基于随机森林的发动机振动问题分析

Analysis of Engine Vibration Based on Random Forest Technique

针对液体火箭发动机振动问题综合分析相关的测量指标参数,寻找对振动问题有较大影响的指标参数,以对改进设计提供理论依据。首先使用随机森林建模评估,得到测量指标的重要性排序结果;然后使用基于特征选择方法的随机森林模型的递归特征消除选择特征的方法进行验证,二者结果一致,证明了该方法的准确性;最后使用交叉验证模型评估预测的方法,得到重要指标变量的数量理论值,为确定工程重点关注的指标变量数量提供理论依据。

Comprehensively analyzing relevant measurement index parameters for liquid rocket engine vibration problems, it finds that the index parameters have some greater impacts on the vibration problem,which provides the theoretical basis for improving design. In this paper, firstly, random forest modeling evaluation is used to obtain the importance ranking results of measurement indicators;Then the recursive feature elimination method of random forest model based on feature selection method is used for verification.The results of the two methods are consistent, which proves the accuracy of the method;Finally, the cross validation model is used to evaluate the prediction method, and the theoretical value of the number of important index variables is obtained, which provides a theoretical basis for the number of index variables that the project focuses on.