Application of Wolf Swarm Neural Network in Surface Reconstruction

2019 2nd International Conference on Advanced Materials, Intelligent Manufacturing and Automation

Wang, Hai-Jun^1 ; Jin, Tao^1

Department of Mathematics and Computer Engineering, Ordos Institute of Technology, Inner Mongolia Ordos

017000, China^1

关键词: BP neural networks;    Fitting algorithms;    Global search ability;    Hybrid neural networks;    Optimization ability;    Reconstruction accuracy;    Reconstruction error;    Surface reconstruction algorithms;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/569/5/052019/pdf
DOI  :  10.1088/1757-899X/569/5/052019

来源: IOP

【 摘 要 】

Although the traditional BP neural network has strong nonlinear fitting ability, it has poor global search ability, slow convergence speed and easy to be trapped into local minimum value, etc. Based on this, a WPA-BP hybrid neural network surface reconstruction algorithm combining the Wolf pack algorithm and BP algorithm is proposed. WPA-BP hybrid algorithm has both the adaptive ability of BP algorithm and the global optimization ability of WPA algorithm, which can improve the existing problems of BP algorithm model. Compared with the reconstruction results based on BP algorithm and quadratic fitting algorithm, the surface reconstruction model using WPA-BP hybrid algorithm has higher reconstruction accuracy and smaller reconstruction error.