周   航 孙绍国 谷   明（天津大学建筑设计研究院，天津市 300073）

On - line Estimation for State of Charge of EPS Accumulator

ZHOU Hang SUN Shaoguo GU Ming（Tianjin University Research Institute of

Architectural Design and Urban Planning，Tianjin 300073，China）

    Abstract：An on-line estimation method for state of charge of emergency power supply（EPS）accumulator is proposed，in order to increase estimation accuracy of state of charge（SOC）of EPS accumulator. A multi-core relevance vector machine（RVM）with optimized parameters is adopted for establishing a SOC estimation model of EPS accumulator，thereby realizing real-time monitoring for EPS state of charge，supplying an important basis for charging / discharging management of accumulator and guaranteeing safe and stable operation of EPS system.

    Key words：emergency power supply；accumulator；state of charge；on-line estimation；multi-core relevance vector machine；charging

/ discharging management；utility function；SOC estimation error

  摘 要：为了提高应急电源（Emergency Power Supply，EPS）蓄电池的荷电状态（state of charge，SOC）估计精度，提出一种EPS电源蓄电池在线荷电状态估计方法。采用一种参数优化的多核相关向量机（relevance vector machine，RVM）建立EPS蓄电池SOC估计模型，可以实现对EPS荷电状态的实时监测，为蓄电池的充放电管理提供重要依据，保证EPS电源系统的安全稳定运行。

  关键词：应急电源；蓄电池；荷电状态；在线估计；多核相关向量机；充放电管理；实用度函数；SOC估计误差

     中图分类号：TU851                 文献标识码：A

      doi：10. 3969 / j. issn. 1003 - 8493. 2019. 08. 017