Reconstruction of networks of cyclostationary processes
Reconstruction of networks of cyclostationary processes
复制标题
循环平稳过程网络的重建
DOI：
--
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
M. Salapaka
中科院分区：
文献类型：
--
作者：
Saurav Talukdar;M. Prakash;D. Materassi;M. Salapaka
许多复杂的系统可以通过代理来描述，代理可以建模为动态交互的循环平稳过程的网络。此类系统出现在电力系统和气候科学等领域。对于许多此类系统来说，一个关键目标是在不改变系统自然行为的情况下了解各个子系统之间的相互影响。这样的目标意味着仅使用被动方式揭示网络拓扑的互连。大多数现有的相关工作都强调基于相关性的方法，其中可能会忽略不同时刻的相互依赖性。最近的工作将动态影响纳入假设静态统计无法适应电力和气候科学等许多领域中出现的应用。在本文中，设计了一种基于维纳滤波的算法，用于重建动态相关的循环平稳过程的互连性。结果表明，所有现有的相互依赖性都被检测到，并且虚假检测仍然是本地的。事实证明，在微电网中的应用可以产生有用的见解。
Many complex systems can be described by agents that can be modeled as a network of dynamically interacting cyclo-stationary processes. Such systems arise in areas like power systems and climate sciences. For many of these systems a key objective is to understand mutual influences between various subsystems without altering the natural behavior of the system. Such an objective translates to unveiling the interconnection of the topology of the network using only passive means. Most existing related works have emphasized correlation based methods where interdependencies over different time-instants can be missed. Recent work where dynamic influences are incorporated assuming stationary statistics cannot accommodate applications that arise in many areas such as power and climate sciences. In this article an algorithm based on Wiener filtering is devised for the reconstruction of interconnectivity of dynamically related cyclo-stationary processes. It is shown that all existing interdependencies are detected and spurious detection remains local. Application to a microgrid power network is shown to yield useful insights.