A dynamic model for the bed temperature prediction of circulating fluidized bed boilers based on least squares support vector machine with real operational data. (1st April 2017)

• Record Type: Journal Article
• Authors: Lv, You; Hong, Feng; Yang, Tingting; Fang, Fang; Liu, Jizhen
• Is Part Of: Energy. Volume 124(2017)
• Page Start: 284
• Page End: 294
• Publication Date: 2017-04-01
• Subjects: Bed temperature -- Dynamic model -- Least squares support vector machine -- Circulating fluidized bed boiler -- Operational data; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2017.02.031
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: Circulating fluidized bed (CFB) combustion is a new clean coal technology with advantages of wide fuel flexibility and low pollutant emissions. The bed temperature of CFB boilers is an important factor that influences operating security and pollutant emission generation. An accurate model to describe the dynamic characteristics of bed temperature is beneficial in reducing temperature fluctuations. This study presents a dynamic model for predicting the bed temperature of a 300 MW CFB boiler based on the least squares support vector machine method with real operational data. Coal feed rate and primary air rate are selected as the independent variables. The current values and previous sequences of the variables are considered as the model inputs to describe the dynamic characteristics of bed temperature. In addition, the past values of bed temperature are taken as feedback and then added to the inputs. The particle swarm optimization technique is used to determine optimal delay orders. Several model patterns are also discussed and compared. Comparison results show that the proposed model structure is reasonable and that the model can achieve the accurate prediction of the bed temperature. Highlights: A dynamic model is developed to predict the bed temperature of a CFB boiler. LSSVM is used to construct the model with real operational data of the boiler. Current and past values of independent variables are taken as model inputs. The optimal delay orders are determined by particle swarm optimization algorithm. The model performance is discussed and validated.