Yu Shouhai, Clark O Grant, Leonard Jerry J
Agricultural, Food and Nutritional Science, AF 4-10, University of Alberta, Edmonton, AB, Canada.
Bioresour Technol. 2008 Apr;99(6):1886-95. doi: 10.1016/j.biortech.2007.03.061. Epub 2007 Nov 7.
Temperature is widely accepted as a critical indicator of aerobic microbial activity during composting but, to date, little effort has been made to devise an appropriate statistical approach for the analysis of temperature time series. Nonlinear, time-correlated effects have not previously been considered in the statistical analysis of temperature data from composting, despite their importance and the ubiquity of such features. A novel mathematical model is proposed here, based on a modified Gompertz function, which includes nonlinear, time-correlated effects. Methods are shown to estimate initial values for the model parameter. Algorithms in SAS are used to fit the model to different sets of temperature data from passively aerated compost. Methods are then shown for testing the goodness-of-fit of the model to data. Next, a method is described to determine, in a statistically rigorous manner, the significance of differences among the time-correlated characteristics of the datasets as described using the proposed model. An extra-sum-of-squares method was selected for this purpose. Finally, the model and methods are used to analyze a sample dataset and are shown to be useful tools for the statistical comparison of temperature data in composting.
温度被广泛认为是堆肥过程中有氧微生物活动的关键指标，但迄今为止，在设计一种合适的统计方法来分析温度时间序列方面几乎没有付出努力。尽管非线性、时间相关效应很重要且普遍存在，但此前在堆肥温度数据的统计分析中并未考虑这些因素。本文提出了一种基于修正的冈珀茨函数的新型数学模型，该模型包含非线性、时间相关效应。展示了估计模型参数初始值的方法。使用SAS中的算法将模型拟合到来自被动曝气堆肥的不同温度数据集。然后展示了检验模型与数据拟合优度的方法。接下来，描述了一种以统计严谨的方式确定使用所提出模型描述的数据集时间相关特征之间差异显著性的方法。为此选择了额外平方和法。最后，使用该模型和方法分析了一个样本数据集，结果表明它们是堆肥温度数据统计比较的有用工具。
