搜索空间代换法及其在城市交通控制中的应用

Search-space substitution algorithm and its application to urban traffic control

针对常用遗传算法局部搜索能力差 ,导致计算速度缓慢、精度差等问题 ,提出一种搜索空间代换的新遗传算法。该算法保留了普通遗传算法的全局搜索性能 ,并通过多次搜索空间的代换提高了局部搜索能力。改进后的新方法在每次代换后只需改变解码规则 ,不需重新编码 ,也不需增加编码的长度 ,从而保证了计算效率。采用该方法对城市多车道、四相位的动态交通网络控制策略进行了多目标优化计算。应用结果表明 ,采用新的遗传算法可在同等情况下减少车辆堵塞 。

A new search-space substitution based genetic algorithm is presented to solve the problem that local space search or local convergence produces computational inefficiency and inaccurate optimization results. The new genetic algorithm adopts multi-search-space substitution and thus has whole-space search capability of ordinal genetic algorithm and high local convergence. After each substitution, the new method changes the decoding rule without renewing or lengthening the coding. Thus high efficiency of optimization computation is reached. The new algorithm is applied to the control decision of multi-driveway, four-phases dynamic urban traffic networks. Results show that the capacity of urban traffic will be increased by using the new genetic algorithm.