This article investigates the problem of partial node-based (PNB) recursive state estimation for complex networks (CNs) with unknown nonlinearities and energy harvesting sensors. To mitigate the effects of energy constraints, an energy replenishment mechanism is employed, in which a group of energy harvesting sensors captures energy from the surrounding environment. These sensors transmit measurement outputs to remote state estimators only when their current energy levels are sufficient to cover the transmission energy costs. By exploiting the universal approximation property, neural networks (NNs) are utilized to approximate the unknown nonlinearities of the CNs. An NN-based recursive estimation algorithm is developed to simultaneously generate the estimates of the system state and the unknown nonlinearities. Following a specific set of recursions, the recursive state estimator gains and the NN weight (NNW) tuning parameters are calculated in a unified framework. Finally, the effectiveness of the developed recursive estimation algorithm is demonstrated through a simulation example.
Zhang et al. (Thu,) studied this question.