近年來，在網路以及資料儲存科技的快速發展下，企業也開始也開始使用資料科學的方式來提升自己產品的良率和品質，以滿足顧客當下以及未來的需求。運用資料科學的幫助，可讓人們從巨量的資料裡找出其中關鍵的重要資訊，而類神經網路可以自行建造出一個複雜的模型使其滿足人類所需之預測函數，再將其套入基因演算法此種啟發式演算法中，以找出最佳之製程參數組合以滿足需求。
在本研究中，透過分析各種影響熱塑性聚胺酯產品耐磨耗之因子，並運用類神經網路與基因演算法，成功找出提升耐磨耗之最佳參數組合，使其能有效的降低此產品之磨耗率。相較於傳統的實驗設計、田口方法、反應曲面法，運用神經網路與基因演算法，只需要過去的歷史資料而不需新增其他實驗資料，便可找出最佳的參數組合，使用此方法能獲得比傳統方法更低之磨耗率，使最低磨耗率由34.4提升至10.08，大幅提升了產品之品質，此外，本研究所建議的方法亦將低了公司所需之實驗次數，幫助公司節省成本以及增加生產之彈性。預估每年可幫公司減少360萬新台幣之支出。

In recent years, with the development of the Internet and data storage technology, companies begin to use data science methods to improve the yield and quality of their products to meet the current and future needs of customers. With the help of data science, people can find out the key and important information from a huge amount of data, and the neural network can build a complex model by itself to fit the prediction function required by humans, and then apply it into the heuristic algorithm like genetic algorithm to find the best combination of parameters to meet the demand.
In this study, by analyzing various factors that affect the wear resistance of thermoplastic polyurethane products, and using neural network and genetic algorithms, we successfully found the best combination of process parameters to improve thermoplastic polyurethane’s abrasion, so that it can effectively reduce the product’s abrasion rate. Compared the difference with the neural network combine genetic algorithm and the traditional methods like design of experiment (DOE), Taguchi methods, and response surface method, using neural network and genetic algorithm only need past historical data without adding other experimental data to find the best combination of manufacturing parameters, can achieve and lower abrasion rate than traditional methods, greatly improving the quality of the product. The best performance of abrasion loss improves from 34.4 to 10.08. This method can also reduce the number of experiments required by the company, helping the company save costs and increase production flexibility. The company estimate can save 3,600,000 New Taiwan Dollars (NTD) per year.

Table of Contents
1. Introduction…………………………………………………………………………...8
1.1 Background………………………… …………………………………………8
1.2 Purposes…………………………………………………………..……………9
2. Related Work…...……………………………………………………………………10
2.1 Neural Network…………………...…………………………………………..10
2.2 Genetic Algorithm……...……………………………………………..………13
2.3 Integration of Neural Network and Genetic Algorithm…………………….16
3. Proposed Method………..…………………………………………………...……18
4. Case Study………………………………………………………………...…………24
4.1 Case Description…..………………………………………………..………24
4.2 Data Collection…………………........………………………………..……26
4.3 Data Preprocessing…………….……………………………………………26
4.4 Construction of Neural Network Model..…………………………….……32
4.5 Optimize Factors by Using Genetic Algorithm…………………………..…39
4.6 Results and Discussion……..……………………………………………......41
5. Conclusion ………………………………………………………………….………42
5.1 Conclusion……………………………………………………………………42
5.2 Future Study……………………………………………………………….….43
References………………………………………………………………………………44

