In order to improve the efficiency of traffic management system, reduce the losses caused by traffic incidents, an algorithm for application in urban expressway incident detection is proposed based on the change trend of traffic parameters using gain methodology. The traffic data for research were obtained from a real section of urban expressway. It contains volume, speed and occupancy. The time scale of data is 20 s, 1 min and 5 min. The performance of automatic incident detection (AID) algorithms was evaluated based on detection rate, false alarm rate, mean time difference between different algorithms, mean time difference between different times scale, robustness. The detection performance of the proposed algorithm was compared to California algorithm which has been adequately proven in practical applications. The experimental results indicate that the proposed algorithm is competitive with California algorithm using traffic data of three time scales, and the robustness of the proposed algorithm is very satisfactory.