Study On Short-term Wind Power Combination Forecasting And Its Uncertainty

In recent years,the consumption of non-renewable energy such as traditional petroleum and coal has intensified,and the environmental pollution problems brought about by it have become increasingly prominent.Because of wind power as a green pollution-free renewable energy,it is inexhaustible and has become a hot spot in the field of new energy research.Increasing the grid-connected capacity of renewable energy such as wind energy is the trend of the times.However,the inherent properties of wind energy(randomness,intermittent and volatility)make wind power generation highly uncertain,which is a huge challenge for power systems.A lot of researches on wind power prediction and uncertainty have been carried out at home and abroad.The research shows that the prediction models based on different algorithms have different manifestations on the original data.If they can be combined,the output power of the wind turbine can be more accurately.Make predictions.In this paper,we use the maximum entropy to establish a short-term power combination forecasting model for BP,wavelet and relevance vector machine RVM,and study the short-term prediction of wind power.The cloud model is used to analyze the uncertainty of wind power forecasting.The main research contents are as follows:1.Wind energy is the basis of wind power generation.The wind speed is different for the output power of wind turbines.Based on this,firstly,the principle of wind power generation and the characteristics of power curve are studied.At the same time,the influencing factors of wind power generation and the classification of wind power generation are analyzed.The wind power forecasting has laid a theoretical foundation.2.With the increasing grid-connected capacity of wind power generation,the accuracy and range of wind power forecasting are critical to the stable operation of high-ratio renewable energy power systems.Based on this paper,this paper proposes a combination of multiple model complementary prediction functions,and deeply analyzes the operation principle and characteristics of BP,wavelet and relevance vector machine RVM prediction models,and combines BP,wavelet and relevance vector machine RVM into a whole combination through information fusion strategy.The prediction model uses a single prediction model and a combined prediction model to predict wind power.3.The main factor of uncertainty in wind power generation is that the wind speed has a strong randomness.Secondly,in the process of wind power conversion,the uncertainty caused by the failure of the fan is also caused.Based on the above factors,if the uncertainty of any factor is studied unilaterally,it is one-sided.Therefore,it is possible to extract the uncertain characteristics of the wind speed from the historical data,which is comprehensive.At present,it is generally believed that the uncertainty of wind power should not be described by Gaussian distribution.Based on this paper,the Gaussian cloud distribution is introduced to describe the uncertainty of wind power.4.Based on the cloud model,the uncertainty of wind power prediction is studied.Based on this,the uncertainty of wind power prediction in the next two days is analyzed.At the same time,the confidence interval at 90% confidence level is analyzed.The deterministic evaluation system is evaluated.