Minimizing Cycle Time in Parallel Placement Machines



In the first part of my presentation I will present research work I have done in the field of large-scale combinatorial optimization applied to electronics manufacturing. Placement machines are used to mount components on printed circuits boards (PCBs). For PCB manufacturers it is important to minimize the cycle times for the PCBs they assembly in order to be competitive. Obviously, the product cycle time depends on the mechatronic behavior (like maximum speed and acceleration of robot arms) of the machine. It is however not only the hard- ware that determines the cycle time. In this presentation I will show that finding an optimal (or sometimes even feasible) placement program for producting a PCB on a parallel placement machine requires the solving a chain of NP-hard problems. This means that the strength of the optimization algorithms also have a big impact on the cycle time for mounting components on a PCB. In this presentation I will present the mathematical model and mathematical algorithms that I have used to find good (hopefully near-optimal) solutions. Special attention is given to problem decomposition, local serach and network flow algorithms. In the second part of my presentation I will give a brief overview of the research topics the Business Optimization Group of IBM Research is working on. Special attention will be given to IBM's Dynamic Inventory Optimization Solution (DIOS).

Contact information:

Naima Zerhane