BP神经网络在炼油污水回用于循环水系统中腐蚀率的预测

摘要

The structure and principle of Back-Propagation (BP) neural network were introduced. The method and steps for building the BP neural network by the Graph User Interface (GUI) of MATLAB were also discussed. The BP neural network prediction model was established by using conductivity and pH as input vectors and corrosion rate as output vector, and the model was utilized to predict the corrosion rate of refinery wastewater reused as circulating cooling water. The results indicate that the established prediction model with three-layer structure has the higher forecast accuracy to the corrosion rate of circulating water. So application of the artificial neural network in corrosion rate prediction of circulating cooling water is feasible, and also suggested that it has extensive practicability.%介绍了BP神经网络的构造及基本原理，阐述了利用MATLAB的GUI建立BP模型的方法和步骤，并将其应用于炼油污水回用于循环冷却水系统腐蚀率的预测，建立一个以电导率和pH为输入向量、腐蚀率为输出向量的BP神经网络预测模型。结果表明，采用GUI建立的三层结构的BP神经网络模型，对炼油污水循环冷却水系统的腐蚀率的预测具有较高的预测精度。说明人工神经网络在循环水腐蚀预测中的应用是可行的，具有一定的应用价值。