Performance assessment of a boiler combustion process control system based on a data-driven approach

Abstract

For the requirements of performance assessment of the thermal power plant control process, the combustion control system of a 330 MW generator unit in a power plant is studied. Firstly, the five variables that affect the process control performance are determined by the mechanism analysis method. Then, a data-driven performance assessment method based on the operational data collection from the supervisory information system was proposed. Using principal component analysis technique, we found that five different variables have different degrees of effect on the performance of the combustion process. By means of qualitative and quantitative analysis, five contribution rates of different variables affecting the performance index of the system were obtained. After that, the data is normalized to the non-dimensional variable, the performance assessment index of the boiler combustion process is defined, and the classification and assessment criterion of it are given. Through using the proposed method on the operation data of the 1# boiler and 2# boiler within 1 day, the performance indexes are calculated and achieved during different time periods. Analysis of the results shows that this method will not generate additional disturbance to the normal operation of the system, and it can achieve a simple, reliable, accurate and rapid qualitative and quantitative analysis of the performance of the boiler combustion control system, and also it can be extended and applied to other multivariable control systems.