In order to achieve fast and effective control for large time-delay system, the single neuron adaptive controller is used to control the large time-delay system. The controller can realize online real-time adjustment of controller parameters when the structure of the controlled object is changed. The single neuron PID control based on grey prediction which combines the single neuron adaptive controller with the grey prediction method overcomes the limitation of large delay in system by using the grey prediction value of the output instead of the measured value of the output and achieving a "pre-conditioning". Simulation results verify that the controller has excellent adaptability and robustness. The control performances are greatly improved.