State prediction method for power plant fans based on the CNN-LSTM-AM dynamic integrated model

被引：0
作者：
Wei W. [1 ]
Lyu Y. [1 ,2 ]
Qi X. [1 ]
Liu J. [1 ,2 ]
Fang F. [1 ,2 ]
机构：
[1] School of Control and Computer Engineering, North China Electric Power University, Beijing
[2] State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources, North China Electric Power University, Beijing
来源：
Yi Qi Yi Biao Xue Bao/Chinese Journal of Scientific Instrument | 2023年 / 44卷 / 04期
关键词：
attention mechanism; convolutional neural network; ensemble learning; long short-term memory network; state prediction;
D O I：
10.19650/j.cnki.cjsi.J2311119
中图分类号：
学科分类号：
摘要：
To solve the problem of low accuracy of the fan state prediction model when the power plant load changes, a dynamic integrated state prediction method based on convolutional neural network (CNN), long short-term memory (LSTM) network and attention mechanism (AM) is proposed. Firstly, the CNN is used to divide the sample data into different classes with overlapping boundaries to achieve soft classification of wind turbine operating conditions. Then, the AM layer is introduced into the traditional LSTM network. LSTM-AM networks as sub-learners are established under different work conditions. The soft classification labels output by CNN are used as the initial weights, and the genetic algorithm is used to search for the optimal weight bias. Finally, the output of each sub-learner is multiplied with corresponding weights and summed to obtain the integrated prediction value, which could improve the prediction accuracy under different operating conditions of power plant fans. The experimental results show that, compared with each LSTM-AM sub-model and signal LSTM-AM model, the proposed CNN-LSTM-AM dynamic integrated model can reduce the relative mean square error by 11.5% and 22.3% when power plant fans are operating under variable loads. Results indicate that the model has better robustness and applicability. © 2023 Science Press. All rights reserved.
引用
收藏
页码：19 / 27
页数：8
相关论文
共 21 条
  • [11] REN J T, TANG B P, YONG B, Et al., Wind turbine gearbox fault warning based on depth variational autoencoders network fusion SCADA data, Acta Energiae Solaris Sinica, 42, 4, pp. 403-408, (2021)
  • [12] LI X B, NIU Y G, GE W CH, Et al., Early fault warning of power plant auxiliary engine based on improved stacked autoencoder neural network, Chinese Journal of Scientific Instrument, 40, 6, pp. 39-47, (2019)
  • [13] SMITH W., Time series forecasting: Analysis of LSTM neural networks to predict exchange rates of currencies, Instrumentation, 7, 4, pp. 25-39, (2020)
  • [14] TAN F, LI CH N, XIAO H, Et al., A thermal error prediction method for CNC machine tool based on LSTM recurrent neural network, Chinese Journal of Scientific Instrument, 41, 9, pp. 79-87, (2020)
  • [15] ZHENG T X, HE J, ZHANG L B., Combustion timing estimation of the HCCI engine with mixed fuel under complex operating conditions based on LSTM neural network, Chinese Journal of Scientific Instrument, 41, 10, pp. 100-110, (2020)
  • [16] SHU T X., Sensory data prediction using spatiotemporal correlation and lstm recurrent neural network [J], Instrumentation, 6, 3, pp. 10-17, (2019)
  • [17] WEI SH R, ZHANG X, FU Y, Et al., Early fault warning and diagnosis of offshore wind DFIG based on GRA-LSTM-Stacking model, Proceedings of the CSEE, 41, 7, pp. 2373-2383, (2021)
  • [18] ZHANG CH, BAI J B, LAN K, Et al., Photovoltaic power generation prediction based on data mining and genetic wavelet neural network, Acta Energiae Solaris Sinica, 42, 3, pp. 375-382, (2021)
  • [19] XU ZH F, LI X D, ZHANG CH CH, Et al., Fault identification for rolling bearing by self-calibrated convolutional neural network under small samples conditions, Chinese Journal of Scientific Instrument, 43, 10, pp. 241-251, (2022)
  • [20] LEI CH L, XIA B F, XUE L L, Et al., Fault identification for rolling bearing by self-calibrated convolutional neural network under small samples conditions, Chinese Journal of Scientific Instrument, 43, 9, pp. 122-130, (2022)
123