Tian Gan

Image Segmentation of Thermal Waving Inspection based on Particle Swarm Optimization Fuzzy Clustering Algorithm

The Fuzzy C-Mean clustering (FCM) algorithm is an effective image segmentation algorithm which combines the clustering of non-supervised and the idea of the blurry aggregate, it is widely applied to image segmentation, but it has many problems, such as great amount of calculation, being sensitive to initial data values and noise in images, and being vulnerable to fall into the shortcoming of local optimization. To conquer the problems of FCM, the algorithm of fuzzy clustering based on Particle Swarm Optimization (PSO) was proposed, this article first uses the PSO algorithm of a powerful global search capability to optimize FCM centers, and then uses this center to partition the images, the speed of the image segmentation was boosted and the segmentation accuracy was improved. The results of the experiments show that the PSO-FCM algorithm can effectively avoid the disadvantage of FCM, boost the speed and get a better image segmentation result.

Numerical Analysis of Influencing Factors and Capability for Thermal Wave NDT in Liquid Propellant Tank Corrosion Damage Detection

Due to the disadvantages of traditional NDT methods for liquid propellant tank corrosion detection, Thermal Wave Nondestructive Testing (ITWNDT) technology was applied. The heat exchange process of thermal wave in corrosion tank was simulated by the numerical method. Parameters of TWNDT as the best detection time (tbest), the maximum surface temperature difference (ΔTmax), and the temperature difference holding time (τΔT>0.1) were discussed as the targets. Based on these parameters, factors influencing the detection results of tank materials, dressed liquid (also considered as the corrosion product), pit characters (depth and size), heat flux and thermal excitation time length (pulsed width), environmental conditions and other factors were analyzed. Simulation results show that ITWNDT can identify the defect depth, size and position rapidly and effectively. Material properties of the tank were influencing the tbest, ΔTmax and τΔT>0.1, while the dressed liquid, thermal excitation parameters and the conditions of environment do not influence the tbest. Pit characters of the depth and size have close relationship with tbest and ΔTmax, therefore, for a tank with certain material and certain liquid dressed in, the pit corrosion damage can be accurately evaluated.