In the modern thermal power generating units, supercritical units are increasingly becoming the main units. Due to adopting once-through furnace rather than the traditional drum boiler, the supercritical units make a higher feedwater control requirements. For Lanxi 600MW simulator, this paper uses such a program which makes use of the mid-point enthalpy value as a reference object to control feedwater, and designs the related water logic. The use of look-up table for the power plants to obtain enthalpy approach requires more modules and has a calculation of a long cycle. In response to these defects, this paper has taken the method of making use of intelligent neural network to fit the enthalpy curve, avoiding the above-mentioned shortcomings of look-up table method.At the same time, this paper has improved the conventional PI controller with fuzzy neural network, allowing it to automatically adjust parameters with the change of conditions to improve the quality of regulation. Based on the simulation running in the STAR-90 simulation platform, the results show that the control system can complete the task well for controlling the water supply, its dynamic quality is clearly superior to conventional PI control.
|