| With the improvement of the quality of life,people begin to pay attention to the driving pleasure in the driving process.Driving pleasure is reflected in the driver’s feeling of being easy to operate,responsive and comfortable.At present,the subjective evaluation method is mainly used in domestic driving evaluation.The disadvantage of subjective evaluation is that it is easy to produce deviation,high cost and difficult to meet the development needs.Therefore,it is necessary to establish an objective driving evaluation system.In this paper,the structured and index thinking in data analysis is introduced into the driving research process,and the pyramid driving evaluation dimension is established.The driving evaluation dimension of hybrid electric vehicle with 8 driving driving conditions and 42 driving evaluation indexes is formulated.On this basis,a set of driving experiment scheme is developed for hybrid electric vehicle.It solves a big problem in the process of driving objective evaluation: driving condition identification.Using the method of rule + fuzzy logic to establish the driving condition recognizer,the structure of the classifier,the method of de fuzziness,membership function and fuzzy rules are studied,and the condition recognizer of the starting condition and the other 7 conditions is established.The result shows that the recognition accuracy of the established condition recognizer is more than 90%.In this paper,a new method of constructing the driving scoring model is proposed,which uses the structure of "multi-level neural network" to build the driving scoring model,uses BP neural network to build the scoring model of all levels of working conditions,uses linear neural network to build the whole vehicle scoring model,and determines the number of nodes,hidden layers and transfer function of the network.At the same time,the driving score model is verified.The results show that the sample mean square error is small,the correlation between the target value and the actual value is more than 0.9,and the error range between the actual output and the expected output is small,so the driving score model has enough accuracy to meet the use requirements.This paper solves the two technical problems of driving condition identification and scoring model construction in the process of driving objectivity evaluation,which is conducive to further research on driving objectivity. |
