一种面向丁羟推进剂燃速预测的机器学习融合模型  

A machine learning fusion model for burningrate prediction of HTPB propellant

作　　者：孙娜 林可仁 沈坚 王心怡 许志钦 张斌[3] SUN Na;LIN Keren;SHEN Jian;WANG Xinyi;XU Zhiqin;ZHANG Bin(Shanghai Space Propulsion Technology Research Institute,Shanghai 201109,China;Beihang University,Beijing 100191,China;Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]上海航天动力技术研究所,上海201109 [2]北京航空航天大学,北京100191 [3]上海交通大学,上海200240

出　　处：《固体火箭技术》2024年第5期738-744,共7页Journal of Solid Rocket Technology

基　　金：航天八院-上海交大联合基金(USCAST2022-42)。

摘　　要：建立了以端羟基聚丁二烯(HTPB)推进剂原材料参数和组分含量为输入变量、燃速值为输出的机器学习预测模型。首先,通过经验性单因素规律对包括随机森林回归、支持向量回归、高斯过程回归、核岭回归和径向基函数模型等进行了初筛;然后,以最大绝对误差(AE_(max))、最大相对误差(RE_(max))、均方根误差(RMSE)及平均绝对误差(MAE)作为模型的性能评价指标,对不同模型的预测性能进行对比,筛选出最大相对误差RE_(max)<5%的模型,对通过筛选的模型进行融合获得最终的燃速预测模型,并将融合模型与单一模型的预测性能进行对比;最后,利用样本集外的10组数据对融合模型的预测精度进行了验证。结果表明,融合模型的RE_(max)和RMSE小于任何单一模型,其中AE_(max)=0.68、RE_(max)=3.97%、RMSE=0.3214、MAE=0.253,可实现对此类HTPB推进剂燃速的高精度、高效率预测。A machine learning model with raw material parameters and the propellant component content as input variables and burning rate value as output was established.Firstly,the models were screened by prior knowledge,including Random Forest reqression,Support Vector Regression,Gaussian Process Regression,Kernel Ridge Regression and Radial Basis Function model.Then,by using_(max)imum absolute error(AE_(max)),maximum relative error(RE_(max)),root mean square error(RMSE)and mean absolute error(MAE)as the performance evaluation indicators of the model,the prediction performance of different models was compared,and the models with the_(max)imum relative error less than 5%were screened.The final burning rate prediction model was obtained by fusing the screened models.Moreover,the prediction performance of the fusion model is compared with that of the single model.Finally,10 groups of data outside the sample set were used to verify the prediction accuracy of the model.The results show that the RE_(max)and RMSE of the fusion model is less than any single model;the AE_(max),RE_(max),RMSE and MAE of the model are 0.68,3.97%,0.3214 and 0.253,respectively.Therefore,the model is reliable and can accurately predict the burning rate of HTPB propellant with high precision and high efficiency.

关 键 词：HTPB推进剂 燃速预测 机器学习 多模型融合 

分 类 号：V512[航空宇航科学与技术—航空宇航推进理论与工程]