Adaptive Robust Control for the Steam Temperature of Thermal Power Plant in Sliding Pressure Operation

为研究火电厂滑压运行方法，采用模块化建模系统（MMS）开发了一个仿真器。MMS提供了许多模块，这些模块都是根据热力学定律开发的。这些模块包括锅炉、过热器、温度器、汽轮机、省煤器、管道、冷凝器、除氧器、给水加热器等。利用200多个MMS模块，结合实际工厂和制造厂采集的数据，开发了仿真系统。为了验证所开发的模拟器的准确性，进行了一些研究。仿真结果表明，在额定负荷下，仿真结果与实际装置吻合较好。虽然仿真实例表明该仿真器是稳定的，但在计算方面仍存在一些问题。模拟器在计算上是“僵硬的”，因为有非常快的模式和非常慢的模式。快模式来源于汽轮机，慢模式来源于锅炉和再热器。在负载需求变化较大的情况下，仿真需要耗费大量的CPU时间。计算刚度给仿真器的动态精度验证带来了困难。为了提高计算效率，开发了一个简化的仿真器。在研究了实际电厂现有的基于PI的控制系统的基础上，提出了一种新的控制系统结构，该控制系统由主蒸汽负荷和压力对燃料流量控制系统和主蒸汽温度控制的FF（前馈）控制作用组成。所提出的FF控制可以插入到当前基于PI的控制系统中。为了设计FF控制器，建立了锅炉厂的时变模型。该模型是由多个锅炉厂在不同负荷和压力需求水平下的线性化模型插值而成的。这些线性化模型是通过用伪随机二值序列（PRBS）刺激模拟器来识别的。根据这一时变模型，设计了一种基于负载和压力需求的增益调度的动态滤波器控制器。对于负荷压力需求的小阶跃变化和负荷压力需求的大坡道变化情况进行了大量的仿真研究。仿真结果令人满意。与无FF控制器相比，主蒸汽温度的控制误差降低到50%。少

A simulator has been developed for the purpose of studying of the sliding pressure operation method for thermal power plants by means of MMS (Modular Modeling System). There are a lot of modules supplied by MMS,which are developed based on the laws of thermodynamics. These modules are boiler, superheater, attemperator, turbine, economizer, piper, condenser, deaerator, feedwater heater, and so on. Two hundreds of MMS modules are used to develop the simulator together with the data collected from real plant and manufactory.Several studies have been done in order to validate the accuracy of the developed simulator. It is shown that the results of the simulator at the rated load are agreed with that of the real plant well. Although the simulator is ahown to be stable by simulation examples, there still exist some problems in calculation aspects. The simulator is 'stiff' in calculation, because there are very fast modes and very slow modes. The fast modes originate from the turbine, and slo … More w modes originate from the boiler and reheater. The simulation costs amount of CPU time if demand of load is changed largely. The calculating stiffness makes it difficult for us to validate the dynamic accuracy of the simulator. A simlified simulator has then been developed in order to improve the calculating efficiency.After studying the current PI based control systems of the real plant, we have developed a new control system structure which is composed of a FF (feedforward) control action from the demands of load and pressure of main steam to fuel flow rate control system and temperature control of main steam. The proposed FF control can be plugged in the current PI based control systems. In order to design the FF controller, a time varying model for boiler plant is developed. This model is interpolated by several linearized models of the boiler plant at several levels of demand of load and pressure. These linearized models are identified by stimulating the simulator with PRBS (pseudo random binary sequence). According to this time varying model, a FF controller that is gain scheduled by demands of load and pressure is developed. A lot of simulation studies have been carried out for cases of small step change of demand of load and pressure and large ramp change of demand of load and pressure. The simulation results are satisfactory. The controlled error of main steam temperature has been reduced to 50% comparing with that of without FF controller. Less