基于PCA-BP神经网络的随钻参数岩性智能感知方法研究

Research on Lithology Intelligent Sensing Method of Drilling Parameters Based on PCA-BP Neural Network

针对地下金属矿山掘进过程中岩层信息参数获取难度大、岩性判别精度低等问题,提出一种基于主成分分析法(PCA)和BP神经网络的岩性智能感知方法:首先,通过理论分析,构建基于PCA-BP神经网络的随钻参数岩性智能感知模型;然后,通过相似材料模拟3种不同岩性的“三层一体”岩样,利用特制的凿岩机钻进试验台获取钻机钻进速度、回转速度、钻压、回转压力、回转扭矩和泥浆泵压力等6种随钻参数;最后,对理论模型进行训练和检验。结果表明,PCA-BP神经网络岩性智能感知方法与传统的BP神经网络岩性智能感知模型相比,减少了模型计算量,且岩性智能感知的准确率得到了有效提升。

There are complex geological conditions and poor underground production conditions in underground metal mines,leading to the difficulty of obtaining rock information parameters and the low accuracy of lithology identification in the process of underground tunneling.Thus,a lithology intelligent sensing method based on principal component analysis(PCA)and BP neural network was proposed to improve the intelligent level of tunneling.Firstly,through theoretical analysis,an intelligent lithology perception model with drilling parameters based on PCA-BP neural network was constructed.Then,the three-layers-in-one rock samples with different lithologies were simulated by similar materials,and six kinds of drilling parameters such as drilling speed,rotary speed,drilling pressure,rotary pressure,rotary torque and slurry pump pressure were obtained by using a special drilling rig.Finally,the theoretical model was trained and tested.The results show that compared with the traditional BP neural network lithology intelligent sensing model,the lithology intelligent sensing method based on PCA-BP neural network reduces the amount of model calculation,and effectively improves the sensing.