SVR-GA-Based adaptive power coal rate modeling and optimization for large coal-fired power units

Power coal rate is an important index to evaluate the overall economic performance of coal-fired power plant. It is however difficult to describe and optimize this feature in different operation conditions because of higher dimension, nonlinear and complex system configuration. An optimized support vector regression (SVR) model was built to predict the power coal rate of power unit, in which the prediction performance of SVR model was optimized by introducing genetic algorithm (GA) to optimize the parameters of SVR model. Considering different boundary parameters, load demand and operation conditions, we built the GA-SVR-based power coal rate model of large coal-fired power unit. The main factors contributing to such model such as the sampling scale, attribute number and specific operators in GA were discussed. The results indicate that the modeling performance is significantly improved in accuracy, searching efficiency and model simplicity; in addition, the model can be conveniently generalized for different types of power units.