隨著科技的日新月異，對於電子產品的功能以及攜帶性要求越來越嚴苛。為了因應上述兩種要求，除了在IC 設計技術上投入巨大的研發成本以及人力之外，電子封裝技術也從傳統的DIP (Dual In-line Package)發展出具有高密度I/O (Input and Output) 以及較小覆蓋區 (Footprint)的覆晶 (Flip Chip)封裝、晶片級封裝(Chip Scale Package, CSP)、晶圓級封裝 (Wafer Level Package, WLP)以及扇出型(Fan-out)封裝。另外也更進一步發展3D 封裝技術來維持摩爾定律(Moore's Law)的發展並提出More than Moore象徵此技術將使半導體工業的發展將超越摩爾定律。
熱循環負載測試(Thermal Cycling Test, TCT)為其中一項取得電子元件可靠度的實驗方法，然而因為進行實驗需要花費相當龐大的成本。因此近年業界採用有限元素分析來減少實驗次數，同時減少產品開發成本以及時程，然而採用有限元素分析依然需要耗費相當多時間，當研發人員未受過充足訓練時則會導致不同的研發人員所獲得的計算結果不一致。現今隨著電腦硬體設施的效能不斷進步，機器學習方法的應用蓬勃發展，因此本研究目地為探討是否能利用支援向量回歸(Support Vector Regression ,SVR)方法預測晶圓級封裝之可靠度，其中包含單核心支援向量迴歸 (Single Kernel SVR)、多核心支援向量迴歸 (Multiple Kernel SVR) 以及新型支援向量迴歸 (ν-SVR)，並設法透過不同的訓練資料量、不同的測試資料集詳細探討前述支援向量迴歸模型之預測效能及其穩定性，最後希望能提供一個快速且有效的預測模型給前端設計研發人員檢視設計參數的可行性。
本研究將依照下列步驟執行，首先利用實驗驗證本研究所使用的有限元素分析方法以及Coffin-Manson 壽命預估公式所得到的結果；接著在固定材料和負載的情況下，給定不同的幾何結構參數值，建立WLCSP之錫球可靠度數據庫；再來透過支援向量迴歸方法預測WLCSP之錫球可靠度，並評估此預測模型之預測效能及其穩定性；最終除了探討是否能利用增加訓練資料樣本數或是不同的支援向量迴歸方法來增進預測效能之外，也探討對於特定測試資料預測效能不足之問題，是否能透過增加與該測試資料相似的訓練樣本來獲得改善。本研究顯示支援向量迴歸模型預測效能十分穩定，增加訓練樣本數與採用不同支援向量回歸方法設定皆能增進預測效能，對於特定測試資料預測效能不足之問題也能透過增加與該測試資料相似的訓練樣本來得到改善。

With the rapid development of technologies, the electronic devices are asked to be thinner, smaller, and more powerful to meet the market demands. The electronic packaging technology evolves from a conventional package structure such as Dual In-line Package (DIP) to the structure with a smaller footprint and higher I/Os (Input and Output), such as Flip Chip (FC), Chip-Scale Package (CSP), Wafer Level Package (WLP) and Fan-out (FO) Packages. Moreover, 3D packaging technologies are developed to break the limitation of the existing Moore’s Law, the term “More than Moore” is proposed to represent that these developing technologies will help the development of the semiconductor industry to go beyond the limitation of Moore’s Law.
