Research and Application of Short-term Power Load Based on Large Data Analysis

Abstract

We want to carry out an effectively distributed method to process large power data. Large data of electricity is first classified by big industries, and then by small industries, and finally put to use clustering to distinguish electro-sort of customers. This paper employed a computational framework named Spark based memory computing to conduct distributed computing, and adopted supporting vector machine on the Spark's, lastly the results were analyzed. Ensuring the final effect of the test, solution speed is greatly improved.