Wind Turbines Yaw Control Strategy Based On ART2wNF Network Combined With Bayesian Inference

With the development of society and economy, energy conflicts around the world have become increasingly prominent. In view of the wind energy is safe, clean, abundant, stable, etc., to increase the use of wind energy will effectively alleviate the energy crisis and reduce environmental pollution. The development of wind power is the most important use of wind energy in the form, the control system will have a direct impact on the efficiency and service life of the wind turbine, wind power is the key component of the yaw system to maximize the capture wind energy and avoid the frequent rotation, which makes the effective control of wind turbine yaw system becomes particularly important.On the basis of the yaw wind power systems and control technology works, the paper presented in conjunction with the Bayesian inference ART2wNF (Adaptive resonance theory with neoteny feature) network and vane control and hill-climbing algorithm combines yaw control strategy. Randomness against the wind, in the normal distribution of wind model, simulation got wind of the sample data, the establishment of a least squares fit under the wind model based filter out wind noise in the signal data for the realization of ART2wNF wind Signal network for clustering data preprocessing done. Since ART2wNF network when the sample self-organizational learning and clustering, the alert value is fixed, while the number of directly affect the number of categories of alert value, in order to achieve automatic adjustment ART2wNF network security value, the paper applied the shellfish Yates classifier principle, under normal wind model to calculate the probability of the sample subject to a number of new wind sample distribution, in order to adjust the reference posterior probability thereafter as the alert value, designed based on Bayesian inference adjustment mechanism ART2wNF network security value and improve the clustering effect of the wind signal samples, the wind appeared to resolve within a small range of variation of wind yaw questions focused foundation. By having the characteristics of juvenile continuation ART2wNF network pretreated wind data from organizational learning and clustering, combined with vane control and hill-climbing algorithm for vigilance parameter ART2wNF network is adjusted after each sample obtained clustering together type of center, the yaw position,yaw automatically completed.Through building simulation model to simulate the wind turbine yaw system in Matlab simulation to verify the feasibility and effectiveness of the combination of Bayesian network inference and ART2wNF wind turbine yaw system control strategy presented by the paper which not only can effective solution when the wind direction within a small range of variation (such as plus or minus 15°) yaw problem appears concentrated wind energy, but also effectively avoid frequent yaw motor rotation, it has important significance to improve the utilization of wind energy and the life of wind turbines.