Time-Series Performance Evaluation Across Changing Operating Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing plant performance evaluation systems struggle to accurately assess the deterioration tendency of devices when operation conditions change, as they primarily focus on the relationship between sensors rather than individual sensor values over time.
Innovation Solution
A data processing device and method that acquire time-series data related to a target device, divide the data into operation periods based on conditions, specify standard conditions for each period, and calculate representative performance values within constant performance periods, while correcting these values using overlap periods to account for changes in operation conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If performance evaluation is performed using sensor relationship models, then device deterioration tendency can be detected, but accurate evaluation becomes difficult when operation conditions change
Solution Approach 1:
The time-series data is divided into multiple operation periods based on operation conditions, and each period is further segmented into constant performance periods. This segmentation allows the system to evaluate device performance under specific stable conditions while accounting for variations across different operation periods, thereby maintaining evaluation accuracy despite changing conditions.
Solution Approach 2:
The system identifies standard conditions for each operation period and calculates representative values only during constant performance periods where parameters remain stable. By adapting the evaluation to account for parameter variations across periods and using correction values derived from overlap periods, the system maintains measurement precision under varying operation conditions.
2Adaptability or versatility
If operation periods are divided based on operation conditions, then performance evaluation under varying conditions improves, but data processing complexity increases
Solution Approach 1:
The division into operation periods and constant performance periods creates a structured framework that simplifies processing by isolating stable data segments. While segmentation increases initial processing steps, it enables more efficient calculation of representative values during constant performance periods and systematic correction using overlap periods, ultimately managing complexity through organized data handling.
Solution Approach 2:
The system performs preliminary division of time-series data into operation periods and identifies standard conditions before calculating representative values. This preliminary organization of data and identification of stable periods prepares the dataset for more efficient subsequent processing, reducing the complexity of the actual performance evaluation by pre-structuring the data according to operation conditions.
3Reliability
If representative values are calculated for each operation period, then performance trends can be tracked, but inconsistencies arise at period boundaries
Solution Approach 1:
The system sets overlap periods that include boundaries between adjacent operation periods before calculating representative values. This preliminary establishment of overlap regions ensures that boundary areas are specifically handled with corrected values derived from both adjacent periods, preventing inconsistencies and maintaining stable performance trend tracking across the entire time-series data.
Solution Approach 2:
The overlap periods act as intermediary regions between adjacent operation periods, where representative values are calculated and used to correct data from both sides. This intermediary approach smooths transitions at period boundaries and ensures consistency in performance trend tracking by mediating the connection between different operation periods through corrected representative values.
Data Source
AI summary
A data processing device comprises a division unit that divides a target period into a plurality of operating periods according to an operating condition; an identification unit that identifies a standard condition for each operating period; a calculation unit that divides each operating period into a plurality of constant performance periods in each of which the performance of a target device is considered to be constant, narrows down time series data for each constant performance period on the basis of the standard condition, and calculates a certain representative value on the basis of the time series data extracted through the narrowing down; and a correction unit that sets an overlapping period containing a boundary between two adjacent operating periods, narrows down, on the basis of each standard condition identified for each of the two operating periods, time series data in the overlapping period for each standard condition.