Electricity Demand Forecasting for a Smart City

Abstract

With 3.3 billion people living in cities across the globe, a number that is expected to get double by 2050, the need for cities that can drive sustainable economic growth and prosperity has never been more apparent. Therefore, the world is shifting towards the “Smart Cities” i.e. a city that takes a holistic approach towards spanning infrastructure, operations and people. Power generation and distribution infrastructure of Smart City should be built on Smart Grid technologies, which will integrate with local power demand patterns, grid supply variations and a well-defined operational process. For optimal operation of electric power plants, electricity demand must be followed by electrical generation. The production, transmission, and distribution of electricity requires possible forecasting of the electricity demand for efficient, secure and economic utilization of electrical infrastructure. Demand forecasting is also very important in order to reduce input costs and the prices of electricity. A reliable and accurate electricity demand forecasting systems are required. This thesis presents a solution methodology using artificial neural network for monthly electricity demand forecasting. It is implemented on historical weather related data i.e. temperature, rainfall, humidity, wind speed, precipitation and historical demand data for demand forecasting and the accuracy of the result is estimated by comparing the values generated by model with actual demand. Amloh, Patiala city load data is used for training and testing collected from Amloh grid. The results obtained are compared with the results from regression model of forecasting and accuracy of neural network is better as compared to existing modules