EXPERIMENTAL DETERMINATION OF ELECTRICAL AND MECHANICAL PARAMETERS OF DC MOTOR USING GENETIC ELMAN NEURAL NETWORK

The most suitable and generalized neural network that represents the control system dynamics is the Elman Neural Network (ENN). This is due to its ability to memorize and emulate the system states. Moreover, ENNs learned by Genetic algorithms are found to be more representative to system order in terms of its structural complexity in comparison to those learned by back propagation algorithm. This facility is utilized efficiently to find the minimum ENN structure that represents the discrete-time state space model of the DC motor. Then by comparing the ENN weights with the well-known discrete-time state space equation in terms of the motor physical parameters (moment of inertia, torque constant, armature inductance, etc.), these parameters can be obtained.