Predicting the Residual Life of Injecting Water Pipeline with the Artificial Neural Network

In this paper, self-adapting and self-organizing study ability of the artificial neural network is used to predict the residual life of injecting water pipeline. By means of field example, four different predicting methods, for example, CVDA—84 standard, conventional BP neural network, improved Rumelhart and MBP neural network, are adopted to predict residual life of injecting water pipeline. The predicted results show that CVDA—84 standard is somewhat conservative, BP neural network, the improved Rumelhart and MBP neural network methods are good agreement with field data, while iterate number is great. The improved Rumelhart and MBP neural network methods can overcome the shortcomings exsited in the rest methods.