Thermal cycling test (TCT) is one of the important experimental approaches to obtain the reliability of the electronic packages. However, the experimental approach will take great amounts of time and cost to obtain the result of reliability. Therefore, finite element analysis (FEA) is introduced to reduce the number of experiments. While implementing finite element analysis still needs to take a lot of time and effort to apply appropriate simulation techniques to construct the model and simulates the process of TCT. Furthermore, different results will be obtained by the different researchers if the researchers are lacking proper training. Thus, the purpose of this research is applying the machine learning techniques, which benefit from the improvement of computer infrastructures that provides high-speed computation and the huge amount of storage, to predict the reliability of wafer level chip scale package (WLCSP), to provide an efficient and a reliable predictive model for front-end designers to check the feasibility of their design. This research adopts the algorithm of support vector regression (SVR), which is one of the machine learning techniques, including the single kernel, multiple kernel, and nu-SVR (ν-SVR) techniques. The evaluation for predictive performance and the stability of the predictive performance of the predictive model obtained by using this approach are discussed in this research. As for discussion of the approaches to improve the predictive performance are also included in this research, such as the implementation of the different training dataset, increasing the number of similar training sample for the specific test sample that cannot be predicted well by the predictive model, and application of multiple kernel SVR and ν-SVR.
In order to accomplish the research goal, this study will be carried out accordance to the following steps: first, verifying the result obtained by finite element analysis method used in this study and Coffin-Manson model by reference experimental data; second, constructing the reliability database for WLCSP by the validated finite element analysis approach with fixed material properties and thermal cycling loading but different geometry design parameters; third, obtain the predictive reliability of WLCSP through the predictive model obtained by using single kernel SVR, multiple kernel SVR, and ν-SVR, evaluating the prediction accuracy and stability of the predictive performance of predictive model; fourth, discuss the effect of adopting following approaches to improve the performance of the predictive model such as increasing the number of training sample, increasing the number of similar training sample for the specific test sample that cannot be predicted well by the predictive model, and implementing the multiple kernel SVR and ν-SVR. The results of this study indicate that the predictive performance of the predictive model obtained by using SVR is stable. Moreover, it can be improved by adopting approaches such as increasing the number of training samples, increasing the number of similar training sample for the specific testing sample that cannot be predicted well by the predictive model. Considering the predictive performance and training time, the ν-SVR technique is more suitable with small amount of training samples in this research, while the single kernel SVR technique is suitable with larger amount of training samples in this study.

摘要………………………………………………………………………………………………………………….I
Abstract……………………………………………………………………………………………………………..II
目錄………………………………………………………………………………………………………………..IV
圖目錄……………………………………………………………………………………………………………VII
表目錄……………………………………………………………………………………………………………..XI
第一章 緒論 1
1.1. 電子封裝發展簡介 1
1.2. 研究動機 2
1.3. 文獻回顧 3
1.4. 研究目標 9
第二章 基礎理論 10
2.1. 錫球外型預測 10
2.2. 有限元素基礎理論 11
2.2.1. 線彈性有限元素理論 12
2.2.2. 材料非線性理論 15
2.2.3. 數值方法及收斂準則 17
2.3. 材料硬化法則 18
2.3.1. 等向硬化法則 19
2.3.2. 動態硬化法則 19
2.3.3. Chaboche 動態硬化模型 20
2.4. 電子封裝結構可靠度之預測模型 21
2.4.1. Coffin-Manson 應變法 (Coffin-Manson Model) 21
2.4.2. Darveaux 能量密度法 (Darveaux Energy Based Method) 22
2.4.3. 修正型能量密度法 (Modified Energy Based Method) 23
2.5. 機器學習 (Machine Learning) 23
2.5.1. 監督式學習 (Supervised Learning) 24
2.5.2. 非監督式學習 (Unsupervised Learning) 26
2.5.3. 資料前處理方法 27
2.5.4. 支援向量機 (Support Vector Machine) 29
2.5.5. 傳統支援向量迴歸方法( Classical Support Vector Regression) 33
2.5.6. 新型支援向量迴歸方法 (New Support Vector Regression) 40
2.5.7. 增量學習以及遷移學習 (Incremental Learning and Transfer Learning) 45
2.5.8. 迴歸模型預測效能評估方法 55
第三章 有限元素分析與驗證 56
3.1. 有限元素模型建立與基本假設 56
3.2. 材料參數設定 63
3.3. 邊界條件設定 64
3.4. 循環溫度負載設定 65
3.5. 有限元素分析結果驗證 65
第四章 結果與討論 67
4.1. 訓練資料庫以及測試資料庫之建立 67
4.2. 支援向量迴歸 77
4.2.1. 支援向量迴歸模型之參數設定 77
4.2.2. 資料前處理方法與核函數種類對預測模型預測效能影響之探討 81
4.2.3. 支援向量迴歸模型之預測效能及穩定性評估 84
4.2.4. 增加訓練資料個數對支援向量迴歸模型預測效能影響之探討 159
第五章 結論與未來工作 163
參考文獻. 165

[1] M.-K. Shih, H.-C. Shih, Y.-C. Lee, D. Tamg, and C. Hung, "Solder joint reliability analysis for large size WLCSP," in 2017 International Conference on Electronics Packaging (ICEP), 2017: IEEE, pp. 61-65.
