Automation of Power Theft Detection Using PNN Classifier

Research areas:

• Volume 3,Issue 4, Jan. 2015

Year:

2015

Type of Publication:

Article

Keywords:

Classifier, Pattern Classification, Power Theft, Probabilistic Neural Network, SVM, Training

Authors:

Srihari Mandava; Vanishree J; Ramesh V

Journal:

IJAIM

Volume:

3

Number:

4

Pages:

118-122

Month:

Jan.-Feb.

ISSN:

2320- –

Abstract:

Recently, detecting the fraudulent electricity customer has been an active area of research. This is a problem faced by almost all power utility companies. This paper presents a novel idea of detection of power theft using probabilistic neural networks (PNN). This model selects the suspected customer on the basis of its electricity consumption behaviour. The probabilistic neural network algorithm represents the likelihood function of given class as the sum of same isotropic Gaussians. Twelve monthly power consumption details are given as input to neural network, which is trained to give the desired output.

Full text: IJAIM-373_final.pdf