Estimation Device for Rotary Electric Machine Crack Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing crack size estimation methods for rotary electric machine rotators are prone to inaccuracies due to ill-posed inverse problems, leading to unstable and non-unique solutions, which can result in undetected cracks and reduced structural lifetime.
Innovation Solution
An estimation device and method that utilize a measurement device to capture surface changes and an estimator to determine a candidate crack surface within the structure by solving a norm minimization problem, forming a sparse solution based on a shape model and measurement data, ensuring uniqueness, existence, and stability of the solution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If inverse analysis is used to estimate crack position and size based on surface strain measurements, then crack detection capability is improved, but solution stability deteriorates when the inverse problem is ill-posed
Solution Approach 1:
The patent transforms the ill-posed inverse problem into a well-posed optimization problem by changing the mathematical formulation. It introduces a cost function with regularization terms that modify the parameters of the inverse analysis, converting unstable solutions into stable ones through parameter transformation while maintaining crack detection accuracy.
Solution Approach 2:
The patent implements an iterative optimization process where the estimation results are continuously fed back to adjust the model parameters and refinement the crack detection. This feedback mechanism ensures convergence to stable and accurate solutions by repeatedly adjusting the inverse analysis based on measurement residuals and model predictions.
2Measurement precision
If inverse analysis is used to estimate crack position and size, then crack detection capability is improved, but solution uniqueness deteriorates
Solution Approach 1:
The patent resolves the non-uniqueness issue by transforming the inverse problem into an optimization problem with a cost function that includes regularization terms. This parameter transformation introduces constraints that eliminate ambiguous solutions, ensuring that the crack detection results are unique and well-defined while preserving detection accuracy.
Solution Approach 2:
The patent segments the crack detection problem into multiple independent components by dividing the crack occurrence surface into discrete elements. This segmentation allows the optimization process to independently determine crack parameters for each element, ensuring unique and unambiguous solutions while maintaining overall detection accuracy.
3Device complexity
If existing crack size estimation methods are used, then apparatus size is reduced, but measurement accuracy deteriorates due to ill-posed inverse problems
Solution Approach 1:
The patent improves measurement accuracy while maintaining compact apparatus size by changing the mathematical parameters of the inverse analysis. It introduces regularization parameters and optimization criteria that enhance the accuracy of crack size estimation without requiring additional physical measurement devices, thus preserving the compactness advantage while resolving the accuracy issue.
Data Source
AI summary
A measurement device is configured to set an observation surface on a surface of a structure as a measurement surface to measure a change of the measurement surface as a measurement surface change vector. An estimator is configured to generate an estimation model based on a shape model obtained by modeling a shape of the structure. The estimator is configured to acquire a coefficient vector by solving a norm minimization problem by setting, as parameters, a measurement surface change vector and a part of the estimation model. The coefficient vector forms a sparse solution. The estimator is configured to estimate a change of a crack occurrence surface by determining a candidate surface, which is inside the structure and assumed to have a crack, as the crack occurrence surface, based on the coefficient vector and another part of the estimation model.