Notes of seamless steel pipe manufacturing process

There are many types of defects that may occur in the manufacturing process of seamless steel pipe. Depending on the production process, the types of defects also show differences. It is precisely because of these potential defects that the pipeline is greatly reduced in strength and premature failure during use, and even causes unpredictable losses. Therefore, taking effective measures to monitor defects in the production process in real time and identify the types of defects has far-reaching significance for improving product quality, improving processes, and rationally utilizing resources.

For a very large number of seamless steel pipe production lines, relying on manual intervention to identify the type of defects is obviously not in line with the requirements of modern mass production. In order to realize online automatic detection of defect type identification, many artificial intelligence algorithms show strong vitality in today's manufacturing informationization, such as genetic algorithm, expert system, empirical heuristic algorithm, signal processing and pattern recognition, adaptive learning, artificial neural network. Network, etc. Artificial neural networks have once been highly anticipated. However, artificial neural networks have the disadvantages of large number of training samples required, easy to fall into local minimum points, slow convergence rate, and poor generalization ability in small samples, which makes it difficult to obtain a large number of samples (such as for power station boilers, Nuclear reactor containers, aero engine rotors, ship turbochargers and other products are unpredictable. As an intelligent optimization algorithm, Particle Swarm Optimization (PSO) has the characteristics of global optimization.

Clustering analysis using particle swarm optimization to identify the type of seamless steel tube defects must first extract the characteristic parameters of the defect. The information that can be obtained by ultrasonic non-destructive testing technology can only be echo signals, and the echo signals are often rich. Information. Therefore, based on the defect echo signal, feature extraction can be performed.