system, it was not consistent with the actual road condition。 Therefore, this research may not have much significance in engineering。 According to the above analysis, the research based on the actual road condition will be the most suitable。

3。2。 Establishment and verification of multi-body dynamic model of suspension

The front suspension of the vehicle was simplified and modeled by a five-link mechanism in MotionView。 The model of the link mechanism was built by rigid body。 Models of the damper and the jounce bumper were built by flexible body。 The link was a rigid hinge。 The five-link mechanism could save space and had a better dynamic performance [10, 11]。 The multi-body dynamic model of the front suspension is shown in Fig。 4。

Fig。 4。 Multi-body dynamic model of the front suspension

It was shown in Fig。 4 that the multi-body dynamic model of suspension was very complicated。 It was necessary to verify reliability of the model through experiments。 In this way, the accuracy of subsequent analysis could be ensured。

An acceleration sensor was mounted on the suspension shaft, as shown in Fig。 5。 When the vibration displacement of vehicle wheel center was tested, the multi-channel data collection equipment was also used to obtain the vibration acceleration of the suspension shaft at the same time。 And then it was imported into test。lab to be processed, the corresponding process was shown in Fig。 2。 Testing results were compared with simulation results of multi-body dynamic model in Fig。 4, and comparison results were shown in Fig。 6。

a) Testing equipments b) Acceleration sensor

Fig。 5。 Testing equipment of vibration acceleration on the suspension shaft

It was shown in Fig。 6 that trends and peaks of both experiment and simulation results coincided well within the frequency of 0-220 Hz。 Within some frequency ranges, the experiment

values were different from simulation values because the experiment only discussed vibration transmission and neglected structure-acoustic coupling。 It was shown that the simulation model was reliable to analyze characteristics of suspension system。

Fig。 6。 Experiment and simulation comparison of the vibration accelerations at mounting point

Based on the multi-body dynamic model, the vibration isolation ratio of suspension system was extracted for subsequent research as shown in Fig。 7。 It was shown in the figure that the vibration isolation ratio changes with frequency in the whole research frequency range。

Fig。 7。 Vibration isolation ratio of suspension system

3。3。 Establishment of finite element model for jounce bumper

As an important component in a suspension system, a jounce bumper had to bear vibration transmitted by pavements and attenuates it。 Therefore, it was very necessary to optimize and analyze its dynamic characteristics and fatigue life。 The elasticity modulus of the aluminum-cast jounce bumper is 70,000 MPa。 Poisson’s ratio is 0。33, and the density is 2。72×10-9 t/mm3。 Finite element mesh needed to be generated for the jounce bumper in order to analyze its fatigue life。 The mesh was hexahedral mesh。 The finite element model of jounce bumper was shown in Fig。 8。

4。 Optimization design of the front suspension

The suspension system was a very complicated structure, and it had a lot of structural parameters。 As a result, when optimization design was carried out, it was very necessary to which parameters were set as the design variables。 Design of experiment (DOE) was one of the most widely method to determine the design variables。 According to DOE, we can not only determine which variables had the largest influence on performance of the suspension system, but also control the variables in a certain range。 As a result, response of the suspension system can be infinitely close to the actual value。 It can construct an approximate model to replace the multi-body