In order to ensure the power safety of important power customers, a new evaluation of power consumption status of important power customers based on the AHP(Analytic Hierarchy Process)-TOPSIS(Technique for Order Preference by Similarity to an Ideal Solution) algorithm is proposed by fully mining and applying the power big data. Firstly, a power consumption big data analysis platform based on the Hadoop architecture is built to provide a high-performance platform support for big data analysis. Secondly, nine evaluation indexes are constructed from the three dimensions of voltage, load and synthesis, which objectively and scientifically describes the power consumption status of important power customers. Finally, the AHP-TOPSIS algorithm is used to evaluate and analyze the voltage, load and comprehensive indicators respectively, thus, obtaining the evaluation values of three kinds of indicators. The power consumption status scores of important power customers are determined by the variable weight weighted summation. The rationality and feasibility of the method and algorithm are proved by example analysis and field verification. This method helps to promote the transformation from post fault emergency repair to warning beforehand. It has the multiple effects of ensuring safe power consumption, supporting accurate patrolling and active emergency repair serving.

This article starts with the importance of electric power marketing systems, introduces the technical characteristics of data mining and its application status in electric power marketing systems, thereby providing decision-making basis for the economic operation of power grids. And propose using C5.0 decision tree algorithm to deeply analyze the marketing data of the electric power marketing management system. The original C5.0 decision tree algorithm is improved by introducing information entropy, which improves its classification speed and accuracy. Experimental results on UCI machine learning dataset and power marketing dataset show that the proposed improved C5.0 decision tree algorithm has good classification performance and can meet the classification and prediction requirements in power marketing work.

Aiming at issues such as the inability of intelligent grid business data to be integrated across disciplines, and the inability of data resources to be intelligently analyzed and managed, a knowledge map construction method based on a unified data center for the entire business is proposed. On the basis of using big data technology to collect power grid multi-source data in the unified data center of the whole business, knowledge extraction is performed on structured, semi-structured, and unstructured data using semantic annotation methods to obtain knowledge entities, attributes, and relationships. Knowledge maps are constructed through knowledge fusion technology, and intelligent recommendation results and related information are recommended according to user search content. Experiments show that this method improves the precision and recall rate, and has better intelligent search and analysis capabilities.