Time-Series Waveform Display for Machine Failure Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting machine failures in production machines are inefficient due to the high volume of time-series data collected at high sampling rates, making it difficult for workers to extract and compare relevant data for learning models, leading to low efficiency and accuracy in maintenance scheduling.
Innovation Solution
An information processing method and apparatus that acquires time-series data from machine sensors, extracts partial data related to specific events, and displays it in a format where time information is aligned, allowing for easier comparison and analysis of waveforms by arranging partial time-series data adjacent to each other on a horizontal axis, reducing redundancy and deformation of waveforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If time-series data is collected at high sampling rates to ensure measurement precision, then measurement precision is improved, but the volume of data increases making it difficult to extract and compare relevant data
Solution Approach 1:
The patent segments the large volume of time-series data by extracting multiple pieces of partial time-series data at different time points and organizing them into groups based on elapsed time. This segmentation makes the data more manageable and comparable while preserving the high sampling rate measurement precision.
Solution Approach 2:
The patent extracts only the necessary partial time-series data from the complete high-volume dataset. By taking out specific segments at predetermined time points and organizing them with elapsed time information, it reduces the effective data volume for analysis while maintaining measurement precision.
2Ease of operation
If multiple pieces of partial time-series data are plotted on a single graph with elapsed time axis for comparison, then ease of operation is improved, but waveform deformation and redundancy occur reducing measurement precision
Solution Approach 1:
The patent adds a new dimension to the display by incorporating an elapsed time axis that indicates when each partial time-series data was acquired. This allows multiple waveforms to be displayed without overlapping or deforming, as each can be positioned according to its acquisition time, thereby maintaining measurement precision while improving ease of operation.
3Reliability
If detailed data analysis is performed to determine suitability for learning data, then reliability is improved, but loss of time increases due to the complexity of checking and comparing waveforms
Solution Approach 1:
The patent performs preliminary organization of time-series data by extracting partial data at predetermined time points and grouping them with elapsed time information before the actual analysis. This preliminary action prepares the data in a ready-to-analyze format, reducing the time required for subsequent detailed checking and comparison while maintaining reliability.
4Reliability
If maintenance work frequency is increased to improve reliability, then reliability is improved, but productivity decreases due to machine downtime
Solution Approach 1:
The patent creates a feedback mechanism by systematically analyzing time-series data to detect changes in machine operation states and predict failures. This feedback allows maintenance to be performed based on actual machine condition rather than fixed schedules, improving reliability while avoiding unnecessary maintenance that would reduce productivity.
Data Source
AI summary
A display apparatus includes a processing portion. The display apparatus is configured to display physical quantity related to a state of a machine apparatus. The processing portion is configured to display an image in which a plurality of pieces of partial time-series data extracted from time-series data related to the physical quantity are arranged in a state where time information is provided in the image, the time information being related to time in which the plurality of pieces of partial time-series data has been acquired.