[2] L. S. Goldmann, "Geometric optimization of controlled collapse interconnections," IBM Journal of Research and Development, vol. 13, no. 3, pp. 251-265, 1969.
[3] S. M. Heinrich, M. Schaefer, S. A. Schroeder, and P. S. Lee, "Prediction of solder joint geometries in array-type interconnects," American Society of Mechanical Engineers Journal of Electronic Packaging, vol. 118, pp. 114-121, 1996.
[4] K. A. Brakke, "Surface evolver manual," Mathematics Department, Susquehanna Univerisity, Selinsgrove, PA, vol. 17870, no. 2.24, p. 20, 1994.
[5] L. F. Coffin Jr, "A study of the effects of cyclic thermal stresses on a ductile metal," Transactions of the American Society of Mechanical Engineers, New York, vol. 76, pp. 931-950, 1954.
[6] S. S. Manson, "Behavior of materials under conditions of thermal stress," National Advisory Committee for Aeronautics, vol. 2933, pp. 317-350, 1953.
[7] R. Darveaux, K. Banerji, A. Mawer, G. Dody, and J. Lau, Reliability of plastic ball grid array assembly. New York: McGraw-Hill, 1995.
[8] R. Darveaux, "Effect of simulation methodology on solder joint crack growth correlation," in 2000 Proceedings. 50th Electronic components and technology conference (Cat. No. 00CH37070), 2000: IEEE, pp. 1048-1058.
[9] C. M. Liu, C. C. Lee, and K. N. Chiang, "Enhancing the reliability of wafer level packaging by using solder joints layout design," IEEE Transactions on Components and Packaging Technologies, vol. 29, no. 4, pp. 877-885, 2006.
[10] K.-C. Wu, S.-Y. Lin, T.-Y. Hung, and K.-N. Chiang, "Reliability assessment of packaging solder joints under different thermal cycle loading rates," IEEE Transactions on Device and Materials Reliability, vol. 15, no. 3, pp. 437-442, 2015.
[11] K.-N. Chiang and W.-L. Chen, "Electronic packaging reflow shape prediction for the solder mask defined ball grid array," 1998.
[12] C. Tsou, T. Chang, K. Wu, P. Wu, and K. Chiang, "Reliability assessment using modified energy based model for WLCSP solder joints," in 2017 International Conference on Electronics Packaging (ICEP), 2017: IEEE, pp. 7-15.
[13] L. Zhang, L. Sun, S.-j. Zhong, J. Ma, and L. Bao, "Reliability of Pb-free solder joints in FCBGA using finite element simulation and Taguchi method," in 2015 16th International Conference on Electronic Packaging Technology (ICEPT), 2015: IEEE, pp. 197-200.
[14] S. Aich, K. Younga, K. L. Hui, A. A. Al-Absi, and M. Sain, "A nonlinear decision tree based classification approach to predict the Parkinson's disease using different feature sets of voice data," in 2018 20th International Conference on Advanced Communication Technology (ICACT), 2018: IEEE, pp. 638-642.
[15] D. Wu, C. Jennings, J. Terpenny, R. X. Gao, and S. Kumara, "A comparative study on machine learning algorithms for smart manufacturing: tool wear prediction using random forests," Journal of Manufacturing Science and Engineering, vol. 139, no. 7, 2017.
