Air-induced separation of flexible rod-like particle mixtures in a specific separator is numerically investigated using a coupled Discrete Element Method (DEM) and Computational Fluid Dynamics (CFD) approach. In the separator, the mixture of flexible rod-like particles of different lengths and material densities deposits under the effect of gravity, and a horizontal airflow stream intersects the particle flow, blowing lighter particles in the mixture to translate horizontally and allowing the heavier ones to fall downwards. The model particles represent flexible biomass materials, specifically tobacco and stem particles. The initial packing density of the particle mixture is 8% by volume. The physical mechanism that causes particle segregation is analyzed. Subsequently, parametric studies are performed to examine the effects of some critical parameters on the extent of segregation, including inflow air velocities, initial particle packing density, volume fraction of heavier particles, particle size distribution, and flow field geometry. Finally, a suggestion is proposed to promote particle segregation in such a type of separator.

Ensuring safe and stable vehicle operation during high-speed emergency obstacle avoidance and low-adhesion roads is the core priority of active research. This paper proposes a novel stability control method for distributed drive vehicles based on neural network adaptive terminal sliding mode control (NNATSMC), aiming to establish the upper layer control architecture for direct yaw moment control (DYC). Firstly, to address the problem that conventional sliding mode control (SMC) relies on the unknown upper bound, an adaptive switching gain based on barrier function (BF) is proposed under sliding mode state constraints to alleviate sliding mode chattering. Secondly, a terminal complementary sliding mode surface is designed based on complementary sliding mode theory, which ensures the terminal sliding mode to achieve rapid converges within finite time. In addition, the Radial Basis Function Neural Network (RBFNN) estimator can compensate for lumped uncertainties online, reducing the dependence of controller design on model accuracy, avoiding control performance degradation caused by the small convergence region of the sliding surface and further improves the robustness of the system. Furthermore, lower controller considering multiple constraints is established to output the four wheel torque. Finally, comparative simulations through four groups of operating conditions with different severity levels are carried out for verification; the proposed NNATSMC in this paper can better guarantee vehicle stability under extreme conditions and possesses better adaptability to multiple operation conditions.