摘要
针对油田抽油机井故障的特点,提出了基于T-S模糊神经网络的抽油机井故障诊断方法。即将神经网络的学习能力引入到模糊系统中,将模糊系统的模糊化处理、模糊推理、精确化计算通过分布式的神经网络来表示,从而提高系统的学习能力和表达能力。提出了基于LM优化的BP算法以提高网络收敛速度,利用MATLAB神经网络工具箱建立模糊神经网络诊断模型,经仿真测试表明,所提出的故障诊断方法能有效地对抽油机故障识别,正确率较高、效果较稳定,可提高网络训练及诊断速度。
This paper proposes a method of fault diagnosis base on T-S fuzzy neural network for pumping well, according to the fault characteristics of pumping well, which brings the learning ability of neural network into fuzzy system and uses distributed neural network to express fuzzy processing fuzzy reasoning and precise calculation. So that it can improve learning and expressing ability of system. Put forward improved BP algorithm based on LM and the next is to establishing fuzzy neural network diagnosis model depend on neural network toolbox of MATLAB, the simulation tests show that the fault diagnosis method proposed can recognize the pumping well faults effectively and accurately. Improve training and the diagnosis rate of network.
出处
《系统仿真技术》
2013年第2期141-146,共6页
System Simulation Technology
关键词
模糊神经网络
抽油机
故障诊断
fuzzy neural network
pumping well
fault diagnosis
