Process control of charging and discharging of magnetically suspended flywheel energy storage system

Flywheel energy storage system (FESS) is an energy conversion device designed for energy transmission between mechanical energy and electrical energy. There are high requirements on the power capacity, the charging ef-ficiency and the output precision of FESS. Active magnetic bearings are used to suspend the flywheel (FW) rotor of the FESS in air to eliminate friction. A high rotating speed of the flywheel can increase the power capacity but it also increases the disturbance load torque on the FW rotor. An observation control model of load torque is therefore proposed to mitigate the disturbance load torque acting on magnetically suspended FESS (MS-FESS) during the charging process. Moreover, for the discharging process of MS-FESS, a compound control model combing the sliding model control and the extend state observer is proposed to improve the response speed and the output voltage precision. Simulations and experiments are conducted to testify the control performances of proposed control models during the charging and discharging processes of a MS-FESS. The charging efficiency is improved by 17.6% with 46.6% reduction of the output voltage error after using the proposed control models for the charging and discharging process control. The proposed control method has high potential to be applied for process control of charging and discharging of practical MS-FESS.

Automated summary

Flywheel energy storage system (FESS) is an energy conversion device with high requirements on power capacity, charging efficiency and output precision. This study proposes an observation control model to mitigate disturbance load torque during the charging process and a compound control model to improve response speed and output voltage precision during the discharging process of magnetically suspended FESS (MS-FESS). Simulations and experiments show improved charging efficiency and output voltage precision using the proposed control models.