Sliding-Mode Control of an Ultrahigh-Speed Centrifugal Compressor for the Air Management of Fuel-Cell Systems for Automotive Applications

This paper presents the modeling and control of an ultrahigh-speed centrifugal compressor for the air management of proton exchange membrane fuel-cell (PEMFC) systems. Centrifugal compressors have the advantages of compactness, high efficiency, and low noise, compared with other kinds of displacement compressors. Moreover, ultrahigh-speed technology can also reduce the size and weight of the compressor, which makes it more feasible for automotive applications. However, the adoption of a centrifugal compressor results in control being difficult because of coupling between mass flow and pressure. In this paper, a neural network model of the compressor is developed, and a decentralized sliding-mode controller based on twisting and super twisting algorithms is proposed and tested to control the compressor pressure and mass flow. The experimental results show good dynamic characteristics and faster response compared with conventional proportional-integral (PI) control.