Oil Pipeline Detection Technology Research Based On Multi-sensor Data Fusion

This paper studies domestic and international oil pipeline detection techniques of present situation, development trend and existing problems, analyzes the application of the multi-sensor technology in the fields of superiority and necessity in theory, proposes the necessity of subject research. We analyze the principle the testing system, test process of magnetic flux leakage testing technology and ultrasonic testing technology of commonly pipeline detection technology in detail, and mainly introduce the relationship of the pipeline defects and the magnetic flux leakage signals, the relation between pipeline flaws and magnetic flux leakage signals and the characteristic extraction, and comparatively analyze the characteristics and deficiencies of the ultrasonic testing methods and the magnetic flux leakage testing methods.Secondly, the article describes the multi-sensor data fusion technology definition, principle, fusion system structure, classification and integration of processes and applications in detail, focusing on the multi-sensor data fusion algorithm in neural network technology and D-S theory. This paper analyzes the characteristics of BP neural network and studied the BP neural network based data fusion method, and adopts the more effective L - M optimization algorithm for BP algorithm was improved. Examples were analyzed. The results shows that the neural network Information Fusion for Damage Identification of pipeline defects is an effective method and the improved algorithm has better recognition effects. On this basis, this paper briefly introduces the basic concepts of D-S evidence theory, combination of rules and decision-making criteria. The article proposed a fusion model and integration algorithm combination of neural networks and D-S evidence theory data for multi-sensor data for the uncertainty .The theory is analyzed through cases.Finally, the combination of the neural network and D-S evidence theory data fusion algorithm is applied to pipeline inspection system. It was tested by the use of pipeline defects simulation data. The test results showed that the effect of this method was good. A better defect recognition rate can provide more accurate information for the pipeline maintenance and management.