Pitch Optimization System Based On Multi-scale Wind Power Prediction

Due to the volatility,randomness and intermittency of wind power generation,the integration of wind power into the grid will have an impact on the safe and stable operation of the power system.At the same time,the wind farm will have to maintain a high rotating reserve capacity to stabilize the grid voltage and seriously reduce the economic benefits of wind power generation.Predicting wind speed and wind power,and multi-scale forecasting wind energy output can effectively solve this problem.At the same time,when the wind speed is too low,it will cause the wind turbine to be in an under-power running state,and when the wind speed exceeds the rated wind speed,it will cause the output power of the fan to be greater than the rated power.In order to ensure that the wind turbine can convert the random wind energy in the case of variable wind speed into stable output of electrical energy,the pitch angle of the wind turbine needs to be adjusted.However,adjusting the pitch angle too frequently will greatly increase the failure rate of the fan and shorten the service life of the fan.Therefore,this subject combined with deep belief network to carry out research to improve the accuracy of wind power prediction,and use the predicted wind power data to optimize the wind turbine pitch angle,and developed a wind power wind power forecasting and pitch angle optimization system,Has important theoretical and practical significance.In order to improve the prediction accuracy of wind power,reduce the rotation reserve capacity,and at the same time reduce the pitch angle adjustment load,this subject conducts the following research on the prediction model of wind power and the optimization of pitch angle:First,taking the actual production data of a wind farm in Inner Mongolia as an experimental object,the original data obtained in the Supervisory Control And Data Acquisition(SC AD A)system is mixed with strong noise,and the complete set of empirical modalities through adaptive noise Decomposition(CEEMDAN)method and correlation coefficient method are combined to reduce the noise of the original data with strong noise,use CEEMDAN to decompose historical data,and use the correlation coefficient method to filter the IMF component obtained by CEEMDAN decomposition to remove the pseudo IMF component And reconstruct the remaining IMF components;the power spectrum analysis method and the Lyapunov exponent method are used to analyze the chaotic characteristics of the wind power data,and the embedding dimension and delay time of the phase space reconstruction are obtained;Secondly,sort the importance of each variable to wind power in the future through the classification regression tree algorithm(CART),and select relevant variables that have a significant impact on the wind power forecasting effect through data analysis;in order to ensure that the selected data is valid data,The input data finally selected is normalized to eliminate the influence of the dimensional gap between the data on the prediction results.Then,the enumeration method is used to optimize the parameters of Deep Belief Networks(DBN)to establish a wind power prediction model of wind farms based on DBN;a least squares support vector machine(least squares support The error correction strategy of veotor maohine(LSSVM)algorithm implements dynamic correction to the prediction model to further reduce the model prediction error.Aiming at the problem of wind power prediction in wind farms,a wind power prediction method based on deep confidence network is proposed.Experimental results show that the proposed wind power forecasting method efectively improves the forecasting accuracy compared to other models.Finally,the minimization of the difference between the actual power and the set value and the minimization of the pitch angle adjustment load are the optimization goals,the pitch angle is the control variable,and the non-dominated sorting multi-objective evolution is used according to the prediction results of the wind power prediction model.Algorithm(NSGA-II)algorithm optimizes the fan pitch angle.And using the established wind power prediction model and pitch angle optimization method,using C#to develop a set of wind farm wind power prediction and pitch angle optimization control system,has certain practicality.