A Backstepping Approach to Decentralized Active Disturbance Rejection Control of Interacting Boost Converters

In this paper, the local trajectory tracking control problems ascribed to two interacting plants, considered as agents, are formulated as adaptive control problems, which involve online interaction estimation and interaction elimination. This approach gives rise to a robust decentralized collaborative control with virtually no information on the part of the agents. It is shown that when the interconnection effects are viewed as exogenous unstructured disturbances, such disturbance can be actively estimated and canceled from each individual subsystem model dynamics. The case presented deals with two agents, two interconnected boost dc-dc power converters, powered each one by a nonidentical photovoltaic module that represents a time-varying power supply. Then, a backstepping-based control together with an extended state observer is developed by each agent. The mutual goals of agents are to maintain a desired behavior (time-varying current demand) of the entire system in order to maintain the desired output voltage value and an equitable current sharing in each converter in order to supply a dc motor. The proposed distributed control technique is implemented in two TMS320F28335 digital signal processor (each one per agent) and its performance is experimentally evaluated in real time. It is shown that the proposed scheme is robust with respect to interaction, unmodeled nonlinearities, and unmodeled dynamics.