人工智慧為當今模擬複雜之物理問題的方法之一，其中又以機器學習為主之應用也逐漸受到注目。本研究採機器學習之技術應用於生質料氣化產生之合成氣基於連續紀錄之特徵之預測、以及配合田口實驗法建立之L16直交表的離散特徵進行最大氫氣產量與最大氫氣與一氧化碳之比值並探討具備相似特徵並增加兩個連續溫度變數條件下機器學習之表現。最後亦探討在燃燒過程中產生之灰份之可燃分含量分析，透過影像作為輸入特徵進行當下操作條件產生之灰分所具備之可燃分含量比例區間，利用分類問題定義此項研究。在基於連續變數之氣化合成氣組成比例預測中，本研究採具不同複雜度之多種機器學習模型進行模擬，包含線性回歸、不同階數之多項式回歸與XGBoost等等，並驗證機器學習於氣化領域之可行性。而在基於田口實驗法設計之離散特徵研究中，離散變數之特徵處理與模型建立需經過更進一步的分析，且由於離散特性，機器較難以學習其輸入與輸出間之特徵，故本研究提出以類神經網路並搭配模型結構之細部調整，成功建立具備一定之泛化能力之模型。而在灰份分析之研究中，生質料經燃燒後產生之灰份透過照片之形式作為輸入特徵，進行燃燒後所含之可燃份區間預測，其不同區間之差異主要基於顏色差異，利用卷積神經網路進行特徵萃取與過濾進行學習，本研究亦探討多種方式提升模型表現，其結果顯示資料之純度對整體影響大於針對演算法之調整，並成功提出具備判斷灰份所含可燃份比例之模型。

The application of machine learning (ML) has been attracting more and more attention nowadays as the artificial intelligence (AI) is one of the methods to simulate the complicated physical problems. In this study, machine learning techniques were applied to the syngas prediction of biomass gasification based on continuous variables, and the maximum H2 and H2/CO ratio based on the discrete features of L16 orthogonal table described by Taguchi method. The similar conditions with the addition of 2 continuous variables were also discussed to observe the performance of the model. Moreover, the combustible matter of the combustion ash was also explored in this study via image features, aiming at predicting the interval of the combustible matter content after the combustion by classification techniques. In the case of the syngas prediction with continuous variables, a variety of models with different complexity including linear regression, polynomial regression with different orders and XGBoost, were adopted and the feasibility of the application of machine learning on the biomass gasification was successfully validated. However, in the case of the discrete variables based on Taguchi method, the features and the construction of models were required to be processed carefully and it was difficult for the machine to learn the relationship between inputs and outputs. The artificial neural network with fine-tuning of hyperparameters were proposed in this study, and eventually a model was successfully created with validated generalization. In addition, in the case of combustion ash analysis, the ash images from the combustion of biomass were used as the input feature to predict the interval of combustible matter, which was mainly based on the difference of color and for this task the convolutional neural network was adopted. Thus, in this study, different methods were proposed to enhance the performance and the results showed that the data purity contributed the most to the proposed model with the ability to predict the combustible matter content accurately.

[1] Scripps Institute of Oceanography, NOAA Earth System Research Laboratory, https://www.esrl.noaa.gov/gmd/ccgg/trends/.
[2] S. Mishra and R. Upadhyay. Review on biomass gasification: Gasifiers, gasifying mediums, and operational parameters, Materials Science for Energy Technologies, Vol. 4, pp. 329-340, 2021.
[3] P. H. Pandey, T. M. UkardeJyoti, S. Mahale, A. Vasishta, and H. S. Pawar, Biomass. Chapter 9 - Syngas production via biomass gasification, Biofuels, Biochemicals, pp. 211-261, 2022.
[4] 經濟部能源局, 110年能源發展綱領, [Online] Available: https://www.moeaboe.gov.
tw/ecw/populace/content/ContentDesc.aspx?menu_id=61
[5] 行政院環境保護署,鍋爐空氣污染物排放標準, [Online] Available: https://enews.epa.
gov.tw/DisplayFile.aspx?FileID=E85B6C6C26A3D755
[6] R. Saidur, E. A. Abdelaziz, A. Demirbas, M. S. Hossaina, and S. Mekhilef. A review on biomass as a fuel for boilers, Renewable and Sustainable Energy Reviews, Vol. 15, pp. 2262-2289, 2011.
[7] 經濟部能源局, 再生能源發展條例, [Online] Available: https://www.re.org.tw/infor
mation/more.aspx?cid=193&id=3392
[8] P. Basu. Biomass Gasification, Pyrolysis and Torrefaction - Practical design and theory, 2013.
[9] X. Jiang, D. Chen, Z. Ma, and J. Yan. Models for the combustion of single solid fuel particles in fluidized beds: A review, Renewable and Sustainable Energy Reviews, Vol. 68, pp. 410-431, 2017.
