摘要
介绍了RBF神经网络的结构和特点,进而讨论了遗传算法与RBF神经网络相结合的方法。以某300MW电站锅炉燃烧调整试验数据为基础,利用RBF神经网络对锅炉效率与NOx排放混合建模,并用遗传算法优化RBF神经网络的性能,使其预测精度大幅提高。同时RBF神经网络具有收敛速度快的独特优点。因此,优化后的RBF神经网络模型为下一步的锅炉运行参数优化和燃烧优化系统的建立奠定基础。
The structure and feature of RBF neural network are introduced then the method of combining genetic algorithm with RBF neural network is discussed. Based on the experimental data of 300MW boiler combustion adjustment in a power plant, RBF neural network is used in hybrid modeling of efficiency and NOx emission. Genetic algorithm is used to optimize RBF neural network performance so that the model prediction accuracy is greatly enhanced. RBF neural network has a unique advantage of fast convergence. Therefore the optimized RBF neural net- work model lays the foundation for the optimization of boiler operating parameters and the establishment of combustion optimization system.
出处
《电力科学与工程》
2012年第5期37-41,共5页
Electric Power Science and Engineering
关键词
RBF神经网络
遗传算法
混合建模
预测精度
RBF neural network
genetic algorithm
hybrid modeling
prediction accuracy