A WP-based nonlinear vibration sensing method for invisible liquid steel slag detection

Abstract Vibration-based liquid slag sensing method is widely used in steel continuous casting production (CCP) due to its long service life and low cost. However, it is apt to be influenced by the surrounding interferences, and the slag detection accuracy (SDA), working stableness and debugging time require to be improved. Aiming at the matter, this paper puts forward a novel liquid slag sensing method based on wavelet packet (WP). A mechanical dynamic model of CCP machine is set up to analyze the vibration attributes of pouring structure. Based on the above model, an improved WP-based nonlinear vibration recognition method is proposed to identify the pouring state. The simulated experiment results show that the SDA of this method can arrive at 100%. CCP industrial field experiments prove that this method can recognize the typical pouring states, and the practical SDA can reach more than 98%. The key scientific contribution of the paper is providing a sensing method for nonlinear vibration signal of invisible objects, which can offer universal references to the measurement of fluid state in invisible chemical vessels, such as pressure containers, reactors or capsules.