NEURAL NETWORKS TECHNIQUE OF LAYER F2 CRITICAL FREQUENCY FORECASTING ABOVE STATION GAKONA (HAARP) AT THE ACCOUNT OF NEAR-EARTH SPACE PARAMETERS AND GEOMAGNETIC DISTURBANCE

In these work nonlinear correlation dependences of ionospheric layer F2 critical frequency above station Gakona (HAARP) from a number of solar-magnetospheric parameters are established. For this purpose the special technique based on creation of the "locking" block of an entrance package of the data, directed in Elman artificial neural network (ANN) is developed. As a result of neural network numerical experiments the forecast of critical frequency values for intervals from 0.5 till 2 hours is executed. Maximal effectiveness of ionospheric layer F2 critical frequency forecasting in sub-auroral region at use of the offered method are 93% and 83% for 0.5 and 1 hour forecast, accordingly. For long-term forecast on 1.5 and 2 hours corresponding efficiency are 75% and 67%.