[16] A. Rabe, S. van der Linden, and P. Hostert, "Simplifying support vector machines for regression analysis of hyperspectral imagery," in 2009 First Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, 2009: IEEE, pp. 1-4.
[17] I. Takigawa, K.-i. Shimizu, K. Tsuda, and S. Takakusagi, "Machine learning predictions of factors affecting the activity of heterogeneous metal catalysts," in Nanoinformatics: Springer, Singapore, 2018, pp. 45-64.
[18] W. S. McCulloch and W. Pitts, "A logical calculus of the ideas immanent in nervous activity," The bulletin of mathematical biophysics, vol. 5, no. 4, pp. 115-133, 1943.
[19] 周佩勲, "以人工神經網路回歸模型評估晶圓級封裝之可靠度," 碩士論文, 國立清華大學動力機械工程學系, 2019.
[20] T. K. Ho, "Random decision forests," in Proceedings of 3rd international conference on document analysis and recognition, 1995, vol. 1: IEEE, pp. 278-282.
[21] J. R. Quinlan, "Induction of decision trees," Machine learning, vol. 1, no. 1, pp. 81-106, 1986.
[22] 蕭翔云, "以隨機森林回歸模型評估晶圓級封裝之可靠度," 碩士論文, 國立清華大學動力機械工程學系, 2019.
[23] C. Cortes and V. Vapnik, "Support-vector networks," Machine learning, vol. 20, no. 3, pp. 273-297, 1995.
[24] T. Yun, K. Sim, and H. Kim, "Support vector machine-based inspection of solder joints using circular illumination," Electronics Letters, vol. 36, no. 11, pp. 949-951, 2000.
[25] H. Wu, X. Zhang, H. Xie, Y. Kuang, and G. Ouyang, "Classification of solder joint using feature selection based on Bayes and support vector machine," IEEE Transactions on Components, Packaging and Manufacturing Technology, vol. 3, no. 3, pp. 516-522, 2013.
[26] A. J. Smola and B. Schölkopf, "A tutorial on support vector regression," Statistics and computing, vol. 14, no. 3, pp. 199-222, 2004.
[27] A. Aljouie and U. Roshan, "Prediction of continuous phenotypes in mouse, fly, and rice genome wide association studies with support vector regression SNPs and ridge regression classifier," in 2015 IEEE 14th International Conference on Machine Learning and Applications (ICMLA), 2015: IEEE, pp. 1246-1250.
[28] M. Gönen and E. Alpaydın, "Multiple kernel learning algorithms," The Journal of Machine Learning Research, vol. 12, pp. 2211-2268, 2011.
[29] S. Qiu and T. Lane, "A framework for multiple kernel support vector regression and its applications to siRNA efficacy prediction," IEEE/ACM Transactions on Computational Biology and Bioinformatics, vol. 6, no. 2, pp. 190-199, 2008.
[30] C.-Y. Yeh, C.-W. Huang, and S.-J. Lee, "A multiple-kernel support vector regression approach for stock market price forecasting," Expert Systems with Applications, vol. 38, no. 3, pp. 2177-2186, 2011.
[31] Z. Xiang-rong, H. Long-ying, and W. Zhi-sheng, "Multiple kernel support vector regression for economic forecasting," in 2010 International Conference on Management Science & Engineering 17th Annual Conference Proceedings, 2010: IEEE, pp. 129-134.
[32] N. Verma, S. Das, and N. Srivastava, "Multiple kernel support vector regression for pricing nifty option," International Journal of Applied Mathematics Research, vol. 4, no. 4, p. 488, 2015.
[33] J. Xiao, C. Wei, and Y. Liu, "Speed estimation of traffic flow using multiple kernel support vector regression," Physica A: Statistical Mechanics and its Applications, vol. 509, pp. 989-997, 2018.
[34] 沈奕廷, "以支援向量迴歸模型評估晶圓級封裝之可靠度," 碩士論文, 國立清華大學動力機械工程學系, 2019.
