Interaction prediction optimization in multidisciplinary design optimization problems.

Area of Science:

• Engineering Optimization
• Systems Engineering
• Computational Methods

Background:

• Collaborative Optimization (CO) is a distributed strategy for large-scale engineering systems.
• Traditional CO faces convergence challenges with high-dimensional coupled systems and struggles to incorporate discipline objectives.
• The Interaction Prediction Method (IPM) is a control strategy for large-scale systems.

Purpose of the Study:

• To enhance Collaborative Optimization (CO) for Multidisciplinary Design Optimization (MDO) problems.
• To address convergence issues in high-dimensional coupled systems.
• To enable simultaneous optimization of individual discipline objectives within a hierarchical framework.

Main Methods:

• Introduction of the Interaction Prediction Optimization (IPO) method, combining IPM with CO.
• IPO employs a hierarchical strategy with system and subsystem levels.
• Interaction design variables are managed at the system level and passed to the subsystem level; enhanced compatibility constraints reduce design variable dimensions.

Main Results:

• The proposed IPO method effectively integrates CO and IPM for MDO.
• IPO allows for simultaneous optimization of discipline objectives at the subsystem level.
• Enhanced compatibility constraints help reduce the dimensionality of design variables.

Conclusions:

• The Interaction Prediction Optimization (IPO) method offers a robust approach for solving large-scale MDO problems.
• IPO enhances the convergence and optimization capabilities of traditional CO.
• The method demonstrates potential applications in complex engineering design scenarios.