Pylon Anti-Vandalism Monitoring System using Machine Learning Approach

A country's economic growth is largely dependent on the stability of its energy sector. The growth of Kenya's power sector is greatly supported by Vision 2030 development blueprint, which is geared at transforming the country into a “middle-income country that is able to provide a high quality of life to all of its citizens by the year 2030”. Though major investments have so far been made in the generation and transmission of power, the distribution sub-sector is yet to reap the full benefits due to a number of challenges ranging from financial to operation and maintenance. Transformers are key components of any power distribution network and despite their specified long operational life, they often fail within a short period of commissioning due to various reasons including vandalism. This has led to great losses to the power distribution sub-sector and consumers alike. This paper proposes an anti-theft system that uses machine learning technique to monitor the pylon's environs against vandals. Results of the study reveals that the system is capable of detecting human images with confidence levels between 68% and 97%.