基于BP神经网络的管道失效状态预测

针对管道系统历史数据缺乏、失效机理非线性的特点,选用具有良好自学习性、鲁棒性等特点的BP神经网络对管道失效状态进行预测.在对管道外表面涂层检测数据预处理的基础上,采用BP神经网络进行建模分析,通过样本的反复训练,得到预测集的相对误差最大为8.3％,预测结果比较理想.结果表明：用BP神经网络能够较好地预测管道的失效状态值,为管道的预防性维修提供理论依据.

Prediction of the pipeline failure state based on BP neural network model

BP neural network model with the characteristic of self-study habits, robustness and so on, is chosen for the prediction of the pipeline failure, for lack of historical data, the nonlinear characteristics of failure mechanism of pipeline system. On the basis of data preprocessing in the pipe outside surface coating, BP neural network model are adopted to analyze, a relative error of the prediction set is got by the repeated training samples, the maximum is 8.3%, the forecasting results would be ideal. The results show that BP neural network can be used to predict pipeline failure status value well, provide the theoretical basis for the preventive maintenance of pipeline.