Federated Filtering Revisited: New Directions to Distributed Systems Estimation and Filtering – a Case Study

Abstract

The paper discusses the application of decentralized filtering to state estimation of distributed nonlinear systems. The proposed filter is operated in a novel federated architecture in an attempt to adaptively balance estimation accuracy and sensor fault tolerance in the presence of both state and measurement uncertainty when the observations of the system are subject to a
variety of sensor faults. The fault management logic built into the federated structure is accountable for the control of the distributed filter architecture to achieve sensor fault accommodation in a seamless way. The performance of the traditional centralized filtering is contrasted with the proposed federated idea, and its ability to trade-off between fault tolerance and estimation accuracy are examined. Benefits from the application of the method to a chemical distillation process are demonstrated by computer simulations .