This work presents the introduction of a neural network-based optimization approach in the tuning of voltage-controlled circuits (such as active filters). A custom calibration system has been already presented by the same Authors. It was realized with a hardware interface and a dedicated software based on a modified version of a Differential Evolution algorithm. In this paper the implemented algorithms are described in detail together with a possible integration of the neural network synthesis to further enhance performance of the proposed system. As the first step in exploiting neural networks, in this paper they are used as a tool for speeding up the choice of initial values of the filter control voltages. Neural networks are used to replace a look-up table representing the relationship between filter parameters, the central frequency and the corresponding attenuation, and the control voltages. According to the obtained results, in such a way, the optimization time is shortened significantly.
|Titolo:||On the Introduction of Neural Network-based Optimization Algorithm in an Automated Calibration System|
|Data di pubblicazione:||2019|
|Appare nelle tipologie:||4.1 Contributo in Atti di convegno|