Multi-Objective Evolutionary Algorithm Constraint Minimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large-scale value-based non-linear models with numerous decision variables and constraints pose computational challenges for Multi-Objective Evolutionary Algorithms (MOEAs), as they require significant bandwidth to search large decision spaces and evaluate constraint violations, often exceeding the resources needed for objective evaluations.
Innovation Solution
A multi-tiered optimization system using a secondary optimization engine for constraint minimization, where items are allocated into categories through negotiation-based driving, allowing for scalability to millions of decision variables, and enabling the discovery of feasible solutions by trading items between categories to reduce constraint violations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If MOEA searches large decision variable space and evaluates constraint violations, then solution feasibility is improved, but computational bandwidth requirement increases significantly
Solution Approach 1:
The patent segments the optimization process into two distinct engines: a primary MOEA engine that focuses on objective optimization without constraint evaluation, and a secondary constraint minimization engine that专门 handles constraint violations. This segmentation allows each engine to specialize in one aspect, reducing the computational bandwidth required for constraint checking in the main optimization loop.
Solution Approach 2:
The patent extracts the constraint evaluation and minimization functionality from the traditional MOEA framework into a separate secondary engine. This extraction removes the computational burden of constraint checking from the main optimization process, allowing the primary engine to focus solely on objective optimization while the secondary engine handles feasibility separately.
2Reliability
If MOEA evaluates constraint violations for each potential solution, then solution feasibility is improved, but computational time increases
Solution Approach 1:
The secondary constraint minimization engine performs preliminary constraint checking and minimization on potential solutions before they are fully evaluated by the primary MOEA engine. This preliminary action filters out infeasible solutions early, reducing the computational time required for full evaluation of problematic solutions.
Solution Approach 2:
The patent implements a feedback mechanism where the secondary engine provides constraint violation information back to the primary MOEA engine. This feedback allows the primary engine to adjust its search strategy and focus on regions of the decision space that are more likely to produce feasible solutions, reducing overall computational time.
3Adaptability or versatility
If MOEA handles large number of decision variables, then optimization scope is improved, but computational bandwidth requirement increases
Solution Approach 1:
The patent segments the handling of large-scale decision variables by assigning different roles to the two engines: the primary MOEA engine handles the broad search across the large decision variable space for objective optimization, while the secondary engine focuses specifically on constraint minimization. This segmentation allows the system to scale to millions of decision variables without proportionally increasing the computational bandwidth required for constraint checking.
Data Source
AI summary
Systems and methods are provided to engage in multi-tiered optimization where there may be a first multi-objective optimization and a second constraint optimization. The multi-objective optimization may be used to drive to one or more goals of the optimization problem. The constraint optimization or minimization may be used to drive towards a reduced and/or no constraint situation where the solution to the overall problem is feasible or near-feasible.