Research on Downhole Tubing Leakage Identification Method Based on Extreme Learning Machine

After the leakage of downhole tubing in gas wells, the annular space pressure between tubing and casing will be caused, which poses a certain threat to the safe production of gas wells. In this paper, the problem of downhole tubing leakage identification in gas wells is studied experimentally. The downhole tubing leakage simulation experiment device is used to simulate the tubing leakage under various working conditions. The time and frequency domain features of acoustic signals under three working conditions of no leakage, leakage location above liquid level and below liquid level are compared and analyzed. By extracting multiple features from acoustic signals as input of the model, a downhole tubing leakage identification model based on extreme learning machine (ELM) is proposed, the accuracy rate can reach 99 %.