摘要
将遗传算法和传统的BP神经网络有机结合起来,提出了一种对注水管道水质腐蚀影响因素进行大小排序的新方法,即二层改进遗传神经网络法。对某实验区注水水质腐蚀的影响因素进行示例分析,结果表明,二层改进遗传神经网络法得到的影响因素排序结果比灰色关联得到的排序准确一些,更能反映腐蚀的实际情况。提出的二层改进遗传神经网络法和灰色关联分析法具有相同的应用范围。该试验区注水水质的主要影响因素排序为:溶解氧(0.877)>pH值(0.856)>SRB(0.84)>温度(0.811)>压力(0.78)>CO_2(0.76)>流速(0.736)>0.7。
A kind of new method is put forward to sort for injecting pipeline corrosion influence factors and determine main influence factors, namely two layers modified Genetic Algorithm (GA) neural network method. Together with given experimental zones injecting water quality corrosion influence factors examples, it shows that prediction results of two layers modified GA neural network method are better than those of gray relation analysis method. The new method in the paper has same application range as gray relation analysis method, and has great breakthrough in mathematical theory and engineering application. In given experimental zones, corrosion main influence factor are sorted below, O2(0. 877) >pH(0. 856) >SRB (0.84) >temperature (0.811) >pressure (0.78) >CO2(0.76) >flow velocity (0.736) >0.7.
出处
《中国海上油气（工程）》
2003年第2期23-28,共6页
China Offshore Oil and Gas
