论文部分内容阅读
为了更加快速、精确地对混合生物质灰熔点进行预测,利用交叉验证(cross validation,CV)方法进一步优化了前人提出的经遗传算法(genetic algorithm,GA)优化的支持向量机(support vector machine,SVM)回归模型。以灰成分作为输入量,灰熔点为输出量,以单生物质数据训练该模型,对混合生物质灰熔点进行了预测;并与仅经GA优化模型的预测结果进行了比较。研究结果表明:经GA与CV优化的SVM模型对混合生物质灰熔点进行预测,平均绝对误差为25.0℃,平均相对误差为2.7%,比仅经GA优化的SVM模型预测结果更为精确;适当地设置相关参数可以节省程序运行时间。
In order to predict the melting point of mixed biomass ash more rapidly and accurately, the cross validation (CV) method was used to further optimize the previous genetic algorithm (GA) -based support vector machine , SVM) regression model. Taking ash content as input and ash melting point as output, this model was trained by single biomass data, and the melting point of mixed biomass ash was predicted. The results were compared with those obtained by GA-only optimization model. The results show that the average absolute error of the hybrid biomass ash melting point predicted by the GA and CV SVM model is 25.0 ℃, the average relative error is 2.7%, which is more accurate than the GA-optimized SVM model. The appropriate Setting related parameters can save program running time.