Research On Intelligent Injection Allocation Strategy Of Water Injection Well Based On AGA-BP Neural Network

As the second form of oilfield exploit,layered water injection is the crucial to stabilizing reservoir pressure,solving inter-layer conflicts,adjusting the uneven distribution of injecting water on the oil layer plane,and increasing oil production.However,oilfield injecting water wells have been facing two important problems in the injection of intervals.One is fine injecting water and accurate prediction of the injection rate of intervals.This is very important for keeping the energy counterpoise of the formation and decreasing the inefficient cycle of water.The second is to stabilize the injecting water flow.In the process of injecting water,controlling the flow rate of the interval of the injecting water well is the crucial to satisfying the injecting water qualification rate.Aiming at the problem of inaccurate prediction of the layer allocation volume,this article analyzed the influencing elements the interval injection rate of the injecting water well based on the real yielding data of the oilfield,and established a forecasting pattern of interval injection volume in view of BP neural network,and illustrated in detail the topological structure of the forecasting pattern,the number of neurons and the learning steps.In order to relieve the weak points of BP neural network that are easy to fall into local optima and slow astringent speed,this article uses AGA to improve the basic forecasting pattern of BP neural network,and found an AGA-BP neural network layer allocation volume forecasting pattern.The real production data of the water well are used to train the interval injection rate pattern,and the real interval injection rate is compared with the pattern predicted amount to test the effectuality of the forecasting model.The designed three interval injection rates are allocated the forecasting model is compared and analyzed by experiments at last,and it is concluded that the forecasting accuracy of the AGA-BP neural network layer distribution fluence forecasting pattern is higher and the astringent velocity is faster.There are unequal in the nature and pressures of each interval for injecting water wells.A reasonable injection rate is the prerequisite for improving oil recovery.However,the intervals are not independent of each other,and the phenomenon of moving one layer and moving the whole well may occur,So it is must have to ensure the smoothness of the flow control of the interval.This article aims to maintain the stable injecting water pressure of the interval flow control system of the injecting water well,and devise an injecting water flow controller based on HSIC PID,which can be accurately controlled when the interval flow and the set injection amount deviate.Through the Simulink control platform,build a pattern of the injecting water flow control system,and perform simulation verification and flow control test.The comparison with regular PID and fuzzy PID control confirms the utility and excellence of the HSIC PID control arithmetic used in the injecting water flow control system.