基于BP神经网络的汽轮机调阀流量特性校正

Correction of Flow Characteristic of Steam Turbine Governing Valve Based on BP Neural Network

汽轮机调阀实际流量特性与理想流量特性偏差大,将直接影响机组一次调频和负荷控制能力,甚至影响机组运行稳定性和安全性。传统调阀流量特性的校正多依靠技术人员的经验,校正效果不理想。为了获得更好的校正效果,基于试验测得的实际调阀流量特性,运用最小二乘法对实测数据进行辨识建模,确定最优调阀流量特性曲线,并利用BP神经网络模型预测出修正量,以对综合阀位进行校正。试验结果表明,校正后的汽轮机调阀流量曲线具有良好的线性度,从而可提高机组网源协调能力。

There is a big deviation in the flow characteristics between the adjusted valve of a turbine and actual situa- tion, which can directly affect the unit primary frequency regulation, load control ability, and even stability and safety of the unit operation. Traditional optimization methods to adjust valve flow characteristics mainly rely on the experience of professional technicians, but the effect is not satisfactory. In order to obtain better results of correction, based on the actual flow characteristics, the least - square method is applied to identify the measured data and determine the optimal curve. And then the integrated valve position can be adjusted according to the corrections predicted by the BP neural network. The optimized flow curve of the adjusted turbine valve has a good linearity, which can greatly improve the net- work - source coordination ability of the unit.