Развитие методов сэмплирования в задаче сокращения продолжительности проекта с ограниченными ресурсами

Научные исследования и разработки. Российский журнал управления проектами. 2014. Т. 3. № 3. С. 14–24.

Tsarkov I.

The problem of minimizing project makespan under resource constrained is one of the most popular problem within mathematical models of project management more over 50 years. Nowadays applying exact optimization methods is impossible in real practice of project manager. So heuristic methods are used in solving such problems. The vast majority of such methods are x-pass heuristics based on applying several rules to one project. If any new schedule produced by this method use random numbers such methods called sampling methods. In this paper, the review of sampling methods will be done and research of these methods will be performed on the base of PSPLIB project library. In addition, new scheduling sampling method based on combining several scheduling schemes and priority rules will be produced. The main idea of this research is to build a system, which could help to choose the set of schema and priority rule combinations to make a best sampling method on the base of RBRS. It was established that there are no reasons to apply “bad priority rules” such as SPT, LPT, GRD and so on. It’s enough to use WCS, SLK, LST, MTS with different directions and serial and parallel schemes. Another interesting result is that after approximately 70 iterations of sampling, the best priority rule is WCS or RND (if we exclude WCS, IRSM and ACS rules). So new method based on combining MTS, SLK, LST, LFT with different schemes are produced and efficiency of this method is depicted in fig.7. The method used RBRS with rules which cyclic replace each other on each iteration. List of these rules depicted in left side of fig.7.

Keywords: ограниченные ресурсы, расписание проекта, project schedule, resource constrained, sampling, project makespan, сэмплирование, продолжительность проекта