采用序贯学习的改进型Elman神经网络预测研究

Research on Chemical Equipment Forecast Using Elman Network Based on Sequential Learning

针对石油化工领域监测系统中,广泛存在的非线性时间序列预测问题,将Elmann神经网络方法引入化工过程监测的在线预测。并在其理论框架基础之上,改进了Elman神经网络的内部结构,引入了串行学习机制,可以根据实时数据对网络进行在线训练,提高网络预测精度。通过对某芳烃厂实时数据在线预测仿真,表明该方法能够准确地在线预测未来数据,同时具有训练速度快、结构简单、适应性强的优点。

The forecasting of nonlinear time series model is the most probable situation in the online monitoring system of petrochemical industry. In this article, the internal structure of basic Elman neural network has been improved by the introduction of the serial learning mechanism, thus enabling online model training by using real-time data to improve the network prediction accuracy. The prediction simulation using real-time online data from one aromatic hydrocarbon plants shows that the method can accurately predict the future data online, and has the advantages of fast training speed, simple structure and good adaptability.