PostHeaderIcon Scheduling of production lines in automotive factories with the modeling, simulation and optimization technology

    Az esemény címe:
    Scheduling of production lines in automotive factories with the modeling, simulation and optimization technology

Az esemény műfaja:
Előadás

Tudományterület:
Matematika, Műszaki tudományok 

Kezdés: 
2014. November 25. 16:15

Befejezés:
2014. November 25. 16:15 

Program:  

Zoltán Horváth, János Jósvai (Széchenyi István University, Győr)
Scheduling of production lines in automotive factories with the modeling, simulation and optimization
technology
Scheduling of production lines with many jobs and machines is one of the most serious task of an
automotive factory. Several commercial software products serve decision makers on the production in
their job providing them with simulation and some optimization tools. However, even the state-of-theart
software tools has serious limitations, in particular those for the optimization. For supporting the
scheduling decisions of engine producing segments, Audi Hungaria Motor Ltd, Győr and the Széchenyi
István University formed cooperations to make simulations of real production lines and their supply
chains and do research for making optimization methods better.We remark that these lines are among
the largest engine production lines of vehicle industry over the world. In this talk we present our
achievements in simulation, modeling and optimization for the scheduling of large production lines.

Namely, we made contributions as follows:

  1. validated models and their simulations using Siemens’ software Plant Simulation for real production lines and their material supply,
  2. optimization under Plant Simulation via built in genetic algorithms,
  3. set up of mathematical models of different levels (i.e. by considering more and more physical features of the lines) and implementing corresponding solvers of the type of black-box heuristics and MIP (mixed integer programming),
  4. fast computational optimization of the large scale problems in the mathematical models (several heuristics and MIP solvers) with interface from and to Plant Simulation,
  5. applications to real lines of industry.

Szervező intézmények:
BME Matematikai Intézet

Helyszínek:
BME K épület I. em. 50. terem

Régió: 
Közép-Magyarország

Kapcsolattartó:
Gergely Madi-Nagy, gnagy@math.bme.hu

Az esemény honlapja:
http://www.math.bme.hu/~gnagy/mmsz/HorvathZoltan2014.htm

Szinopszis:
Scheduling of production lines with many jobs and machines is one of the most serious task of an automotive factory. Several commercial software products serve decision makers on the production in their job providing them with simulation and some optimization tools. However, even the state-of-theart software tools has serious limitations, in particular those for the optimization. For supporting the scheduling decisions of engine producing segments, Audi Hungaria Motor Ltd, Gy˝ or and the Széchenyi István University formed cooperations to make simulations of real production lines and their supply chains and do research for making optimization methods better.We remark that these lines are among the largest engine production lines of vehicle industry over the world. In this talk we present our achievements in simulation, modeling and optimization for the scheduling of large production lines.

Namely, we made contributions as follows:

  1. validated models and their simulations using Siemens’ software Plant Simulation for real production lines and their material supply,
  2. optimization under Plant Simulation via built in genetic algorithms,
  3. set up of mathematical models of different levels (i.e. by considering more and more physical features of the lines) and implementing corresponding solvers of the type of black-box heuristics and MIP (mixed integer programming),
  4. fast computational optimization of the large scale problems in the mathematical models (several heuristics and MIP solvers) with interface from and to Plant Simulation,
  5. applications to real lines of industry.

The main conclusion was that solvers for some MIP models converge fast for large scale problems as well and give optimum in industrial cases. 
This is a joint lecture with Sándor Kálmán (Audi Hungária Motors Ltd.).

 

Tudomány Ünnepe 2014