Real-Time Control of Tank Water Level Using Gradient-Free Lyapunov-Based Neural-PID Controller

This study addresses liquid level control in nonlinear systems using a neural-PID controller for a laboratory water tank. Conventional PID controllers often underperform due to nonlinearity and time delays. Our proposed controller adapts PID gains via a neural network, validated through simulation and real-time experiments. The system’s mathematical model and Lyapunov stability ensure robust performance. Simulations show the NN-PID controller outperforms a conventional fuzzy controller, reducing overshoot, rise time, and mean square error. Real-time experiments confirm superior stability and robustness under varying conditions, indicating that the gradient-free neural-PID controller is a promising solution for nonlinear system control.