融合数据与机理的燃煤发电机组协调系统建模

Modeling of Coal-fired Generating Unit Coordination System With Data and Mechanism

为掌握灵活运行条件下燃煤发电机组动态特性,通过融合传统机理建模与数据建模方法,建立了亚临界机组协调系统模型。模型基础构架通过机理分析构建,针对宽负荷下机理模型中部分环节非线性程度高、建模难度大,采用极限学习机进行辅助建模。以某660 MW亚临界机组为例,利用稳态运行数据数据训练获得了基于极限学习机的燃料系数、汽轮机系数、饱和蒸汽流量及过热蒸汽流量辅助子模型,采用粒子群寻优算法对机理模型动态参数进行辨识建立了协调系统动态模型,并对模型进行了开环阶跃响应测试和闭环验证。结果表明:模型能够准确反映机组动态特性,相比于传统机理模型具有更高的精度,模型对宽负荷下机组的汽包压力、主蒸汽压力、输出功率预测误差都在1.5%左右,可用于控制器改进与设计。

In order to master the dynamic characteristics of coal-fired generating units under flexible operating conditions,a sub-critical unit coordination system model was established by a modeling method combining traditional mechanism and data.The basic framework of the model was constructed by mechanism analysis,and for some links with high nonlinearity and large modeling difficulty in the mechanism model under wide load,extreme learning machines were used for auxiliary modeling.Taking a 660 MW subcritical unit as an example,a dynamic model of a coordinated system was established.The steady state operating data was employed to train the submodels of fuel coefficient,turbine coefficient,saturated steam flow rate and superheated steam flow rate based on the ultimate learning machine,and the dynamic parameters of the mechanism model were identified based on the particle swarm optimization algorithm.Furthermore,open-loop simulation and close-loop simulation upon history operational data were conducted to verify the developed model.The results demonstrate that the model is capable to catch the dynamic characteristics of the actual unit and has higher accuracy than the traditional mechanism model.The prediction errors of drum pressure,main steam pressure and output power of the unit under wide load are about 1.5%,which can used for controller improvement and design.