Soft computing based techniques for comparative analysis of wind speed and power prediction

Abstract

Currently industrial, agricultural, residential and economical development of any country mostly depends on energy availability and per capita energy consumption by its peoples. Today power demand in various sectors are increasing with increasing population, so it is very challenging task for power generating company to maintain proper balance of demand & supply . Currently most of the countries of the world uses fossil fuel as a main source for power generation, so pollution / Carbon dioxide emissions into the atmosphere from burning of fossil fuels has become a very serious concern all around the world, to counter this problem there is strongly need of such types of energy sources that are capable to generate adequate amount of power with lowest possible pollution, cost and highly reliable to insure the balance between supply and demand. Additional power requirement and integration of renewable energy sources into smart grid, forced the power generating companies to divert their attention towards power generation from renewable sources along with conventional sources, as power generation from the conventional sources faces many challenges such as pollution, adequate availability, storage and security, to mitigate the above problems renewable energy sources may be better alternative. So, many countries are expensing major portion of their energy fund into the development of renewable setup for power generation. In India currently major sources of renewable power are wind and solar out of which wind power contribution is slightly more as compare to solar power, but power generation from wind puts many barriers in term of intermittent nature, frequency instability and availability with certain speed range that is capable for power generation from wind mills. Although above challenge cannot eliminate completely, but it can be minimize with the help of forecasting technique / method that have accuracy / error up to certain level which will be acceptable / reliable for power generation. In this regard I have used the different soft computing based techniques such as NARX, SVM tuned NARX, GA-NFIS, PSO-NFIS, for comparative analysis of wind speed and power forecasting and their results are compare with benchmark method like Persistence method.