Abstract: Firstly, the crew shifting card in current crew scheduling plan was abstracted into several work units, and the steering quality was defined as the weighted average sum of four indicators, i.e. the accuracy of the benchmark, the punctuality, the comfort and the energy consumption. Furthermore, the calculation formulas of various indicators were given. Secondly, the calibration error, time error and interval v-t curve data of each driver in each unit of work were measured by simulation experiment and the steering quality was calculated. Thirdly, using the no estimate model fitness software 1stOpt and the calculus formula, the three-variable function relationship between steering quality and interval rest time, attendance time as well as end time was deduced. Then, the problem was transformed into a general constrained non-linear programming model maximizing the function, and the interval rest time, attendance time and end time in the crew scheduling were optimized respectively. Finally, in order to verify the effectiveness of the proposed optimization model, comparative experiments were designed. The results show that the average steering quality is improved by 14.09%, 8.17%, 6.41%, 9.45%, 6.31%, 7.64% and 7.57% respectively.

Key words: traffic and transportation engineering；urban rail transit, crew schedule optimization, no estimate function fitness, steering quality, nonlinear programming with constraints

摘要： 将已有乘务排班计划中的轮乘卡抽象为若干工作单元，将驾驶质量定义为由对标精度、准时度、舒适度和能耗度4个指标加权平均和 ，并给出各种指标计算公式；通过模拟实验测出每位驾驶员在每个工作单元的对标误差、时间误差、区间v-t曲线数据，计算驾驶质量，再 利用无预估模型拟合软件1stOpt,并借助微积分公式推导出驾驶质量同间休时间、出勤时刻、结束时刻的三元函数关系；然后把问题转化为 以该函数最大化的一般约束非线性规划模型，分别对七种情况下乘务排班间休时间、出勤时刻、结束时刻进行优化；最后，为了验证优化模 型的有效性，分别设计了对比试验，结果表明优化后平均驾驶质量分别提高14.09%、8.17%、6.41%、9.45%、6.31%、7.64%、7.57%。

关键词: 交通运输工程, 城市轨道交通, 乘务排班优化, 无预估函数拟合, 驾驶质量, 约束非线性规划