改进粒子群算法求解广义Usher油田开发动态预测模型

Using Modified PSO Algorithm to Solve Generalized Usher Model for Oilfield Development Performance Prediction

广义Usher累积产油量和含水率预测模型属于超越函数,涉及未知参数多,求解具有一定的难度。针对该模型特点,对标准粒子群优化算法进行了改进,提出了一种基于Sigmod函数的非线性惯性权重的改进粒子群优化算法,这在非线性函数优化中能够较好地保证算法收敛性,避免算法陷于局部最优。通过实际油田开发数据对广义Usher预测模型进行了求解,结果表明,用改进的粒子群优化算法求解非线性函数具有较强的可靠性和实用性。

The standard generalized Usher model for prediction of cumulative oil production and water cut belong to transcendental functions involving a lot of parameters to be solved,and has a certain difficulty.In this paper,the particle swarm optimization(PSO) algorithm is improved,and a modified PSO algorithm of non-linear inertia weight based on Sigmod function is proposed.This modified PSO algorithm can well guarantee the convergence and avoid falling into local optimal solution during non-linear function optimization.The case study of using real oilfield development data to solve the generalized Usher prediction model indicates that the modified PSO algorithm has good reliability and applicability for the non-linear function solution.