中文摘要：

为研究摆式列车倾摆控制信号预测方法,建立“动车＋拖车＋拖车”3辆车编组的摆式列车机电耦合系统动力学模型。建模中考虑了列车系统中存在的轮轨蠕滑力非线性、钩缓作用力非线性和悬挂力非线性。摆式列车通过安装于头车前转向架的陀螺仪在线检测曲线,对测出的横向加速度信号进行滤波和实时生成倾摆控制信号。为了补偿加速度信号的滤波延时,对倾摆控制信号的预测分别采用线性插值法和线性BP神经网络预测,并仿真研究摆式列车曲线通过性能。数值仿真结果表明：线性插值法预测和神经网络预测均能有效补偿加速度信号的滤波延时,使头车及时倾摆,大幅度降低未平衡横向加速度;在输入信号波动较大和预测时间较长时,神经网络预测效果更好;倾摆控制信号的预测方法对车辆动力性能影响不大。

英文摘要：

In order to study the forecast method of tilting control signal of tilting train, the electrical-mechanical coupled model of tilting train is set up. The train model includes a motor car and two trailer cars. During the modeling process, various kinds of nonlinear factors are considered, including nonlinear wheel/ rail creep force, nonlinear coupler force, nonlinear suspension forces, etc. The tilting control signal is based on the filtered lateral acceleration of the sensor on the first bogie frame. The curved track information is obtained from the filtered signal of the gyroscope on the first bogie frame. The linear interpolation and BP neural network forecast methods are used to reduce the time delay of the tilting signal. The curve negotiation dynamic performance of tilting train is studied by numerical method. It can be known from the numerical results that the linear interpolation and the neural network forecast methods all can reduce the time delay of tilting control signal of the first car, the unbalanced lateral acceleration can be decreased. The neural network forecast is better when the forecast time is long or the fluctuation of the input signal is big. The forecast method of tilting control signal has little influence on the dynamic performance of the tilting train.