Statistical Approach for Wind Speed Forecasting Using Markov Chain Modelling as the Probabilistic Model

Abstract

Electricity supply and demand have increased significantly over the years largely due to population and economic growth. However, the use of energy has witness tremendous challenge in recent years because of high cost of raw materials for generation (fossil fuels), environmental concerns and sustainability of resources. In recent years, the use of renewable energy resource takes a centered stage with wind energy as the front runner. This is largely due to its availability and maturity in technology. The biggest challenge for generating electricity from wind is its inherent property of speed intermittency; it also affects correct forecasting for future planning. Statistical approach using Markov modelling method is used to predict wind speed using dataset from a location collected over 24hrs period at an interval of 10mins. Results from this method shows it performs well during the early periods with Individual Absolute Error between 0.3-1.43 during the first five (5) periods of the forecast and for the last five periods the absolute error is from 1.3 to 3.0 making Markov probabilistic model good for very short term forecasting of wind speed.