Yanhui Zhou

A distributed ESO based cooperative current-sharing strategy for parallel charging systems under disturbances

In this paper, a distributed extended state observer (ESO) based cooperative control is proposed to achieve current-sharing for parallel chargers under disturbances in energy storage type light rail vehicle systems. By treating the parallel charging system as a nonlinear and non-identical multiagent system, the current-sharing objective is converted as a distributed leader-follower consensus problem. A distributed ESO is designed to observe total disturbances with the known information from the established mathematical model of subchargers. A tracking differentiator (TD) is established to provide a transitional process for the reference signal to increase the dynamical performance and robustness. Based on ESO and TD, a distributed cooperative control protocol is proposed. Simulation and experimental results are provided to verify the disturbance rejection ability, feasibility and practicability of the proposed distributed current-sharing control method.

An adaptive energy allocation strategy for battery/supercapacitor hybrid energy storage system

Battery/supercapacitor hybrid energy storage system is widely used in renewable energy grid since its excellent power and energy performances. However, the load and grid output power may fluctuate drastically, resulting in the dc bus voltage rise or fall sharply. The power demand is composed of both high-frequency and low-frequency component that brings some difficulties to energy distribution. Therefore, it is still a challenge to design a safe and high-efficiency energy allocation strategy to respond to the power demands and stabilize the bus voltage in time. In this paper, an adaptive energy allocation strategy based on the isolation frequency is proposed to address this challenge. The isolation frequency is adjusted adaptively that can improve the energy utilization ratio and ensure the energy storage element safety. In addition, a practical control method based on frequency-domain analysis is adopted since its less computation. Simulation results validate the potential advantages and effectiveness of the proposed strategy.

A generalized extended state observer for supercapacitor state of charge estimation under disturbances

In this paper, a generalized extended state observer (GESO) is proposed to accurately observe supercapacitors state of charge (SOC) by counteracting the modeling deviation and disturbances. With the consideration of dynamic modeling uncertainty and disturbances, a nonlinear mathematical model is established based on the three branch equivalent circuit model for supercapacitor (SC). A GESO is designed to observe SCs internal capacitance voltage and modeling error with the known information from the established mathematical model. With the estimated modeling error information, a compensation control law is designed to eliminate deviation, which results in an accurate internal capacitance voltage observation. An accurate SOC for SC is obtained with the estimated parameters. Experimental results from a synchronous buck circuit based test platform are provided to verify the feasibility and superiority for SOC observation by GESO.

A hybrid model predictive charging control strategy for ultracapacitors of urban rail vehicles

Ultracapacitors have been widely used in urban rail vehicles as the power source because of its high power density. However, it is still a challenge to design a rapid and reliable charging control system for ultracapacitors since the DC-DC converters are highly nonlinear along with hard constraints. In this paper, the piecewise-affine (PWA) model of the ultra-capacitor charging system and a hybrid model predictive control strategy are proposed to address this challenge. The hybrid nature of the converter is taken into account in v-resolution model which can reflect the dynamics of the charging system precisely. Based on the discrete-time model, we formulate and solve a constrained optimal control problem, and the control law is calculated offline using multi-parameter programming method. Simulation and experiment results validate the potential advantages of the proposed scheme.