Machine Learning Device for Dynamic Abnormal Load Threshold
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques face challenges in calculating a suitable threshold value for detecting abnormal load torque in machine tools, as the magnitude of load torque varies with factors like cutting speed and cut depth, making it difficult to accurately determine when an abnormal load is applied.
Innovation Solution
A machine learning device that observes state variables such as tool information, spindle revolution rate, coolant amount, workpiece material, and cutting speed to learn a threshold value for detecting abnormal loads, using training data and teacher data to update a learning model and calculate errors, enabling the detection of abnormal loads in machine tools.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a fixed threshold value is set for detecting abnormal load torque, then the detection method is simple, but the detection accuracy deteriorates because load torque magnitude varies with cutting speed and cut depth
Solution Approach 1:
The patent implements dynamic threshold adjustment by using a neural network to calculate appropriate threshold values based on real-time machining conditions (cutting speed, cut depth, workpiece material). The threshold is no longer fixed but dynamically adapts to varying operational parameters, resolving the contradiction between simplicity and accuracy by automating the adjustment process through machine learning.
Solution Approach 2:
The patent changes the threshold parameter from a fixed value to a variable that depends on machining conditions. The neural network learns the relationship between machining parameters and appropriate threshold values, allowing the system to adjust the threshold dynamically based on cutting speed, cut depth, and other operational factors, thereby maintaining high detection accuracy across different operating conditions.
2Measurement precision
If the threshold is set to account for various machining conditions, then the detection accuracy improves, but the device complexity increases due to need for neural network and multiple sensors
Solution Approach 1:
The patent makes the neural network serve multiple functions: it learns from historical data, predicts normal load torque under various conditions, and determines appropriate threshold values. This multi-functionality reduces the need for separate systems for each task, thereby limiting the increase in device complexity while achieving high detection accuracy across diverse machining conditions.
Solution Approach 2:
The system performs self-learning and self-adjustment through the neural network, which automatically adapts to different machining conditions without requiring manual intervention or complex configuration. The network learns from accumulated data and autonomously optimizes threshold values, reducing the operational complexity despite the increased structural complexity of incorporating the neural network.
3Adaptability or versatility
If machine learning is used to learn threshold value, then the adaptability to various machining conditions improves, but the loss of time increases due to learning process and data processing
Solution Approach 1:
The patent performs preliminary learning during idle periods or between machining operations, accumulating training data and refining the neural network model when the machine is not actively cutting. This allows the system to prepare threshold values in advance for upcoming machining conditions, reducing the time penalty during actual production operations while maintaining high adaptability.
Solution Approach 2:
The neural network continuously learns and updates threshold values during machine operation, processing data in real-time without interrupting the machining process. The system maintains continuous adaptation to varying conditions while minimizing time loss by performing learning and prediction operations concurrently with normal machining activities.
Data Source
AI summary
A machine learning device for learning a threshold value of detecting an abnormal load in a machine tool, includes a state observation unit, and a learning unit. The state observation unit observes a state variable obtained based on at least one of information about a tool, main spindle revolution rate, and amount of coolant of the machine tool, material of a workpiece, and moving direction, cutting speed, and cut depth of the tool, and the learning unit learns the threshold value of detecting an abnormal load based on training data created from an output of the state observation unit and data related to detection of an abnormal load in the machine tool and on teacher data.