[10] K. Ishii. Advanced pulverized coal injection technology and blast furnace, 2000.
[11] A. A. A. Abuelnuor, M. A. Wahid, S. E. Hosseini, A. Saat, K. M. Saqr, H, H, Sait, and M. Osman. Characteristics of biomass in flameless combustion: A review. Renewable and Sustainable Energy Reviews, Vol. 33, pp. 363-370, 2014.
[12] W. -H. Chen, B. -J. Lin, Y. -Y. Lin, Y. -S. Chu, A. T. Ubando, P. L. Show, H. C. Ong, J. -S. Chang, S. -H. Ho, A. B. Culaba, A. Pétrissans, and M. Pétrissans. Progress in biomass torrefaction: Principles, applications and challenges, Progress in Energy and Combustion Science, Vol. 82, 100887, 2021.
[13] K. D. Wey. Refuse-derived fuel as a secondary Energy in Taiwan-using hotelling space allocation model. Resources Policy, Vol. 31, pp. 204-210, 2006.
[14] 戴文堅, 李宗憲, 固態廢棄物衍生燃料技術之應用評估, 資源與環境研討會, 2003.
[15] Y. Niu, H. Tan, and S. Hui. Ash-related issues during biomass combustion: Alkali-induced slagging, silicate melt-induced slagging (ash fusion), agglomeration, corrosion, ash utilization, and related countermeasures, Progress in Energy and Combustion Science, Vol. 52, pp. 1-61, 2016.
[16] A. A. Khan, W. D. Jonga, P. J. Jansens, and H. Spliethoff. Biomass combustion in fluidized bed boilers: potential problems and remedies, Fuel Processing Technology, Vol. 90, pp. 21-50, 2009.
[17] W. S. McCulloch and W. Pitts. A logical calculus of the ideas immanent in nervous activity, The Bulletin of Mathematical Biophysics, Vol. 5, pp. 115-133, 1943.
[18] L. Wang, J. Liu, and F. Qian. Wind speed frequency distribution modeling and wind energy resource assessment based on polynomial regression model, International Journal of Electrical Power and Energy Systems, 106964, 2021.
[19] S. M. H. Luk, E. C. Ford, M. H. Phillips, and A. M. Kalet. Improving the quality of care in radiation oncology using artificial intelligence, Clinical Oncology, Vol. 34, pp. 89-98, 2022.
[20] P. Chhajer, M. Shah, and A. Kshirsagar. The applications of artificial neural networks, support vector machines, and long–short term memory for stock market prediction, Decision Analytics Journal, Vol. 2, 100015, 2022.
[21] A. A. M. Abadi, P. Barham, E. Brevdo, Z. Chen, C. Citro, G. -S. Corrado, A. Davis, J. Dean, M. Devin, S. Ghemawat, I. Goodfellow, A. Harp, G. Irving, M. Isard, Y. Jia, R. Jozefowicz, L. Kaiser, M. Kudlur, J. Levenberg, D. Man, R. Monga, S. Moore, D. Murray, C. Olah, M. Schuster, J. Shlens, B. Steiner, I. Sutskever, K. Talwar, P. Tucker, V. Vanhoucke, V. Vasudevan, F. Vigas, O. Vinyals, P. Warden, M. Wattenberg, M. Wicke, Y. Yu, and X. Zheng. {TensorFlow}: Large-Scale Machine Learning on Heterogeneous Systems, 2015.
[22] F. Chollet. Keras, GitHub repository, 2015, [Online] Available: https://github.com/fcho
llet/keras.
[23] D. Adams, D. -H. Oh, D. -W. Kim, C. -H/ Lee, and M. Oh. Prediction of SOx-NOx emission from a coal-fired CFB power plant with machine learning: Plant data learned by deep neural network and least square support vector machine, Journal of Cleaner Production, Vol. 270, 122310, 2020.
[24] A. Saxena, M. Prasad, A. Gupta, N. Bharill, O. P. Patel, A. Tiwari, E. M. Joo, D. Weiping, and L. C. -Teng. A review of clustering techniques and developments, Neurocomputing, Vol. 267, pp. 664-681, 2017.
[25] L. van der Maaten, E. Postma, and J. van den Herik. Dimensionality Reduction: A Comparative Review, Journal of Machine Learning Research, 2007.
[26] N. Hanchate, S. Ramani, C. S. Mathpati, and V. H. Dalvi. Biomass gasification using dual fluidized bed gasification systems: A review, Journal of Cleaner Production, Vol. 280, 123148, 2021.
