汽车空调自动控制算法路试标定数据的聚类分析研究

Research on Cluster Analysis of Road Test Calibration Data of Automatic Control Algorithm of Automotive Air-conditioning

为了解决传统汽车空调自动控制算法路试实验效率低和成本高的问题,本文采用聚类分析方法对路试实验数据进行处理,并通过夏季路试数据进行了验证。将原始数据切分为1,030个片段,归类为32个簇,采用重复率分析、离群值分析和动态特征数据提取的方法进行分析,结果表明,在一定的判据下,数据重复率高达49.2%,实验优化尚存大量空间;得出数据中的离群片段等隐含信息。本文为瞬态模型标定提供了仅占原数据14.2%的动态特征明显的训练数据,大大提升了汽车空调瞬态模型的标定效率。

In order to solve the problems of low efficiency and high cost of the traditional road test for automotive control algorithm calibration,the cluster analysis method is used to process the road test experimental data,which is verified by the summer road test data.The original data are segmented into 1,030 segments and classified into 32 clusters.The results are analyzed by using repetition rate analysis,outlier analysis and dynamic feature data extraction.The results show that,under a certain criterion,the data repetition rate is up to 49.2%,so that the experimental optimization needs further research;the hidden information such as outliers in the data is also obtained in this paper;only 14.2%of the training data with dynamic characteristics are provided for the transient model calibration,which greatly improves the calibration efficiency of the transient model.