论文部分内容阅读
利用人工神经网络的自适应、自组织学习能力,通过对训练样本集的学习,使用传统的CVDA—84规范、传统的BP神经网络、改进的Rumelhart和MBP神经网络,对注水管道的剩余寿命进行了预测。结果表明,CVDA—84规范偏保守,采用BP以及改进的BP神经网络预测的剩余寿命和观测值基本一致。但采用BP人工神经网络预测时,迭代次数比CVDA多得多。采用改进的 Rumelhart和 MBP神经网络能有效地提高预测速度,改善网络的收敛性,并且使预测精度有所提高。
Using the adaptive and self-organizing learning ability of artificial neural network, the residual life of water injection pipeline is studied by using the traditional CVDA-84 standard, the traditional BP neural network, the improved Rumelhart and MBP neural network by learning the training sample set Predicted. The results show that CVDA-84 is conservative and the remaining life predicted by BP and BP neural network are basically the same. However, with BP artificial neural network prediction, the number of iterations is much more than that of CVDA. The improved Rumelhart and MBP neural networks can effectively improve the prediction speed, improve the convergence of the network, and improve the prediction accuracy.