Dynamic Inspection of Latent Variables in State-Space Systems
Dynamic Inspection of Latent Variables in State-Space Systems
复制标题
DOI：
10.1109/tase.2018.2884149
复制
发表时间：
2019-01
影响因子：
5.6
通讯作者：
Tianshu Feng;Xiaoning Qian;Kaibo Liu;Shuai Huang
中科院分区：
文献类型：
--
作者：
Tianshu Feng;Xiaoning Qian;Kaibo Liu;Shuai Huang
状态空间模型（SSM）被广泛应用于各种领域，其中一组可观测变量被用来跟踪一些潜在变量。虽然大多数现有的工作集中在统计建模的潜变量和可观察变量之间的关系或基于可观察变量的潜变量的统计推断，它来我们意识到，一个重要的问题已在很大程度上被忽视。在许多应用中，虽然潜变量不能被常规地获取，但它们可以偶尔被获取以增强对状态空间系统的监视。因此，在本文中，新的动态检查（DI）的一般框架下的SSM的方法来识别和检查的潜变量是最不确定的。进行了大量的数值研究，以证明所提出的方法的有效性。从业者注意-SSM旨在估计表征系统状态的关键潜在变量，但无法常规或直接测量。传统的方法一直完全基于专用于观测变量的测量能力。然而，我们意识到，在某些情况下，虽然潜变量不能常规测量，但可以在给定的频率下检查一小部分潜变量。因此，问题是如何分配检查资源，以帮助最佳地监控状态空间系统的潜在变量，条件是建立SSM的统计机制，用于模型估计和推理。我们提出了一种DI方法来选择和部分测量潜变量，并通过结合测量的潜变量和观测值来提高估计精度。
The state-space models (SSMs) are widely used in a variety of areas where a set of observable variables are used to track some latent variables. While most existing works focus on the statistical modeling of the relationship between the latent variables and observable variables or statistical inferences of the latent variables based on the observable variables, it comes to our awareness that an important problem has been largely neglected. In many applications, although the latent variables cannot be routinely acquired, they can be occasionally acquired to enhance the monitoring of the state-space system. Therefore, in this paper, novel dynamic inspection (DI) methods under a general framework of SSMs are developed to identify and inspect the latent variables that are most uncertain. Extensive numeric studies are conducted to demonstrate the effectiveness of the proposed methods. Note to Practitioners—The SSM aims to estimate crucial latent variables that characterize the states of a system but cannot be measured routinely or directly. The conventional way has been solely based on a measurement capacity dedicated to observed variables. However, we realize there are situations that, although latent variables cannot be measured routinely, it is possible to inspect a small portion of latent variables at a given frequency. Thus, the problem is how to allocate the inspection resources to help monitor the latent variables of the state-space system optimally, conditioning on the established statistical machinery of the SSM for model estimation and inference. We propose a DI method to select and partially measure the latent variables and improve the estimation accuracy by combining the measured latent variables and observations.