3D Weld Appearance Inspection Across Changing Scan Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing weld appearance inspection systems face challenges in maintaining accurate inspection accuracy when inspection conditions such as speed and resolution are changed, leading to inconsistent detection of weld defects due to varying production takt times and workpiece materials and shapes.
Innovation Solution
An appearance inspection apparatus and method that includes a shape measurement unit, data processor, and determination model generator to generate multiple types of determination models based on learning data sets, allowing accurate evaluation of weld shape despite changes in inspection conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If inspection speed is increased to reduce production takt time, then productivity is improved, but measurement resolution deteriorates making small weld defects undetectable
Solution Approach 1:
The patent applies preliminary action by pre-processing shape data through noise removal and generating multiple learning data sets with different data densities before creating determination models. This preparation ensures that when high-speed inspection occurs, the pre-processed data and multiple models are ready to maintain accuracy without requiring slower inspection speeds.
Solution Approach 2:
The patent changes parameters by generating multiple determination models corresponding to different data densities and resolutions. Instead of using a single fixed model, the system adapts by selecting or generating models matched to the actual inspection conditions, allowing high-speed inspection with lower resolution to use appropriate low-density models while maintaining accuracy.
2Measurement precision
If measurement resolution is increased to detect small weld defects, then measurement precision is improved, but production takt time increases reducing productivity
Solution Approach 1:
The patent enables parameter changes by creating multiple determination models with different data densities and resolutions. This allows the system to select the appropriate model resolution based on the specific inspection needs and production requirements, rather than being locked into a single high-resolution model that would slow down inspection.
3Reliability
If determination models are generated using shape data acquired at fixed inspection conditions, then reliability is improved for those specific conditions, but adaptability deteriorates when inspection conditions change
Solution Approach 1:
The patent applies universality by generating multiple determination models that can handle different inspection conditions. Instead of a single model limited to fixed conditions, the system creates a family of models with varying data densities and resolutions, making the inspection system universally applicable across different speeds, resolutions, and production requirements.
Solution Approach 2:
The patent enables parameter changes in determination models to match varying inspection conditions. The system generates models corresponding to different data densities and resolutions, allowing the inspection accuracy to be maintained whether inspecting at high speed with low resolution or at lower speed with high resolution.
4Adaptability or versatility
If multiple determination models are generated for different data densities and resolutions, then adaptability to inspection condition changes is improved, but device complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-processing shape data through noise removal and generating multiple learning data sets with different data densities before model creation. This preliminary preparation organizes the data systematically, making the subsequent generation of multiple determination models more manageable and less complex.
Solution Approach 2:
The patent applies segmentation by dividing the learning process into distinct stages: noise removal, generating first learning data sets, generating second learning data sets with different data densities, and creating determination models. This segmented approach breaks down the complex task of creating multiple models into manageable steps.
Data Source
AI summary
An appearance inspection apparatus includes a shape measurement unit configured to measure the three-dimensional shape of a weld and a data processor configured to process sample shape data and shape data acquired by the shape measurement unit. The data processor includes a first learning data set generator configured to generate a plurality of first learning data sets based on the sample shape data, a second learning data set generator configured to generate a plurality of second learning data sets based on the first learning data sets, a determination model generator configured to generate multiple types of determination models using the second learning data sets, and a first determination unit configured to determine whether the shape of the weld is good or bad based on the shape data and one of the determination models.