Predicting First-Order System Parameters Using Neural Networks Trained on Multiple Test Signals

Determining the parameters of First Order System, such as the system Gain (k) and the Time Constant (τ) is crusial for control system identification. Traditional parameter estimate methods are frequently dependent on the manual assessment of specific test signals, such as step, ramp, or sinusoidal inputs. However, when trained on different test signals, neural networks provide numerous key advantages that make them especially well-suited for this task. This paper describes a neural network-based method for predicting the gain (k) and time constant (τ) of a first-order system using multiple test signals. A fully connected neural network was trained to predict these characteristics using time-series data. The results reveal that the network can effectively predict system parameters, proving the viability of using deep learning approaches to solve system identification challenges.