摘要
目前离心式压缩机透平转速预测难以实现,虽然引入了基于Elman神经网络的离心式压缩机透平转速预测方法,但当离心式压缩机转速变化比较大时,该预测方法的预测精度就明显下降。针对这一现象,提出了一种基于外反馈Elman的离心式压缩机透平转速预测方法。在标准Elman神经网络的基础上加一个由输出到输入的外反馈,通过带外反馈的Elman神经网络实现对离心式压缩机透平转预测。仿真实验结果表明,所提出的带外反馈的Elman神经网络预测方法预测精度较高,特别在转速变化较大时,收敛速度快,稳定性高。
In view of the present centrifugal compressor turbine speed prediction difficult to achieve,the centrifugal compressor turbine rotational speed based on Elman neural network forecasting method has introduced.But when the centrifugal compressor speed change is relatively large,the accuracy of the forecasting method is significantly decreased,so the method of centrifugal compressor turbine speed prediction by a kind of external feedback based on Elman is put forward.Based on the Elman neural network with a feedback which is between the output and input,this paper proposes a method to predict centrifugal compressor turbine rotating speed by Elman neural network with feedback.Experimental results show that Elman neural network with external feedback has high prediction accuracy,especially when the rotating speed changes greatly.The improved Elman neural network has better dynamic performance.
出处
《微型机与应用》
2016年第2期54-56,62,共4页
Microcomputer & Its Applications
基金
宁夏自然科学基金(NZ14050)
关键词
离心式压缩机
转速预测
外反馈
神经网络
centrifugal compressor
rotating speed prediction
external feedback
neural network
作者简介
张彦清（1989-），通信作者，女，硕士研究生，主要研究方向：智能仪器与检测技术。E-mail：396989181@qq．com。
刘大铭（1969-），男．硕士，教授，硕士生导师，主要研究方向：智能仪器与过程控制。
白冰（1990-），男，硕士研究生，主要研究方向：图像处理与模式识别。
