Bad Data Identification And Correction Of Power Grid Planning Based On Data Mining

The scale of grid is becoming larger and larger with the rapid construction of smart grid. There are large amounts of unreasonable data in the process of grid planning snice BPA,the tool which is used for grid planning has limitations and shortcomings. Planners feel helpless facing such a scale planning grid. However, wrong and unreasonable data has a great impact on the convergence and results of BPA. On one hand,it will increase the workload of grid planning because wrong and unreasonable data will affect the convergence, on the other hand, it will reduce the accuracy of the results and could lead to farther deviation between planning and actual results and resulting in huge economic waste or more than this,it could leave hidden danger on system safety. Therefore, it becomes more and more important to find potential unreasonable data quickly and effectively in large-scale planning grid.For this problem, this paper studies a method to find and correct bad data in planning grid. First of all, on the basis of consulting the domestic and abroad document literature widely, this paper gives a detailed description on the current situation and level of development about the method of finding and correcting unreasonable data. Then, based on analyzing the characteristics and limitations of BPA which is used in power flow calculation, this paper forms a library of standard parameter using K-means clustering. Then the model of judging line type and model of calculating length of line are built in view of the problem of losing line type and line length in planning grid. A concept of boundaries is put forward and a method of judging line type through the included angle cosine between two straight lines is proposed. The BP neural network algorithm is used in the model of calculating line length by training neural network using correlation among line parameters. At last this paper proposes an algorithm suitable for identifying and correcting bad data in actual planning power grid and validates the effectiveness of the algorithm.