Research And Application Of Advanced Load Control System In Gas-Steam Combined Cycle Unit

Relying on its advantage of high efficiency and low pollution, gas-steam combined cycle unit has been widely used in many countries. According to the new load control index, the traditional load control system cannot satisfy it. Due to the combined cycle unit load controlled object is a large inertia, multivariable coupling process, conventional control scheme is difficult to meet the control requirements. Therefore, it is urgent to design a set of suitable for the control system. Based on the problems above, the following study has been done in this paper:1. Re-established gas-steam combined cycle unit load model, think the gas turbine rotor speed as a security parameter and establish the model of fuel valve and unit power directly. An open loop dynamic characteristic test is performed at high and low load points. Test data is collected and processed. Identify the model by the least square method and validate it at two load points.2. In view of the problem that the original fuel computing circuit is too complex, design the new fuel calculating loop without PID controllers, so it can simplify the load control system. Meanwhile, in order to speed up the response rate of primary frequency, fuzzy feedforward compensation of primary frequency is introduced, using Mamdani inference method. And the simulation researches are carried out at two load points. The results show that the improved load control system is better than the original one in the aspects of rapidity and stability.3. In practical application, the original accumulative controller is replaced by the rapid integral controller without load deviation dead zone. And the improvements above are applied in the combined cycle unit. The results show that the improved load control system has an obviously improved effect.4. On the basis of analyzing the limitations of conventional control strategy, design the multivariable generalized predictive controller. Optimize the related parameters of the generalized predictive control algorithm by the simulated annealing genetic algorithm to avoid the blindness of parameters setting. At the same time, the simulation is tested at two load points, and make a comparison between the traditional PID control system and the generalized predictive control system. Of course, the latter further improves the control quality and has a wide application prospect.