References
[1] Balasubramanian, M., Paglicawan, M. A., Zhang, Z. X., Lee, S. H., Xin, Z. X., & Kim, J. K. (2008). Prediction and optimization of mechanical properties of polypropylene/waste tire powder blends using a hybrid artificial neural network-genetic algorithm (GA-ANN). Journal of Thermoplastic Composite Materials, 21(1), pp. 51-69.
[2] Bengio, Y. (2007). Learning deep architectures for AI. Technical Report 1312, Université de Montreal.
[3] Boger, Z., & Guterman, H. (1997). Knowledge extraction from artificial neural network models. In 1997 IEEE International Conference on Systems, Man, and Cybernetics. Computational Cybernetics and Simulation, 4, pp. 3030-3035.
[4] Bouktif, S., Fiaz, A., Ouni, A., & Serhani, M. A. (2018). Optimal deep learning lstm model for electric load forecasting using feature selection and genetic algorithm: Comparison with machine learning approaches. Energies, 11(7), p. 1636.
[5] Brynjolfsson E. and McAfee A. (2012). Big data: the management revolution. Harvard business review, 90(10), pp. 60-68.
[6] Fausett, F. (1994). Fundamental of Neural Network: Architectures, Algorithms, and Applications.
[7] Gaidhane, R., Vaidya, C., & Raghuwanshi, M. (2014). Intrusion detection and attack classification using back-propagation neural network. International Journal of Engineering Research & Technology, 3(3), pp. 1112-1115.
[8] Gröger, C., Niedermann, F., & Mitschang, B. (2012). Data mining-driven manufacturing process optimization. In Proceedings of the world congress on engineering, 3, pp. 4-6.
[9] Jain, A. K., Mao, J., & Mohiuddin, K. M. (1996). Artificial neural network: A tutorial. Computer, 29(3), pp. 31-44.
[10] Karimi, H., & Yousefi, F. (2012). Application of artificial neural network–genetic algorithm (ANN–GA) to correlation of density in nanofluids. Fluid Phase Equilibria, 336, pp.79-83.
[11] Kobayashi, T., & Simon, D. L. (2005). Hybrid neural-network genetic-algorithm technique for aircraft engine performance diagnostics. Journal of Propulsion and Power, 21(4), pp. 751-758.
[12] Mitchell, M. (1998). An introduction to genetic algorithms. MIT press.
[13] Panchal, G., Ganatra, A., Kosta, Y. P., & Panchal, D. (2011). Behaviour analysis of multilayer perceptrons with multiple hidden neurons and hidden layers. International Journal of Computer Theory and Engineering, 3(2), pp. 332-337.
[14] Shen, C., Wang, L., & Li, Q. (2007). Optimization of injection molding process parameters using combination of artificial neural network and genetic algorithm method. Journal of materials processing technology, 183(2-3), pp. 412-418.
[15] Su, C. T. (2013). Quality engineering: off-line methods and applications. CRC press.
[16] Su, C. T., Lin, C. M., & Chang, C. A. (2012). Optimization of the bistability property for flexible display by an integrated approach using Taguchi methods, neural networks and genetic algorithms. Microelectronics Reliability, 52(7), pp. 1492-1500.

[17] Su, C. T., Chiu, C. C., & Chang, H. H. (2000). Parameter design optimization via neural network and genetic algorithm. International Journal of Industrial Engineering: Theory, Applications and Practice, 7(3), pp. 224-231.
[18] Su, C. T., Yeh, C. J., Chou, C. J., & Chang, C. A. (2014). An experimental approach to enhance Cu wire bonding yield through parameter optimization. Experimental Techniques, 38(3), pp. 29-36.
[19] Tsai, J. T., Chou, J. H., & Liu, T. K. (2006). Tuning the structure and parameters of a neural network by using hybrid Taguchi-genetic algorithm. IEEE Transactions on Neural Network, 17(1), pp. 69-80.
[20] Whitley, D. (1994). A genetic algorithm tutorial. Statistics and computing, 4(2), pp. 65-85.
[21] Widrow, B., Winter, R. G., & Baxter, R. A. (1987). Learning phenomena in layered neural network. In Proceedings of the IEEE First International Conference on Neural Network, 2, pp. 411-430.
[22] Zhang, H-C., and S. H. Huang. (1995). Applications of neural network in manufacturing: a state-of-the-art survey. The International Journal of Production Research, 33(3), pp. 705-728.
[23] Zhou, C. C., Yin, G. F., & Hu, X. B. (2009). Multi-objective optimization of material selection for sustainable products: artificial neural network and genetic algorithm approach. Materials & Design, 30(4), pp. 1209-1215.