摘要
基于冷蜡沉积实验装置所得实验数据,在分析BP、RBF神经网络结构原理的基础上,采用BP神经网络和RBF神经网络分别建立蜡沉积速率模型,计算预测蜡沉积速率,并且对比相同实验数据下两类神经网络模型对蜡沉积预测的精度。结果表明,BP神经网络和RBF神经网络预测精度均满足要求。BP神经网络预测时间要比RBF神经网络更长,而且当神经网络维数增加时预测值的精度不一定会增加;在模拟时要反复尝试隐含层节点个数和其他参数,而RBF神经网络在数据的训练过程中就已给出隐含层节点个数,学习速度更优于BP神经网络,对新数据的适应性更好,在满足精度条件下更易得到最优解。
Based on experimental data obtained cold wax deposition experimental equipment, and on the basis of the analysis of BP,RBF neural network structure principle, establish the wax deposition rate model with BP neural network and RBF neural network and prediction of wax deposition velocity, then contrast the prediction accuracy of these two neural network model of wax deposition under the same experimental data. The results show that the prediction accuracy of the BP neural network and the RBF neural network can meet the requirements. The BP neural network has a longer prediction time than the RBF neural network,and the accuracy of the predicted value is not necessarily increased when the dimension of the neural network is increased;The number of nodes and other parameters of the hidden layer should try repeated during the simulation.While the RBF neural network has given the number of hidden layer nodes in The learning speed is better than the BP neural network,It is easier to get the optimal solution under the condition of satisfying the precision.
作者
张煜
王力
刘鹏
李星雨
ZHANG Yu;WANG Li;LIU Peng;LI Xing-yu(College of Petroleum Engineering, Northeast Petroleum University, Daqing 163348, China)
出处
《化学工程师》
CAS
2018年第4期16-19,共4页
Chemical Engineer
关键词
神经网络
相对误差
蜡沉积速率
网络结构
neural network
relative error
wax deposition rate
network structure