,040 min with εi > 0. The new initial defect moments are 40 min, 0 min and 35 min earlier than that determined by Kurtosis factor. Which initial defect time is more accurate? The contrast experiments shown in Figures 9–11 are carried out to verify them.


A monotonic degradation assessment index of rolling bearings using fuzzy support vector data description and running time.

Shen Z, He Z, Chen X, Sun C, Liu Z - Sensors (Basel) (2012)

The Time-domain waveforms and Hilbert spectrums of test 1 failure bearing: (a) Time-domain waveform on 4,000 min; (b) Hilbert spectrum on 4,000 min; (c) Time-domain waveform on 9,860 min; (d) Hilbert spectrum on 9,860 min; (e) Time-domain waveform on 10,000 min; (f) Hilbert spectrum on 10,000 min; (g) Time-domain waveform on 12,250 min; (h) Hilbert spectrum on 12,250 min.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3472819&req=5

f9-sensors-12-10109: The Time-domain waveforms and Hilbert spectrums of test 1 failure bearing: (a) Time-domain waveform on 4,000 min; (b) Hilbert spectrum on 4,000 min; (c) Time-domain waveform on 9,860 min; (d) Hilbert spectrum on 9,860 min; (e) Time-domain waveform on 10,000 min; (f) Hilbert spectrum on 10,000 min; (g) Time-domain waveform on 12,250 min; (h) Hilbert spectrum on 12,250 min.
Mentions: The monitoring coefficient ε is a comprehensive index which is sensitive to the initial defect and steadily increases with the damage development. The monitoring coefficient ε is utilized to find the initial defect time. The degradation beginning threshold is defined as ε > 0. The samples with ε > 0 are outside of the sphere acquired by training the normal samples. The two assumptions of Equation (18) are necessary in the processes of the initial defect time determined by the threshold ε > 0. They assure that the initial defect time is free from the influence of the vibration randomness, so the moments of 0 ∼ 220 min of test 1 failure bearing are not the degradation beginning time though εi > 0, because of the bearing in the run-in period. Likewise, the moments of 5,585 min, 7,340 min, 7,830 min are not the initial defect time because they are the exceptional points. Their εi > 0 but their neighbor points are all εi < 0. The similar situations appear in the tests 2 and 3 failure bearings. The degradation beginning times of three bearings are determined as 9,860 min, 10,710 min and 3,040 min with εi > 0. The new initial defect moments are 40 min, 0 min and 35 min earlier than that determined by Kurtosis factor. Which initial defect time is more accurate? The contrast experiments shown in Figures 9–11 are carried out to verify them.

Bottom Line: DSI inherits all advantages of ε⁻ and overcomes its disadvantage.A run-to-failure test is carried out to validate the performance of the proposed method.The results show that DSI reflects the growth of the damages with running time perfectly.

View Article: PubMed Central - PubMed

Affiliation: State Key Laboratory for Manufacturing System Engineering, School of Mechanical Engineering, Xi'an Jiaotong University, Xi'an 710049, China. zjshen.2007@stu.xjtu.edu.cn

ABSTRACT
Performance degradation assessment based on condition monitoring plays an important role in ensuring reliable operation of equipment, reducing production downtime and saving maintenance costs, yet performance degradation has strong fuzziness, and the dynamic information is random and fuzzy, making it a challenge how to assess the fuzzy bearing performance degradation. This study proposes a monotonic degradation assessment index of rolling bearings using fuzzy support vector data description (FSVDD) and running time. FSVDD constructs the fuzzy-monitoring coefficient ε⁻ which is sensitive to the initial defect and stably increases as faults develop. Moreover, the parameter ε⁻ describes the accelerating relationships between the damage development and running time. However, the index ε⁻ with an oscillating trend disagrees with the irreversible damage development. The running time is introduced to form a monotonic index, namely damage severity index (DSI). DSI inherits all advantages of ε⁻ and overcomes its disadvantage. A run-to-failure test is carried out to validate the performance of the proposed method. The results show that DSI reflects the growth of the damages with running time perfectly.

No MeSH data available.


Related in: MedlinePlus