Aiming at the heterogeneity of mixed bicycle flow, the density and flow of three mixed bicycle flow states were calculated from the concepts of space split and perceived density, and the traffic wave formula of mixed bicycle flow was derived from the cell transmission model (CTM). In this paper, the actual survey data of six different bicycle lanes width in Kunming are used, the stopping wave is taken as the contrast parameters, and the calculated results are compared with the single class LWR model. The results show that the calculation accuracy of the proposed model is higher than that of the single class LWR model under different bicycle lane widths, as can be seen from MAPE, the calculation accuracy of stopping wave and starting wave of the proposed model is 9.25% and 13.64% higher than that of the single class LWR model, respectively, which indicates that the proposed model can describe the running characteristics of the mixed bicycle flow well. It provides a new way to reveal the traffic mechanism of mixed bicycle flow. Copyright © 2019 by Science Press.