基于神经网络的电参数反演载荷算法  

Algorithm for Electrical Parameters and Payload Based on Neural Network

作　　者：王卫江[1] 史玥婷 刘箭言[1] 任桂山[2] 董汉苑 

机构地区：[1]北京理工大学信息与电子学院,北京100081 [2]大港油田采油工艺研究院,天津300280

出　　处：《北京理工大学学报》2015年第7期706-710,共5页Transactions of Beijing Institute of Technology

摘　　要：提出一种基于神经网络的抽油机载荷反演算法,从电机输入电功率和载荷输出非线性逼近的角度建立高维输入输出系统模型,通过对网络误差的负梯度反馈,自动调整权值参数,求解电参数到载荷的映射网络.通过实测数据软件仿真,算法能实现电参数到载荷之间的非线性拟合,误差小于10-3,相关系数达到0.95以上.A BP neural network based algorithm for payload calculation was proposed. A highdimensional input-output system was formed to describe the non-linear relation between electricalparameters and payload. By the use of negative gradient of the network error feedback, networkparameters were automatically adjusted, therefore the mapping connection between electricalparameters and payload can be solved. Experimental results and simulations show that thealgorithm can solve the problem of the non-linear approximation, and achieve an error probabilityof e-3 with correlation coefficient up to 0.95.

关 键 词：电参数 BP神经网络 悬点载荷 示功图 

分 类 号：TN710[电子电信—电路与系统]