论文部分内容阅读
本文叙述了我们在对非平稳、非线性和非高斯环境声与振动系统进行状态估价时,对通用贝叶斯滤波器设计的基本考虑。利用贝叶斯定理的正交或非正交展开式,讨论了几种贝叶斯滤波器的基本要点。这几种滤波器可以用来估价任意函数类型状态变量的统计量,包括在环境声与振动系统中与L_x评价指数有关的低阶和高阶的统计量。这里,我们着重给出设计原则的基本观点,包括一些尚未发表的新估算方法和新的结果。
This article describes the basic considerations for designing universal Bayesian filters for state estimation of non-stationary, nonlinear and non-Gaussian ambient sound and vibration systems. Using the orthogonal or non-orthogonal expansion of the Bayesian theorem, the basic points of several Bayesian filters are discussed. These filters can be used to estimate the statistic of state variables of any function type, including the low-order and high-order statistics related to the Lx evaluation index in ambient sound and vibration systems. Here, we focus on the basic concepts of design principles, including some unpublished new estimation methods and new results.