Least Squares Support Vector Machine In Steam Forecast

Resource conservation, technology upgrading, and production optimization scheduling must be done by every enterprise for China is a per capita country of resource-pool. These factors are also very important for the cigarette factory. The stream needed by Tobacco redrying of the cigarette factory is produced through energy conversion of coal, water, electricity. Coal, water, and electricity etc is scarce resource, so short term predicting of stream in cigarette factory has very important research significance for resource conservation, technology upgrading and production optimization scheduling.Special least squares support vector machine (LS-SVM) constraint is a linear equation. It can get support vector that only requires very little computation compared with the standard support vector machine. Most traditional forecasting methods are based on the empirical risk minimization principle developed so traditional forecasting methods can not meet the requirements. Support vector machine is a developed emerging learning method on statistical learning theory, it has good generalization ability. So three improved support vector machines is introduced in this paper to solve the existing problems in the steam data, steam data is predicted by application of three support vector machines and error analysis is done. The research work of this paper as follows,1, Model online LS-SVM based on differential evolution algorithm is proposed, problem of parameter optimization and small sample is solved by the model. Uncertainty problem in steam forecast is solved by real-time grey LS-SVM that formed by LS-SVM combined with grey system. The forecast risk in steam predicting is solved by the algorithm that formed by LS-SVM combined with combination predicting. The steam is predicted by three improved methods, these is compared and analyzed by error analysis.2, Conclusions three improved methods can significantly improve the accuracy of steam predicting is drawn by error analysis. It provide technical basis for optimization scheduling, resource conservation, security of enterprise market. Then establish future research plans and implementation steps.