In this paper, a general theoretical framework is established for the robust fusion filtering problem of discrete time-varying stochastic multisensor systems under energy harvesting constraints. The energy harvesting technology is utilized to provide the needed energy for persistently maintaining the operation of the multisensor systems. The energy level at the energy harvester is characterized by a random variable obeying a certain probability distribution. For the communication between sensors and filters, we consider a scenario where the measurements received by sensors are broadcasted via networks and then obtained by filters according to a set of preassigned communication links. The aim of this paper is to design the fusion filter over a multisensor system with locally minimized variance of the estimation error. Specifically, the local filter is firstly designed such that, in the presence of energy harvesting constraints and parameter uncertainties, an upper bound on the filtering error covariance is guaranteed and subsequently minimized by appropriately choosing the filter parameters at each time instant. Then, all the local estimates obtained by local filters are fused by using the covariance intersection fusion strategy for fusion estimation purposes. Finally, an illustrative simulation is carried out to demonstrate the usefulness of the proposed fusion filtering scheme.

本文针对能量收集约束下离散时变随机多传感器系统的鲁棒融合滤波问题建立了一个通用的理论框架。能量收集技术用于提供持续维持多传感器系统运行所需的能量。能量收集器的能量水平以随机变量为特征服从一定的概率分布。对于传感器和过滤器之间的通信，我们考虑这样一种场景：传感器接收到的测量结果通过网络广播，然后根据一组预先分配的通信链路由过滤器获取。本文的目的是在具有局部最小估计误差方差的多传感器系统上设计融合滤波器。具体而言，局部滤波器首先被设计为在存在能量收集约束和参数不确定性的情况下，保证滤波误差协方差的上限，随后通过在每个时刻适当地选择滤波器参数来最小化。然后，通过使用协方差交集融合策略融合估计的目的，通过局部滤波器获得的所有局部估计被融合。最后，进行了说明性模拟以证明所提出的融合滤波方案的有用性。