WIND POWER DENSITY PREDICTION METHOD USING NEURAL NETWORK TECHNIQUE

摘要

PURPOSE: A wind power density predicting method using a nerve network analyzing technique is provided to supply a scientific base for establishing a complex position strategy for wind power generation and ground wind power resource potential calculation by developing a geographical statistics model about ground wind power density analysis. CONSTITUTION: Wind power density which is an output variable and input variables selected from among surface roughness, altitude, a relative altitude difference, wide area openness, a relative gradient, a gradient, an average altitude, the highest altitude, the lowest altitude, a relative relief, and a distance form a coast (S100). Nerve network analysis is performed by using the output variable and the input variables. An RMSE value and the number of hidden nodes are calculated by each nerve network analyzing model (S200). A correlation coefficient is calculated by using the RMSE value and the number of the hidden nodes. A nerve network analyzing model having the closest correlation coefficient to 1 and a small number of hidden nodes and RMSE values is estimated (S300). [Reference numerals] (AA) Start; (BB) End; (S100) Step of inputting variables; (S210) Step of first analyzing; (S220) Step of second analyzing; (S230) Step of third analyzing; (S240) Step of fourth analyzing; (S250) Step of fifth analyzing; (S300) Step of estimating a nerve network model