基于机理和数据驱动建立牛粪厌氧发酵产气预测混合模型

Prediction Hybrid Model for Biogas Production in Cattle Manure Anaerobic Fermentation Established by Mechanism and Data-driven

厌氧发酵过程作为一种微生物参与的质能传递过程,其复杂性、时变性以及严重非线性使得纯机理或纯数据方式所建立预测模型的精确性和实用性大大降低。为解决该问题,以牛粪发酵产甲烷过程为研究对象,提出一种厌氧发酵产气预测混合模型:首先建立水解、酸化、产氢产乙酸阶段的清晰动力学方程,再结合随机森林回归(RFR)构建产甲烷及液-气传质过程的数据驱动模型,并运用麻雀搜索优化算法(SSA)对RFR算法的超参数进行最优估计,最后将发酵机理动力学模型和随机森林回归模型两个子模型级联,形成厌氧发酵产气预测全过程混合模型。利用该混合模型对牛粪厌氧消化实验进行动态仿真模拟,结果表明:混合模型仿真下的甲烷产率与实验数据吻合良好,且经过麻雀搜索优化算法优化的机理和数据驱动混合模型(SSA-RFR-ADM)的预测精度更接近复杂纯机理模型的预测精度,模型预测均方误差为0.003035。所提出的混合模型有效克服了单一模型存在的局限性,可作为复杂纯机理模型的代理模型,为实际厌氧发酵产甲烷工艺提供正确的理论指导。

As a mass and energy transfer process involving microorganisms,the complexity,time variability and severe nonlinearity of anaerobic fermentation process greatly reduced the accuracy and practicability of the prediction models established on pure mechanism or pure data.In order to solve this problem,taking the methane production process of cow dung fermentation as the research object,a hybrid model for predicting gas production by anaerobic fermentation was proposed.Firstly,the clear kinetic equations of hydrolysis,acidification hydrogen production and acetic acid production were established,and then the data-driven model of methane production and liquid-gas mass transfer process was constructed combined with Random Forest Regression(RFR).And the Sparrow Search Algorithm(SSA)was used to optimally estimate the super parameters of RFR algorithm.Finally,two sub-models,the kinetic model of the fermentation mechanism and the random forest regression model,were cascaded to form a hybrid model for the whole process of anaerobic fermentation biogas production prediction.The hybrid model was used to simulate the anaerobic digestion experiment of cow manure.The results showed that the methane production rate using the hybrid model simulation was in good agreement with the experimental data,and the prediction accuracy of the mechanism and data-driven mixed model(SSA-RFR-ADM)was close to the prediction accuracy of the complex pure mechanism model.The mean squared error of the model prediction was 0.003035.The proposed hybrid model effectively overcame the limitation of single model,and could be used as a proxy model of complex pure mechanism model,providing correct theoretical guidance for practical anaerobic fermentation methane production process.