Comparison Of Characteristic Features By Implementing The Hindmarsh-Rose (Hr) Neuron Model With Google Colab, Op-Amp And Arm Controllers

Today, artificial intelligence technologies are advancing day by day, and neurons, which are the building blocks of artificial neural networks, one of the artificial intelligence technologies, are supporting this advancement. By studying the biological structures of neurons found in the human brain, dynamic models are created using mathematical differential equations. These models are implemented using analogue circuit elements or digital controllers such as computers to investigate their dynamic properties. Within the scope of this study, time series, phase portraits, bifurcation diagrams, and Lyapunov exponent spectrum analyses were performed using Google Colab, a new generation numerical analysis development environment, to reveal the dynamic properties of the Hindmarsh-Rose (HR) neuron model, which is effectively used in neuron modelling. In Google Colab, solutions were obtained using the Euler, Heun, RK4, solve_ivp, Adams-Bashforth/Moulton, and Z-transform methods in the numerical analysis solutions of the HR model; it was determined that the RK4 method, which has sufficient speed and high accuracy, is more suitable for microcontroller applications. Subsequently, an HR analogue circuit model was designed using Op-Amp operational elements in the OrCAD PSpice program, and simulations were performed to validate the Colab dynamic numerical analysis results. Finally, simulation results were obtained using a microcontroller with a Cortex-A72 (ARM v8) processor, which confirmed the Colab dynamic RK4 numerical analysis results of the HR neuron model. The applicability of membrane potential, fast recovery, and slow adaptation conditions to mini-systems for data processing was demonstrated.