This paper classified and defined traffic flows from various directions to achieve a refined estimation of vehicle inflow rates at the downstream intersections. Using sampled trajectory data from these flows, the paper analyzed vehicle arrivals, selected an exponential distribution as the prior distribution, and constructed the likelihood function for vehicle arrivals. The Maximum A Posteriori (MAP) estimation method was then applied to estimate the traffic arrival rate. Results have indicated that the Mean Absolute Percentage Error (MAPE) for arrival rate estimations from the proposed model has fluctuated around 13% under different penetration rates of connected vehicles, which is significantly lower than the average arrival rate estimation error. This has demonstrated the model's strong estimation performance. The proposed model has effectively adapted to varying penetration rate scenarios, achieving precise arrival rates and traffic volume estimates for traffic flows from different directions.