[35] V. Cherkassky and Y. Ma, "Practical selection of SVM parameters and noise estimation for SVM regression," Neural networks, vol. 17, no. 1, pp. 113-126, 2004.
[36] L. Liu, B. Shen, and X. Wang, "Research on kernel function of support vector machine," in Advanced Technologies, Embedded and Multimedia for Human-centric Computing: Springer, 2014, pp. 827-834.
[37] M. Castro-Neto, Y. Jeong, M. K. Jeong, and L. D. Han, "AADT prediction using support vector regression with data-dependent parameters," Expert Systems with Applications, vol. 36, no. 2, pp. 2979-2986, 2009.
[38] H. Shafizadeh-Moghadam, A. Tayyebi, M. Ahmadlou, M. R. Delavar, and M. Hasanlou, "Integration of genetic algorithm and multiple kernel support vector regression for modeling urban growth," Computers, Environment and Urban Systems, vol. 65, pp. 28-40, 2017.
[39] L. Xiangdong, L. Bin, and C. Zhaoqian, "Optimal Model Selection for Support Vector Machines [J]," Journal of Computer Research and Development, vol. 4, p. 006, 2005.
[40] C.-H. Wu, G.-H. Tzeng, and R.-H. Lin, "A Novel hybrid genetic algorithm for kernel function and parameter optimization in support vector regression," Expert Systems with Applications, vol. 36, no. 3, pp. 4725-4735, 2009.
[41] B. Schölkopf, A. J. Smola, R. C. Williamson, and P. L. Bartlett, "New support vector algorithms," Neural computation, vol. 12, no. 5, pp. 1207-1245, 2000.
[42] S. J. Pan and Q. Yang, "A survey on transfer learning," IEEE Transactions on knowledge and data engineering, vol. 22, no. 10, pp. 1345-1359, 2009.
[43] J. Blitzer, R. McDonald, and F. Pereira, "Domain adaptation with structural correspondence learning," in Proceedings of the 2006 conference on empirical methods in natural language processing, 2006, pp. 120-128.
[44] H. Daumé III, "Frustratingly easy domain adaptation," arXiv preprint arXiv:0907.1815, 2009.
[45] B. Asefisaray, A. Haznedaroğlu, M. Erden, and L. M. Arslan, "Transfer learning for automatic speech recognition systems," in 2018 26th Signal Processing and Communications Applications Conference (SIU), 2018: IEEE, pp. 1-4.
[46] Z. Huang, S. M. Siniscalchi, and C.-H. Lee, "A unified approach to transfer learning of deep neural networks with applications to speaker adaptation in automatic speech recognition," Neurocomputing, vol. 218, pp. 448-459, 2016.
[47] Y. Cui, Y. Song, C. Sun, A. Howard, and S. Belongie, "Large scale fine-grained categorization and domain-specific transfer learning," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 4109-4118.
[48] P. Zhou, B. Ni, C. Geng, J. Hu, and Y. Xu, "Scale-transferrable object detection," in proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 528-537.
[49] A. Bouchachia, B. Gabrys, and Z. Sahel, "Overview of some incremental learning algorithms," in 2007 IEEE International Fuzzy Systems Conference, 2007: IEEE, pp. 1-6.
[50] L. Carbonara and A. Borrowman, "A comparison of batch and incremental supervised learning algorithms," in European Symposium on Principles of Data Mining and Knowledge Discovery, 1998: Springer, pp. 264-272.
[51] G. Cauwenberghs and T. Poggio, "Incremental and decremental support vector machine learning," in Advances in neural information processing systems, 2001, pp. 409-415.
[52] J. Ma, J. Theiler, and S. Perkins, "Accurate on-line support vector regression," Neural computation, vol. 15, no. 11, pp. 2683-2703, 2003.
[53] Z.-W. Li, J.-P. Zhang, and J. Yang, "A heuristic algorithm to incremental support vector machine learning," in Proceedings of 2004 International Conference on Machine Learning and Cybernetics (IEEE Cat. No. 04EX826), 2004, vol. 3: IEEE, pp. 1764-1767.
