Mix-Zone is one of the most effective real-time location privacy preserving techniques over road networks. By breaking the continuity of location exposure and changing pseudonyms, this strategy can effectively keep users from tracking attacks. Existing Mix-Zone evaluation mechanisms are mainly divided into two categories – Calculation and vehicle tracking. Calculation based evaluation methods like K-anonymity and Entropy are more suitable for qualitative analysis and comparison, but lacks accuracy and quantitative criteria, whereas existing tracking methods are too simple to accurately quantify the privacy of the Mix-Zone. Accordingly, it is urgent to bring up more accurate tracking methods to get a better view of the Mix-Zone's privacy. In this paper, we propose two categories of Mix-Zone tracking methods based on the basic BP and the customized artificial neural networks, where the former is used as verification and based on the result analysis of which we designed the latter, which greatly improves the tracking result, revealing the privacy preserving level of the Mix-Zone more reasonably.