The prediction for output of blast furnace gas based on genetic algorithm and LSSVM

Generated fluctuation of blast furnace gas has a great influence on gas scheduling and optimization. Generating blast furnace gas changes non-linearly, so the traditional prediction method is difficult to deal with it. Least squares support vector machine (LSSVM), which is a machine learning method based on statistical learning theory, can be good to solve non-linear problems. It will be established a LSSVM model to predict the number of blast furnace gas in this article. However, LSSVM forecasting precision for the model parameters are very sensitive. In order to achieve more ideal effect, this paper is introduced a kind of improved genetic algorithm to optimize the parameters of LSSVM model. Instruct a short-term forecasting model of blast furnace gas based on genetic algorithm and LSSVM so as to improve the prediction accuracy of forecasting model.