臺科大化工系所建立的試驗廠級乙酸異丙酯反應蒸餾，在進行啟動與傳統PID控制產能擾動實驗時，發現達到穩定狀態需花費很長的時間，且產生許多次級品。因此本研究將以縮短此兩種實驗達穩定狀態之時間為目標進行操作策略的研究。
對於啟動實驗，本研究在相同的啟動程序下，同時探討：再沸器凝液罐液位為40%與15%的情況，並依達穩定狀態最短時間液位為最適啟動個案。而試驗廠達穩態條件為產品濃度達98wt%且維持兩小時以上。結果顯示，起始液位40%達穩態所需時間約10小時，而以15%作為啟動起始液位能於6小時內達到穩態。
為了縮短產能擾動後的排除時間，本研究建立一套模型預測控制(Model Predictive Control, MPC)架構，根據試驗廠當前的數據提供溫度設定點操作建議值。本研究採用GRU模型作為MPC之內部預測模型，但試驗廠實驗數據量不足於模型訓練，因此先建立與試驗廠數據相符之Aspen Plus穩態與Aspen Plus Dynamics動態模型，並利用動態模型生產大量虛擬數據於GRU模型訓練。此GRU的模型輸入變數為：異丙醇進料流率、第2板與第12板溫度設定點、第2板與第12板溫度當前值，輸出變數為側流產品之乙酸異丙酯濃度。先將MPC連接於動態模型進行驗證，再連接於試驗廠並探討MPC與PID排除擾動之差異。以模擬驗證結果顯示，GRU模型之測試結果MPAE可達0.02%以下，MPC於產能擾動中，產品濃度與濃度設定值積分誤差IAE約為0.0018，達到穩定時間相較於PID控制縮短5小時以上。從實驗結果顯示，於產能擾動中，PID控制達到穩定時間為10小時且過程中產品最低濃度下降至97wt%，MPC可於5小時內穩定並維持產品濃度在98wt%以上。

For a reactive distillation pilot plant, it is found that the time span for start-up operation and to reach steady-state (S.S.) after disturbances is quite long. Hence, this study aims to find out the optimal strategies for these two situation to shorten the operation time.
Based on the same start-up procedure, this study starts from two different initial sunp level: 40 % and 15 %, respectively. The results show that only 6 hours is needed to finish the start-up procedure with 15 % of initial sump level while the other one takes 10 hours.
To shorten the time to reach S.S. under disturbances, Model Predictive Control (MPC) is developed to provide the operation guidance. To build MPC, Gated Recurrent Unit (GRU) is selected as the internal prediction model. Due to the insufficient experimental data, Aspen Plus Dynamics is used to produce virtual data which will be used for training in GRU. The input variables are: the feed flow rate of isopropanol, the 2nd and 12th stage set points of the temperature controllers, the 2nd and 12th stage process variable of temperature controllers. The output variable is IPAc mole faction of the product. Next, Python is used for building MPC. GRU model as the input of MPC connected with the dynamic models and pilot plant, and the results would be compared with that of traditional PID control.
The results show that the mean absolute percentage error (MAPE) of the testing set of GRU model can be less than 0.02 %. After ±10 % throughput disturbances, the result of dynamic model with MPC shows a low integral absolute error (IAE) of the product putity, which is about 0.0018. The differences of the product purity after disturbances with MPC are half of those only with PID controllers, and the time needed to reach S.S. can be shortened at least 5 hours. For the pilot plant with PID control, it takes 10 hours to stabilize system under dusturbances, and the product purity decreases to 96.9 wt%. As for the plant with MPC, the system approaches stable within 5 hours, and IPAc mole faction of the product maintains above at 98 wt%.

【English】
1. Backhaus, A. A. (1921). U.S. Patent No. 1,400,849. Washington, DC: U.S. Patent and Trademark Office.
2. Benz, S. J., & Scenna, N. J. (1997). The influence of different models in obtaining multiple solutions in distillation columns. The Canadian Journal of Chemical Engineering, 75(6), 1142-1152.
