Comparison of Support Vector Machine and K-Means Algorithms Performance in Extracting the Real Driving Cycle of Combined Tehran-Amol

Driving cycles represent the vehicle speed as a function of time and are used in vehicle design, fuel management, and the improvement of standard indicators. In this study, four combined driving cycles were extracted using real data. The data was collected from a passenger car with a gasoline engine under real driving conditions while driven from Tehran to Amol based on the car chasing method. A code was generated in MATLAB software to create the desired cycle using support vector machine and K-means algorithms considering mid-range and mean values as group centers. The characteristic parameters of the cycles such as the average speed and the percentage of the car travel time at idle, cruise, accelerating, and decelerating conditions were also calculated. These cycles were compared based on the mean relative error, the root-mean-square error, and the Chi-square test. The results showed that the cycles extracted by the support vector machine were closer to the allowable time interval (less than 1800 seconds); however, the cycle extracted by the K-means algorithm with the mean value as the centers of the generated categories, recorded the least errors. This cycle, in addition to spending most of its time in accelerated motion, represented a greater amplitude of acceleration and velocity fluctuations than other cycles.