| With the rapid development of intelligent power grid and explosive growth of the power system’s information, during the function of the power system, it will produce huge amounts of data, in the process of production and transmission, the data is easy to be disturbed by various factors and generates bad data, such as lightning, magnetic field and the fault of the transmission equipment and so on. The most basic demand is that the basic data must be accurate in the power system state estimation. Because the data is massive, the cost of artificial screening is very high, so there is an urgent need for a new solution.In this thesis, after studying the characteristics of the bad data in power system, it gives a solution that using the data mining algorithm to solve the bad data problem firstly, through the deep learning of the traditional KMeans algorithm, it gives a new method that the improved KMeans algorithm based on particle swarm method optimized by inertia weight. The new method can solve the problem that the traditional KMeans clustering method couldn’t determine the center of the clustering efficiently after the integration. Combining the PSO algorithm which has the characteristic that can determine the center of swarm quickly with traditional KMeans algorithm, it makes itself more efficient and accurate, and it proves its feasibility by simulation in MATLAB. Then in order to deal with the large amount of data in the power system and the low computational efficiency of the traditional algorithm, it puts forward a parallel algorithm of particle swarm optimization of the KMeans solutions, next using the Hadoop technology to realize the integration algorithm., it uses Map function to realize the improved solution of KMeans algorithm optimized by PSO, finally, we do the experiment on the Hadoop platform, and demonstrates the validity of the results. In this experiment, we use IEEE-14 as the basic data set to generate the simulated data set, do the contrast test by way of single mode and cloud mode, we build a cloud cluster in the laboratory to test the performance of the new algorithm, the result of the experiment shows that the cloud computing mode of the KMeans algorithm optimized by PSO is better than traditional algorithm and has better performance. |
