Motor Drive Control Using Environment-Specific Training Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motor drive control systems struggle to perform appropriate motor drive control in actual usage environments due to differences in environmental conditions such as air pressure and temperature, which can lead to inaccurate abnormality detection and suboptimal control.
Innovation Solution
A motor drive control device equipped with a machine learning function that generates training data based on measurement data from the actual usage environment and performs machine learning to create a learned model for determining the motor's operation state, enabling accurate abnormality detection and adaptive control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If training data is generated using measurement data from mass production environment, then machine learning can be performed, but the learned model cannot accurately detect abnormalities in actual usage environment due to environmental differences
Solution Approach 1:
The system performs preliminary actions by generating training data in the mass production environment before actual usage, and stores this training data for later use. The learned model is created in advance based on mass production data, and then this pre-created model is utilized in the actual usage environment to detect abnormalities.
Solution Approach 2:
The system creates a copy of the mass production environment characteristics by generating training data that replicates the operational patterns and conditions of the mass production phase. This copied data structure and patterns are then used to train the model, allowing the model to learn from the mass production environment without requiring the actual physical environment to be present during operation.
2Productivity
If a learned model is generated using mass production environment data, then machine learning can be performed, but appropriate motor drive control cannot be achieved in actual usage environment
Solution Approach 1:
The system performs preliminary machine learning during the mass production phase when efficient data collection is possible. The learned model is created in advance using this efficiently collected data, and then the model is stored and reused during actual usage, maintaining both learning efficiency and operational reliability.
Solution Approach 2:
The motor drive control device performs self-learning by generating its own training data from its internal measurement data during mass production. The system uses its own operational data to train the model, eliminating the need for external data sources and enabling autonomous model creation that maintains reliability across different usage environments.
Data Source
AI summary
In a motor drive control device including a machine learning function, appropriate motor drive control in accordance with the usage environment of a motor is realized. A motor drive control device 10 includes: a measurement data generation unit 23 that generates measurement data 300 relating to operation of a motor 50; a training data generation unit 24 that attaches predetermined identification information indicating the operation state of the motor 50 to the measurement data 300 and generates training data 310; a machine learning unit 25 that generates a learned model 320 for determining the operation state of the motor 50 by performing machine learning using the training data 310; and a monitor control unit 26 that monitors the operation state of the motor 50 using the learned model 320. In the motor drive control device 10, the training data generation unit 24 starts generation of the training data 310 when the training data generation unit 24 receives a command ordering acquisition of the training data 310 from a host device 4.