期刊名称：Lecture Notes in Engineering and Computer Science
印刷版ISSN：2078-0958
电子版ISSN：2078-0966
出版年度：2018
卷号：2233&2234
页码：23-28
出版社：Newswood and International Association of Engineers
摘要：The non-preemptive resource-constrained project
scheduling problem is considered in this work. It is assumed
that each activity has many ways of execution and the objective
is to find a schedule that minimizes the project’s completion
time (multi-mode RCPSP). Methods that are based on priority
rules do not always give the needed very good results when
used to solve multi-mode RCPSP. In solving large real-life
problems quickly though, these methods are absolutely
necessary. Hence good methods based on priority rules to get
the primary results for metaheuristic algorithms are needed.
This work presents a novel method based on priority rules to
calculate the primary solutions for metaheuristic algorithms. It
is a machine learning approach. This algorithm first of all uses
Preprocessing to reduce the project data in order to speed up
the process. It then employs a mode assignment procedure to
obtain the mode of each job. After which the algorithm uses
machine learning priority rule to get the precedence feasible
activity list of the project’s tasks. Finally, it then uses the Serial
Schedule Generation Scheme to get the total completion time
of the project. In our experiments, we use our algorithm to
solve some problems in the literature that was solved with
metaheuristic procedures. We compared our results with the
initial solutions the authors started with, and our results
competes favorably with the initial solutions, making our
algorithm a good entry point for metaheuristic procedures.
