Ultrasonic Indication Classification via Pattern Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing ultrasonic testing methods face challenges in identifying reflected ultrasonic waves within measurements, providing limited information on target characteristics without additional equipment or requiring trained personnel, increasing complexity and cost.
Innovation Solution
A method and system using pattern recognition to classify ultrasonic indications by comparing received ultrasonic measurements with pre-defined patterns, employing an analyzer connected to ultrasonic sensors to identify target characteristics, and a trainer to improve classifier accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing ultrasonic analysis methods are used, then equipment complexity is reduced, but measurement precision and information completeness deteriorate
Solution Approach 1:
The system performs preliminary action by pre-training classification models with labeled ultrasonic data before actual inspection. The trainer component prepares classification rules in advance that can automatically identify target characteristics during operation, eliminating the need for complex real-time analysis equipment while improving measurement precision through pre-computed classification algorithms.
Solution Approach 2:
The patent replaces complex mechanical/electronic analysis equipment with software-based pattern recognition. Instead of using additional sophisticated hardware to improve measurement precision, the system substitutes computational algorithms that analyze ultrasonic signal patterns, thereby maintaining simple equipment while achieving high identification accuracy through intelligent software processing.
2Measurement precision
If trained personnel are used for ultrasonic analysis, then measurement precision improves, but device complexity and operational complexity increase
Solution Approach 1:
The system implements self-service by enabling automatic classification of ultrasonic indications without requiring trained personnel intervention. The trainer component automatically learns from labeled data and the analyzer automatically applies classification rules, making the system self-sufficient in performing accurate target characteristic identification while greatly simplifying operation for end users.
Solution Approach 2:
The system uses feedback mechanisms where the trainer component continuously improves classification accuracy by learning from labeled ultrasonic data and known target characteristics. This automated feedback loop replaces the need for human expert judgment, maintaining high measurement precision while eliminating the operational complexity associated with requiring trained personnel for analysis.
3Measurement precision
If additional equipment is used to improve analysis capability, then measurement precision improves, but device complexity and cost increase
Solution Approach 1:
The patent applies universality by designing a system where the trainer and analyzer components can handle multiple classification tasks using the same core infrastructure. The pattern recognition framework is versatile and can classify different types of ultrasonic indications (defects, material properties, geometric features) without requiring specialized equipment for each task, thereby improving measurement precision across multiple applications while maintaining equipment simplicity.
Solution Approach 2:
The system uses copying by creating digital models of target characteristics through training with labeled data. Instead of using additional physical equipment to detect and analyze target characteristics, the system creates virtual representations (classification patterns) that can be repeatedly applied to new ultrasonic measurements, achieving high measurement precision without the cost and complexity of additional hardware.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances the accuracy of target characteristic identification, reduces the need for trained personnel, and simplifies ultrasonic testing by automating the classification process, thereby decreasing costs and complexity.
Implementation Method 1
Ultrasonic testing is one type of NDT. Ultrasound is acoustic (sound) energy in the form of waves that have an intensity (strength) which varies in time at a frequency above the human hearing range. In ultrasonic testing, one or more ultrasonic waves can be directed towards a target in an initial pulse.
Implementation Method 2
As the ultrasonic waves contact and penetrate the target, they can reflect from features such as outer surfaces and interior defects (e.g., cracks, porosity, etc.). An ultrasonic sensor can acquire ultrasonic measurements, acoustic strength as a function of time, that include these reflected ultrasonic waves.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A pattern recognizing ultrasonic testing system 200 and methods for using the same are provided. The system can include an ultrasonic probe 202 and an analyzer 204. The ultrasonic probe 202 can be configured to acquire ultrasonic measurements from target 216, where the ultrasonic measurements can contain one or more ultrasonic patterns representing a target characteristic. The analyzer 204 can be in communication with the ultrasonic probe 202. The analyzer 204 can also be configured to receive the ultrasonic measurements, receive a first classifier configured to identify a predetermined target characteristic based upon one or more predetermined first ultrasonic patterns, examine the received ultrasonic measurements using the first classifier, and identify ultrasonic patterns of the received ultrasonic measurements that correspond to the predetermined ultrasonic patterns as representing the predetermined target characteristic. With pattern recognition, it can be possible to not only identify but distinguish between different target characteristics contained within the received ultrasonic measurements.