In this paper, a novel continuous gradient-Zhang neuronet (GZN) model and three discrete GZN algorithms are proposed to solve time-variant overdetermined linear equation system (TVOLES) problems, and are developed on ...
详细信息
In this paper, a novel continuous gradient-Zhang neuronet (GZN) model and three discrete GZN algorithms are proposed to solve time-variant overdetermined linear equation system (TVOLES) problems, and are developed on the basis of a gradient neuronet and Zhang neuronet. Specifically, the continuous GZN model is proposed and derived in the form of an explicit-dynamic equation, and its stability is analysed and proven using Lyapunov theory. To ensure that the problems can be solved and implemented on digital platforms, three discrete GZN algorithms are established by using three different discretization formulas. The numerical simulative results verify that, compared with the Zhang neuronet model, gradient neuronet model and variant-parameter neuronet model, the proposed continuous GZN model converges faster in some cases. Other computer experiments further substantiate that the proposed continuous model and its corresponding discrete GZN algorithms are effective and efficient. Finally, the proposed discrete GZN algorithms are applied to mobile object localizations, and the simulative results verify their applicability and effectiveness.
[1] A new physical analogy for the discrete approximation of the vertical diffusion equation is presented. It is based on the resistor-capacitor chain and incorporates the classical resistance approach to dry depositi...
详细信息
[1] A new physical analogy for the discrete approximation of the vertical diffusion equation is presented. It is based on the resistor-capacitor chain and incorporates the classical resistance approach to dry deposition simulation in a natural way. It is shown that in the simplest case the finite chain of resistors and capacitors is equivalent to a second-order discrete representation of the diffusion equation resulting from the K theory closure. However, the scheme provides several features, which either cannot be obtained from straightforward discretization at all, or would require a sophisticated construction for their justification and realization. The examples discussed include a thick model layer with a highly variable vertical diffusion coefficient and an internal emission source, interaction with a chemical transformation unit, and spectral features of the model's vertical structure.
暂无评论