Grinder Selection Using Vibration Data for Grinding Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The selection of an appropriate grinder for grinding machining is often time-consuming and requires empirical knowledge, as existing methods rely on grinder condition files that need to be generated for each grinding machine, which is labor-intensive and prone to errors due to vibration issues.
Innovation Solution
A grinder selection device and method that utilize a learned model from supervised learning to independently select an optimum grinder based on input grinding conditions, including geometry and vibration data, to form an adequate combination with the grinding machine, eliminating the need for machine-specific files.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a grinder condition file is generated for each grinding machine based on empirical knowledge, then the selection accuracy of grinder can be improved, but the time and labor required for file generation increases significantly
Solution Approach 1:
The patent replaces the manual expert-based file generation process with an automated machine learning system. The learning device automatically creates grinder condition files by analyzing vibration data and grinding conditions, eliminating the need for experts to manually generate files for each grinding machine. This substitution of manual mechanical work with automated computational processing resolves the contradiction between accuracy and time consumption.
Solution Approach 2:
The system enables self-service automation where the learning device independently generates grinder condition files without requiring expert intervention. The automated generation process uses collected vibration data and grinding conditions to create selection criteria automatically, allowing the system to serve itself rather than relying on external expert knowledge for file creation.
2Reliability
If expert knowledge is used to generate grinder condition files, then the reliability of grinder selection can be improved, but the complexity and difficulty of the selection process increases
Solution Approach 1:
The patent extracts the essential selection criteria from complex expert knowledge and encapsulates it in automated learning algorithms. By separating the core decision-making logic from the complex manual evaluation process, the system maintains reliability while reducing operational complexity. The learning device extracts patterns from vibration data and grinding conditions to automatically determine suitable grinders.
Solution Approach 2:
The patent replaces complex manual expert evaluation with automated machine learning processing. The learning device substitutes human experts in analyzing vibration data and determining grinder suitability, thereby maintaining reliable selection while eliminating the complexity associated with manual expert-based decision-making processes.
3Adaptability or versatility
If grinder selection is based on vibration data and geometry, then the adaptability to different grinding conditions can be improved, but the measurement and detection difficulty increases
Solution Approach 1:
The patent introduces vibration sensors and data processing systems as intermediaries between the grinding machine and the grinder selection process. These intermediaries automatically capture and analyze vibration data, transforming complex physical measurements into usable selection criteria. The intermediary systems handle the measurement complexity, allowing the core selection process to focus on adaptability to different grinding conditions.
Data Source
AI summary
A grinder selection device includes: an input unit that inputs a grinding condition for a workpiece as a grinding target of grinding machining including at least the geometry of the workpiece and vibration data indicating vibration of a grinding machine, and grinder information about one or more grinders as grinder candidates to be used for the grinding machining; a learned model acquired through supervised learning using training data containing input data and label data, the input data containing an arbitrary grinding condition for a workpiece as a grinding target of grinding machining by an arbitrary grinding machine including at least the geometry of the workpiece and vibration data indicating vibration of the grinding machine, and grinder information about an arbitrary grinder, the label data being data indicating the adequacy or inadequacy of a combination between the grinding condition and the grinder information about the grinder; and a judgment unit.