Soft-sensing Technology of SO₂ Emission for Coal-fired Power Plants

To solve the problem of high installation and maintenance cost of continuous emission monitoring system (CEMS) for flue gas pollutants from coal-fired power plants, a soft-sensing technology is proposed for prediction of the SO₂ emission. The specific way is to firstly build up a model using BP neural network, then configure and optimize the link weights and threshold value of BP network using genetic algorithm, and finally establish a new soft-sensing model. Predicted results based on the new soft-sensing model were compared with actual measurements. Results show that the new model has a high prediction accuracy and stability, which may be used to predict SO₂ emission from coal-fired power plants.