Data Processing Device Correlating Sensor and Job Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing methods struggle to accurately estimate the cause of changes in data relations, particularly in systems with numerous data items, such as real-time monitoring systems for trains, where changes in correlations between sensors are difficult to interpret.
Innovation Solution
A data processing device and method that combine past and real-time sensor data to identify data relations, calculate similarities between these relations, and output associations between past and real-time data relations, enabling the estimation of cause changes by grasping correspondence between sensor and job data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If only sensor data is used for correlation analysis, then the data processing is simple, but the ability to estimate the cause of changes is insufficient
Solution Approach 1:
The patent combines sensor data with job data (operational data) to perform correlation analysis. The correlation analysis unit calculates correlation coefficients between sensor data items and job data items, enabling more accurate cause estimation by integrating multiple data sources rather than relying solely on sensor data.
Solution Approach 2:
The patent introduces a correlation analysis unit as an intermediary that bridges sensor data and job data. This unit calculates correlation coefficients and identifies correlated data items, serving as a mediator that connects different data types to improve cause estimation without directly complicating the core monitoring system.
2Loss of information
If all data items are monitored individually, then complete information is obtained, but the burden on data analyzing persons increases greatly
Solution Approach 1:
The patent extracts and highlights only the correlated data items from the complete set of monitored data. The display unit presents job data items correlated with sensor data items, allowing analysts to focus on relevant relationships rather than examining all individual data items, thus reducing analysis burden while maintaining information completeness.
Solution Approach 2:
The patent segments the large volume of monitored data into meaningful correlations between sensor data items and job data items. By organizing data according to their correlation relationships rather than presenting all data flatly, the system helps analysts navigate and understand data relationships more efficiently.
3Measurement precision
If correlation changes are presented without context, then the change detection is effective, but the cause estimation becomes difficult
Solution Approach 1:
The patent merges sensor data with job data (operational context) in the correlation analysis. By calculating correlations between sensor data items and job data items together, the system preserves cause information while detecting changes, allowing analysts to understand both what changed and why it changed.
Data Source
AI summary
Provided are: a past relation identification unit that combines past sensor data with job data based on information on an item relating to data combination, and identifies a data relation between items based on an item of the past sensor data and an item of the job data; a real-time relation identification unit that identifies a data relation between items of real-time sensor data; a similarity calculation unit that calculates a similarity between the data relation identified by the past relation identification unit and the data relation identified by the real-time relation identification unit; and an output controller that outputs the data relation identified by the past relation identification unit and the data relation identified by the real-time relation identification unit while associating with each other by the item of the past sensor data and the item of the real-time sensor data, the similarity between those data relations exceeding a threshold, the similarity being calculated by the similarity calculation unit.