在煤矿循环流化床锅炉仿真中RBF神经网络的应用

RBF Neural Network Application in the Coal Mines of Circulating Fluidized Bed Boiler Simulation

为了解决煤矿循环流化床锅炉燃烧实时动态数学模型的高阶、多变量微分方程不易求解等问题,提出利用RBF神经网络实现该模型的参数辨识,并提出利用小生境克隆选择算法提高RBF网络学习算法的收敛性。通过SNCC循环流化床仿真系统的数字仿真验证,算法具有良好的收敛性和逼近效果,并避免了传统模型的复杂微分方程求解过程。

In order to solve the problem of difficultly solved high-level, multi-variable differential equation for mine circulating fluidized bed boiler real-time dynamic model, RBF neural network is used to identify the model pa- rameter, and maken use of clonally selection algorithm to improve the RBF network learning convergence. Digital simulation results by SNCC CFB simulation system show that the proposed algorithm has good convergence and approximation results, and to avoid the complex process of solving differential equations of the traditional model.