Ultrasonic in-line inspection of pipeline corrosion based on support vector machine multi-classifier

Because of the complicated condition in pipeline, ultrasonic detection echoes are affected by many factors such as the line defect orientation, the defect width, the branching-point geometry, wall roughness, working state and the interaction between the different echoes. So it is difficult to distinguish ultrasonic detection echoes. Analyze ultrasonic detection echoes from experiments, use radial basis kernel function to process the signal, and achieve automatic recognition by using one-against-rest algorithm and tree algorithm. The experimental results show that the method not only improves the recognition correct rate, but also improves the computing speed, and achieves good results.