臺灣博碩士論文加值系統

本論文的研究目的是以模糊理論為基礎的誤差修正機制 (Fuzzy-based Error Correction Mechanism, FECM) 改善預測演算法對空間中移動物體軌跡的預測誤差。FECM為二輸入一輸出的模糊推論系統 (Fuzzy Inference System, FIS)，能依據運動物體軌跡不同的量測尺度而自動調整其論域範圍，且透過模糊推論過程修正預測誤差值。本論文對移動物體軌跡的預測分為兩部份，首先以軌跡預測演算法得到物體移動軌跡的結果，最後以模糊理論為基礎的誤差修正機制修正其預測誤差值，使最後的預測結果更接近實際物體移動的軌跡。
本論文以所提出的誤差修正機制結合預測演算法與未結合誤差修正機制的預測演算法進行比較，結果可知所提出的誤差修正機制能獲得較佳的預測值。另外此方法為一個不需要經過訓練及沒有一般預測演算法建模 (modeling) 過程的後置處理單元，能夠在不改變原預測演算法架構下增進預測軌跡的準確度。

The purpose of this thesis is using Fuzzy-based Error Correction Mechanism (FECM) to improve the forecasting error of prediction algorithm. FECM is a two-input-one-output fuzzy inference system (FIS), it can adjust the universe of discourse according to the different measuring scale of moving object’s trajectory and reduce the prediction error through FIS. The prediction of moving object’s trajectory includes two parts in this thesis, first calculate the forecasting error of moving objective’s trajectory by prediction algorithm, and then correct the error with FECM. At last, the prediction result is closer to real moving objective’s trajectory.
In this thesis, we compare the prediction result of FECM with original prediction algorithm and the results show that FECM can obtain better prediction value. In addition, the proposed method can improve the precision of trajectory prediction without training and modeling process of general prediction algorithm.

致 謝 I
摘 要 II
Abstract III
目 錄 IV
圖目錄 VII
表目錄 X
第一章 緒論 1
1.1 前言 1
1.2 論文回顧 2
1.3 研究目的 4
1.4 研究方法 5
1.5 論文架構 6
第二章 預測方法的探討 8
2.1 卡爾曼濾波器 8
2.1.1 卡爾曼濾波器的整體架構及流程 8
2.1.2 卡爾曼濾波器的預測流程 11
2.2 類神經網路 12
2.2.1 結合最適權值選擇機制之TDNN網路架構 14
2.2.2 結合最適權值選擇機制之TDNN演算法流程 18
2.3 以最大熵值原理為基礎的模糊聚類 20
2.3.1 模糊聚類演算法使用最大熵值原理 20
2.3.2 以最大熵值原理為基礎的模糊聚類演算法流程 23
2.4 本章總結 23
第三章 以模糊理論為基礎的誤差修正機制 25
3.1 誤差修正機制的基本架構及原理 25
3.1.1 輸入變數的原理說明 26
3.1.2 依據輸入誤差自動調整語言變數的論域 29
3.1.3 模糊函數原理說明 31
3.2 自動誤差修正機制演算法的流程 34
第四章 實驗結果 36
4.1 實驗平台 36
4.2 軌跡預測的實驗結果 39
4.2.1 卡爾曼濾波器的軌跡預測 39
4.2.2 時間延遲類神經網路的軌跡預測 43
4.2.3 以最大熵值原理為基礎的模糊聚類的軌跡預測 47
4.2.4 綜合檢討 51
4.3 混沌時間序列的軌跡預測 52
第五章 結論與未來研究方向 54
5.1 結論 54
5.2 未來研究方向 54
參考文獻 56
作者簡歷 59

[1]D. Zhen, H. Leung and K. Chan, “Model-set adaptive using a fuzzy Kalman Filter,” in 3rd International Conference on Information Fusion, Vol. 1, pp. MOD2/3 – MOD2/9, July. 2000.
[2]J. Z. Sasiadek and J. Khe, “Sensor fusion based on fuzzy Kalman Filter,” in Proceedings of the Second International Workshop on Robot Motion and Control, pp. 275-283, Oct. 2001.
[3]N. Promkajin and S. Noppanakeepong, “An Improvement to the Adaptive Kalman Filter with the Feedback of Estimation Error,” in Proceeding of Student Conference on Research and Development, Aug. 2003.
[4]J. Sang, K. Kubik and L. Zhang, “Prediction of DGPS Corrections with Neural Networks,” in Proceeding of Knowledge-Based Intelligent Electronic Systems, Vol. 2, pp. 355-361, May. 1997.
[5]J. A. Marshall and V. Srikanth, “Curved Trajectory Prediction Using a Self-Organizing Neural network,” International Journal of Neural System, 10(1):59-70, Feb. 2000.
[6]G. Beni and X. Liu, “Online Data-Driven Fuzzy Clustering With Applications to Real-Time Robotic Tracking,” IEEE Trans. Fuzzy System, Vol. 12, pp. 516-523, Aug. 2004.
[7]A. Elnagar and A. M. Hussein, “An adaptive prediction model for trajectory planner systems,” in Proceedings of IEEE International Conference on Robotics and Automation, Vol. 2, pp. 2442-2447, Sept. 2003.
[8]N. Kehtarnavaz and S. Li, “A collision-free navigation scheme in the presence of moving obstacles,” in Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 808-813, June 1988.
[9]R. C. Luo and T. M. Chen, “Target tracking by gray prediction theory and look-ahead fuzzy logic control,” in Proceedings of IEEE International Conference on Robotic and Automation, Vol. 2, pp. 1176-1181, May 1999.
[10]T. M. Chen and R. C. Luo, “Mobile target tracking using hierarchical gray-fuzzy motion decision-making method,” in Proceedings of IEEE International Conference on Robotics and Automation, Vol. 3, pp. 2118-2123, Apr. 2000.
[11]H. A. Xie, B. H. Zhang, Y. H. Li, G. L.Yu, D.C. Zhou, F. N. Yao and L. Y. Wang, “A Variable Structure Trajectory Prediction Algorithm Based on Transform For Complex Exponential Times Series,” in Proceedings of IEEE Transmission and Distribution Conference and Exhibition: Asia and Pacific 2005, pp.1-5, Aug. 2005.
[12]M. S. Grewal and A. P. Andrews, Kalman Filtering, Wiley Interscience, 2001.
[13]K. Rose, E. Gurewitz and G. C. Fox, “Statistical mechanics and phase transition in clustering,” Physical Review Letters, Vol. 65, no. 8, pp. 945-948, 1990.
[14]E. T. Jaynes, “Information theory and statistical mechanics I,” Physical Review, Vol.106, no.4, pp. 620-630, 1957.
[15]M. Mackey and L. Glass, “Radial basis function for multivariable interpolation: A review,” Science, vol. 197, pp. 287-289. 1987.
[16]F. Gers, D. Eck and J. Schmidhuber, Applying LSTM to Time Series Predictable Through Time-Window Approaches, New York: Springer-Verlag, 2001.
[17]C. C. Liu, T. Y Sun, Y. C. Tseng, and C. L. Lin, “The Time-Delay Neural Network with Suitable Weight Selection Mechanism for Moving Object Prediction,” in Proceeding on CACS Automatic Control Conference, Nov. 2005.
[18]葉怡成, 類神經網路模式應用與實作, 儒林圖書, 2003.
[19]蘇木春, 張孝德, 機器學習-類神經網路、模糊系統以及基因演算法則, 全華科技, 1999.
[20]曾英璟, 使用最適權值選擇的時間延遲類神經網路於移動物體預測的研究, 國立東華大學碩士論文, 2004.