关键词: Animal waste Environmental effect Machine learning Methane Regression XGBR algorithm

Mesh : Animals Cattle Biofuels Manure Poultry Ruminants Turkey

来  源:   DOI:10.1007/s11356-024-32666-7   PDF(Pubmed)

Abstract:
Animal waste can be converted into a renewable energy source using biogas technology. This process has an impact on greenhouse gas emissions and is a sustainable source of energy for countries. It can reduce the effects of climate change and protect the planet for future generations. Tier1 and tier2 approaches are commonly used in the literature to calculate emissions factors. With boosting algorithms, this study estimated each animal category\'s biogas potential and CH4 emissions (tier1 and tier2 approach) for 2004-2021 in all of Turkey\'s provinces. Two different scenarios were created in the study. For scenario-1, the years 2020-2021 were predicted using data from 2004 to 2019, while for scenario-2, the years 2022-2024 were predicted using data from 2004 to 2021. According to the scenario-1 analysis, the eXtreme Gradient Boosting Regressor (XGBR) algorithm was the most successful algorithm with an R2 of 0.9883 for animal-based biogas prediction and 0.9835 and 0.9773 for animal-based CH4 emission predictions (tier1 and tier2 approaches) for the years 2020-2021. When the mean absolute percentage error was evaluated, it was found to be relatively low at 0.46%, 1.07%, and 2.78%, respectively. According to the scenario-2 analysis, the XGBR algorithm predicted the log10 values of the animal-based biogas potential of five major cities in Turkey for the year 2024, with 11.279 for Istanbul, 12.055 for Ankara, 12.309 for Izmir, 11.869 for Bursa, and 12.866 for Antalya. In the estimation of log10 values of CH4 emission, the tier1 approach yielded estimates of 3.080, 3.652, 3.929, 3.411, and 3.321, respectively, while the tier2 approach yielded estimates of 1.810, 2.806, 2.757, 2.552 and 2.122, respectively.
摘要:
可以使用沼气技术将动物废物转化为可再生能源。这一过程对温室气体排放有影响,是各国可持续的能源来源。它可以减少气候变化的影响,为子孙后代保护地球。文献中通常使用Tier1和Tier2方法来计算排放因子。使用增强算法,这项研究估计了土耳其所有省份2004-2021年每种动物类别的沼气潜力和CH4排放量(tier1和tier2方法)。在研究中创建了两种不同的场景。对于情景1,使用2004年至2019年的数据预测2020年至2021年,而对于情景2,使用2004年至2021年的数据预测2022年至2024年。根据情景1分析,极限梯度提升回归器(XGBR)算法是最成功的算法,在2020-2021年期间,基于动物的沼气预测的R2为0.9883,基于动物的CH4排放预测(tier1和tier2方法)的R2为0.9835和0.9773.当评估平均绝对百分比误差时,发现相对较低,为0.46%,1.07%,和2.78%,分别。根据情景2分析,XGBR算法预测了2024年土耳其五个主要城市基于动物的沼气潜力的log10值,伊斯坦布尔为11.279,12.055forAnkara,12.309用于伊兹密尔,布尔萨11.869,和12.866的安塔利亚。在估计CH4排放的log10值时,Tier1方法分别得出3.080、3.652、3.929、3.411和3.321的估计值,而Tier2方法的估计值分别为1.810、2.806、2.757、2.552和2.122。
公众号