Safety-related applications in Vehicular Ad-hoc Networks (VANETs) can help to reduce the number of traffic accidents by periodically broadcasting Basic Safety Messages (BSMs). But considering that the density and topology of VANET change frequently, the Quality of Service (QoS) of safety applications with the fixed transmission parameters would not always meet the requirements of safety-related applications. In this paper, we firstly propose an analytical model to evaluate the performance of the BSM broadcast in VANET at the intersections. The effect of the traffic light is also taken into account by introducing the non-homogeneous Poisson process (NHPP) vehicle distribution, which is simulated by microscopic traffic simulator SUMO and validated by Kolmogorov-Smirnov (K-S) test. Secondly, the number of vehicles in the hidden terminal area and the concurrent transmission area need to be computed by complex integral computations for the proposed analytical model for evaluating the QoS of the safety applications, while the vehicle BSMs provide the vehicle location, speed, and transmission parameters, and these BSMs could be measured and collected in a timely manner to integrate into the analytical model to save the complex integral calculations. Based on the cross-validation QoS metrics between the NS2 (Network Simulator-ns-2) simulation and the analytical model, we employ the vehicle entity of the NS2 simulation model to represent the actual vehicle BSMs, to obtain the number of vehicles in the hidden terminal area and the concurrent transmission area. Finally, to maximize the transmission capacity and minimize the delay under the constraint of maintaining a high application-level QoS of safety applications, a multi-objective optimization scheme with Bare Bones Particle Swarm Optimization (BBPSO) is proposed to dynamically adjust multi transmission parameters. The accuracy of the proposed analytical model is validated by the NS2 simulation. The experimental results also show that the optimized ones could get better results compared with the real test-bed used transmission parameters. Furthermore, the comparisons with slotted-p(or 1)-persistent protocol and CSMA/CA with retransmission strategy show that the proposed solution could make VANET work with better performance at various vehicle densities.