基于遗传变异与BP混合算法的抽油井流入动态预测

针对BP神经网络中存在的局部极小问题,提出了一种基于遗传变异和BP混合的网络算法。在综合考虑了油层渗透率、地层压力、井底流压等对油井产液量产生影响的因素的基础上,建立了基于上述网络算法的抽油井流入动态预测模型。实验结果表明,该模型具有较高的预测精度,预测结果可靠,可以作为抽油井流量估计的一种有效手段。

Application of GA and BP algorithm in oil well drifting forecast

In accordance with the local minimum question that exists in BP's neural network, the paper presented a network model based on the features of genetic algorithm and BP network. Considering oil penetrable rate, stratum pressure, liquid pressure, etc, which affect oil well yield, a network model of oil well drifting forecast was established. Experimental results indicate that the method possesses high forecasting precision and reliable calculating result. It can be regarded as an effective method for estimating oil well volume of flow.