潜油电泵机组及井下传感器状态监测技术研究

Status montoring of electric submersible pump and down hole sensors

建立了基于电力线载波通信的井下多参数状态监测系统结构,提了基于RBF网络时序预测的传感器状态监测方法,引入最近邻聚类算法的思想,提出了一种K-均值聚类算法初始中心选取方法。该方法能够根据样本序列的变化,自动调整初始聚类中心半径的大小,保证精确度的同时可提高计算速度,提出了一种根据故障传感器输出特性参数的变化对井下传感器故障类型进行分类的新方法,对提出的状态监测系统及传感器监测方法及技术的有效性进行了实验研究。实验结果验证了所提出方法的有效性。

The multi-sensor gauge that based on power line cattier communication is presented. The status monitoring method that based on RBF networks time series prediction was proposed for sensors of the gauge. Referring to the nearist neighbor clustering algorithm, an initial center choosing method was given, which could adjust the radius of the initial clustering center automatically with the changing of the sample series. A new fault classification method for down hole sensors fault identification was presented with the changing of the sensor output signal. Experiments were carried out on the down hole gauge and temperature sensors to verify the efficiency of the status monitoring system. Experiment results testify the validity of the proposed method.