In order to solve the negative effect of random disturbance on the height switching, the air suspension vehicle height adjustment system model was established with the air spring model. A neural network PID adaptive height adjustment controller was designed based on single neuron self tuning gain algorithm. To test performance of the controller, vehicle height control model was set up by Matlab/Simulink, and a real vehicle test platform was built with main equipment of Yaxing YBL6891 passenger car of air suspension, independently developed controller, inductance height sensor, assembled solenoid valve, debug system of electronic-controlled air suspension. Simulation and test were done in a continuous process of firstly increased and then decreased height. The simulation results showed that the height adjustment, pitch angle and lateral dip angle of PID controlled system had some strong oscillation, and the vibration phenomenon was obviously relieved in the control of the adaptive neural network PID controller. The improved height control system was relatively stable, and the control precision was improved. Vehicle test results showed that the open loop control system reached the target height with multiple oscillations. It was obvious that the control effect was not satisfactory over charge and discharge phenomenon, and the reaction time was time delay obviously. The system was stable to the target position after inflating 20 s and deflating 20 s separately. The response of the designed height controller was significantly accelerated, and stable to the target control position after inflating 5 s and deflating 3 s separately. The oscillation was significantly reduced, which was not obvious over charge and discharge. The designed controller meets the vehicle height switching.