基于改进的K-均值聚类算法构建柴油车运行工况

Construction of Diesel Vehicle Driving Condition Based on Improved K-Means Clustering Algorithm

摘要: 针对传统的K-均值聚类算法在聚类时容易陷入局部最优的问题，提出一种改进的K-均值聚类算法，并结合主成分分析法将其应用到柴油车运行工况的构建中。首先通过主成分分析完成运动学片段特征参数的降维，再利用改进的K-均值聚类算法进行运动学片段的聚类，最后选择具有代表性的片段合成运行工况。对自主构建的运行工况的特征参数值和总体数据的特征参数值进行比较，各参数间的相对误差在2%~5%之间，说明用改进后的K-均值聚类算法结合主成分分析法构建的运行工况具有很高的精度，并且对车辆的运行状况具有很强的代表性。

Abstract: Aiming at the problem that the traditional K-mean clustering algorithm is prone to fall into the local optimum during clustering, an improved K-mean clustering algorithm is proposed, and combined with the principal component analysis method, it is applied to the construction of diesel vehicle op-erating conditions. Firstly, dimensionality reduction of characteristic parameters of kinematic seg-ments was completed through principal component analysis, and then the improved K-means clus-tering algorithm was used to cluster kinematic segments. Finally, representative segments were selected to synthesize operating conditions. By comparing the characteristic parameter values of the independently constructed operating conditions with those of the overall data, the relative er-ror of each parameter is between 2% and 5%, indicating that the operating conditions constructed by the improved K-mean clustering algorithm combined with the principal component analysis method have high precision, and have a strong representation of the vehicle operating conditions.