No abstract

Pulse coupled neural network (PCNN), a well-known class of neural networks, has original advantage when applied to image processing because of its biological background. However, when PCNN is used, the main problem is that its parameters aren’t self-adapting according to different image which limits the application range of PCNN. Considering that, this paper proposed a new method based on particle swarm optimization (PSO) to determine automatically the parameters of PCNN.  In this method, the algorithm of PSO is applied to search automatically optimum in the solution space of PCNN’s parameters until finding global optimal solution. Experimental results demonstrate that the proposed method is accurate and robust for image segmentation, and its performance is better than the methods of Otsu, manual adjustment of parameters when mutual information is adopted as evaluation criteria

No abstract

Load-controlled fatigue tests were conducted on dual-phase X80 pipeline steel to investigate the effects of stress ratio (R-ratio) on the fatigue crack growth behaviour. Dual-phase X80 pipeline steel showed a non-linear relationship between fatigue crack growth rate (da/dN) and the stress intensity factor range (ΔK ) at each R-ratio. Fatigue crack propagation curves of X80 pipeline steel were evaluated using the conventional Paris equation and a new exponential equation named αβ model. In addition, the electron back-scattered diffraction technique was used to study the effects of stress ratio on the fatigue crack growth behaviour. The results indicated that the corresponding ΔK of the transition point decreased with the increase of R-ratio. That was attributed to the variation of the crack path and the fracture mode because of the changes in the size of monotonic plastic zone and cyclic plastic zone at crack tip. Compared to the overall fitting, piecewise fitting by Paris equation and αβ model, piecewise fitting was the most accurate method, and αβ model is more convenient and efficient than the conventional Paris-based equations.

No abstract