Boiler Feedforward Learning for Faster Power Command Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The accuracy of the boiler input rate (BIR) for compensating response delays in thermal power generation plants is user-experience dependent, leading to suboptimal tracking characteristics of power generation quantity with respect to command values.
Innovation Solution
A learning device constructs a model to generate a prior acceleration command value by mechanically learning from previous operational data, including power generation command values and process parameters, to improve the tracking characteristic of power generation quantity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the boiler input rate (BIR) is set by user experience, then the system is easy to operate, but the tracking characteristic of power generation quantity with respect to command values deteriorates
Solution Approach 1:
The system performs self-learning by automatically acquiring operational data from the boiler and using machine learning to optimize the BIR settings. The learning device autonomously improves tracking characteristics without requiring user intervention or manual tuning, allowing the system to serve itself in optimizing its control parameters.
Solution Approach 2:
The patent replaces the manual, experience-based mechanical tuning of BIR parameters with an automated machine learning system. The learning device uses algorithms to analyze operational data and automatically determine optimal BIR values, substituting human expertise with computational intelligence to improve tracking accuracy.
2Manufacturing precision
If the prior acceleration command value is optimized for tracking accuracy, then the tracking characteristic improves, but the device complexity increases
Solution Approach 1:
The learning device is designed to be universally applicable to various boiler types and operational conditions. By using generic machine learning algorithms that can adapt to different scenarios, the system achieves improved tracking characteristics without requiring complex, boiler-specific customization. The same learning framework handles diverse operational data and generates appropriate BIR values for different situations.
3Manufacturing precision
If mechanical learning from operational data is implemented, then the tracking characteristic improves, but the loss of time for data processing increases
Solution Approach 1:
The system performs preliminary learning by continuously acquiring and processing operational data in the background during normal boiler operation. The machine learning model is trained incrementally using historical data, so when optimization is needed, the system already has pre-processed knowledge to quickly generate improved BIR values without requiring extensive real-time computation that would delay control responses.
Data Source
AI summary
A learning device, for constructing a learning model for generating a prior acceleration command value for controlling a control target in advance of the time a load on a boiler used for thermal power generation changes, includes: a learner configured to generate the learning model by mechanically learning, as learning data, a data set including a power generation command value for the thermal power generation and the prior acceleration command value which have been used in a previous operation of the boiler.