Abstract
A multistage approach is proposed merging a deterministic and evolutionary algorithm for PI controller tuning. This technique is formulated through design of a multi-objective optimization procedure, to ensure the construction of Pareto frontier that guarantee well distribution and exclude the non- Pareto and local Pareto points. This procedure focuses on reliability-based optimization instances. To validate the approach, we will consider the Boiler Control Benchmark. The results of its usefulness for controller tuning is demonstrated.