Research Of The Intelligent Method In The Analysis Of Track Dynamics

The environmental vibration resulting from traffic load directly influences people's life quality. Now people pay more attention to this problem. Generally, track structure is consisted of track, rail pad, sleeper and ballast. When the train is running, there has interaction between wheels and track, and track will bend because of the weight of wheels. Through the component of track structure, the bending waves propagate to soil, which causes the second vibration on building. Due to the complexity of vehicles, tracks and ground, it's difficult to construct mathematical models to simulate the real conditions of the track system. The artificial intelligence has offered an efficient tool to evaluate the environmental vibration. However, the study is still under a kind of path-breaking period.In this paper, on the basis of traditional track analysis method, it integrate genetic algorithms and neural network technology, to make a preliminary research into the modal parameter of track, the characteristic parameter of ground's surface layer and the ground's response forecast affected by moving load. The key points of this paper can be concluding as follows:1. The main purpose is to optimize BP and RBF neural network globally. Matlab, one of the programming language which is simple directly efficient and open is selected to fulfill this model.2. The basic modal parameter (modal frequency and modal damping) is the most important dynamic characteristics of the track system, which is closely related to its structural property. Regarding to the simplified Euler Beam modal, this paper will make use of the finite element method to calculate the FRF, with considering the compress transaction of response data by using PCA as the input for optimizing the BP network model, and Euler Beam's four exponent modal frequency and modal damping as network output. By means of such simulated calculation, obviously identification is feasible.3. The ground's topmost parameter is of vital importance to the analysis of environmental vibration caused by tracks. This paper will marry the improved genetic algorithms to traditional forward analysis, to conduct the inversion aboutground's elastic modulus, density and attenuation factor. There is a clear evidence that inversion method has been provided with overall searching capability, and perfect anti-noise capability.4. On the basis of traditional wave theory, the complicated nonlinear relationship is constructed among the train velocity, the ground's topmost parameter (including elastic modulus, density and attention factors), the linear distance from the observation point to the track center, the response velocity of ground's observation point and the virtual value of acceleration. Furthermore, it is expected that this RBF network modal optimized by such kind of practice will effectively forecast the ground's response, velocity and acceleration, with the influence of the moving load.