Process Control Model Adaptation Across Changing Target Set Values
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Trained models used in process control lack versatility and practicality as they are not capable of performing effective control when conditions change, as they are specific to the initial learning conditions and target set values.
Innovation Solution
A controller that acquires process data for the actual controlled object, converts it using a trained model, and calculates a manipulated variable, allowing the model to approximate process data to a target set value, even when the controlled object or target set value changes, by using parameters that specify the relation between manipulated variables and process data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a trained model is obtained through learning under conditions of a specific controlled object and a specific target set value, then the model can perform appropriate control for that specific condition, but the model lacks versatility and practicality because it is not capable of performing appropriate control when conditions are changed
Solution Approach 1:
The patent applies parameter changes by converting process data using conversion expressions that incorporate parameters representing characteristics of the actually used controlled object. This allows the trained model to adapt to different controlled objects by changing the parameters in the conversion expressions, rather than retraining the model for each new object. The conversion expressions transform process data based on object-specific parameters, enabling the same trained model to perform appropriately across different conditions.
Solution Approach 2:
The patent achieves universality by creating a single trained model that can be applied to multiple different controlled objects and target set values. Instead of having separate models for each specific condition, the system uses one trained model combined with conversion expressions that adapt to different objects. This makes the control system versatile and practical for various applications without requiring multiple specialized models.
2Productivity
If process data is directly input to a trained model without conversion, then the model processing is simple and fast, but the model cannot accurately handle process data from different controlled objects or different target set values
Solution Approach 1:
The patent applies preliminary action by converting process data before inputting it to the trained model. The conversion expressions prepare the process data in advance by transforming it according to the characteristics of the actually used controlled object. This preliminary conversion ensures that the data is in the appropriate format and scale for the trained model to process accurately, without requiring complex modifications during the control process itself.
Solution Approach 2:
The conversion expressions act as intermediaries between the raw process data and the trained model. They transform the process data into a format that the trained model can effectively process, bridging the gap between different controlled objects and the fixed-trained model. This intermediary conversion layer enables accurate control without requiring the model to directly handle raw data from various sources.
Data Source
AI summary
A controller includes: an acquisition unit configured to acquire process data for an actually used controlled object; and a calculation unit configured to, based on at least one of a target set value and parameters for the actually used controlled object, convert the process data acquired by the acquisition unit, and calculate a manipulated variable for the actually used controlled object by use of the converted process data and a trained model. Upon receiving input of process data for a specific controlled object, the trained model outputs a manipulated variable for approximating process data for the specific controlled object to a specific target set value. The parameters include parameters for specifying the relation between manipulated variables for the actually used controlled object and process data obtained by use of the manipulated variables.