Researches On The Key Technologies Of Railway Line Information Measurement System Of Tilting Train

Running tilting train is an effective way for increasing the speed of train on the existent railways. Combined with the researches of the first tilting train of our country, the main tasks of this paper are researches about railway line information measurement system and corresponding signals processing methods so as to establish an effective real-time measurement system that can provide important measurement information for tilting control accurately and reliably.To control car body tilting, corresponding measurement information, especially the unbalanced centrifugal acceleration, with which the tilting control instructions are produced, should be obtained on line while train running through curves. In view of current various railway line measurement methods and their disadvantages, such as the time delay of acceleration signal after filtering, two kinds of new measurement methods based on single-axis gyroscope platform and mathematical platform respectively are proposed and developed for the first time in our country. Both railway tests and theoretical analysis show that this two methods are effective tilting control measurement modes, which can avoid the time delay of accelerations caused by filtering and satisfy the time requirement of tilting train.With the single-axis gyroscope platform mounted on the floor of the first car and two displacement sensors set between the car and the second suspension, a measuring horizontal datum line is established up. This measurement system not only can measure superelevation of railway, but also curvature of curves, from which the unbalanced centrifugal acceleration can be computed to provide the instruction information for tilting control. The measurement principle and errors are analyzed in a detail and corresponding mathematical models are constructed. Railway tests show that this tilting control measurement method can satisfy the time requirement of tilting train.Measurement method based on the single-axis gyroscope platform requires that the single-axis gyroscope platform must be of good dynamic characteristics. A differential mathematical model, which describes dynamic characteristics of gyroscope platform, is established up with least square and its improving algorithms. Dynamic characteristics of the single-axis gyroscope platform are analyzed. Accordingly with zero-pole matching method, corresponding digital filters for dynamic compensation are designed, and dynamic compensation method based on neural network is also studied and corresponding dynamic compensation networks are designed. Simulations show that dynamic characteristics of the single-axis gyroscope platform can be effectively improved and then meet the needs of the real-time control of tilting train.For further increasing the reliability of the measurement system and decreasing its cost, the concept of math platform is introduced into the tilting train for the first time. The mathematical models of math platform system are established up. According to the principle of rigid body centering movement, computation models about attitudes and positions of train through curves are established up and corresponding computation methods are comprehensively studied. Coning motion affection of tilting train running through curves and its affection on attitude computation are analyzed in a detail. Corresponding "multi-samples" algorithms for reducing coning error are also researched deeply. Simulation shows that this compensation algorithm can effectively increase the accuracy of attitude computation, which is of important sense to make tilting train correctly tilt.Because of integration computation, measurement system of tilting train based on math platform would produce errors that are gradually accumulated with time. So whether can diminish or reduce these errors is the key of the successful application of math platform in the titling train. Thereby, an error compensation method based on Kalman filter principle is put forward and researched. Corresponding error factors are analyzed in a detail and complete error equations of math platform are established. Observability and degree of observability of state variables directly influence the state estimation of Kalamn filters. So based on so called PCWS and SVD observability analysis methods, the affections of the train mobility to the observability and degree of observability of state variables are analyzed, discussing the election of external observing variables and its affection to the estimate accuracy of Kalamn filter. Considered sensor errors and vibration interferences of railways, for increasing the accuracy of Kalman filter model to compensate for the dynamic errors more effectively, a new adaptive Kalman filter algorithm that can simultaneously estimate state variables and model parameters is also studied. Finally, simulations show that this compensation method is feasible and effective.For eliminating or minimizing influences of disturbance noises caused by irregularities of railways, an effective Kalman dynamic adaptive filtering method, which combines adaptive estimations of time-vary parameters of non-stationary random signals model and statistical characteristics of noises with Kalman filtering algorithm together, and can effectively realize the filtering treatment to the railway line measurement signals, is put forward and studied.For this Kalman filtering method, parameter models of signals must be firstly acquired and real-time modeling accuracy of non-stationary random signals is key to the results of this dynamic filtering. So various modeling methods for non-stationary stochastic signals are researched. By comparing and analyzing these algorithms, it is shown that time-varying parameters differential model combined with RLSAF algorithm and state space model combined with time-varying fading factor algorithm can realize strong track to the non-stationary random signals. They are effective methods for real-time modeling of non-stationary random signals.Finally, railway line information measurement system of tilting train is established up and corresponding railway tests are implemented. Analysis and treatment to mounts of testing data show that this two measurement methods and the designed measurement system can satisfy the requirement of tilting control of tilting train.