Classification of Power Quality Disturbances using Hilbert Huang Transform and a Multilayer Perceptron Neural Network Model

Disturbances in power quality have increased due to the use of electronic equipment, causing deviations in current and voltage waveforms, which can cause many failures and damage to equipment used in different demand points. Therefore, an efficient disturbance detection method is required in order to provide relevant information regarding its occurrence. However, there are many difficulties detecting disturbances throughout traditional data extraction methods. These methods have not been able to perform the detection process with the efficiency, speed and accuracy required for this type of work, due to the non-stationary and non-linear behavior of these disturbances. In this study, the Hilbert-Huang Transform and the Multilayer Perceptron Neural Network model are implemented in order to detect and classify disturbances in power quality. Eight common types of disturbances were analyzed based on the parameters stated in the IEEE 1159 standard. By means of instantaneous frequencies and intrinsic mode functions of each disturbance, the neural network is trained for the classification of these disturbances. The implemented method reached a precision percentage of 94.6, demonstrating the versatility and great potential that this method provides when detecting disturbances in power quality.