摘要
工程应用中的许多问题,如控制理论、信号处理、安全通信和图像重建,都可以归结为可行性问题,其中许多问题涉及时变环境,短时间内提供解决方案是至关重要的。当需要实时获得可行解决方案时,特别是存在不确定性时,时间依赖约束可行性问题的难度明显增大。在这样的应用中,与传统的数值可行性算法相比,神经动力学方法可以在指定的硬件中物理实现,因此,提出了一种基于精确罚函数法的时间依赖凸约束可行性问题的神经动力学模型。对于罚参数的适当值,证明了所提出的神经动力系统的任何状态解在有限时间到达移动集合且随时间跟踪移动集。此外,估计了惩罚参数的下界和收敛时间。
Many problems in engineering applications, such as control theory, signal processing, security communications, and image reconstruction, can all be attributed to feasibility issues. Many of these issues involve time-varying environments and it is crucial to provide solutions in a short time. When a feasible solution is needed in real time, especially when there is uncertainty, the difficulty of the time-dependent constraint feasibility problem increases significantly. In this kind of application, compared with the traditional numerical feasibility algorithm, the neural dynamics method can be physically implemented in the specified hardware/firmware. Therefore, a neurodynamic model for time-dependent convex constraint feasibility problem based on the exact penalty function method is proposed. For the appropriate values of penalty parameters, it is proved that any state variable of the proposed neurodynamic system can reach the moving set within a limited time. After that, the state tracks the moving set. In addition, the lower bound and convergence time of the penalty parameter are estimated.
作者
张宸宁
李国成
ZHANG Chenning;LI Guocheng(School of Applied Sciences,Beijing Information Science&Technology University,Beijing 100192,China)
出处
《北京信息科技大学学报（自然科学版）》
2019年第1期12-17,38,共7页
Journal of Beijing Information Science and Technology University
基金
国家自然科学基金项目(61473325)
关键词
时间依赖约束
神经动力学方法
有限时间收敛
可行性问题
动态分析
time dependent constraint
neurodynamic approach
finite time convergence
feasibility issues
dynamic analysis
