Operation Model Evaluation for Adaptive Facility Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current systems for controlling and optimizing the operation of complex facilities, such as industrial plants, face challenges in efficiently adjusting control parameters to meet target performance metrics like energy efficiency and product quality, often requiring extensive trial-and-error and lacking adaptive learning capabilities.

Innovation Solution

An apparatus and method that utilize an operation model and a target setting model, integrated with learning processing units, to adjust control parameters based on real-time data and predefined reward functions, enabling adaptive optimization of facility operations to achieve target performance metrics through reinforcement learning and supervised learning techniques.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional trial-and-error methods are used to adjust control parameters, then the system can eventually find optimal settings, but the process requires extensive time and manual intervention

Engineering Contradiction:
Improveoptimization effectivenessVSAvoidadjustment time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system employs reinforcement learning models that automatically adjust control parameters through self-learning from environmental feedback. The operation model and target setting model continuously optimize facility operations without human intervention, allowing the system to serve itself in the optimization process while significantly reducing adjustment time

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system implements closed-loop feedback mechanisms where the reinforcement learning model receives reward signals based on performance metrics (energy efficiency, product quality). This feedback drives continuous optimization of control parameters, enabling the system to learn from outcomes and improve over time without manual trial-and-error

Inventive Principle:
Principle #23Feedback

2Adaptability or versatility

If traditional control systems are used, then the system structure remains simple, but the system lacks adaptive learning capabilities

Engineering Contradiction:
Improveadaptive learning capabilityVSAvoidsystem structure
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system introduces an operation model as an intermediary layer between the control system and the facility. This model acts as a mediator that processes state data, generates control parameters through reinforcement learning, and translates high-level objectives into actionable control commands, enabling adaptive learning without overwhelming system complexity

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces traditional mechanical trial-and-error adjustment mechanisms with intelligent software-based reinforcement learning models. The operation model and target setting model use algorithms to automatically optimize control parameters, substituting physical experimentation with computational learning and optimization

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Manufacturing precision

If manual intervention is used for parameter adjustment, then the system maintains simplicity, but energy efficiency and product quality optimization are insufficient

Engineering Contradiction:
Improveproduct qualityVSAvoidmanual intervention level
Core Design Contradiction:
Manufacturing precisionVSExtent of automation

Solution Approach 1:

The reinforcement learning model autonomously optimizes control parameters to maximize energy efficiency and product quality without manual intervention. The operation model continuously learns from operational data and automatically adjusts facility operations, enabling the system to self-optimize manufacturing precision while minimizing human involvement

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system dynamically changes control parameters based on learned patterns and real-time state data. The operation model adjusts multiple parameters simultaneously (temperature, pressure, flow rates) to optimize both energy efficiency and product quality, achieving superior manufacturing precision through automated parameter optimization

Inventive Principle:
Principle #35Parameter changes

4Reliability

If extensive trial-and-error is performed, then comprehensive optimization can be achieved, but operational variability increases

Engineering Contradiction:
Improveoptimization completenessVSAvoidoperational variability
Core Design Contradiction:
ReliabilityVSStability of the object's composition

Solution Approach 1:

The reinforcement learning model uses continuous feedback from performance metrics to guide optimization, reducing operational variability. By learning from reward signals and systematically adjusting parameters based on observed outcomes, the model achieves comprehensive optimization while maintaining stable and predictable operations compared to random trial-and-error approaches

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4174589B1Apparatus, method and computer readable medium for evaluating an operation model
Publication Date: 2024.07.03 YOKOGAWA ELECTRIC CORP
  • EP4174589B1 patent drawingFigure 1
  • EP4174589B1 patent drawingFigure 2
  • EP4174589B1 patent drawingFigure 3

AI summary

Provided is an apparatus including a supply unit configured to supply a value of a state parameter to an operation model configured to output a recommendation value of a control parameter of a piece of equipment in response to a value of a state parameter relating to the piece of equipment being input; a control parameter acquisition unit configured to acquire a recommendation value of a control parameter that is output from the operation model in response to the supply unit supplying a value of a state parameter to the operation model; an acquisition unit configured to acquire a model evaluation value corresponding to a result of having operated the piece of equipment according to the recommendation value acquired by the control parameter acquisition unit; and an evaluation unit configured to evaluate the operation model, based on the model evaluation value and a reference evaluation value corresponding to a result of having operated the piece of equipment through a manipulation by a human.