A Variable Selection Method for Pulverizing Capability Prediction of Tumbling Mill Based on Improved Hybrid Genetic Algorithm

Tumbling mill of thermal power plant, grinding the raw coal for the boiler, has high energy consumption, and pulverizing capability is usually used for representing the efficiency of tumbling mill. In the paper, a variable selection method for pulverizing capability prediction of tumbling mill based on improved hybrid genetic algorithm is proposed. Based on the tradition GA, the proposed method adopts the multi-population mechanism, the elites sharing mechanism and the heterogeneity mechanism for avoiding the premature convergence. The support vector machine is used for building the prediction model of the pulverizing capability with the selected variables. The proposed method is performed on the real field data. The results of the experiments verify that the proposed method has faster convergence speed and the model of pulverizing capability built with the variables selected by the proposed method has higher prediction precision. In addition, the proposed method has been put into practice and the field operation curve verifies that the pulverizing capability could be predicted successfully.