Research On Fault Identification System Of Casing Cutter

A casing cutter is a workover tool for repairing damaged casing wells.During its downhole operation,various failures often occur,resulting in large economic losses.Therefore,it is imperative to study the fault identification system of the casing cutter.However,the vibration signal of the casing cutter contains a lot of noise,and there is an incomplete measurement phenomenon,and the problem of submerging the fault characteristic information caused by the problem needs to be solved urgently.In addition,due to the scarcity of fault data and a large number of parameters of complex network models,it also brings great challenges to the recognition accuracy of the model and the realtime performance of fault recognition.First of all,this article fully considers the downhole working environment temperature index of the casing cutter,and designs and realizes the high-temperature environment casing cutter vibration acceleration data acquisition device.Then,to obtain the characteristics of the vibration signal,the model of the casing cutter in the state of no failure and the other two common failures were established,and the vibration characteristics of the model were analyzed.The analysis results showed that the environmental noise would cause the failure characteristic information to be submerged.This thesis designs a fault feature enhancement algorithm based on signal-to-noise ratio enhancement and sparse representation model to solve this problem.The specific implementation of the signal-to-noise ratio enhancement method and the mathematical model of sparse representation and a fast solution method are given.Use this method The actual operating data of the cutter is processed,and a twodimensional time-frequency vector graph that can uniquely represent different fault characteristics is obtained.This image is used as the input of the lightweight convolutional neural network for fault identification.Then,because of the complex structure and redundancy,slow calculation speed,and the huge number of parameters in the large-scale convolutional neural network model,this thesis designs an improved fault recognition model based on the MobileNet-V3 Small lightweight convolutional neural network model.The goal is to improve The calculation speed of the model can be more accurate for fault identification under the premise of reducing the parameter amount and resource occupation of the model to the greatest extent.Finally,because of the problem of the small amount of faulty data,this thesis uses a semi-supervised learning-pseudo-label-based model performance improvement method,and incorporates unlabeled data into the model training process,which further improves the performance of the model.The experimental results show that the improved model has a recognition accuracy rate of 96.14%,a high degree of lightweight,and a better realization of the fault recognition and classification of the high-temperature casing cutter.