VHDL Modeling of an Artificial Neural Network for Classification of Power Quality Disturbance

This paper describes the design and modeling of an artificial neural network (ANN) classifier using VHDL. This classifier is targeted primarily to classify the six different types of power quality disturbance. The high level architecture comprises of a control unit and a neural network datapath. The control unit is further divided into five interconnected sub modules: bus master, ram, pseudo random number generator, error calculator and trainer. Univariate randomly optimized Neural Network (uronn) algorithm is employed to model the neural network. Proper simulation is carried out to verify the functionality of the individual modules and the system. In addition, the algorithm was also implemented in Matlab and C as comparison with the hardware implementation in VHDL. Comparisons, verification and analysis made validate the advantage of this approach. Currently, the classification average accuracy is 77.53%. The classifier also has the potential of being extended to classify other kinds of power quality disturbances. Key-Words: Artificial neural network, power quality, VHDL, FPGA, classification, modeling