The Research On The BP-based Optimization And Control Algorithms For The Injection-production Model System

The operating efficiency is an important index of injection-production system of offshore oil platform.How to realize the efficiency optimization and control have always become research problems because the efficiency is influenced by the characteristics of the water injection pump and pipeline.Several optimization and control schemes based on BP algorithm are presented in this paper in order to reach the optimization and control goal.And they are verified on the injection-production dynamic simulation system of offshore oil platform.(1)The plant properties of injection-production system are analyzed.The injection production system is chosen as the control plant in this paper,and the process flow and data communications of the system are analyzed.The factors influencing the system efficiency are summarized and the methods to improve the system efficiency are put forward,according to the operation characteristics of the pumps and system.The paper also builds the automatic control system of the injection-production dynamic simulation system according to the practical requirements.Then,the control scheme is designed based on automatic island Advantys STB,King view,Matlab and the experimental device.(2)The neural network optimization and control scheme based on BP algorithm is designed.The system efficiency is set as the optimization and control goal,and the structure of neural network controller and learning algorithm are determined.Based on the practical application,the sum of several control errors is determined in input layer to improve control accuracy.To increase stability of the experiment,the filter is designed to overcome the fluctuation influence.(3)The neural network optimization and control scheme based on improved BP algorithm is designed.Several improved BP algorithms,such as changing the excitation function,adding the momentum and changing the learning rate,are proposed in this part to overcome the disadvantages of the traditional BP algorithm,such as falling into local minimum easily,converging of the error slowly and long learning time.The experiment results verify that the improved schemes have better control effects than the traditional control schemes.(4)The integrated control schemes based on BP algorithm and improved BP algorithm are designed.In the integrated control scheme,the producing system is set as the main unit and the injection system is the servo one.The system efficiency is chosen as the control goal in the producing system,and the injection system may reach the efficiency optimization working point by the output command of the producing system.The experiment results show that the BP control scheme and improved BP control scheme have better effects than the traditional control schemes.Analyzing the provided optimization and control schemes,the accurate mathematical model is not needed for the neural network control method.The neural network optimization and control method has the characteristics of real-time and parallelism,and it can obtain to the optimal working point under different initial conditions.Compared with the traditional control scheme,the neural network method has better performances,such as the adjusting time,the peak time,the overshoot.