Plant Control Parameter Setting for Multi-Loop Stability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
It is challenging to predict and stabilize the behavior of complex plant processes with multiple control systems due to unpredictable disturbances and mutual interference between control loops, making it difficult to set optimal control device adjustment parameters for stable operation.
Innovation Solution
A plant operating condition setting support system that uses deep reinforcement learning to determine control device adjustment parameters for multiple control devices, enabling integrated support for feedback control tasks and stabilizing plant operations by learning policies from measured values and optimizing PID parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple control systems are used to control complex plant processes, then control coverage and functionality are improved, but mutual interference between control loops and system instability increase
Solution Approach 1:
The patent introduces a non-interference element as an intermediary component between multiple control loops. This element calculates and applies compensation signals that cancel out the mutual interference between control loops, allowing multiple control systems to operate simultaneously without destabilizing each other. The non-interference element acts as a mediator that enables coexistence of multiple control systems while maintaining system stability.
2Ease of operation
If mathematical approximation methods are used to model control system behavior, then analysis and control design are simplified, but prediction precision deteriorates
Solution Approach 1:
The patent transforms the complex, non-linear control system into a simplified first-order lag system model by changing the representation parameters. Instead of attempting to accurately model the complex original system, the patent approximates the overall behavior using standard first-order dynamics with identifiable parameters (gain and time constant). This parameter transformation enables straightforward control design while accepting a controlled loss of prediction precision in exchange for analytical tractability.
3Reliability
If control device adjustment parameters are set by experienced operators, then control performance is improved, but operator dependency and labor requirements increase
Solution Approach 1:
The patent enables the control system to automatically determine its own optimal parameters through self-identification of the first-order lag characteristics. By measuring the step response and automatically calculating the time constant and gain, the system eliminates the need for experienced operators to manually tune parameters. The control device performs self-service in determining its adjustment parameters, reducing operator dependency while maintaining reliable control performance.
Data Source
AI summary
A plant operating condition setting support system for supporting the setting of an operating condition of a plant that performs a process formed by devices includes: control devices that subject controlled devices to feedback control respectively; and an operating condition setting support device that provides integrated support for the setting of the control devices, which perform feedback control tasks respectively and independently. The operating condition setting support device includes: a measured value multiple acquisition unit that acquires measured values indicating states of the controlled devices, respectively; and a control device adjustment parameter determination unit that determines, based on the measured values acquired, control device adjustment parameters used by each of the control devices to determine manipulation variables for control that should be input to the controlled devices, according to a policy learned by deep reinforcement learning.