[54] J. Liu and E. Zio, "An adaptive online learning approach for Support Vector Regression: Online-SVR-FID," Mechanical Systems and Signal Processing, vol. 76, pp. 796-809, 2016.
[55] M. Martin, "On-line support vector machine regression," in European Conference on Machine Learning, 2002: Springer, pp. 282-294.
[56] B. Gu, V. S. Sheng, Z. Wang, D. Ho, S. Osman, and S. Li, "Incremental learning for ν-support vector regression," Neural networks, vol. 67, pp. 140-150, 2015.
[57] B. Gu, J.-D. Wang, Y.-C. Yu, G.-S. Zheng, Y.-F. Huang, and T. Xu, "Accurate on-line ν-support vector learning," Neural Networks, vol. 27, pp. 51-59, 2012.
[58] R. Cook, D. Malkus, M. Plesha, and R. Witt, "Concepts and Applications of Finite Element Analysis, Wiley," 2002.
[59] K. J. Bathe, Finite element procedures in engineering analysis. Prentice-Hall, 1982.
[60] W. N. Findley, J. Lai, and K. Onaran, "Creep and relaxation of nonlinear viscoelastic materials (with an Introduction to Linear Viscoelasticity). ," Amesterdam: North-Holland publishing Company, 1976.
[61] W. F. Chen and D. J. Han, Plasticity for structural engineers. J. Ross Publishing, 2007.
[62] N. E. Dowling, Mechanical behavior of materials: engineering methods for deformation, fracture, and fatigue. Pearson, 2012.
[63] J. L. Chaboche, "Constitutive equations for cyclic plasticity and cyclic viscoplasticity," International journal of plasticity, vol. 5, no. 3, pp. 247-302, 1989.
[64] J. L. Chaboche, "On some modifications of kinematic hardening to improve the description of ratchetting effects," International journal of plasticity, vol. 7, no. 7, pp. 661-678, 1991.
[65] A. Mahmoudi, S. Pezeshki-Najafabadi, and H. Badnava, "Parameter determination of Chaboche kinematic hardening model using a multi objective Genetic Algorithm," Computational Materials Science, vol. 50, no. 3, pp. 1114-1122, 2011.
[66] M. Awad and R. Khanna, Efficient learning machines: theories, concepts, and applications for engineers and system designers. Springer Nature, 2015.
[67] N. S. Altman, "An introduction to kernel and nearest-neighbor nonparametric regression," The American Statistician, vol. 46, no. 3, pp. 175-185, 1992.
[68] S.-L. Developers, "Scikit-Learn User Guide," Release 0.19, vol. 2, pp. 214-215, 2018.
[69] C.-C. Chang and C.-J. Lin, "LIBSVM: A library for support vector machines," ACM transactions on intelligent systems and technology (TIST), vol. 2, no. 3, pp. 1-27, 2011.
[70] B. Rogers and C. Scanlan, "Improving WLCSP reliability through solder joint geometry optimization," in International Symposium on Microelectronics, 2013, vol. 2013, no. 1: International Microelectronics Assembly and Packaging Society, pp. 546-550.
[71] M.-C. Hsieh and S.-L. Tzeng, "Solder joint fatigue life prediction in large size and low cost wafer-level chip scale packages," in 2014 15th International Conference on Electronic Packaging Technology, 2014: IEEE, pp. 496-501.
[72] M.-C. Hsieh, "Modeling correlation for solder joint fatigue life estimation in wafer-level chip scale packages," in 2015 10th International Microsystems, Packaging, Assembly and Circuits Technology Conference (IMPACT), 2015: IEEE, pp. 65-68.
[73] M. Motalab et al., "Thermal cycling reliability predictions for PBGA assemblies that include aging effects," in International Electronic Packaging Technical Conference and Exhibition, 2013, vol. 55751: American Society of Mechanical Engineers, p. V001T05A008.