Sequential Multi-Objective Optimization for Chemical Mixture Degeneration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current system optimization methods for chemical mixture formulations face challenges in balancing multiple conflicting objectives and constraints, such as physicochemical properties, manufacturing costs, and toxicity, which complicates the design process and often results in degenerated models that limit the variability of target variables.
Innovation Solution
A computer-implemented method that performs multi-objective optimization by associating design parameters with objective parameters, using a system model to explore design configurations and iteratively or parallelly optimize primary and secondary objective parameters to achieve a non-degenerated multi-objective optimal design, providing a recipe profile for chemical mixtures that meets specified requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multi-objective optimization is performed with many conflicting objectives and constraints, then the design process can balance multiple requirements, but the system optimization procedure becomes complicated and may yield degenerated models
Solution Approach 1:
The patent segments the optimization process into two distinct phases: a first multi-objective optimization step that identifies a preliminary optimal design, and a second multi-objective optimization step that refines this design while preventing degeneracy. This segmentation allows the complex problem to be broken into manageable stages, reducing overall procedure complexity while maintaining the ability to balance multiple conflicting objectives.
Solution Approach 2:
The first multi-objective optimization step serves as a preliminary action that establishes a baseline optimal design before the second optimization step begins. This preliminary optimization provides a starting point that guides subsequent optimization efforts, preventing the system from exploring unnecessarily complex solution spaces and reducing the likelihood of generating degenerated models.
2Productivity
If dimension reduction algorithms are used in model-based prediction, then computational efficiency is improved, but the model becomes degenerated with reduced variability in target variables
Solution Approach 1:
The patent applies dynamic optimization by performing two sequential multi-objective optimization steps rather than a single static optimization. The first step operates on the full system model with dimension reduction, while the second step introduces additional objective functions that dynamically adjust the optimization criteria. This dynamic approach allows the system to maintain computational efficiency through dimension reduction while recovering variability in target variables through the second optimization pass.
Solution Approach 2:
The patent changes optimization parameters between the two steps by introducing different objective functions and constraints. The second multi-objective optimization step uses modified parameters compared to the first step, including additional objectives that specifically target variables that may have become degenerate. This parameter change allows the system to maintain the computational benefits of dimension reduction while restoring variability where needed.
Data Source
AI summary
The present invention generally relates to a system optimization procedure. The method includesa) providing, via an input channel, a system model for modelling the chemical mixture, which associates a set of design parameters with a plurality of objective parameters that represent design characteristics of the chemical mixture, wherein the set of design parameters comprises a chemical mixture recipe having two or more ingredients, and the plurality of objective parameters comprises two or more physicochemical properties of the chemical mixture;b) defining, via the input channel, a set of primary optimization objective parameters,c) performing, by a processor, a multi-objective optimizing process on the system model by exploring a plurality of design configurations by assigning specified values to the set of design parameters; andd) determining, by the processor, if the multi-objective optimizing process yields a degenerated multi-objective optimal design.