遗传神经网络在随钻测斜仪传感器补偿中的应用  

Application of Sensor Compensation in MWD Instrument Based on Genetic Elman Neural Network

作　　者：王秀芳[1] 句莉莉[1] 齐元俊[1] 魏春明[2] 

机构地区：[1]东北石油大学电气信息工程学院,黑龙江大庆163318 [2]大港油田定向井公司,天津300280

出　　处：《化工自动化及仪表》2010年第10期70-72,共3页Control and Instruments in Chemical Industry

基　　金：黑龙江省教育厅科技研究项目(11551027);黑龙江省研究生创新科研资金项目(YJSCX2008-082HLJ)

摘　　要：为了进一步提高随钻测斜仪的测量精度和工作稳定性,采用Elman神经网络进行建模,将自适应遗传算法作为Elman网络的权阈值学习算法,克服了传统方法训练时间长、容易陷入局部最优解等缺点。仿真结果表明,与采用Elman网络算法相比,补偿精度提高两个数量级,补偿效果稳定。In order to improve the measurement precision and stability of MWD instrument,Elman neural network model was created and self-adaptive genetic algorithm was utilized to optimize weights threshold value of the right of Elman network,which overcame the disadvantages of traditional method,such as training for a long time,easy to fall into local optimal solution.Simulation results show that the error accuracy increases 2 orders of magnitude,compared with Elman network,the compensation effect is very stable.

关 键 词：自适应遗传算法 ELMAN网络 补偿 随钻测斜仪 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]