基于灰色理论和遗传算法的单井产油量预测方法

Grey Theory and Genetic Algorithm for Single Well Oil Production Prediction

提出了一种基于灰色理论和遗传算法的时间序列预测模型。该模型首先将时间序列进行一次累加,再对累加后的新序列进行相空间重构,将重构后的样本作为遗传算法的训练样本。用波兰表达式的二叉树形式构造种群中的染色体,通过不断交叉、变异和筛选,得到最佳的表达式组合。将输出值进行累减还原,得到时间序列的预测值。将仿真实验与神经网络和多基因表达式编程算法进行对比,实验结果表明,新预测模型具有更强的泛化性能。在混沌时间序列预测和油田单井产量预测的实例分析中,新算法较其他算法都表现出了更强的泛化性能。因此,新算法具有较好的应用前景和一定的研究价值。

This paper presents a time series forecasting model based on grey theory and genetic algorithm.Firstly,the time series was accumulated at a time,and then the new sequence was reconstructed.The reconstructed samples were used as training samples of the genetic algorithm.The chromosomes in the population were constructed by the two fork tree forms of the Polish Expression,and the best combination of expressions was obtained through continuous crossover,mutation and selection.Through inverse accumulated generating operation(IAGO)of the output value,the predicted values of the time series were obtained.The simulation results were compared with the neural network and the multi gene expression programming algorithm.The experimental results show that the new prediction model has better generalization performance.