Research On Detection And Recognition Of Measurement While Drilling Mud Pulse Signals In Complex Noise

At present,forms of energy supply and demand in the world is increasingly serious.Greater oil and gas demand makes exploration and development activities gradually move to complex areas with more harsh environments.The application of measurement while drilling(MWD)technology in the exploration activities under complex conditions is more common.The drilled data are mainly transmitted to the ground by mud with the aid of the mud inside the drill string,and the mud pulse signals are very weak in complex noise background after long distance transmission.The effective detection and recognition of the mud pulse signals directly affects the precision control of drilling and target rate.But the current detection and recognition method is inefficient,which makes the effective detection and recognition of mud pulse signals in complex noise background becomes a research hotspot.This paper researched on the mud pulse signals,and the methods of mud pulse signals encoding in current drilling measurement are studied,and two coding formats of PLM encoding and Manchester encoding are emphatically studied.Secondly,the characteristics of mud pulse signals transmission are analyzed,the interference signal source and mud pulse signals caused by attenuation and attenuation factors in the transmission process,and the complicated noise background of mud pulse signals is affected by the low amplitude of the original signals and the noise interference.In view of the problem of low recognition rate for existing of noise in the mud pulse signals,a signal detection method based on adaptive stochastic resonance and a signal recognition method based on deep learning are proposed.This paper studies adaptive stochastic resonance and deep neural network technology.In this paper,the adaptive stochastic resonance method is used to detect the mud pulse signals.The simulated annealing algorithm is used to optimize the parameters of the stochastic resonance system.Then,the stochastic resonance system is used to detect the mud pulse signals in the complex noise background.And based on deep neural network framework,this paper puts forward the stacked wavelet autoencoder,and the structure of stacked wavelet autoencoder is described in detail.Recognition experiments of PLM encoding and Manchester encoding signals are made and the accuracy and recognition timings are analyzed.In this paper,stochastic resonance and deep neural network technology are used to improve the accuracy of mud pulse signals under complex noise background.Experimental results show that the proposed methods have better effect.