The nonlinear mechanical properties of ballasted tracks along railways result in complex dynamics of the vehicle–track systems. Employing localized characterization of ballast and a simplified model might underestimate the vehicle–track system’s dynamical responses and safety. This paper presents a new dynamical stiffness model of railway ballast by incorporating the ballast’s time-varying, nonlinear viscoelastic, and elastoplastic properties. The new nonlinear stiffness model is a versatile tool that comprehensively characterizes the ballast characteristics of displacement-dependent stiffness, frequency-dependent stiffness, hysteresis, and time/space-varying features. These features, widely reported in previous experimental research, can now be effectively understood. Conventionally, to characterize time/space-varying characteristics of ballast along the track, ground-penetrating radar (GPR) has been used as the most efficient approach to survey railway networks quickly and to infer track dynamical properties. Aiming to improve the present technique for characterizing time/space-varying properties of ballast stiffness by using a GPR signal, the adaptive optimal kernel time–frequency representation (AOKTFR) method is used to process a typical GPR signal from a railway ballast scanning. It is found that the results of AOKTFR exhibit a clear time-varying pattern and precise frequency modulation. In contrast, the conventional time–frequency methods failed to give a clear time-varying pattern. The results illustrate that AOKTFR is a practical approach for processing the time-varying nonlinear signal of GPR and correlating it with the time-varying nonlinear stiffness of ballast.
To enhance the balance between lateral stability and energy efficiency, we propose an adaptive compound controller based on phase plane analysis for four-wheel independent drive electric vehicles (4WID-EVs). The adaptive stability and energy-saving controller (SEC) is designed with a three-layer structure. The upper-layer controller employs model predictive control (MPC) to compute the external yaw moment based on the desired yaw rate and side slip angle derived from a reference model. The adaptive-layer controller utilizes a phase plane diagram to evaluate vehicle stability and reduces unnecessary external yaw moment consumption by accounting for the vehicle’s steering state and battery’s state-of-charge (SOC) level. The lower-layer controller implements an optimal torque distribution algorithm to minimize an objective function that considers tire workload, energy consumption, and smooth motor control. Numerical simulations are performed in MATLAB/Simulink using three distinct steering angles to evaluate the performance of the proposed control strategy. At each steering angle, the SEC’s stability and energy efficiency are compared to those of the energy-saving controller (EC) and stability controller (SC) under varying battery charge levels. The results indicate that, at small steering angles, the vehicle operates in a highly stable state, enabling a reduction in the external yaw moment to achieve substantial energy savings. As the steering angle increases, the vehicle approaches a critical stability state, where the external yaw moment is applied to maintain lateral stability. Furthermore, as the SOC decreases, the SEC strategy will increasingly prioritize energy savings. Simulation results verify that the SEC strategy effectively balances lateral stability and energy savings while maintaining consistent performance across a range of operating conditions.
We propose a compound control framework to improve the path tracking accuracy of a four-wheel independent steering and driving (4WISD) vehicle in complex environments. The framework consists of a deep reinforcement learning (DRL)-based auxiliary controller and a dual-layer controller. Samples in the 4WISD vehicle control framework have the issues of skewness and sparsity, which makes it difficult for the DRL to converge. We propose a group intelligent experience replay (GER) mechanism that non-dominantly sorts the samples in the experience buffer, which facilitates within-group and between-group collaboration to achieve a balance between exploration and exploitation. To address the generalization problem in the complex nonlinear dynamics of 4WISD vehicles, we propose an actor-critic architecture based on the method of two-stream information bottleneck (TIB). The TIB method is used to remove redundant information and extract high-dimensional features from the samples, thereby reducing generalization errors. To alleviate the overfitting of DRL to known data caused by IB, the reverse information bottleneck (RIB) alters the optimization objective of IB, preserving the discriminative features that are highly correlated with actions and improving the generalization ability of DRL. The proposed method significantly improves the convergence and generalization capabilities of DRL, while effectively enhancing the path tracking accuracy of 4WISD vehicles in high-speed, large-curvature, and complex environments.