Validation of railway vehicle dynamic models in training simulators

Training simulators play an important role for sustaining safety, efficiency and cost effective railway transportation. Dynamic modelling of train systems is one of the main modules of training simulators. Validation of the dynamic models with collected real data ensures the fidelity of the simulator utilising the respective models. In this study, a validation process (Dynamic Modelling Validation Process (DyMVaP)) which is developed to support the validation of railway dynamic models is introduced. However, the proposed process can also be used in validating other dynamic models as well. The developed process is based on five steps including the preparation of validation scenarios, sensor deployment, real data collection, data preparation, and comparison of simulated and measured data. Note that the proposed DyMVaP was used for the validation of a full-mission training simulator so called TRENSIM, which was developed for Turkish State Railways. During the study it is realised that the current speed, travelled distance, acceleration (in x, y, z directions), rotation angles (around x, y, z axes), air pressure, in-train pressure/tension forces, traction motor currents, catenary voltage, positions of controllers must be collected synchronously by using proper sensors in order to ensure simulation validation. The required data was collected from locomotive body, bogies, wheel sets and connection of railway cars. The data (～200?GB) collected from the field by applying 27 different scenarios and transformed into appropriate data for utilising the generated dynamic models within the simulator. The measured and simulated data were also compared visually using graphical representation of the parameters as well as performing computations regarding the magnitude, phase and comprehensive error factors.     

基于强化学习的自主式水下潜器障碍规避技术（英文）

Obstacle avoidance becomes a very challenging task for an autonomous underwater vehicle(AUV) in an unknown underwater environment during exploration process. Successful control in such case may be achieved using the model-based classical control techniques like PID and MPC but it required an accurate mathematical model of AUV and may fail due to parametric uncertainties, disturbance, or plant model mismatch. On the other hand, model-free reinforcement learning(RL) algorithm can be designed using actual behavior of AUV plant in an unknown environment and the learned control may not get affected by model uncertainties like a classical control approach. Unlike model-based control model-free RL based controller does not require to manually tune controller with the changing environment. A standard RL based one-step Q-learning based control can be utilized for obstacle avoidance but it has tendency to explore all possible actions at given state which may increase number of collision.Hence a modified Q-learning based control approach is proposed to deal with these problems in unknown environment.Furthermore, function approximation is utilized using neural network(NN) to overcome the continuous states and large statespace problems which arise in RL-based controller design. The proposed modified Q-learning algorithm is validated using MATLAB simulations by comparing it with standard Q-learning algorithm for single obstacle avoidance. Also, the same algorithm is utilized to deal with multiple obstacle avoidance problems.