Load abnormal value identification method

Abstract

The invention relates to the technical field of power load data mining, in particular to a load abnormal value identification method. The method comprises the following steps: carrying out power utilization mode classification on a load curve; carrying out load level classification on the load curve belonging to the normal power utilization mode; constructing an abnormal load data domain, and identifying abnormal values of a load curve belonging to a normal power utilization mode; and constructing an abnormal load data domain for identifying a negative abnormal load value in the abnormal powerconsumption mode by utilizing the maximum upper limit and the minimum lower limit of the abnormal load data domain, and identifying the abnormal load value of the load curve belonging to the abnormalpower consumption mode. Based on a spatial density clustering method and a K-center clustering method, load data can be reasonably classified, and different power utilization modes and different loadlevels in the same power utilization mode can be conveniently processed. According to the method for constructing the abnormal load data domain through the central limit theorem and the quartile difference, the range of the abnormal load data domain can be flexibly adjusted according to the required confidence coefficient.

Description

technical field [0001] The invention relates to the technical field of power load data mining, in particular to a load abnormal value identification method. Background technique [0002] The installation of smart meters makes the load records of power users change from electric quantity to time-series load curve. Compared with the quantity of electricity, the power time-series load curve contains the user's power consumption behavior and load level information at each moment. As an important basis for power suppliers to provide energy management services and users themselves to perform demand-side response, abnormal load values ​​in the power load curve often have a significant impact on the above decisions. Since the collected power load curve contains a large amount of data, it is difficult to accurately identify abnormal load values ​​manually, and it is necessary to develop an abnormal load value identification method for large-scale data. [0003] Since the distributio...

AI Technical Summary

Problems solved by technology

The main disadvantage of the boxplot is that the abnormal data domain constructed does not consider the local density characteristics of the load value at each moment

[0004]The above solutions have the following problems: For the situation of directly assuming the distribution of load data, due to the large randomness of the user's power consumption behavior, the power consumption behavior will vary with the It is difficult to accurately give the distribution of load data due to the passage of time and changes in the external environment; for the assumption of distribution first, and then the distribution test, because there may be load outliers in the load data, before the outlier identification , the distribution test process may be affected by abnormal data, so the test results are not reliable

Images