摘要
分类介绍近年来国内外各种数据预测方法,重点论述各种人工神经元网络的数据预测应用情况,包括BP神经网络、Fuzzy神经网络、径向基神经网络(RBF)、自联想神经网络(AANN)、基于神经网络的偏最小二乘法(NNPLS)以及组合预测方法。针对火电厂测量数据的特点及实时监控系统的要求,可以考虑将NNPLS法运用在实时数据预测方面,利用其应用简单、收敛迅速、估计值对噪声和坏点敏感度低等优点提高预测的效率和精度。
Variety of classified forecasting methods recently used in China and abroad are introduced, especially kinds of ANN models and their improvements and applications which include BP, Fuzzy Neural Network (FNN), Radial Based Function (RBF), Autoassociative Neural Network (AANN), Neural Network Partial Least Square (NNPLS) and combined forecasting model. Considering the features of measured data and requirements of real-time monitored control system, the NNPLS model is highly recommended in real-time data forecasting. Forecasting efficiency and accuracy can be promoted due to its merits of easily application, rapidly convergence and low sensitivity to noises and bad points.
出处
《电站系统工程》
北大核心
2005年第2期1-4,共4页
Power System Engineering
