Process Data Stage Modeling for Multi-Facility Pipe Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The causal relationship between multiple production facilities connected by pipes is complex, making it difficult to associate process data effectively.
Innovation Solution
An information processing device that acquires and analyzes process data from multiple production facilities to determine their operation stages, using a regression model to predict the state or cause of substances flowing between them, including predicting the concentration of components in waste liquids or emissions, and adjusting operation schedules based on these predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If process data is acquired from multiple production facilities to determine operation stages, then the ability to reflect influences between facilities is improved, but the complexity of data association and causal relationship analysis increases
Solution Approach 1:
The patent segments the complex causal relationship analysis into distinct operation stages (starting, operating, ending, stoppage) for each production facility. By dividing the continuous process data into discrete stages, the system simplifies the association of causal relationships between facilities while maintaining accurate influence reflection through stage-based regression models.
2Measurement precision
If operation stages are determined using correspondence relationships between operation stages and sensor data features, then the precision of operation stage determination is improved, but the complexity of predetermined correspondence relationships increases
Solution Approach 1:
The patent changes the parameter representation by transforming continuous sensor data into discrete operation stage parameters. Instead of directly using complex continuous data relationships, the system defines predetermined correspondence relationships between operation stages and sensor data features, simplifying the determination process while maintaining precision through the structured stage classification.
3Measurement precision
If regression models use explanatory variables that differ depending on operation stages, then the accuracy of state prediction is improved, but the complexity of model management increases
Solution Approach 1:
The patent implements dynamic model management where the regression model automatically selects and uses different explanatory variables based on the current operation stage. This dynamic approach allows the system to maintain high prediction accuracy by adapting to stage-specific conditions while managing model complexity through automated variable selection rather than manual configuration of multiple static models.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
To perform processing in which an influence between a plurality of production facilities is appropriately reflected. An information processing device includes a processing unit configured to: acquire process data for determining an operation stage of a production facility from each of a plurality of production facilities on an upstream side connected by a pipe; determine the operation stage for each of the plurality of production facilities on the upstream side based on the acquired process data; and perform, by using a combination of the operation stages of the plurality of production facilities on the upstream side, at least one of predicting a state of a predetermined substance flowing into a production facility on a downstream side, or a predetermined reaction intermediate, product, or emission generated in the production facility on the downstream side, or predicting a cause of an irregularity generated in any of the plurality of upstream production facilities and the downstream production facility.