Study on Prediction of Gas Emission by Data Fusion

In order to improve the accuracy of prediction of gas emission, a novel nonlinear combined prediction method using support vector machines(SVM) was introduced. SVM, which was based on the rule of structural error minimization, was adopted to build a multi-input and single-output nonlinear prediction model. The model parameters were tuned by training samples sets and evaluated by the principle of the minimum standard deviation. Three original predictive values(hyperbola regression prediction , exponential regression prediction and grey prediction) were combined to get the prediction results in this model. The experimental results showed that this model was far more superior to other prediction models.