基于数据驱动的水泥企业碳排放预测模型研究

Study on the Data-driven Prediction Model for CO_(2)Emissions in Cement Enterprises

为解决目前水泥企业碳排放预测中存在的影响因素多、预测精度低问题,借助大数据机器学习技术,尝试构建了多种预测模型。结果表明:线性回归模型对企业碳排放预测误差达12.78%,机器学习模型可降低至9%,而通过智能烟花算法改进的BP(Back Propagation)网络模型可将误差降低至6%,能够较好地满足实际应用需求。进一步分析发现:对于企业碳排放量,“熟料产量和净购入电量”两因素影响最为显著,而提高替代燃料使用率则是当前实现节能减排的主要途径。

To address the problems of multiple influencing factors and low prediction accuracy in the current carbon emission prediction of cement enterprises,various prediction models have been constructed with the help of big data machine learning technology.The results show that the prediction accuracy error of the linear regression algorithm model for cement enterprises carbon emissions is 12.78%,which can be reduced to 9% by machine learning model.However,by improving the BP(Back Propagation)network model with intelligent fireworks algorithm,the error can be reduced to only 6%,which can better meet the practical application requirements.Further analysis found that for enterprise carbon emissions,"clinker production and net electricity purchase"are the two factors with the most significant impact,while increasing the use rate of alternative fuels is currently the main way to achieve energy conservation and emission reduction.