ANN-based Soft Sensing of Oxygen Content in Boiler Air-flue Gas System

Oxygen content in the air-flue gas system of a large-scale coal-fired power unit is an important factor affecting the boiler efficiency, and its accurate measurement plays an important role in evaluating the boiler operation economy. However, due to the complex combustion process and many factors affecting the oxygen content, on-site direct measurement accuracy of the oxygen content is often poor and lagged. In recent years, Soft-sensing method based on modelling with correlation parameters has been gradually applied to coal-fired power plants. By investigating on the factors influencing the oxygen content and the structure of the boiler air-flue gas system, an oxygen content soft-sensing model based on artificial neural network is established, trained and validated with historical operating data collected from an actual 1000-MW power plant. The test results show that the method can predict the oxygen content in the air-flue gas accurately, favourable for optimal boiler combustion adjustment for saving energy and reducing coal consumption.