Research On Reliability Evaluation Of Power System Based On Renewable Energy Resource

With the development of society,the traditional fossil fuel has been unable to meet the needs of the human society,so the pollution-free renewable energy has been paid more and more attention,such as solar energy,wind energy,tidal energy and so on.After the study on the technology for long,and with the continuous improvement of equipment,the wind power has made great progress by using the existence of fixed nature conditions.Wind power is more suitable for the requirements of green production,and it has become one of the important methods of the current power generation system.The main condition depended by the wind power generation system is the wind energy.The wind is random in nature,with the characteristics of strong instability and large volatility.After access to the power grid,the penetration will appear great fluctuation,causing the instability of the power system and affecting the actual industrial production.Therefore,the accurate prediction of wind power output has become an important factor in the reliability of wind power generation system.The accurate prediction results can not only reduce the spinning reserve capacity of power grid,but also reduce the cost of wind power grid operation,to achieve the purpose of reasonable scheduling.At present,one of the main research direction of wind power prediction is to control it with neural network,and to forecast the direct output power with artificial neural network.The neural network model is an important condition for wind power prediction.In order to improve the reliability of power system,determine the scientific prediction model,and process data,the concrete process is as follows:Firstly,taking the complex data of the factors affecting the wind power into consideration,the first procedure is a preliminary processing of the initial sample data calculated by the ant colony clustering algorithm.With global search ability,ant colony clustering could firstly determine the preliminary data rules and categories and numbers,to provide scientific basis for the classification of follow-up data processing and the necessary basis for construction of network data model.Secondly,the fuzzy neural network is created largely dependent on the data clustering and the membership degree.Taking the number of category obtained by the initial colony of Poly as initialization conditions of fuzzy C mean clustering,the membership data of each data relative to the whole clustering can be calculated through fuzzy C mean clustering processing.The result is the fuzzy processing data as well as the initialconditions of the network.Then,simulation training should be taken through fuzzy neural network.Concerning certain fuzziness of the measure of the things in the real production process,the forecast of fuzzy neural network can reflect realistic features better and get accurate result coming up to the logical thinking.In order to improve the practicability of the system,the system of management connecting database should be developed,which enables to manage and search data and even database.It has preferable practicability that input data to the database constantly and analyze it through the data in the database.