Bayesian Evaluation Weighting for Constrained Experimental Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Bayesian optimization methods face inefficiencies when applied to optimization problems with constraint conditions, particularly in industrial product development and manufacturing processes, as they may exclude candidate experimental points with low probability of falling within standard ranges, leading to suboptimal solutions and increased calculation costs.
Innovation Solution
An evaluation device that uses Bayesian optimization to calculate evaluation values for candidate experimental points based on experimental result data, objective data, and constraint condition data, incorporating a region reduction rule to divide the characteristic space and adjust weighting according to conformity with constraint conditions, allowing for efficient search within standard ranges and optimizing objective characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional Bayesian optimization is used to search for optimal experimental conditions, then the optimization process can be automated and quantitative evaluation can be performed, but candidate experimental points with low probability of falling within standard ranges are excluded, leading to suboptimal solutions and increased calculation costs
Solution Approach 1:
The patent changes the parameter of probability threshold from a fixed value to a dynamically adjustable parameter. By allowing the threshold to be modified based on problem characteristics and requirements, the system can balance between evaluation accuracy and search efficiency, avoiding premature exclusion of potentially optimal candidate points while maintaining automated quantitative evaluation.
2Reliability
If the probability threshold for accepting candidate experimental points is set high to ensure constraint satisfaction, then reliability of meeting standard ranges improves, but the search space is overly restricted and true optimal solutions may be missed
Solution Approach 1:
The patent introduces dynamic adjustment of the probability threshold parameter throughout the optimization process. The threshold can be adapted based on the current state of the search, the characteristics of the objective function, and the requirements of the problem, allowing the system to maintain reliability while preserving adaptability to find true optimal solutions.
Solution Approach 2:
By treating the probability threshold as a可调 parameter rather than a fixed constant, the system can modify it to balance constraint satisfaction and solution optimality. This parameter change allows the optimization process to explore the solution space more effectively while still maintaining reliability in meeting standard ranges when needed.
3Adaptability or versatility
If multiple objective characteristics are optimized simultaneously with constraint conditions, then comprehensive optimization of product development goals is achieved, but the calculation complexity and time costs increase significantly
Solution Approach 1:
The patent reduces calculation time by optimizing the parameter selection and probability threshold settings when dealing with multiple objective characteristics. By carefully choosing parameters and adjusting thresholds based on the specific multi-objective problem, the system achieves comprehensive optimization without incurring excessive calculation time costs.
Data Source
AI summary
Evaluation device (100A) is a device that evaluates, by Bayesian optimization, an unknown characteristic point corresponding to a candidate experimental point based on a known characteristic point corresponding to an experimented experimental point, the evaluation device including: reception controller (10A) that acquires experimental result data (222) indicating the experimented experimental point and the known characteristic point, objective data (212) indicating an optimization objective, constraint condition data (213) indicating a constraint condition, and region reduction rule data (214); evaluation value calculator (12A) that calculates an evaluation value of the unknown characteristic point based on experimental result data (222), objective data (212), constraint condition data (213), and region reduction rule data (214); and evaluation value output unit (13) that outputs the evaluation value, in which evaluation value calculator (12A) gives a weighting according to a degree of conformity of the constraint condition to the evaluation value for at least one objective characteristic.