Aiming at the control of nonlinear and time-delay and uncertainty plant，a gain-adaptive single neural PID controller is presented，which is independent of the model of plant. Due to the links of neural adjustable online，the arithmetic has the ability of self-learning and self-adaptability，and adjusts the proportion coefficient K which based on the self-adaptive PSD arithmetic，then makes the gain-adaptive single neural PID controller. The simulation in MATLAB/SIMULINK shows that the controller having the ability of self-learning and self-adaptive function and strongly robust，is a simple and applicable controller. The arithmetic is especially applied to uncertain object for placidity control.