Nondestructive Testing Of Oil-well Tubing And Quantitative Recognition Of Defects Based On Multi-sensor Fusion

Nondestructive testing of oil-well tubing is of vital significance to the safety of oil extraction. At the same time, oil-well tubing is very expensive, while a guarantee could be provided to the safety of oil-extraction and the reusing of oil-well tubing by a method of nondestructive testing and quantificational analysis to the defects. Aiming at the fact that domestic researches of defects detection on oil-well tubing are mostly by far qualitative while the quantificational analysis is still on its early stage, this thesis develops a method, as well as an inspection system, of nondestructive testing and quantitative recognition of defects on oil-well tubing, which is based on a multi-sensor fusion. All in one, the specific works finished and main innovative contributions of this dissertation are as follows:Typical defects of oil-well tubing are analyzed and classified, according to which a series of sample defects are designed and manufactured on oil-well tubing. By experiments magnetic flux leakage signals of defects are collected with an approach of multi-sensor detection, and the relation model of magnetic flux leakage signals and the size of defects is developed and analyzed according to a large amount of experimental data.Method of short-time processing of magnetic flux leakage detection is applied. Magnetic flux leakage signals of defects of oil-well tubing are uneven and random processes. Thus, the signals could be divided into a certain amount of short parts to be processed. With the forward analysis and comparison, a group of characteristics reflecting the size of defects and based on multi-sensor fusion are proposed.With the good time & frequency characteristic of wavelet transform, signals of defects are decomposed and reconstructed the signals .the signals with gradual change caused by partial abrasion are separated from signals with abrupt change caused by defects such as hole, etc. Noise is removed from signals got by each sensor by means of reconstructing the high frequency parts of the signals. As a result of which, the SNR(Signal-to-Noise Ratio) of signals of defects is increased. With a method of enhancing the magnetic flux leakage signals, the SNR of signals of defects and background is increased effectively, and relative examples are given in the dissertation.The method of quantitative recognition of oil-well tubing defects based on multi-sensor fusion is proposed in this dissertation. Different characteristics and algorithms are adopted in accordance with different types of defects. According to the deference between characteristics of signals of partial abrasion and hole-shaped defects, the two kinds of signals are separated with each other by hardware. The size of partial abrasion is worked out quantificationally with a method of interpolation. The crack-shaped and hole-shaped defects are identified respectively by a modal identification method based on classification of characteristic and minimum distance. This two kinds of defects are analyzed quantificationally based on neural network algorithms, and the analysis results are given.A system for on-line oil-well tubing inspection and data analysis with computer is developed. According to the magnetic flux leakage model the technique of quantificational analysis of magnetic flux leakage signals developed forwardly, with the aid of computer technology, this inspection and analysis system is able to, by means of multi-thread programming technology, control data collection and storage and alerting a sound-light alarm. And by the application of USB interface technology, the system could collect data with a high speed and perform the function of plug and play very well. With this system, the veracity and efficiency of defects testing are increased and the defects are analyzed quantificationally.