| In this dissertation,an improved niching genetic algorithm of self-organizing feature mapping is proposed for the problem of “premature convergence” caused by the difficulty of determining the initial radius of traditional niching genetic algorithm.Firstly,using self-organized feature mapping network algorithm to adaptive clustering for the population,and according to the distance between particle and clustered center to determine whether or not generated a new cluster.Then,using the roulette strategy to select the excellent individual after initial clustering of different niching to perform genetic operation.In order to search as many local optimal solutions as possible to provide more possibilities for the final solution,and then find global optimal solution of the problem.Finally,to evaluate performance of the algorithm by Benchmarks function.The results show that the improved niching genetic algorithm has good convergence and can maintain the characteristic diversity of population samples well.At present,easy to develop oilfields are increasingly scarce,how to choose a reasonable well layout,which has become the focus of attention of the people before oilfield put into operation.And by the inspiration of some specific problems in oilfield development process encountered,the improved niching genetic algorithm is applied to the evaluation of reservoirs.Taking the position of injection wells as the decision variables to establish well optimization model,then to determine objective function of the net present value.Based on the principle of “less well and efficient”,the well-Bit,achieving the maximum net present value of reservoir development. |
