Engine Life Prediction based on Degradation Data

doi:10.23940/ijpe.18.12.p1.29052914

Abstract

Abstract:

The motor hour (working time) of an armored vehicle’s engine reflects its technical state to a certain extent. However, even the same type of engine with the same motor hour shows very different technical states in different working environments. At the same time, it is difficult to obtain the full life data or physical failure mechanism required by the traditional life prediction method. In view of the above problems, a model of engine life prediction based on degradation data and neural networks is built in this paper. Firstly, the degradation parameters are selected according to certain principles, and the sample data are standardized. Then, the principal component analysis method is used to simplify multiple parameters to a comprehensive parameter, and the interpolation method is applied to get the parameter’s time series data as the train data of the neural network. Finally, the life prediction model of the engine based on the neural network is established. The validation results indicate that the model runs accurately. It is also practical and worthy of being used abroad.

Key words: degradation data, neural network, life prediction, principal component analysis

Yanhua Cao, Jinmao Guo, Yong Li, and Huiqiang Lv. Engine Life Prediction based on Degradation Data [J]. Int J Performability Eng, 2018, 14(12): 2905-2914.

Add to citation manager EndNote|Reference Manager|ProCite|BibTeX|RefWorks

Figures/Tables 12

Figure 1

Figure 2

Table 1

Table 2

Table 3

Table 4

Table 5

Table 6

Figure 3

Figure 4

Figure 5

Table 7

References 16

[1]	Y. B. Liu, J. M. Liu, X. Y. Qiao , “Application of Supportive Vector Machine in Technical States Evaluation of Diesel Engine,” Journal of Academy of Armored Force Engineering, Vol. 23, No. 2, pp. 38-40, 2009
[2]	F. Zhao and H. W. Wang , “Research on Condition based Maintenance for Aero-Engine Using Hidden Markov,” Aeronautical Computing Technique, Vol. 40, No. 5, pp. 15-19, 2010
[3]	X. Liang, X. S. Li, L. Zhang , “Survey of Fault Prognostics Supporting Condition based Maintenance,” Measurement & Control Technology, Vol. 26, No. 6, pp. 5-8, 2007 doi: 10.1631/jzus.2007.A1596
[4]	Y. H. Cao , “Research on Autonomic Logistics Key Technologies for Armored Equipment,” Ph.D. Dissertation, Academy of Armored Forces Engineering, 2012
[5]	H.W. Wang and K. N. Teng , “Review of Reliability Evaluation Technology based on Accelerated Degradation Data,” Systems Engineering and Electronics, Vol. 39, No. 12, pp. 2877-2885, 2017
[6]	Z. G. Guo , “Reliability Analysis of Barrel’s Life based on Performance Degradation Data,” Master’s Thesis, Nanjing University of Science and Technology, 2011
[7]	X.X. Kou and X. P. He , “Time Series Prediction based on RBM Neural Network,” Mathematics in Practice and Theory, Vol. 46, No. 9, pp. 173-178, 2016
[8]	W. B. Chen , “The Principle and Practice of Artificial Neural Network, ”Xidian University Press, Xi'an, pp. 44-64, 2016
[9]	Y. H. Cao, S. X. Zhang, and Y. Li , “Design of Engine Condition Detection for a Certain Type of Armored Vehicle,” in Proceedings of OSEC2017, pp. 15-18, 2017
[10]	W.B. Zhang and H.Y. Chen. , “Statistical Analysis of Practical Data and Application of SPSS12.0,” The Posts and Telecommunications Press, Beijing, 2006
[11]	X. P. Zheng, J. J. Gao, M. T. Liu , “Accident Prediction Theory and Method, ” Tsinghua University Press, Beijing, 2009
[12]	D. Zhang, C. S. Zu, C. C. Zhao , “Principal Component and Neural Network Combined Fuel Consumption Forecast,” Agricultural Equipment & Vehicle Engineering, Vol. 53, No. 6, pp. 47-52, 2015
[13]	J. J. Jiang , “Research on Prediction of Lead-Acid Batteries’ Remaining Discharge Time”, Electrical & Energy Management Technology, No. 7, pp. 73-78, 2017
[14]	Z. L. Huang, Z. B. Wang, J. W. Wang , “Review of Reliability Evaluation Methods based on Performance Degradation Data,” Electrical & Energy Management Technology, No. 19, pp. 35-40, 2017
[15]	Y. H. Cao, Y. Li, Y. Zheng, X. Zan , “Runtime Forecast of Military Vehicle Diesel Engine based on BP Neural Network,” in Proceedings of QR2MSE 2016, pp. 509-512, 2016
[16]	H. Y. Chen, J. M. Zhu, Z. N. Ding , “A Survey of Researches on Combination Forecasting Models and Methodologies,” College Mathematics, Vol. 33, No. 4, pp. 1-9, 2017

Number	Use time/h	${{\hat{p}}_{\max }}$/MPa	${{\hat{\theta }}_{fd}}$/oCA	${{\hat{V}}_{p}}$/g²
1	11	2.8950	34.9500	54.3540
2	38	2.8744	33.9796	48.7860
3	59	2.8647	33.9000	46.8470
4	103	2.8228	32.8462	42.8991
5	140	2.8090	32.1142	42.0632
6	187	2.7958	31.1880	38.8610
7	190	2.7847	31.1671	36.0951
8	250	2.7736	31.0942	31.3503
9	300	2.7521	30.3144	32.9560
10	320	2.7439	29.5521	32.7874
11	350	2.7330	29.0114	30.8780
12	390	2.7132	28.9751	31.7270
13	450	2.6739	27.5521	28.4621
14	497	2.6647	26.6285	26.5810
15	507	2.6527	26.4883	27.8581
16	550	2.6429	25.8854	25.1580

Number	Eigenvalue	CR/%	CCR/%
1	2.936	97.880	97.880
2	0.060	1.985	99.865
3	0.004	0.135	100.000

Parameters	Eigenvectors
Parameters	No.1	No.2	No.3
${{{\hat{p}}'}_{\max }}$	0.581	-0.333	-0.743
$\hat{\theta }_{fd}^{'}$	0.579	-0.472	0.665
${{{\hat{V}}'}_{p}}$	0.572	0.816	0.082

Number	Use time/h	Standardized data and the PCX
Number	Use time/h	${{{\hat{p}}'}_{\max }}$	$\hat{\theta }_{fd}^{'}$	${{{\hat{V}}'}_{p}}$	X
1	11	1.6627	1.6318	2.1069	3.1160
2	38	1.4047	1.2874	1.4641	2.3989
3	59	1.2832	1.2591	1.2402	2.1840
4	103	0.7582	0.8850	0.7845	1.4017
5	140	0.5854	0.6252	0.6880	1.0956
6	187	0.4200	0.2964	0.3183	0.5977
7	190	0.2809	0.2890	-0.0010	0.3300
8	250	0.1419	0.2631	-0.5488	-0.0790
9	300	-0.1275	-0.0137	-0.3634	-0.2898
10	320	-0.2302	-0.2843	-0.3829	-0.5173
11	350	-0.3668	-0.4762	-0.6033	-0.8339
12	390	-0.6148	-0.4891	-0.5053	-0.9294
13	450	-1.1071	-0.9942	-0.8822	-1.7235
14	497	-1.2224	-1.3220	-1.0994	-2.1046
15	507	-1.3727	-1.3718	-0.9520	-2.1364
16	550	-1.4955	-1.5858	-1.2637	-2.5100

Number	Use time/h	X
1	550	-2.5100
2	550	-2.5723
3	550	-2.5510