[27] L. M. Quan, H. Kamyab, A. Yuzir, V. Ashokkumar, S. E, Hosseini, B. Balasubramanian, and I. Kirpichnikova. Review of the application of gasification and combustion technology and waste-to-energy technologies in sewage sludge treatment, Fuel, Vol. 316, 123199, 2022.
[28] A. Ramosa, E. Monteirob, V. Silva, and A. Roubo. Co-gasification and recent developments on waste-to-energy conversion: A review, Renewable and Sustainable Energy Reviews, Vol. 81, pp. 380-398, 2018.
[29] M. P. -Gamero, D. T. Pio, L. A. C. Tarelho, P. S´anchez, and L. S. -Silva. Simulation of biomass gasification in bubbling fluidized bed reactor using aspen plus®, Energy Conservation and Management, Vol. 235, 113981, 2021.
[30] M. Niu, J. Xie, S. Liang, and L. Liu. Simulation of a new biomass integrated gasification combined cycle (BIGCC) power generation system using Aspen Plus: Performance analysis and energetic assessment, International Journal of Hydrogen Energy, Vol. 46, pp. 22356-22367, 2021.
[31] Z. Košnář, F. Mercl, I. Perná, and P. Tlustoš, Investigation of polycyclic aromatic hydrocarbon content in fly ash and bottom ash of biomass incineration plants in relation to the operating temperature and unburned carbon content, Science of The Total Environment, Vol. 563-564, pp. 53-61, 2016.
[32] Standard Methods for Instrumental Determination of Carbon, Hydrogen, and Nitrogen in Laboratory Samples of Coal and Coke, [Online] Available: https://archive.org/detail
s/gov.law.astm.d5373.1993/page/n3/mode/2up
[33] ASTM-E1131, Standard Test Method for Compositional Analysis by Thermogravimetry, [Online] Available: https://www.document-center.com/standards/
show/ASTM-E1131
[34] J. Chen, Y. Wang, X. Lang, X. E. Ren, and S. Fan. Evaluation of agricultural residues pyrolysis under non-isothermal conditions: Thermal behaviors, kinetics, and thermodynamics, Bioresource Technology, Vol. 241, pp. 340-348, 2017.
[35] Q. Wang, E. Wang, K. Li, N. Husnain, and D. Li. Synergistic effects and kinetics analysis of biochar with semi-coke during CO2 co-gasification, Energy, Vol. 191, pp. 116528, 2020.
[36] T. Kujirai and T. Akahira. Effect of temperature on the deterioration of fibrous insulating materials. Sci Paper Inst Phys Chem Res., Vol. 2, pp. 23–52, 1925.
[37] J. H. Flynn and L. A. Wall. A quick, direct method for determination of activation energy from thermogravimetric data, J Polym Sci B, Vol. 4, pp. 323-328, 1966.
[38] T. Akahira and T. Sunose. Method of determining activation deterioration constant of electrical insulating materials, Res Rep Chiba Inst Technol (Sci.Technol), Vol. 16, pp. 22-31, 1971.
[39] S. R. Naqvi, R. Tariq, Z. Hameed, I. Ali, S. A. Taqvi, M. Naqvi, M. B. K. Niazi, T. Noor, and W. Farooq. Pyrolysis of high-ash sewage sludge： Thermo-kinetic study using TGA and artificial neural networks, Fuel, Vol. 233, pp. 529-538, 2018.
[40] S. Vyazovkin, A. K. Burnham, J. M. Criado , L. A. Perez-Maqueda, C, Popescu, and N. Sbirrazzuoli. ICTAC Kinetics Committee recommendations for performing kinetic computations on thermal analysis data, Thermochim Acta, Vol. 520, pp. 1–19, 2011.
[41] C. D. Doyle. Series approximations to the equation of thermogravimetric data, Nature, Vol. 207, pp. 290-291, 1965.
[42] D. Baruah, D. C. Baruah, and M. K. Hazarika. Artificial neural network-based modeling of biomass gasification in fixed bed downdraft gasifiers, Biomass and Bioenergy, Vol. 98, pp. 264-271, 2017.
[43] W. Zeng, T. Lu, Z. Liu, Q. Xu, H. Peng, C. Li, S. Yang, and F. Yao. Research on a laser ultrasonic visualization detection method for human skin tumors based on pearson correlation coefficient, Optics & Laser Technology, Vol. 141, 107117, 2021.
[44] T. -L. Fang, The study of co-gasification characteristics of sewage sludge and palm kernel shell, [Unpublished master’s thesis]. National Cheng Kung University, 2019.
[45] G. -B. Chen, F. -H. Wu, T. -L. Fang, and T. -H. Lin. A study of co-gasification of sewage sludge and palm kernel shells, Energy, Vol. 218, 119532, 2021.
[46] D. Berrar. Cross-validation, Encyclopedia of Bioinformatics and Computational Biology Vol. 1, pp. 542-545, 2019.
