Aiming at the problems of low accuracy and poor robustness of speed measurement due to uncertain model parameters in the speed measurement fusion process of multi-sensors for high-speed trains, a robust Kalman/H∞ hybrid filter and fusion method under the constraint of diagonal array weighted error variance is put forward. Firstly, the high-speed train motion model is constructed, and the speed, position, and acceleration of the train are observed by axle speed sensor, radar speed sensor, and acceleration sensor; Secondly, a weighted robust Kalman/H∞ hybrid filter is designed, and the augmented system model of the state error constraint is introduced. Under the filter satisfies the three performance indexes, the filter is obtained by solving two Riccati equations with the error variance constraint; Thirdly, a multi-stage fusion structure is used to determine the optimal weighting coefficients of each local filter by estimating the error covariance matrix, and then multi-speed fusion is carried out; Finally, the designed method and algorithm are simulated, and the simulation results are analysed and compared, The results show that the weighted robust hybrid filtering algorithm proposed in this paper can reduce the detection error of speed and acceleration of high-speed trains, further improve the accuracy of speed measurement and the robustness of the system, and provide technical support for the high-precision speed control and efficient and safe operation of high-speed trains.