Composition Optimization with Backpropagation for Phase Fraction Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing models require significant computational time to predict phase fractions due to a large number of parameters, making it impractical to search for a material composition corresponding to a target phase fraction within a realistic time frame.
Innovation Solution
A composition optimization device utilizing a trained model with backpropagation to update material compositions based on predicted phase fractions until a predetermined error condition is met, employing architectures like RNN, bidirectional RNN, Seq2Seq, Seg2Seq with attention mechanism, GRU, LSTM, or Transformer to efficiently search for compositions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a traditional model with many parameters is used to predict phase fraction, then prediction accuracy is maintained, but computational time becomes excessively long
Solution Approach 1:
The patent changes the parameters of the prediction model by replacing a traditional model with many parameters with a neural network model having fewer parameters. This parameter reduction enables faster computation while maintaining acceptable prediction accuracy for phase fraction calculations across different material compositions and temperatures.
Solution Approach 2:
The patent creates a simplified copy of the traditional model using neural networks that replicates the essential prediction functionality with reduced complexity. The neural network is trained to copy the behavior of the traditional model but executes predictions much faster, enabling practical use in composition optimization loops.
2Manufacturing precision
If exhaustive search of material compositions is performed to find target phase fraction, then comprehensive solution is achieved, but search time becomes unrealistic
Solution Approach 1:
The patent implements feedback mechanisms where the neural network prediction results are used to guide the next set of composition predictions. By using the predicted phase fractions and their errors to inform subsequent search directions, the system efficiently converges to optimal compositions without exhaustive searching, significantly improving productivity while maintaining optimization accuracy.
Data Source
AI summary
A material composition corresponding to a target phase fraction is efficiently searched for. A composition optimization device includes a prediction unit configured to predict, by inputting a predetermined material composition into a trained model, a phase fraction of a material having the predetermined material composition at each temperature within a predetermined temperature range, the trained model being trained using training data in which a material composition of a training target material and a phase fraction of the training target material at each temperature within the predetermined temperature range are associated with each other; and an update unit configured to update the predetermined material composition input into the trained model by performing backpropagation of an error calculated based on a target phase fraction and the predicted phase fraction. A process of the prediction unit predicting a phase fraction based on the updated material composition and a process of the update unit updating a predetermined material composition by performing backpropagation of a calculated error are repeated until a calculated error satisfies a predetermined condition.