基于不同压气机特性曲线预测方法的单轴燃气轮机动态性能仿真研究

Simulation Investigation on Dynamic Performance of Single Shaft Gas Turbine based on Different Compressor Characteristic Curve Prediction Methods

部件精细化建模方法研究一直为燃气轮机动态性能仿真领域的研究热点.以某经典单轴燃机核心部件压气机为对象,基于模块化建模思想利用Matlab/Simulink平台搭建了系统仿真计算平台,并将最小二乘法、三次样条插值法以及BP神经网络法嵌入平台进行了预测应用研究.结果表明:在压气机性能预测中,三种方法均可有效预测部件性能,但BP神经网络法和三次样条插值法的预测结果优于最小二乘法;在整机性能预测中,最小二乘法整机仿真计算结果偏离预设值,而BP神经网络法和三次样条插值法的仿真计算结果具有较高的准确性;在时效性方面,BP神经网络法所需时间成本比其余两种方法高.

The study of component refinement modeling method has always been a hot topic in the field of gas turbine dynamic performance simulation.Here with the compressor core components of a classic single shaft gas turbine as object, based on modular modeling idea using Matlab/Simulink platform a system simulation platform is set up.The least square method, cubic spline interpolation method and BP neural network method are embedded into the platform for this prediction application study.The results show that in the compressor performance prediction, three methods can effectively predict the component performance, but the predictive results of the BP neural network and the cubic spline interpolation method are superior to the least square method.for the performance prediction of the whole machine, the simulation results of the least square method deviate from the preset value, while the simulation results of the BP neural network method and cubic spline interpolation method have high accuracy.For the timeliness, the time cost of BP neural network method is higher than the other two methods.