摘要
炼钢-连铸排程是NP完全问题,文献中遗传算法、模拟退火等算法的计算速度无法满足进度控制的要求.目的就在于为改进现有算法提供标杆(benchmark).将任务委派和任务排序问题描述为黑球和白球的组合问题,大幅改进了从大量组合中找出可行组合的速度.以B企业两座转炉冶炼51炉为例,所给方法在CPU为双核4600+,内存为2 GHZ的主机上运行,仅用1.375 s从6 765 201个组合中,首先找出834个排列可行解,然后找出209个时间可行解,最后找出5个满意解,作为改进算法的标杆.
Steelmaking-eontinuous casting scheduling is a NP-complete problem. The computing speed of algorithms such as the genetic al- gorithm and the simulated annealing algorithm in literatures can not satisfy the requirements of progress control. Based on researches, aimed at improving the existing algorithms for benehmarking,the problem of tasks assigned and tasks sorted was described as black ball and white ball combinations ,which significantly improved the speed of finding the possible combinations from a large number of combinations. Take smelting 51 furnace by two converters of B enterprise for instance ,while running on the host whose CPU is dual-core 4600 ＋ and memory is 2GHZ ,it only took 1. 375 seconds to firstly discover 834 arrangement feasible solutions ,then discover 209 time feasible so- lutions;finally ,5 satisfactory solutions were discovered as a benchmark to improve the algorithm from 6 765 201 comhinations.
出处
《内蒙古科技大学学报》
CAS
2009年第3期270-273,共4页
Journal of Inner Mongolia University of Science and Technology
基金
国家自然科学基金资助项目(70662002)
关键词
炼钢-连铸
优化
排程
穷举法
线性规划
steelmaking-continuous casting
optimization
scheduling
exhaustive method
linear programming
