Deviations in the manufacturing process of electronic components may lead to rejections due to malfunctioning. Uncertain design variables can be modeled as random variables. Then, the probability that a product fulfills its performance feature specifications is called the yield. The aim of our research is to estimate the yield and then maximizing the yield, which corresponds to minimizing the failure probability. In current approaches the design or performance is optimized first (nominal optimization) and the yield is maximized in a second step. In this work we research on solving simultaneously the yield optimization and the nominal optimization problem. Optimization with more than one objective function is called multi-objective or Pareto optimization.
The aim of this thesis is the research, comparison and implementation of Pareto optimization approaches. The major tasks are:
- Literature study on Pareto optimization approaches
- Systematical analysis and structuring on Pareto optimization approaches in order to evaluate their applicability within the framework of yield optimization
- Implementation of selected approaches in Python
- Inclusion of Pareto optimization approaches in the existing yield optimization algorithms
Basis knowledge of electromagnetism, programming and mathematical optimization, some experience with Python, interest in electromagnetic field simulations and stochastics.