Optimization Study Of Grid Data Applied Algorithm Based On Hadoop

The rapid development of modern power grid leads to rapid growth of data,which shows that the power system has entered the era of big data.This increases the power system's data storage and management difficulty,the existing technology and hardware is difficult to meet the demand of power network operation,control and analysis.In order to enrich the lower power system data redundancy and improve resource utilization,this paper develops a new type of efficient data prediction algorithm,which based on Hadoop and improved on the MapReduce network model.And through comparing and analyzing various algorithm results to examine that if this algorithm meets the current grid system development requirements.Based on the research status of big data at home and abroad and the study of power grid,this paper analyses the distribution and characteristics of grid data,combining with big data processing method which based on power grid to proposes a MA BP algorithm,and the algorithm has been simulated experimented and by comparing various algorithms of wind farm output and measuring the absolute error of data to verify whether the proposed algorithm meets the relevant requirements.Analyze whether the algorithm is improved compared with the traditional algorithm by the processing time.Finally,a wind farm in Xinjiang power grid is taken as an example to verify the feasibility of the solution.The following are the main research results of this study:(1)This paper designs a data processing algorithm,which based on big data platform architecture,combined with information fusion technology and choose the Hadoop as the technical support to integrate a large number of data from three aspects.This also improves the processing speed of the data by electric power system and provides a reliable guarantee for the further application of grid data.(2)the research focuses on the existing problem of the traditional prediction algorithm analyzing the data.It puts the BP network mutual information together to improved the MapReduce model,and use the condition attributes and decision attributes of mutual information measurement model to effectively processed data redundancy.Comparing BP network,PCA-BP network,and the actual value and the algorithm proposed by this paper,the wind power output results of absolute error results by BP network,PCA-BP network,the algorithm show that the algorithm proposed in this paper meets the requirements of data grid prediction with high accuracy and a significant advantage.(3)According to the current development of Xinjiang Power Grid,combined with the operation process of power system and the operation characteristics of wind power generation,the information fusion technology is applied to construct the data application integration platform of power grid.By observing the test results,we can see that "MA-BP" algorithm matches the actual capacity better.It can effectively eliminate the redundant variables and meet the demand of big data capacity prediction.It achieves the fusion goal and can be popularized and applied.