基于改进的BP算法的过热汽温建模

Superheated Steam Temperature Modeling Based on Improving BP Algorithm

采用机理建模方法获得热工被控对象的数学模型非常困难,相比之下,利用电厂采集到的大量历史数据,不用人为地加入特殊扰动信号,来建立火电厂热工过程的动态数学模型,是很有发展前途的方向。本文利用电厂采集到的现场数据,应用加入二次动量项和改进激活函数的BP算法,对电站锅炉的过热汽温建模进行了研究。仿真结果表明,改进后的BP算法在误差精度、收敛速度、泛化能力等方面明显优于标准BP算法,对过热汽温的建模结果令人满意。

The mathematics model of thermal process controlled object is obtained by using mechanism modeling method is very difficult, in contrast, it is established by using large amount of history data of power plant not adding the special disturbance signals artificially. The modeling method has perspective.development. In this paper, a modeling method for superheated steam temperature is developed by using field data of power plant and the BP algorithm With second order momentum and the improved activation function. The simulation results show that the improved BP algorithm is much better than standard BP algorithm in error precision, convergence rate and generalization ability and so on. The results of superheated steam temperature modeling are satisfactory.