论文部分内容阅读
考虑到热工对象的动态特性大多从阶跃响应获得,非线性的存在使此种线性动态模型只能在相应稳态点附近有良好的性能,文中采用了基于非线性稳态模型来实现动态模型自适应的策略,该策略用较为精确的非线性稳态模型得到当前输入下的稳态参数,然后由此修正线性动态模型中与稳态相关的参数,实现了动态模型的自适应,进而有效提高了大范围下的动态预测性能。通过对某电厂360MW“W型”火焰强制循环固态排渣煤粉炉的稳态和动态试验,建立了NOx的神经网络稳态模型和线性动态模型,用两个不同工况下的实际数据,验证了结合稳态模型的非线性自适应动态模型比线性动态模型具有更好的NOx排放预测性能。
Considering that most of the dynamic properties of thermal objects are obtained from step response, the existence of nonlinearity makes the linear dynamic model have good performance only in the vicinity of the corresponding steady-state point. In this paper, based on the nonlinear steady-state model, Model adaptive strategy. The steady-state parameters under current input are obtained by the more accurate nonlinear steady-state model, and then the parameters related to the steady-state in the linear dynamic model are modified to realize the dynamic model adaptation. Effectively improve the dynamic prediction performance in a wide range. Based on the steady state and dynamic tests of 360MW “W ” flame forced circulation solid state slagging pulverized coal fired boiler in a power plant, the steady state model and linear dynamic model of NOx neural network were established. Data. It is verified that the nonlinear adaptive dynamic model combining the steady-state model has better NOx emission prediction performance than the linear dynamic model.