Condition Optimization Using Bayesian and Response Surface Convergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Bayesian optimization requires numerous tests to determine if condition values are optimized, and the response surface method necessitates pre-setting condition values for future experiments, making the process inefficient.
Innovation Solution
A condition optimization device that combines Bayesian optimization to calculate new condition values, a response surface method to predict optimal physical property values, and determination units to assess convergence based on actual reaction values and predicted values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If Bayesian optimization is used to determine optimized condition values, then the reliability of optimization is improved, but the number of tests required increases significantly
Solution Approach 1:
The patent merges Bayesian optimization with response surface method prediction. The system uses the response surface method to predict optimal condition values and their reaction values, then uses Bayesian optimization to evaluate these predictions. This combination allows the system to leverage the speed of response surface predictions while maintaining the reliability of Bayesian optimization through convergence determination based on prediction accuracy.
Solution Approach 2:
The patent implements a feedback mechanism where the system determines convergence by comparing predicted reaction values from the response surface method with actual measured values. When the difference between predicted and actual values falls within a predetermined threshold, the system concludes optimization has converged. This feedback loop enables early termination of tests while maintaining optimization reliability.
2Loss of time
If response surface method is used to predict optimal condition values, then the number of experiments is reduced, but the experimenter must pre-set condition values for future experiments
Solution Approach 1:
The patent makes the system self-service by implementing automated convergence determination. The system automatically compares predicted reaction values with actual measured values and determines whether optimization has converged based on predetermined accuracy thresholds. This eliminates the need for experimenters to manually pre-set future condition values, as the system autonomously decides when to terminate experiments based on convergence criteria.
3Reliability
If Bayesian optimization is used without convergence determination, then comprehensive optimization is achieved, but the number of tests increases unnecessarily
Solution Approach 1:
The patent replaces the mechanical trial-and-error approach of continuous testing with an intelligent determination system. Instead of mechanically continuing tests until a fixed number is reached, the system uses convergence determination based on prediction accuracy to intelligently substitute when further tests are unnecessary. This maintains optimization completeness by ensuring convergence while dramatically improving productivity by eliminating unnecessary tests.
Data Source
AI summary
A condition optimization device includes an actual reaction value obtainment unit to obtain an actual reaction value by observing a predetermined physical property value at a predetermined condition value; a condition value calculation unit to calculate a new condition value by Bayesian optimization; an optimum condition value prediction unit to predict, by a response surface method, a best physical property value and an optimum condition value at which the best physical property value is obtainable; a first determination unit to determine whether or not convergence has occurred based on the condition value corresponding to the actual reaction value and based on the new condition value; and a second determination unit to determine, upon the first determination unit determining that the convergence has occurred, whether or not convergence has occurred based on a difference between the physical property value observed at the optimum condition value and the best physical property value.