臺灣博碩士論文加值系統

本論文研究主旨在於將長短期記憶神經網路結合最佳化演算法，進行非線性系統預測。在本文中，使用長短期記憶神經網路，並且利用了兩種最佳化演算法，反向傳播演算法、雜草入侵演算法，對神經網路參數進行最佳化。
在預測非線性系統模型中，使用MATLAB求出神經網路權重。依照所採用的最佳化演算法搜尋最佳化參數。最後將得到的最佳化參數放入長短期記憶神經網路架構，以此來預測非線性系統模型。
本論文使用的非線性系統模型有Mackey-Glass系統、非線性系統、動態時變系統。實驗結果使用均方誤差進行比較與分析

The main purpose of this paper's research is to combine Long Short-Term Memory (LSTM) neural networks with optimization algorithms to predict nonlinear systems. In this study, an LSTM neural network is utilized, and two optimization algorithms, the Backpropagation Algorithm and the Invasive Weed Optimization Algorithm, are used to optimize the neural network parameters.
In predicting nonlinear system models, MATLAB is used to determine the neural network weights. The optimal parameters are found according to the chosen optimization algorithm. Finally, the optimized parameters are applied to the LSTM neural network architecture to predict the nonlinear system models.
The nonlinear system models used in this paper include the Mackey-Glass system, nonlinear systems, and dynamic time-varying systems. The experimental results are compared and analyzed using Mean Squared Error (MSE).

摘要 i
Abstract ii
誌謝 iii
目錄 iv
表目錄 vi
圖目錄 vii
第一章 緒論 1
1.1研究動機 1
1.2文獻探討 1
1.3研究方法 2
1.4論文架構 2
第二章 長短期記憶神經網路 3
2.1引言 3
2.2長短期記憶神經網路(Long Short-Term Memory, LSTM) 3
第三章 最佳化方法 5
3.1反向傳播演算法 5
3.2雜草入侵演算法(Invasive Weed Optimization, IWO) 6
第四章 非線性系統模型 8
4.1 Mackey-Glass 系統 8
4.2 非線性系統(Nonlinear plant) 9
4.3 動態時變系統(Dynamical time-varying plant) 9
第五章 實驗過程與結果 11
5.1實驗方式 11
5.1.1最佳化演算法設定參數 11
5.2 Mackey-Glass 系統預測結果 11
5.2.1 Mackey-Glass反向傳播演算法預測結果 11
5.2.2 Mackey-Glass雜草入侵演算法預測結果 12
5.3非線性系統預測結果 13
5.3.1 非線性系統反向傳播演算法預測結果 14
5.3.2 非線性系統雜草入侵演算法預測結果 15
5.4動態時變系統預測結果 16
5.4.1 動態時變系統反向傳播演算法預測結果 16
5.4.2 動態時變系統雜草入侵演算法預測結果 17
5.5最佳化演算法比較結果 18
第六章 結論與未來展望 20
6.1結論 20
6.2未來展望 20
參考文獻 21

[1]D. M. Sahoo and S. Chakraverty, “Functional link neural network approach to solve structural system identification problems,” Neural Comp Appl 28:1–12, 2017.
[2]C. N. Ko, “Identification of non-linear systems using radial basis function neural networks with time-varying learning algorithm”, IET Sign Proc 6(2):91–98, 2012.
[3]Z. J. Fu, W. F. Xie, X. Han and W. D. Luo, “Nonlinear systems identification and control via dynamic multitime scales neural networks,” IEEE Trans Neural Net Learn Syst 24(11):1814–1823, 2013.
[4]T.A. Tutunji, “Parametric system identification using neural networks,” Appl Soft Comp 47:251–261, 2016.
[5]Z. J. Fu, W. F. Xie and W. D. Luo, “Robust on-line nonlinear systems identification using multilayer dynamic neural networks with two-time scales,” Neurocomp 113:6–26, 2013
[6]Y. Y. Fu, C. J. Wu, J. T. Jeng and C. N. Ko, “Identification of MIMO systems using radial basis function networks with hybrid learning algorithm,” Appl Math Comp 213:184–196, 2009.
[7]C. H. Lee and C. C. Teng, “Identification and control of dynamic systems using recurrent fuzzy neural networks,” IEEE Trans Fuzzy Syst 8:349–366, 2000.
[8]J. F. Qiao and H. G. Han, “Identification and modeling of nonlinear dynamical systems using a novel self-organizing RBF-based approach,” Automatica 48:1729–1734, 2012.
[9]S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997.
[10]F. A. Gers, J. Schmidhuber and F. Cummins, “Continual prediction using LSTM with forget gates,” Springer, London, 1999.
[11]D. E. Rumelhart, G. E. Hinton and R. J. Williams, “Learning representations by back-propagating errors,” Nature 323, pp.533–536 ,1986.
[12]A. R. Mehrabian, C. Lucas, “A Novel Numerical Optimization Algorithm Inspired from Weed Colonization,” Ecological Informatics, pp.355-366, 2006.
[13]M. Misaghi, M. Yaghoobi, “Improved invasive weed optimization algorithm (IWO) based on chaos theory for optimal design of PID controller,” Journal of Computational Design and Engineering, Vol.6, pp.284-295, July 2019
[14]M. C. Mackey, and L. Glass, “Oscillation and chaos in physiological control systems,” Science, Vol. 197, No.4300, pp. 287-289, 1977.
[15]J. Tavoosi, A. A. Suratgar, and M. B. Menhaj, “Nonlinear system identification based on a self-organizing type-2 fuzzy RBFN,” Eng. Appl. Artificial. Intell., Vol. 54, pp. 26-38, 2016.
[16]C. H. Lee, and C. C. Teng, “Identification and control of dynamic systems using recurrent fuzzy neural networks,” IEEE Trans. Fuzzy Sys., Vol. 8, pp.349-366, 2000.
[17]H. Loussifi, K. Nouri, and N. B. Braiek, “A comparative study on initial parameterization methods of fuzzy wavelet neural networks for time delay systems identification,” In: 4th International Conference on Systems and Control, Sousse, Tunisia, pp. 7-12, 2015.