The bus dispatching system is an important part of urban public transportation. We need to be able to promptly detect changes in the current operation status and passenger flow

so as to effectively improve the performance of the Intelligent Bus Scheduling System in a dynamic environment. This paper proposes an adaptive PID and Robust Control Hierarchical Hybrid Architecture. The online adjustment method of PID parameters using BP Neural Network is used to achieve rapid and accurate dispatching of vehicles. Among them

the upper layer uses the Robust Control controller based on LMI Optimization to control the robustness

and is paired with a Disturbance Observer for control

which can effectively suppress external interference and parameter changes. The adaptive module that combines NSGA-II Multi-objective Optimization and Deep Reinforcement Learning can not only perform Dynamic Weight Distribution in real - time

but also pre - adjust the parameter values. Considering factors such as passengers' waiting time

equipment energy consumption

and Safety Redundancy in combination with the real - time passenger flow

and training the dispatching system with historical data

the dispatching system can foresee the occurrence of emergencies and enhance the system's self - learning ability. Three types of scenarios (normal passenger flow scenario

passenger flow fluctuation scenario

and communication delay scenario) are tested through the built Digital Twin platform. Compared with the traditional PID Control

this algorithm has the advantage of small Steady-state Error. Compared with single Robust Control

the recovery time is significantly reduced. When facing multiple conflicting objectives

it achieves a coordinated balance between energy consumption and waiting time. This method can obtain a dispatching method with high accuracy and Disturbance Rejection for application in the Intelligent Transportation Hub.