Predicting the gross calorific value of coal based on support vector machine and partial least squares algorithm

Abstract

In view of the problem of inaccurately measuring the gross calorific value of coal, this paper analyzed the relationship between industry analysis data and calorific value of fire coal into the furnace, and five kinds of industrial analysis components were selected as the original independent variables, which were moisture, ash, volatile matter, sulphur content and fixed carbon in fire coal. In the process of modelling, the latent variable factors were extracted by the partial least square regression method, and the latent variables were selected as the input vectors and coal-fired calorific value as the output vector, then the power plant coal calorific value forecasting model was built based on support vector machine regression algorithm. The predicted results showed that, the accuracy of the coupled model was higher than that of the single model and the forecasting deviations met engineering requirements. Therefore, the model proposed here has practical engineering application value. At the same time, in the process of modelling, the five-point method was used to determine the optimal combination of penalty coefficient and kernel coefficient in the process of modelling. The method could quickly and accurately determine the optimal combination and avoid the blindness of the traditional method in the process of determining the optimal combination of penalty coefficient and kernel coefficient.