火电机组汽轮机系统数字孪生模型研究

Studyon Digital Twin Model of Steam Turbine Systems of Thermal Power Units

以数字孪生为理论指导建立数据驱动孪生模型,通过对机组历史运行数据的分析,利用深度学习中的深度森林算法,合理确定模型的输入输出参数,并采用改进的蝙蝠算法进行超参数优化,构建高精度的数据驱动模型。选取各模型的输出参数,利用相似度函数法预测输出数组与实际输出数组之间的欧几里得距离,实现对故障的准确检测。最后利用数字孪生模型对物理实体进行状态监测和故障检测试验,结果表明借助建立的数字孪生模型和故障检测体系,能准确检测汽轮机系统中加热器管路泄漏和进出水室短路故障,证明了该方法的有效性。

In this paper,a data-driven twin model is established under the guidance of digital twinning theory.Through the analysis of historical unit operation data,input and output parameters of the model are determined reasonably by using deep forest algorithm.Improved bat algorithm is used to optimize hyperparameters to build a high-precision data-driven model.In order to realize fault detection,output parameters of each model are selected,and the Euclidean distance between the predicted output array and the actual output array is calculated by using similarity function method to achieve accurate fault detection.Finally,condition monitoring and fault detection experiments of physical entities are carried out by using the established digital twin model showing its competence and effectiveness in accurately detecting leakage of heater pipeline and short circuit fault of inlet and outlet chamber in steam turbine system.