Virtual speed sensor for DC motor using back-propagation artificial neural networks

The use of artificial neural networks has enabled applications that would be impossible to achieve with conventional electronics, through to versatility that them have, they can be configured as needed and use as required such as classifiers, adaptive filters, controllers and predictors. This paper describes a experimental implementation of virtual speed sensor for DC motor using back-propagation artificial neural networks through voltage and current acquisition. In implementation process a comparative between both logsig and tansig activation function in hidden layer is made. The training of the neural network is performed by acquiring signals of voltage, current and speed sensor, later the response efficiency of the network is analyzed by comparing the result produced by physical sensor and virtual sensor implemented using neural networks. In this experiment training pattern values has a large range, a several process training with both small and large section range is made. Results show a good performance and a minimum error with both logsig and tansig activation function in hidden layer when pattern signal is full training.