Neural Network-Based Reliability Prediction of Free-Running Submarine Model Test System

Abstract:

In response to the issue that FMECA and FTA analysis in the initial experimental and testing stage of free-running submarine model cannot satisfy the actual requirements, a method based on BP neural network is proposed to conduct reliability prediction of free-running submarine models. By inputting the established parameters of free-running submarine models (MTBF) as a BP neural network into neurons and by training the network, the purpose of predicting the reliability of free-running submarine model test system was fulfilled. The simulation results show that the method is effective, feasible and practically meaningful, producing smaller prediction errors.

• Periodical: Applied Mechanics and Materials (Volume 109)
• Pages: 318-322
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.109.318
• Online since: October 2011
• Authors: Ou Liu, Wei Wang, Chun Hui Yang
• Keywords: Free-Running Submarine Model, MTBF, Neural Network (NN), Reliability Prediction

References

[1] CHEN Xiang-guang, PEI Xu-guang: Artificial neural network technology and application (China Power Press, CHINA 2003).

[2] TIAN Jing-wen and so on: Artificial neural network algorithm research and application (Beijing Polytechnic University Press, CHINA 2006).

[3] HUNG Rui-yi, YANG Shao-hua and LI Kun-lan, Application of BP neural network on the prediction of non-operation reliability of electronic, Electronic Product Reliability and Environmental Testing, 2005, 23(5)7-9.