摘要
针对一个375 MW热电厂的锅炉—汽轮机系统仿真模型,采用多层前向神经网络进行离线建模;讨论了网络结构设计、训练算法等神经网络建模问题;采用相同的固定负荷数据分别建立了线性ARX模型和局部神经网络模型并做多步预测比较;通过对基于一层隐层的全局神经网络模型的训练和仿真,结果证实了神经网络在非线性系统建模和辨识上的有效性。
The dynamic model of a simulator of the boiler-turbine system of a 375 MW （megawatt） thermal power plant is built by a feedforward neural network that is trained offline. Network modeling issues such as networks structure design and training algorithms are discussed. A linear ARX model and a local neural network model, all estimated using the same data sampled at a certain operating level, are built for comparing their multi-step prediction performance. A global neural network model based on the multilayer perceptron with one hidden layer is also applied to the simulator modeling. The results of simulation studies show the effectiveness of neural network for the modeling and identification of nonlinear systems.
出处
《计算机测量与控制》
CSCD
2006年第5期622-624,共3页
Computer Measurement &Control
基金
国家自然科学基金资助项目(60443008)