[47] M. K. Kim, Y. -S. Kim, and J. Srebric. Predictions of electricity consumption in a campus building using occupant rates and weather elements with sensitivity analysis: Artificial neural network vs. linear regression, Sustainable Cities and Society Vol. 62, 102385, 2020.
[48] A. M. Abubaker, A. D. Ahmad, B. B. Singh, N. K. Akafuah, and K. Saito. Multi-objective linear-regression-based optimization of a hybrid solar-gas turbine combined cycle with absorption inlet-air cooling unit, Energy Conversion and Management Vol. 240, 114266, 2021.
[49] M. K. Kim, Y. -S. Kim, and J. Srebric. Predictions of electricity consumption in a campus building using occupant rates and weather elements with sensitivity analysis: Artificial neural network vs. linear regression, Sustainable Cities and Society Vol. 62, 102385, 2020.
[50] A. M. Abubaker, A. D. Ahmad, B. B. Singh, N. K. Akafuah, and K. Saito. Multi-objective linear-regression-based optimization of a hybrid solar-gas turbine combined cycle with absorption inlet-air cooling unit, Energy Conversion and Management Vol. 240, 114266, 2021.
[51] A. Cutler, D. R. Cutler, and J. R Stevens. Ensemble Machine Learning: Methods and Applications, Chap. 5, Springer, pp. 157-176, 2011.
[52] Y. Mutlu and O. Yucel. An artificial intelligence based approach to predicting syngas composition for downdraft biomass gasification, Energy Vol. 165, pp. 895-901, 2018.
[53] T. Chen and C. Guestrin. XGBoost: A Scalable Tree Boosting System, Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 785–794, 2016.
[54] M. Ma, G. Zhao, B. He, Q. Li, H. Dong, S. and Wang, Z. Wang, XGBoost-based method for flash flood risk assessment, Journal of Hydrology Vol. 598, 126382, 2021.
[55] P. Trizoglou, X. Liu, and Z. Lin, Fault detection by an ensemble framework of Extreme Gradient Boosting (XGBoost) in the operation of offshore wind turbines, Renewable Energy Vol. 179, 945-962, 2021.
[56] I. Goodfellow and Y. Bengio, A. Courville. "Deep learning, " MIT press, 2016.
[57] G. Klambauer, T. Unterthiner, and A. Mayr. Self-Normalizing Neural Networks, arXiv:1706.02515 [cs.LG].
[58] D. E. Rumelhart, G. E. Hinton, and R. J. Williams. Learning representations by back-propagating errors, Nature, Vol. 323, pp. 533-536, 1986.
[59] K. O'Shea and R. Nash. An Introduction to Convolutional Neural Networks, arXiv:1511.08458 [cs.NE], 2015.
[60] M. Sandler, A, Howard, M. Zhu, A. Zhmoginov, and L. -C. Chen. MobileNetV2: Inverted Residuals and Linear Bottlenecks, The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 4510-4520, 2018.
[61] S. Hochreiter and J. Schmidhuber. Long Short-term Memory, Neural Computation, Vol. 9, pp. 1735-1780, 1997.
[62] W. Zhang, Y. He, Y. Wang, G. Li, J. Chen, and Y. Zhu. Comprehensive investigation on the gasification reactivity of pyrolysis residue derived from Ca-rich petrochemical sludge: Roles of microstructure characteristics and calcium evolution, Energy Conversion and Management, Vol. 253, pp. 115-150, 2022.
[63] W. Zhang, Y. He, Y. Wang, G. Li, J. Chen, and Y. Zhu. Comprehensive investigation on the gasification reactivity of pyrolysis residue derived from Ca-rich petrochemical sludge: Roles of microstructure characteristics and calcium evolution, Energy Conversion and Management, Vol. 253, 115150, 2022.
[64] A. Ramos, E. Monteiro, V. Silva, and A. Rouboa, Co-gasification and recent developments on waste-to-energy conversion: A review, Renewable and Sustainable Energy Reviews, Vol. 81, pp. 380-398, 2018.
[65] D. Chi, Research on electricity consumption forecasting model based on wavelet transform and multi-layer LSTM model, Energy Report, Vol. 8, pp. 220-228, 2022.
[66] F. Zhou, Z. Huang, and C. Zhang. Carbon price forecasting based on CEEMDAN and LSTM, Applied Energy, Vol. 311, 118601, 2022.
[67] Y. Chen, Y. Wang, Z. Dong, J. Sua, Z. Han, D. Zhou, Y. Zhao, and Y. Bao. 2-D regional short-term wind speed forecast based on CNN-LSTM deep learning model, Energy Conversion and Management, Vol. 244, 114451, 2021.
[68] R. R. Selvaraju, Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization, arXiv:1610.02391 [cs.NE], 2017.