Modeling And Trend Prediction Of Reheat Steam Temperature System Based On LS-SVM

The reheat steam temperature control is an important link to improve the economy and ensure the security of the unit operation, maintaining the reheat steam temperature in rated range is of great significance for the optimal operation of the unit. As the dc type boiler in generating set is blossoming, the difficulty in control of reheat steam temperature is increased. Dc furnace reheat steam temperature object has obvious hysteresis characteristics, the traditional PID control is prone to "overheating" or "under temperature" and unable to achieve satisfactory control quality.Based on the LS-SVM algorithm, a short-term prediction model of reheat steam temperature system is established, using the movement trend of the main influence factors to predict the change of the reheat steam temperature in the next moment. Using the grid method and cross validation method optimizes parameters of LS-SVM. Refactoring sample structure reflects the change trend of input variables, the selection of the sample dimension can not only ensure accuracy of fitting model, but also consider the model complexity. With many internal relations between the thermal parameters of model is finally obtained,.Use the same modeling method to establish for medium-term and long-term prediction of reheat steam temperature prediction model. Select 12 seconds, 36 seconds, 60 seconds as a model of short-term, medium-term and long-term prediction, analyse the accuracy of the prediction method in different prediction time. Simulation study shows that the LS-SVM prediction model of short-term prediction’s accuracy is high, when the predicting time gradually extends, mean square error(mse) has also been gradually increasing, but in the medium term prediction, it is still in the range of allowable error. The model can be well used to calibrate reheat steam short-term, medium-term trend prediction. It can be used to analyze the change tendency of the reheat steam temperature combined with the field data, guide the personnel operation, optimize the running state of the reheat steam temperature.