|M.Sc Student||Olga Schneider|
|Subject||Operational Analysis of Multi-Project Systems Managed|
through Artificial Agents
|Department||Department of Industrial Engineering and Management||Supervisor||Full Professor Golany Boaz|
The purpose of this research is to perform a comprehensive study of the structural effects of multi-projects networks on the performance of agent-based procedures for scheduling activities in resource-constrained environment. This issue was largely ignored in the literature so far. Several basic network parameters were defined, including network complexity, which is a well-known measure in the literature. The rest of the parameters describe other structural dimensions of the network, such as its hierarchical levels, depth, width, etc.
Another aspect that still was not fully studied in a multi-project environment is how to select the proper operational mode for each activity, where modes differ from one another by duration of an activity and by its resource requirements. According to the literature survey, the problem of mode selection belongs to the “NP-Hard” class. As a result, the majority of the existing algorithms deal only with small problems. Lately, an approach that employs artificial agents was developed and proven to be successful in a single project environment (including some very large problems). An artificial agent is a piece of software or hardware, which is able to “think” and “make decisions” according to its intelligence level.
The goals of the current research were: to define multi-mode networks on the basis of structural measurements and to define different levels of agents’ intelligence in order to analyze system performance as a function of physical structure of multi-project, different types of agents, and rules for activity indexing, which influence activity priorities during the scheduling process.