Research On Intelligent Control Of Power Plant Boiler Main Steam Temperature

At present, in the thermal power, as one of the monitoring and control parameters the boiler main steam temperature is very important. If the main steam temperature is not suitable, it will directly influence the safe and economic operation of generating unit. Therefore, generally requires the main steam temperature in the vicinity of a given value.The control of main-steam temperature is one of the problems at the process control in power plant, main reasons are: Firstly, the controlled object(main steam temperature)always has the characteristic of time-delayed. And the larger the unit capacity is, the longer the delayed time is. This situation always makes feedback control too late to act. Secondly, in the actual production process, as the time goes, the unit operation condition is changing, which makes the dynamic characteristic of the controlled object is changing, its mathematical transfer function model is also changing. So the control effect of traditional controller is not ideal. In addition, the dynamic characteristic of the controlled object is nonlinear, which also increases the difficulty of control.For the main steam temperature with large inertia and large delay, some intelligent control strategies have been put forward based on the research of traditional control methods. Using fuzzy control method realizes the effective control of the controlled object though its mathematical model is difficult to obtain. Using neural network control method solves the effective control of the time-varying, nonlinear controlled object. Fuzzy neural network fuses artificial neural network and fuzzy logic by the information fusion technology. And it is an advanced means of control. It can optimize the fuzzy control rules and adjust the scaling factor through the self-learning ability of neural network. This method realizes the effective control of main steam temperature. The fuzzy neural network controller has been improved through joined the compensation part. Finally, simulation experiments are carried out using Matlab, and results show that the compensation fuzzy neural network control has better control quality.