Thermal Fluid Model Accuracy via Time Differential Regression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques for modeling physical phenomena using symbolic regression struggle to improve accuracy and efficiently handle varying time steps and long-term predictions in thermal fluid analysis.
Innovation Solution
An information processing device that calculates time differential values and differences from initial values, estimates coefficients using machine learning with a sparse estimation technique, and generates a linear regression equation by mixing short-term and long-term components, allowing for improved accuracy in thermal fluid analysis models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If symbolic regression is used to obtain a mathematical model from time-series data, then a model can be generated, but the accuracy of generating a model of a physical phenomenon cannot be further improved
Solution Approach 1:
The patent segments the time-series data into multiple time intervals and performs separate regression analysis for each interval. This allows the model to capture different behavioral patterns at different time scales, improving overall model accuracy while maintaining manageable complexity through modular processing
Solution Approach 2:
The patent introduces dynamic time step adjustments and adaptive regression parameters that change based on the characteristics of different time intervals. This dynamic approach allows the model to adapt to varying data patterns, significantly improving accuracy compared to static symbolic regression methods
2Measurement precision
If conventional symbolic regression techniques are applied, then a model can be obtained, but learning time and data requirements are excessive
Solution Approach 1:
The patent performs preliminary data preprocessing and feature extraction before applying regression analysis. By pre-processing the time-series data to extract key characteristics and reduce dimensionality, the subsequent modeling process requires less computational time and fewer data samples while maintaining or improving accuracy
Solution Approach 2:
The patent applies regression analysis selectively to specific time intervals and key variables rather than processing all data uniformly. This partial action approach focuses computational resources on the most influential segments of the data, reducing overall learning time while capturing essential patterns
3Device complexity
If a single model is used for both short-term and long-term predictions, then simplicity is maintained, but accuracy for varied time steps deteriorates
Solution Approach 1:
The patent divides the prediction task into multiple segments corresponding to different time intervals. Each segment has its own optimized regression model parameters, allowing accurate predictions for both short-term and long-term variations without requiring a single complex unified model
Solution Approach 2:
The patent dynamically adjusts model parameters based on the prediction time horizon. By changing regression coefficients, time step sizes, and weighting factors according to whether the prediction is short-term or long-term, the system achieves high accuracy across varied time steps while maintaining a relatively simple model structure
Data Source
AI summary
According to an embodiment, an information processing device includes a memory and one or more processors coupled to the memory. The memory stores therein time-series data including one or more variables. The one or more processors are configured to: calculate one or more time differential values of the one or more variables; calculate one or more differences representing variation of the one or more variables from an initial value; estimate a coefficient of a linear regression equation by machine learning in which the time differential values and the differences are used as learning data; and output the linear regression equation.