3. Cutler, C. R., & Ramaker, B. L. (1980). Dynamic matrix control – a computer control algorithm. In joint automatic control conference (No. 17, p. 72).
4. Chakrabarti, A.,  Sharma, M. M. (1993). Reactive polymers, 20(1-2), 1-45.
5. Cho, K., Van Merriënboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., & Bengio, Y. (2014). Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv preprint arXiv:1406.1078.
6. Doherty, M. F.,  Perkins, J. D. (1978a). On the dynamics of distillation processes the simple distillation of multicomponent non-reacting, homogeneous liquid mixtures. Chemical engineering science, 33(3), 281-301.
7. Elsisi, M. (2019). New design of adaptive model predictive control for energy conversion system with wind torque effect. Journal of cleaner production, 240(10), 118265.
8. Gadewar, S. B., Malone, M. F., & Doherty, M. F. (2002). Feasible region for a countercurrent cascade of vapor‐liquid CSTRS. AIChE journal, 48(4), 800- 814.
9. Ganguly, S., & Saraf, D. N. (1993). Startup of a distillation column using nonlinear analytical model predictive control. Industrial & engineering chemistry research, 32(8), 1667-1675.
10. Garcia, C. E., Prett, D. M., & Morari, M. (1989). Model predictive control: Theory and practice—A survey. Automatica, 25(3), 335-348.
11. Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8), 1735-1780.
12. Henson, M. A., & Seborg, D. E. (1997). Nonlinear process control (pp. 5-8). Upper Saddle River, New Jersey: Prentice Hall PTR.
13. Hussein, A. S., El-amary, N. H., Abdelaziz, A. A.,  Swief, R. A. (2020). Model predictive control for hybrid AC/DC micro-grids. 2020 International Conference on Innovative Trends in Communication and Computer Engineering (ITCE), 365-369.
14. Kumar, A. S., & Ahmad, Z. (2012). Model predictive control (MPC) and its current issues in chemical engineering. Chemical Engineering Communications, 199(4), 472-511.
15. Liu, L., Jiang, H., He, P., Chen, W., Liu, X., Gao, J., & Han, J. (2019). On the variance of the adaptive learning rate and beyond. arXiv preprint arXiv:1908.03265.
16. Liu, W. T., & Tan, C. S. (2001). Liquid-phase esterification of propionic acid with n-butanol. Industrial & engineering chemistry research, 40(15), 3281- 3286.
17. Lai, I. K., Liu, Y. C., Yu, C. C., Lee, M. J., & Huang, H. P. (2008). Production of high-purity ethyl acetate using reactive distillation: experimental and start-up procedure. Chemical Engineering and Processing: Process Intensification, 47(9-10), 1831-1843.
18. Maki-Arvela, P., Salmi, T., Sundell, M., Ekman, K., Peltonen, R.,  Lehtonen, J. (1999). Applied catalysis A：general, 184(1), 25-32.
19. Ma, J., & Yarats, D. (2018). Quasi-hyperbolic momentum and adam for deep learning. arXiv preprint arXiv:1810.06801.
20. Mikael, Y. N., Niklas, A., Bernt, N., Mark, M. H.,  Oleg, P. (2020). Chemical Engineering Research and Design, 155, 12-17.
21. Pham, H. N.,  Doherty, M. F. (1990). Design and synthesis of heterogeneous azeotropic distillations-II residue curve maps. Chemical engineering science, 45(7), 1837-1843.
22. Powell, M. J. (1994). A direct search optimization method that models the objective and constraint functions by linear interpolation. In Advances in optimization and numerical analysis (pp. 51-67). Springer, Dordrecht.
23. Ruiz, C. A., Cameron, I. T., & Gani, R. (1988). A generalized dynamic model for distillation columns—III. Study of startup operations. Computers & chemical engineering, 12(1), 1-14.
24. Reepmeyer, F., Repke, J. U., & Wozny, G. (2003). Analysis of the start-up process for reactive distillation. Chemical Engineering & Technology, 26(1), 81-86.
25. Reepmeyer, F., Repke, J. U., & Wozny, G. (2004). Time optimal start-up strategies for reactive distillation columns. Chemical Engineering Science, 59(20), 4339-4347.
