Abstract: Aiming at the problem that single neuron PID controller contained output noise, which resulted in degradation of the control performance, we proposed an improved single neuron adaptive PID control algorithm based on Kalman filtering theory. By introducing the concept of state space, the algorithm used recurrence method in the time domain to filter data. The output value of the controlled object was processed by Kalman filtering algorithm and then returned to closed loop control system. The experimental results show that the improved algorithm can effectively reduce output noise of the control system, which is close to an ideal state without noise, and improve the control performance.

Key words:  Kalman filtering, denoising, single neuron PID control, state space, recursive estimation