Normal Data Combining for Intermittent Condition Monitoring AI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for creating machine learning models struggle to efficiently generate training data for normal conditions, making it difficult to effectively detect abnormal system conditions, especially in devices with intermittent operations or those requiring combination of data from multiple time periods.
Innovation Solution
An information processing apparatus and method that acquires and combines normal data from multiple intervals of measurement data to generate combined data, which is then used to train a machine learning model to output a score indicating the normal condition of a system, allowing for the detection of abnormal conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If only one time period of normal data is selected for training, then the training process is simple, but the detection accuracy of abnormal conditions deteriorates
Solution Approach 1:
The patent combines multiple time periods of normal data into a single training dataset. The data combining unit merges normal data from different time periods (e.g., morning operation and afternoon operation) to create comprehensive training data, enabling the machine learning model to learn various normal operational patterns and improve abnormal condition detection accuracy.
2Reliability
If multiple intervals of normal data are combined, then the detection accuracy improves, but the data processing complexity increases
Solution Approach 1:
The patent segments the data processing function into distinct units: a data acquiring unit that collects normal data from multiple time periods, and a data combining unit that merges these segments. This segmentation allows systematic handling of multiple data intervals while maintaining manageable processing complexity through modular architecture.
Solution Approach 2:
The data combining unit acts as an intermediary between data acquisition and machine learning training. It receives normal data from multiple time periods, processes and merges them into combined training data, and outputs the integrated dataset to the training unit, simplifying the overall data flow and processing complexity.
3Measurement precision
If comprehensive normal data from multiple time periods is used, then the machine learning model becomes more accurate, but the training time increases
Solution Approach 1:
The patent performs preliminary action by pre-processing and combining normal data from multiple time periods before training the machine learning model. The data combining unit prepares comprehensive training data in advance, including all relevant normal operational patterns, so that the model can be trained efficiently with ready-to-use integrated data, reducing overall training time.
Data Source
AI summary
An information processing apparatus acquires pieces of normal data belonging to a plurality of intervals of measurement data measured by a device, combines the acquired pieces of normal data to generate combined data, and performs machine learning using the combined data to train a machine learning model that outputs a score indicating a normal condition of a system in which the device is installed, in response to an input of measurement data.