DYNAMIC OPTIMIZATION

On AIR acquired a vast experience in developing dynamic optimisation models, applied to decision support systems for industrial applications, traffic management and scheduling in complex logistic networks. Multi-disciplinary knowledge enables On AIR to select, tailor and integrate different methodologies and solutions, on the basis of process and market requirements.

On AIR addresses a large part of industrial applications applying innovative methodologies for optimisation problems:

  • Graph modelling and optimisation
  • Shortest path algorithms
  • Dynamic programming algorithms
  • Constraint programming
  • Evolutive Programming Algorithms (e.g. genetic algorithms)
  • Statistical classification
  • Neural networks and neuro fuzzy
  • Interpolation and extrapolation methods

On AIR offer in the field of optimisation systems is based on two main components, that are specialised for different applications. They can be joined or taken individually, depending on the specific objective.

The first one is a long-term planning system: given an initial situation and a long-term objective, it is able to find the optimal sequence of basic operations, by applying an alternative graph model, in which nodes represent events and arcs describe precedence relations between events. The system can incorporate external constraints and foresee events: it can prevent deadlock situations or intractable delays.

The second one is a control system: given a current situation and a short-term objective, it generates in real-time the control actions needed to match the goals, minimizing the costs. The cost function depends on the application type and can be tailored.