Logo image
Improving combustion performance by online learning
Book chapter

Improving combustion performance by online learning

Andrew Kusiak and Zhe Song
Optimization in the energy industry, pp.131-148
Springer
2009
DOI: 10.1007/978-3-540-88965-6_6

View Online

Abstract

In this chapter, combustion process is improved by computing control settings with clustering algorithms. The framework involves learning from a high-dimensional data stream generated by the combustion process. Thus the system's dynamics is captured. The concepts of virtual age of the boiler and the control settings are introduced. The confidence of applying a control setting to improve boiler performance is quantified. The framework is easy to implement and it handles a large number of process variables. The ideas introduced in this paper have been implemented at a 20 MW boiler controlled with a standard control system. That system makes run-time recommendations to the standard control system.

Details

Metrics

23 Record Views
Logo image