油藏生产优化组合寻优过程往往会陷入局部最优解的陷阱中,无法在短时间内跳出局部最优解,且计算耗时长。首次将遗传—退火混合算法(SG-AG)引入到油藏模拟的注采参数优化中，基于概率机制迭代寻优方向,对种群进行大规模扰动,在交叉变异中引入退火模拟算法,加强扰动产生新群体,避免寻优陷入局部收敛，提升了全局搜索性。同时,通过SPSA随机扰动算法,计算目标函数近似梯度确定单次优化的扰动步长。将算法与油藏数值模拟结合,实现对A油田转注后注采结构参数的优化。优化结果显示,模型能够有效控制并优化油藏生产制度时间,优化后预计累产油提高约1.9×104 m3,达到了增油控水的目的,为类似油田转注开发提供了借鉴。

The optimization process of reservoir production optimization combination often falls into the trap of local optimal solution, while it is impossible to jump out of local solution in a short period of time, and the calculation is time consuming. In this paper, a hybrid genetic-annealing algorithm(SG-AG) is introduced into the optimization of injection production parameters in reservoir simulation for the first time. Based on the iterative optimization direction of probabilistic mechanism, the population was disturbed on a large scale, and the SG algorithm is introduced into the crossover and mutation calculation to strengthen the disturbance and generate a new population, which avoids the local convergence of the optimization and improves the global searching performance. At the same time, SPSA random perturbation algorithm is used to calculate the approximate gradient of the objective function to determine the disturbance step size of the single optimization. By combining the algorithm with reservoir numerical simulation, the injection production structural parameters of oil field A are optimized. The optimization results show that the model can effectively control and optimize the production schedule of the reservoir. After optimization, the cumulative oil production is expected to increase by about 1.9×104m3, achieving the purpose of oil increase and water control, which provides reference for oil fields of the same nature to be converted into injection development.