状态空间下列车区段晚点预测误差控制

Error Control of Delay Prediction in Railway Sections Under State Space

考虑区段间信息级联效应,提出一种状态空间下的列车晚点预测误差控制分析架构.应用拟合回归方法处理运行统计数据以确定常微分方程组,根据现代控制理论设计一种减小误差的线性二次调节器,并通过推导论证为所提方法提供合理性支撑.设计了双温模拟退火算法以获取调节器的优化参数,并结合SIMULINK环境实现仿真计算.结果表明,相较于传统贝叶斯网络预测方法,所提出的状态空间分析方法对各项误差性能指标具有不同程度的改善,且在平均误差值、数据需求量和运算时间等计算指标上具有优势.

Considering the cascading effect of information between sections,an analysis framework of train delay prediction error control in state space was proposed.The fitting regression method was used to process the operating statistics to determine the ordinary differential equations.According to the modern control theory,a linear quadratic regulator was designed to reduce the error,and it provides reasonable support for the proposed method through deduction.The dual-temperature simulated annealing algorithm was designed to obtain the optimized parameters of the regulator,and the simulation calculation was realized in combination with the SIMULINK environment.Numerical experiment results show that compared with the traditional Bayesian network prediction method,the proposed state space analysis method has different degrees of improvement on various error performance indicators,it has obvious advantages in calculation indicators such as average error value,data demand and computing time.