【作者】 刘源； 周宇； 谢磊； 苏宏业； 罗晨； 王涛；

【Author】 Yuan Liu;Yu Zhou;Lei Xie;Hongye Su;Chen Luo;Tao Wang;Zhejiang University;Qinshan Nuclear Power Co.,Ltd.;Shanghai Hexing Nuclear Power Technology Co.,Ltd.;

【机构】 浙江大学； 秦山核电有限公司； 上海核星核电科技有限公司；

【摘要】 核电库存需求的不确定性等各方面因素影响着核电企业的采购策略,导致库存资产逐年增加。本文以核电库存系统为研究对象,建立线性的核电库存控制模型,推导出控制率使得库存系统渐进稳定。模型表明,备件的采购策略主要由系统控制器和确定性需求决定。在此基础上,建立备件需求的高斯过程模型,并推导出等效的状态空间表示。最后,对动态高斯过程进行非参数重建,通过卡尔曼滤波得到对需求分布的最优后验估计。仿真结果表明:基于卡尔曼滤波的高斯回归算法计算效率高,能在有限内存下实现较为精确的需求预测。本研究为核电领域的库存控制提供了具有需求第一性原则的参考依据,有望为降低库存水平提供有效的战略支持。更多还原

【Abstract】 The uncertainty in demand for nuclear power inventory,among other factors,influences procurement strategies of nuclear power enterprises,resulting in a gradual increase in inventory assets.This study focuses on the nuclear power inventory system,establishing a linear control model for nuclear power inventory and deriving control rates to achieve asymptotic stability of the inventory system.The model indicates that procurement strategies for spare parts are mainly determined by system controllers and deterministic demand.Based on this,a Gaussian process model for spare parts demand is developed,and an equivalent state-space representation is derived.Finally,non-parametric reconstruction of the dynamic Gaussian process is conducted,and the optimal posterior estimate of demand distribution is obtained through Kalman filtering.Simulation results show that the Gaussian regression algorithm based on Kalman filtering achieves high computational efficiency,and can accurately predict demand with limited memory.This research provides a reference based on the first principle of demand for inventory control in the nuclear power sector,offering potential strategic support for reducing inventory levels.更多还原