Oleksandr Firsov

Abstract:
Production planning is becoming an increasingly difficult task, since client requirements are numerous in terms of quality, flexibility, availability, as well as acceptable delays of provision. Moreover, because of the product diversity and the variable demand, the production process remains in constant change. Because a considerable investment is needed for such adjustments, they must be backed up with a justification that each resource is indeed exploited with maximum efficiency. Which is why the planning algorithms become an attractive solution. In theory, they would allow the businesses to achieve their objectives in the most efficient manner possible. But in reality, there doesn't exist a universal algorithm, flexible enough to address all these optimization and planning problems in the most efficient manner possible. Every one of them is fit to solve only a very specific problem, thus making it impossible to apply it to different clients. Which is why many businesses still use human experts for production planning. This thesis aims to propose a flexible, generic planner, which would rely on PDDL or HDL programming language, as well as techniques demonstrated by Constraint Satisfaction problem solvers. When faced with a problem, it will either use ad hoc algorithms to provide a partial or complete solution, or it will resort to classical planning techniques.