Numerical Analysis Abnormality Detection via Predictive Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current analysis technologies, such as computational fluid dynamics and structural analysis, often result in abnormal or interrupted results due to errors in grid design, operation conditions, or parameter settings, leading to wasted time as the entire analysis must be repeated.
Innovation Solution
An apparatus and method that predict abnormal analysis results by dividing the analysis space into cells, deriving analytic data through computational fluid dynamics, and using a diagnostic layer to compare this data with predictive data from an analytic model to detect and interrupt abnormal conditions, thereby preventing unnecessary continuation of analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If numerical analysis iteration is performed to ensure accurate design analysis, then analysis reliability is improved, but analysis time increases significantly
Solution Approach 1:
The patent applies preliminary action by performing a preliminary numerical analysis to generate reference data before the main analysis. This reference data is used to create a predictive model that can forecast analysis results, allowing the system to predict whether the main analysis will complete successfully without actually performing the entire time-consuming iteration process.
Solution Approach 2:
The patent uses copying by creating a predictive model that replicates the behavior of the full numerical analysis system. This model is trained on reference data from preliminary analyses and can copy the essential characteristics of the analysis process to predict outcomes, replacing the need to execute the complete time-consuming analysis iteration.
2Reliability
If error checking is performed during analysis to detect abnormalities, then analysis reliability is improved, but device complexity increases
Solution Approach 1:
The patent introduces an intermediary element - the predictive model - that acts as a mediator between the preliminary analysis data and the main analysis process. This model checks for potential abnormalities by comparing predicted results with expected outcomes, providing error detection functionality without requiring complex real-time monitoring of the entire analysis system.
3Reliability
If the entire analysis is repeated when an abnormality is detected, then analysis reliability is maintained, but loss of time increases
Solution Approach 1:
The patent implements feedback by using the predictive model to continuously monitor the analysis process and provide information about potential abnormalities. When the model predicts that an analysis will fail or produce abnormal results, the system can interrupt the process early, providing feedback that prevents wasting time on doomed analyses while maintaining reliability by only accepting results from successfully completed analyses.
Data Source
AI summary
An apparatus and method for diagnosing analysis is provided. The apparatus includes an analytic layer to divide a peripheral space of a target component into a plurality of cells and to derive analytic data by performing a numerical analysis iteration according to computational fluid dynamics for the plurality of cells; a model layer to derive an analytic model that simulates the numerical analysis iteration; a predictive layer to derive predictive data by predicting a result of the numerical analysis iteration by using the analytic model; and a diagnostic layer to diagnose an abnormality condition of numerical analysis by comparing the analytic data and predictive data during the numerical analysis iteration performed by the analytic layer. The diagnostic layer includes an early alarm to generate early alarm information by sorting a cell satisfying an early alarm condition; and an abnormality diagnostic device to determine whether the numerical analysis iteration is abnormal.