Application Of Intelligent Control Technology To All Range Feedwater Control System In Large-Sized Firepower Plant Units

The control system of boiler all range feedwater is one of the main control systems in firepower plant units, which mission is adjusting boiler feedwater flow and maintaining water level of steam drum in definite bound. excellent performance or bad performance of all range feedwater control system has important impact on safety,stabilization and economic operation of units. In recent years coupled with extending of units capability, higher performance of all range feedwater is need. It is because decrease of both steam sluice and vaporize area quicken water level change speed of steam drum, In the same time widening of boiler capacity increase the thermal load of boiler heating surface, which make load change have marked effect on water level and lead to large extent change of system dynamic characteristic. Since boiler is a multiple variable system with non-linear controlled values and large time variation, it is difficult to accurate modeling for boiler water level. So some problem is found while using traditional means to control boiler water level .For example ,it takes more times and labors to adjust parameter of controller. in addition, to adapt the controlled object with non-linear controlled values and Large time variation or change of control equipment , skilled worker have to switch proportion band continually in all range feedwater control course. Even so, it is difficult to achieve desired control effect. So a robust,relatively simple and self-tuning control method is need.Fuzzy-Neural network is a new intelligent control technology. It is combinative production of fuzzy logic control and neural control. Fuzzy-Neural network make use of pellucid expression ability of fuzzy and strong self-learning of neural, transform fuzzy system to relevant neural network, syncretize two intelligent control technology, learn from other's strong points to offset one's weakness, Improve study ability and expression ability of system, Can carry out to control system with non-linear controlled values and Large time variation nicely. The paper improves the structure and study arithmetic of general Mamdani model fuzzy-neural network, designs a self-tuning PID controller based on fuzzy-neural network, replace conventional cascade three-element controller PI1 and cascade one-element controller PI. The controller can get hold of fuzzy rules and optimize its subjection function online by self-learning ability of the neural network ,and it also can adjust PID parameter in operation according to "e" and "ec" at different time by fuzzy reasoning, solve the problem of controller parameter self-tuning in all range feedwater control system and switching proportion band continually in load change course of units, simplify the structure of control system, enhance performance of control system greatly, have very high engineering application value.