轻型汽车实际行驶排放普适性严苛场景的统计识别

Based on a large amount of light-duty fuel vehicle real driving emissions (RDE) test data, the RDE test second-by-second sampling data were divided into a large number of class trip data window subsets using the time-series moving average window method with a window time length of 300 seconds.Then, factor analysis was applied to reduce the multifactorial metrics affecting the RDE test to two metrics, trip dynamics factor and terrain factor, and cluster analysis was used to characterize the statistical class trip data window and to identify and extract a generalized and representative window of severe emissions data.The method is able to eliminate the interference of occasional anomalously high emission values in the RDE test with a high level of confidence, resulting in 41 prescriptive real driving emission severity scenarios.Speed profile and terrain profile data from the scenarios can be accurately reproduced on a laboratory chassis dynamometer to support laboratory calibration of real-world vehicle emissions.