Genetic programming method based on random resource constrained multi-project scheduling

Abstract

The invention discloses a genetic programming method based on random resource limited multi-project scheduling, which comprises the following steps: an initialization stage: collecting and initializing genetic programming parameters to obtain a priority rule expression set, and obtaining an initial mixed priority rule set based on the priority rule expression set; a generation stage: evalualting the population priority rules through an NSGA-II algorithm, so that iterative optimization is achieved, and a non-dominated priority rule set is obtained; a selection stage: weighting and normalizing the traditional priority rule set and the non-dominated priority rule set to obtain a target priority rule scheduling combination which is used for realizing priority rule combination scheduling of the engineering project in multiple states. According to the method, the defect that a traditional genetic programming decision is made by using a single priority rule is overcome, and the comprehensive performance of the decision is further improved. Therefore, the method has important significance and engineering application value for scheduling of multiple projects with limited random resources.

Description

technical field [0001] The invention belongs to the field of item scheduling optimization, in particular to a genetic programming method based on random resource limited multi-item scheduling. Background technique [0002] Resource Constrained Project Scheduling Problem (RCPSP) is the most core and classic NP-hard problem in project management. However, there are a lot of random interference and multi-project collaborative execution problems in actual engineering (for example, 90% of project management is executed under multi-project), the traditional RCPSP model is difficult to adapt, so RCPSP is extended to random Resource Constrained Multi-Project Scheduling Problem (Stochastic ResourceConstrainedMulti-Project Scheduling Problem, SRCMPSP). [0003] Because the SRCMPSP model is more suitable for engineering practice, it has been widely studied. Because it also belongs to the field of combinatorial optimization, and because it shows strong optimization and search capabili...

AI Technical Summary

Problems solved by technology

However, the meta-heuristic technology is an iterative optimization method accompanied by a large number of random operations, which has two defects in the solution of SRCMPSP: 1) SRCMPSP is a random problem, and the meta-heuristic algorithm with a large number of random searches will further deteriorate the robustness. Thus affecting the scheduling effect; 2) The meta-heuristic algorithm has an iterative process, which takes time, and the randomness problem will increase the frequency of rescheduling, further increasing the calculation cost

However, the existing super-heuristic technology is still applied to static problems in the field of project scheduling, and the SRCMPSP model, which is more biased towards engineering applications, has not been applied to solve

At the same time, the priority rule discrimination generated by existing hyperheuristic techniques will remain fixed throughout the decision process

Images