基于神经网络的卡钻预测

The sticking prediction based on neural networks

针对钻井过程的复杂性、不确定性等特点,提出应用神经网络技术预测卡钻事故,建立事故预测模型。选取对卡钻事故的发生有较大影响的变量作为神经网络的输入项,分析钻井现场实时监测的卡钻数据和正常运行的数据,应用钻井现场数据对神经网络进行训练以此建立卡钻事故预测模型,最终通过钻井现场数据证实该网络具有对卡钻事故做出准确预测的能力以及良好的泛化能力。

TCR+FC static var compensator has been designed based on the voltage flicker, harmonic pollution and low power factor caused by drilling equipment. TMS 320F28027 has been used as a main control chip while the electrical energy metering chip—ATT7022B with high-accuracy had been used and the dynamic reactive power compensation has been achieved. The test results show that the controller is characterized by a high-accuracy in calculation, high rate in response, good dynamic performance and good reactive compensation effects. It can improve the power factor, enhance the power-supplying quality of power net and have better promotion value on energy-saving & consumption-reducing.