26. Richalet, J., Rault, A., Testud, J. L., & Papon, J. (1976). Algorithmic control of industrial processes. 4th IFAC symposium on identification and system parameter estimation, 1119-1167.
27. Scenna, N. J., Ruiz, C. A., & Benz, S. J. (1998). Dynamic simulation of start-up procedures of reactive distillation columns. Computers & chemical engineering, 22, S719-S722.
28. Siirola, J. J. (1996). Industrial application of chemical process synthesis. AICHE journal, 23, 1-62.
29. Shalmani, R. A., Rahmani, M.,  Bigdeli, N. (2020). Nash-based robust distributed model predictive control for large-scale systems. Journal of process control, 88, 43-53.
30. Seborg, D. E., Edgar, T. F., Mellichamp, D. A., Doyle, F. J. (2017). Process Dynamics and Control. Wiley.
31. Taylor, R.,  Krishna, R. (2000). Modelling reactive distillation. Chemical engineering science, 55(22), 5183-5229.
32. Tang, Y. T., Chen, Y. W., Huang, H. P., Yu, C. C., Hung, S. B., & Lee, M. J. (2005). Design of reactive distillations for acetic acid esterification. AIChE Journal, 51(6), 1683-1699.
33. Tsatse, A., Oudenhoven, S. R. G., Kate, A. J. B. T.,  Sorensen, E. (2022). An investigation of the interactions between system characteristics and controllability for reactive distillation system. Chemical engineering and processing：process intensification, 171, 108712.
34. Vathna, A., Currie, J.,  Wilson, D. I. (2017). 2017 6th International Symposium on Advanced Control of Industrial Processes, 487-492
35. Wahid, A.,  Putra, I. G. E. P. (2018). Multivariable model predictive control design of reactive distillation column for dimethyl ether production. IOP conference series：materials science and engineering, 334(1), 012018.
36. Woldemicheal, D. E.,  Woldeyohannes, A. D. (2016). ARPN journal of engineering and applied sciences, 11(24), 14264-14268.
37. Yasuoka, H., Nakanishi, E.,  Kunigita, E. (1987). Design of an on-line startup system for a distillation column based on a simple algorithm. International chemical engineering, 27(3), 466-472.
38. Yasuoka, H.,  Nakanishi, E. (1982). Convenient algorithm for the on-line optimum startup operation of binary distillation column. Memoirs of the faculty of engineering, (29), 161-169.
39. Zarzycki, K.,  Lawrynczuk, M., (2021). LSTM and GRU Neural Networks as Models of Dynamical Processes Used in Predictive Control: A Comparison of Models Developed for Two Chemical Reactors. Sensors, 21(16), 5625.

【中文】
1. 林瀚淵，2007石化工業年鑑經濟部 ITIS 專案辦公室，臺北市，2007。
2. 林政綱，「試驗場級乙酸異丙酯反應蒸餾程序啟動策略與擾動排除指引之研究」，國立臺灣科技大學化學工程研究所，碩士論文，2020。
3. 陳建穎，「使用類神經網路及模式預測控制於乙酸異丙酯反應蒸餾製程之控制研究」，國立臺灣科技大學化學工程研究所，碩士論文，2020。
4. 薛宇玲，「使用非均相觸媒Amberlyst39之乙酸異丙酯合成反應動力行為研究」，國立臺灣科技大學化學工程研究所，碩士論文，2019。
5. 黃敬瑋，「試驗廠級乙酸異丙酯反應蒸餾製程之類神經網路模型建置與干擾排除指引研究」，國立臺灣科技大學化學工程研究所，碩士論文，2021。
【網頁】
1. 行政院環境保護署列管汙染源資料查詢系統網頁,
https://prtr.epa.gov.tw/
2. 台灣積體電路製造股份有限公司電-廢棄物管理官方網頁,
https://esg.tsmc.com/ch/focus/greenManufacturing/wasteManagement.html
3. 中國交易平台-生意社價格網頁,
https://www.100ppi.com/monitor/