Research On Construction Method Of Vehicle Driving Cycle Based On Sparse Manifold Clustrering And Embedding Algorithm

With the rapid development of China’s economy,the number of vehicles is also rising steadily year by year.The energy consumption and emission of motor vehicles has become an urgent problem in the field of vehicles in China.The driving condition curve of automobile is one of the important standards to evaluate energy consumption and emission.It is an important technology to develop energy saving and emission reduction technology to make the appropriate driving condition curve for motor vehicles.It is of great significance to develop the driving conditions of the new cars and to make the curve of the driving conditions as representative of the local actual road conditions as possible.This paper takes Fuzhou vehicle condition data set and San Angelo vehicle condition data set as research objects,analyzes the original data collected and preprocesses the data.Based on the processed data set,the feature parameter model of kinematic segment is constructed by using the concept of kinematic segment.According to the divided kinematic segment data set,this paper firstly analyzes the kinematic segment data set by dimension reduction clustering,and then selects the kinematic segment to synthesize the driving condition curve of the vehicle based on the result of the reduced dimension clustering.This paper mainly focuses on how to improve the clustering effect of the dimension reduction of the working condition curve,reduce the error rate of the working condition curve and test data,and how to select the optimal kinematic segment set from the category.The main work is as follows:1.in order to improve the effect of the reduced dimension clustering of the working condition curve,this paper proposes a method of reducing the dimension clustering of the working condition curve based on smce algorithm.Smce algorithm(smce)is a nonlinear manifold learning algorithm,which is based on the assumption that high-dimensional data may be distributed in low-dimensional manifold space.It can automatically find neighborhood and corresponding weight in the process of setting data point neighborhood,and use the neighborhood matrix after dimension reduction to gather data into a specified category through spectral clustering.The experimental results show that the clustering effect and average error rate of the working condition curve based on smce algorithm are better than PCA + k-means algorithm.2.based on the method of reducing dimension clustering of working condition curve based on smce algorithm,this paper proposes a method of reducing dimension of working condition curve based on semi supervised manifold learning algorithm.Compared with unsupervised learning,a small number of distinct data tags are effectively used.Compared with supervised learning,a lot of energy is saved for data labeling.Unknown data categories can be classified according to the known data in the low-dimensional coordinates.The experiment shows that the classification effect and average error rate of the working condition curve based on SS smce algorithm are better than PCA + k-means algorithm,which is better than smce algorithm.3.in order to select the most representative set of kinematic fragments based on the reduced dimension clustering results,this paper summarizes a mathematical model of kinematic segment selection.There are many ways to solve the optimization problem.This paper uses the dynamic programming method to solve the problem,and gives the state transfer equation.The experimental results show that the global optimal solution of dynamic programming is better than the local optimal solution of distance ordering.