Distributed H∞ moving horizon estimation over energy harvesting sensor networks

This paper addresses the distributed $ {H_\infty } $ H∞ moving horizon estimation problem for nonlinear systems over energy harvesting sensor networks in a deterministic framework, where each sensor is able to gather energy from the surrounding environment. A transformed Poisson process model is introduced to describe the energy collected by each sensor, with particular consideration given to the minimum energy collected. In contrast to previous research on distributed state estimation over energy harvesting sensor networks, which solely considered the energy costs associated with transmission and assumed knowledge of the statistical properties of harvested energy, we consider a more comprehensive scenario. Specifically, we account for the energy costs associated with both sensor sensing and data transmission to neighbouring sensors, while only the parameters of the minimum harvested energy are known. Subsequently, a novel energy allocation strategy is proposed to provide energy for each sensor sensing and transmission based on a predetermined fixed circular order, which determines a maximum time interval for each sensor sensing and transmission. Then, local measurement output predictors and prior state predictors are constructed to handle interruptions in sensor sensing and communication caused by the energy harvesting mechanism. Furthermore, we derive sufficient conditions for the existence of a distributed moving horizon estimator, ensuring $ {H_\infty } $ H∞ consensus. To illustrate the proposed method's effectiveness, a single-machine infinite-bus power system is presented.