Voxel-Based Machining Condition Prediction for Complex Workpieces
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Determining machining conditions for complex workpieces is challenging when operators lack experience, as existing methods rely heavily on operator input and know-how.
Innovation Solution
A method and device that convert workpiece shape data into voxels, allowing machine learning to determine machining conditions based on multiple example cases, including tool path patterns and surface quality, and display the results for easy recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If machining conditions are determined based on operator experience and know-how, then machining quality can be maintained for known workpiece patterns, but it becomes difficult to determine desired machining conditions for workpieces with complex shapes or no existing patterns
Solution Approach 1:
The system performs preliminary action by collecting and storing machining conditions for multiple known workpieces in advance, converting them to voxel data and training a neural network model beforehand. This pre-trained model can then automatically determine machining conditions for new workpieces without requiring operator experience, resolving the contradiction between maintaining quality for known patterns and adapting to new patterns.
Solution Approach 2:
The system uses copying by creating voxel representations of known workpieces and their machining conditions, then using these copied data patterns to train a neural network. The network learns from these copies and can generate machining conditions for new workpieces by analogy, enabling the system to handle workpieces with no existing patterns while maintaining accuracy.
2Ease of operation
If traditional CAM software is used to generate machining conditions, then operator control and flexibility are maintained, but automation level remains low and productivity is limited
Solution Approach 1:
The system implements self-service by enabling automatic determination of machining conditions through the neural network model. The system serves itself by learning from historical data and autonomously generating machining conditions for new workpieces, reducing dependency on operator intervention while maintaining ease of operation through simple voxel input and automated output.
Solution Approach 2:
The system replaces the mechanical system of manual operator judgment and CAM software configuration with an intelligent system based on neural networks and machine learning. This substitution automates the machining condition determination process, significantly improving productivity while maintaining operational simplicity through automated workflows.
3Productivity
If machine learning is implemented to automatically determine machining conditions, then productivity and automation are improved, but system complexity increases
Solution Approach 1:
The system applies local quality by processing workpieces in discrete voxel units and determining machining conditions for each voxel independently through the neural network. This localized processing approach simplifies the overall system structure by breaking down complex workpiece analysis into manageable unit operations, reducing system complexity while maintaining high automation and productivity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method is provided with a step of converting shape data (51) into a plurality of voxels (55) for each of a plurality of known objects being processed, a step of setting, for each of the known objects being processed, a processing condition for a voxel (55) constituting a processing surface, a step of using the voxel (55) of the plurality of known objects being processed and the processing condition to perform machine learning in which an input is the voxel (55) and an output is a processing condition, a step of converting shape data (52) of a candidate object being processed into a plurality of voxels (56), a step of setting, on the basis of results of machine learning, a processing condition for a voxel (56) constituting a processing surface of the candidate object being processed, and a step of determining a processing condition for each processing surface of the candidate object being processed, a processing condition that is set for the largest number of voxels (56) in one processing surface being determined as a processing condition for said processing surface.