Discrete Wavelet Transform based Probabilistic Neural Network Technique of Detection and Classification of Power Quality Disturbances Suhail Khokhar 1 , Muhammad Akram Bhayo 2 , Adnan Ahmed Arain 3 , Mohsin Ali Tunio 4 1,2 Electrical Engineering Department Quaid-e-Awam University of Engineering, Science & Technology Nawabshah, Pakistan 3 Computer System Engineering Department Quaid-e-Awam University of Engineering, Science & Technology Nawabshah Pakistan 4 Electrical Engineering Department MUET. S.Z.A.B. Campus Khairpur Mirs, Pakistan Abstract:- Automatic detection and classification of Power Quality (PQ) disturbances plays a vital role for the protection of power system. In this paper, a Discrete Wavelet Transform based Probabilistic Neural Network (DWT-PNN) approach has been proposed for the automatic detection and classification. The DWT is used for the detection of the PQ disturbances. The statistical features such as energy values of the detail and approximation coefficients are obtained using multi-resolution analysis of DWT. Then the statistical features are used as the training data to the PNN classifier. The results demonstrate the proposed DWT-PNN classifier effectively detects and classifies the PQ disturbances with high accuracy. Keywords: Power quality disturbances, discrete wavelet transform, probabilistic neural network I. INTRODUCTION The PQ can be guaranteed by monitoring and classifying the disturbances using measurement instruments [1]. The instruments must be able to accumulate enormous quantity of data measurement such as voltages, currents, frequency and disturbance occurrence time duration. Since, the traditional PQ measuring instruments cannot automatically discriminate the PQ disturbances and require offline analysis from the recorded data. Therefore, in this research, the idea of a computational intelligent based instrumentation is suggested to measure the PQ disturbances automatically. In general, the main reasons for the PQ disturbances are the enormous implementation of switching equipment, capacitor energization, unbalanced loads, lighting controls, computer and data processing equipment as well as inverters and converters [2]. The PQ disturbances are created from the utilities and the customers driven loads. The customers' loads and equipment that create PQ disturbances consist of power electronic converters, pulse modulated loads, fluorescent and gas discharge lightings, machine drives, certain rotating machines and magnetic circuits based components. The grounding and resonance problems in the utility subsystems of transmission and distribution networks cause PQ disturbances [3]. In particular, short circuit faults in power distribution network, switching operation of heavy industrial loads and energization of large capacitor banks may cause PQ disturbances. For instance, voltage sag, swell, interruption and transients disturbances [4]. The application of switching devices and loads such as converters and inverters cause steady-state waveform distortion disturbances in voltage and current signals such as Direct Current (DC) offset, harmonics, inter-harmonics, notch and noise. The utilization of the electric arc furnaces create flicker disturbance [5]. Ferro-resonance, transformer energization, or capacitor switching and lightning lead to spikes disturbances. Although the PQ disturbances are created due to the aforementioned types of loads yet these devices are malfunctioning due to the induced PQ disturbances. The PQ disturbances cause various problems to power utilities and customers; for example, malfunctions, instabilities, short life span and breakdown of electrical equipment. Harmonics disturbances create power losses in transmission lines, power transformers and rotating machines. The most important and the most frequent PQ disturbance is the voltage sag due to short circuits which have a huge economic impact on end users [6]. The traditional methods of PQ monitoring exercised by the utilities are normally based on visual inspections, which are indeed laborious and time-consuming. Therefore, a highly automated hardware and software based International Journal of Computer Science and Information Security (IJCSIS), Vol. 15, No. 5, May 2017 412 https://sites.google.com/site/ijcsis/ ISSN 1947-5500