| The main steam temperature is a very important controlled parameter in the control of power plant. Ensuring its quality is an essential part for the unit plant’s safe and economical operation in the power plant. The controlled object of the main steam temperature in the thermal power units has the features of great inertia and delayed, in which it is beyond possibility to acquire high performance by the traditional PID controller when there are great changes in working condition. Therefore, it’s necessary to do research of the modeling of the main steam temperature system in depth, in order to improve the control quality of the main steam temperature system.The widespread use of DCS and SIS provides a great convenience to modeling based on the field data. This paper systematically introduces the specific modeling steps and methods,and does a feasibility study of modeling based on field data. Based on the data of main steam temperature object in a certain power plant under the disturbances of the three stage superheater spray, this paper puts forward the method of main steam temperature system particle swarm identification method on account of the historical operation data of power plant. The dynamic mathematical model was established on the leading area and the inert area of the main steam temperature, and the actual data during the run was used to validate the model. It turned out that the errors between the simulation result and the real situation are smaller. As a result, the model can reflect the dynamic characteristics of main steam temperature, and it can be taken as a reference for the main steam temperature control system optimization.After got the transfer function model, we designed the control system of feedforward-feedback based on SVM. Firstly, the main parameters were selected as feedforward signals by analyzing influence factors of the main steam temperature. Then SVM was used to do classification and prediction. Finally, the feedforward controller was built and added to the traditonal cascade PID control system. The results showed that the new control system regulated fast and had better stability and control quality, which from the simulation of the feedforward-feedback control system on Matlab platform. |