Improving Quality of Smart Grid Data by Functional Data Analysis

As an important industry to the national economy and people’s livelihood, the power industry has become a dataintensive industry after years of information construction. Among them, electricity data covers the whole industry and tens of thousands of households, so it is of great significance and value to conduct in-depth analysis and research on large power data. The existing power consumption data has the problem of low quality, which is mainly manifested in data missing and anomaly, which has a great impact on the accuracy of data analysis. Therefore, the cleaning of power consumption data is the first problem that industry personnel will face. However, there are many problems in the existing data cleaning methods, which have failed to achieve good results in the business scenario of power consumption data. Therefore, this paper presents a daily power data cleaning model based on FDA, which successfully finds and eliminates abnormal values of power data, and can repair the missing values. The experimental results show that the data cleaning method proposed in this paper has a good effect on the real electricity data scenario.