基于离散Hopfield神经网络的公交调度评价方法研究

摘要

According to characteristics of the evaluation about optimization of bus scheduling, this paper puts forward an evaluation method based on the Discrete Hopfield Neural Network (DHNN). At the course of constructing DHNN, the authors have selected the effective time-use law, average transfer time, transport capacity match and operating profit as evaluation indicators about the bus scheduling, according to running efficiency, smooth convergence, coordination capacity and company running profit. These indicators are independent of each other and quantifiable. At last we design the DHNN programming with MATLAB, and make this method be able to evaluate the bus scheduling conveniently and reasonably; thus it has a broader applicable scope. Finally, the method is verified by an example, and the calculation has been compared with the Delphi, the traditional evaluation method. Results show that the method is reasonable and effective.Compared with Delphi, this method is simple and easy to promote and has superiority to other methods.%根据公交调度评价问题的特点,提出了基于离散Hopfield神经网络的评价方法.在构造Hopfield神经网络时,从公交系统运行效率,衔接的顺畅性,运力协调性,企业经济利润率等四个主要内容出发,选取有效时间利用率、平均换乘次数、运能匹配度和运营利润率四个可量化且彼此相互独立的指标,作为公交调度优化结果的评价指标.并用matlab对Hopfield网络进行设计仿真,使其可以更加合理和方便地对公交调度结果进行评价,从而具有更广泛的工程应用面.最后通过实例计算验证并将计算结果与传统的德尔菲法进行比较.结果表明,该方法合理有效,且相对于传统的方法操作简单、易推广,具有一定的优越性.