Abstract: The large wear of slide plate and instability of current-carrying in pantograph-catenary system is a key problem, which has restricted the development of the electrified railway to high speed, heavy load, safety and stability, and should be solved by adopted effective measures in our country. Under the special current-carrying and sliding speed conditions, the wear rate of slide plate which characterizes the wear speed of the pantograph slider decreases first and then increases, which shows the changing trend of U-type while the normal load increasing. But the stability coefficient of current-carrying which characterizes the stability of the contact current will decrease with the increase of normal load. So there is a suitable normal load to make the wear of slide plate and stability of current-carrying relatively optimal. A number of experiments under different conditions are done to get the data about the wear rate of slide plate and stability coefficient of current-carrying firstly, in which the steeped metal carbon slider and copper conductor are used. Based on these data, a BP neural network nonlinear model is built, whose inputs are normal load, current-carrying and sliding speed, and outputs are wear rate of slide plate and stability coefficient of current-carrying. Then some parameters are optimized by the genetic algorithm. And the particle swarm algorithm is used to optimize the multi-objective problem when the current-carrying and sliding speed are given. At last the Pareto set with the best normal loads are obtained to make the wear rate of slide plate and stability coefficient of current-carrying minimal.

Key words: wear rate of slide plate, stability coefficient of current-carrying, optimal normal load, pareto solution, multi-objective optimization

摘要： 弓网系统中滑板磨耗大及电力机车受流不稳定是制约我国电气化铁路向高速、重载、安全、稳定运行方向发展的一个关键问题，亟待采取有效的措施进行解决。在特定的载流、滑动速度条件下，表征滑板磨损快慢的滑板磨耗率随法向载荷增大呈现出先减小后增大的“U”型变化趋势，而表征载流稳定性的载流稳定系数随法向载荷增大而逐渐减小，因此选择合适的法向载荷可使滑板磨耗和载流稳定性均达到相对最佳。通过使用浸金属碳滑板与铜导线进行对磨实验，得到了不同实验条件下滑板磨耗率和载流稳定系数的大量实验数据。以此建立了以法向载荷、载流和滑动速度为输入量，以滑板磨耗率、载流稳定系数为输出量的BP神经网络非线性模型，进而利用遗传算法优化了相关参数，并在特定的载流和滑动速度条件下，采用基于粒子群算法的多目标优化方法，确定了滑板磨耗率最低、载流稳定系数最小的最佳法向载荷对应的非劣解。

关键词: 滑板磨耗率, 载流稳定系数, 最佳法向载荷, 非劣解, 多目标优化