Multi-objective Dynamic Optimization Of Wheel Tread Curve

When vehicle is running on the track, their mutual connection is achieved through the wheel/rail contact. As a consequent, the tread profile plays an important role on the performance of the vehicle itself. Excellent wheel tread profile does not only contribute a lot to vehicles’stability against hunting and derailment, but also helps to reduce the wear between wheel and rail. In recent years, worn wheel tread is largely used on railway vehicles in China which improves the running stability a lot comparing to the early cone tread, and extends the service life of the wheel effectively. However, uneven wear does happen sometimes. So there is still some room for worn wheel tread to be improved.The model of tread geometry parameters is built according to NURBS(Non-uniform Rational B Spline), in which weights and control point coordinates are chosen as design variables. The range of design variables of multi-objective optimization is verified based on their effects to geometric contact characteristics in case of LM tread. Multi-body dynamic model for simulating stability and curve passing performance is built based on the structure of subway B type body. The agent model including of design variables, non-linear critical speed, wheel-rail wear, derailment, lateral wheel force and vertical wheel force is built with the Approximation module in Isight. Multi-objective optimization model of LM tread is built with the objective function of non-linear critical speed and wheel-rail wear, and the constraint condition of derailment, lateral wheel force and vertical wheel force.3 optimization schemes are designed with different scale factors on the basis of multi-objective optimization model and actual working condition of subway B type body combining with Pareto theory. The 3 schemes are respectively LM01, LM02, LM03. Dynamic verification and adaptability analysis are carried out. The results show that LM03 is suitable for the working condition of small radius curves and low velocity, LM02 is suitable for the working condition of high velocity.According to the research work and results, parameterized model based on NURBS effectively reduces the quantities of design variables and improves the efficiency of optimization; Pareto genetic algorithm deals with the contradiction of stability and curve passing performance as optimization objectives; computational complexity is decreased through agent model. Multi-objective optimization model synthesized by methods mentioned above effectively improves the efficiency of tread design and helps to obtain the tread profile with comprehensive properties.