Abstract: Based on the analysis of the characteristics and classification of the three abnormal traffic parameters data,an evaluation method is designed to evaluate the missing data of the traffic data.An abnormality evaluation method is established using gain amplification theory,and the dynamic threshold method is designed using the 68-95-99.7 rule normal distribution.The results show that this micro-evaluation algorithm is better than classical algorithms.