论文部分内容阅读
根据神经网络的非线性和良好的函数逼近特性,提出了基于人工神经网络的灰色模型、多项式回归模型组合的输气管道腐蚀速率预测模型.此组合模型将最佳组合权重隐含在网络的连接权中,兼具灰色预测、回归预测和神经网络预测的优点,克服了原始数据少,数据波动大对预测精度的影响,也增强了预测的自适应性,在客观地反映输气管道腐蚀速率变化趋势方面具有一定的优势.通过实例分析,表明预测值与实际结果有很好的一致性.
According to the nonlinearity of neural network and the good approximation function, a gray model and polynomial regression model based on artificial neural network is proposed to predict the gas pipeline corrosion rate.The model combines the optimal combination weight with the network connection It has the advantages of gray prediction, regression prediction and neural network prediction. It overcomes the influence of less original data and large data fluctuation on prediction accuracy, enhances the predictive adaptability and objectively reflects the corrosion rate of gas pipeline And has some advantages in the trend of change.Through the case analysis, it shows that the predicted value is in good agreement with the actual result.