open menu
MENU

Optimization Algorithms

ESTECO expertise in numerical solutions equips designers with a complete array of optimization algorithms covering deterministic, stochastic and heuristic methods for both single and multi-objective problems.

The implementation of advanced algorithms in modeFRONTIER offers a reliable tool for making the simulation process more effective.

By exploiting the flexibility of modeFRONTIER in the appropriate optimization strategy, engineers can successfully address:

  • strong non-linearity 
  • high or low constrained problems
  • sizable problem dimensions.
WHAT'S NEW​

New Self-Initializing algorithms 

Teams can now get their optimization projects on the fast track exploiting the single-parameter setup of the classic modeFRONTIER algorithms. MOGA-II, NSGA-II, MOPSO, HYBRID, SIMPLEX, ARMOGA, MOSA and Evolution Strategies can now be exploited in the SELF-I mode, where the algorithms analyze the problem characteristics, automatically generate the most suitable DOE and start searching for optimal solutions. All by setting a single parameter: the number of evaluations.
 

Fine-tuned Multi-Strategy Optimization Algorithms

Besides the traditional methods, modeFRONTIER provides fine-tuned multi-strategy optimization algorithms able to multiply the capabilities of single approaches. Engineers are able to combine powerful optimization methods to further reduce time and effort of the design cycles.

FAST | accelerates the process by exploiting RSM performance over the region of most interest in the Design Space, reaching a high-speed detection of the optimal solutions.

HYBRID | automatically combines the robustness of Genetic Algorithms with the accuracy of Gradient methods providing an unprecedented balance between exploration and refinement capabilities.

SAnGeA | provides an automatic screening phase, coupled with a Genetic Algorithms global search phase, reliably identifying the most meaningful variables to face high-dimension and unconstrained problems.

pilOPT | Combines global and local search and efficiently calibrates the portion of RSM-based evaluations to achieve optimal and robust solutions. Its architecture delivers remarkable results even for problems with elaborate output functions, when RSM modelling requires considerable effort.

Sign in