Machine Tool Control Using Workpiece Accuracy Feedback

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional machine tool control techniques do not consider the quality of the manufactured workpiece, leading to inconsistent results.

Innovation Solution

A machine learning-based control method that generates command data by training models using state data from the machine tool and accuracy data of the workpiece, incorporating multiple linear regression and neural network models to determine optimal machining parameters for consistent quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If conventional machine tool control techniques are used, then the control system is simple and easy to operate, but the quality consistency of manufactured workpieces deteriorates

Engineering Contradiction:
Improvequality consistency of workpieceVSAvoidcontrol system complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The control system implements feedback by collecting state data during machining operations and accuracy data from measured workpieces, then using this feedback to train and update machine learning models that generate command data for the machine tool, creating a closed-loop system that continuously improves quality consistency

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent replaces conventional mechanical control systems with a machine learning-based control system that uses neural networks and regression models to process state data and generate optimal command data, substituting traditional control mechanisms with intelligent algorithms that learn from data

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Manufacturing precision

If machine learning models are introduced to improve quality consistency, then manufacturing precision improves, but device complexity increases

Engineering Contradiction:
Improvequality consistency of workpieceVSAvoidcontrol system complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary action by collecting and storing state data and accuracy data before actual machining operations, training machine learning models in advance with this accumulated data so that when machining occurs, the models are already prepared to generate optimal command data without real-time computation delays

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The control system is made dynamic through the machine learning models that continuously learn and adapt from new data, allowing the system to evolve and improve its control strategies over time rather than relying on static control parameters

Inventive Principle:
Principle #15Dynamics

3Productivity

If adaptive control is used to adjust feed speed based on load torque, then machining efficiency improves, but quality consistency deteriorates

Engineering Contradiction:
Improvemachining efficiencyVSAvoidquality consistency of workpiece
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The system changes multiple machining parameters simultaneously based on state data, not just feed speed. The machine learning model processes comprehensive state data including machining conditions, machine tool status, and workpiece properties to dynamically adjust multiple parameters together, achieving both efficiency and quality consistency

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4035829B1Control system
Publication Date: 2024.07.24 DAIKIN INDUSTRIES LTD
  • EP4035829B1 patent drawingFigure 1
  • EP4035829B1 patent drawingFigure 2~3
  • EP4035829B1 patent drawingFigure 4(a)~4(b)

AI summary

A machine tool control technique that takes into consideration the quality of the manufactured workpiece is provided. An embodiment of the present disclosure relates to a control system comprising a state observation unit configured to collect state data relating to a machine tool, the state data including one or more of machining state data, equipment state data, and material state data, an accuracy result observation unit configured to acquire accuracy data of a workpiece made by the machine tool, a learning unit configured to train a machine learning model by using the collected state data and the acquired accuracy data, and a command determination unit configured to, when machining the workpiece a same portion of which is machined a plurality of times, determine command data, through the trained machine learning model, with respect to the machine tool from the state data collected during machining of the workpiece at a certain point in time before a final process among a plurality of times of processes performed by the machine tool, and cause the machine tool to perform the final process with respect to the workpiece by using the command data.