阅读说明：本技术 一种用户用电行为异常的监测方法 (method for monitoring abnormal electricity consumption behavior of user ) 是由 兰森林 彭奕 庄晓凌 陈吉 何海 于 2019-10-24 设计创作，主要内容包括：本发明的一种用户用电行为异常的监测方法,先行收集各个用户的电器开关的开合状态数据和电器功率数据作为数据集{x<Sup>1</Sup>,x<Sup>2</Sup>,x<Sup>3</Sup>...x<Sup>m</Sup>},再将该数据集分别导入高斯概率分布函数和Percentile检测算法,从而构件了两种检测用户用电行为异常的数学模型进行验证,提高了对客户用电量异常情况检测的效率和准确率,缩小检查范围。(The method for monitoring abnormal electricity consumption behavior of users firstly collects the on-off state data and the electric appliance power data of the electric appliance switch of each user as a data set { x } 1 ，x 2 ，x 3 ...x m And respectively importing the data set into a Gaussian probability distribution function and a Percentile detection algorithm, so that two mathematical models for detecting the abnormal electricity consumption behavior of the user are constructed for verification, the efficiency and the accuracy of detecting the abnormal electricity consumption condition of the client are improved, and the inspection range is narrowed.)

1, kinds of user's abnormal monitoring method of power consumption behavior, characterized by, including the following step:

step 1: collecting the on-off state data and the electric appliance power data of the electric appliance switch of each user as a data set through user electricity consumption data collecting equipment, wherein the data set corresponding to the on-off state of the mth type of electric appliance switch and the current power of the electric appliance is x¹,x²,x³…x^m}；

Step 2: will be provided withData set { x¹,x²,x³…x^mInputting a Gaussian probability distribution function, and carrying out monitoring and judgment on the electricity utilization behavior of a user to obtain a Gaussian model abnormal sample set;

and step 3: will { x¹,x²,x³…x^mInputting a Percentile detection algorithm, and carrying out monitoring and judgment on the power utilization behavior of a user to obtain a Percentile detection algorithm abnormal sample set;

and 4, step 4: marking a union set of all abnormal samples in the abnormal sample set of the Gaussian model or the Percentile detection algorithm, then carrying out power consumption abnormal behavior subdivision on the abnormal samples, and outputting a subdivision behavior sample set.

2. The method for monitoring abnormal behavior of users as claimed in claim 1, wherein the step 2 comprises the following steps:

step 2.1: data using switch on and off states and appliance power { x¹,x²,x³…x^mCalculating the expected value mu and variance sigma of the Gaussian probability distribution function²The calculation formula is as follows:

wherein x is⁽ⁱ⁾For a data set { x¹,x²,x³…x^mData in (c);

step 2.2: constructing a distribution function of Gaussian probability, wherein the distribution function of the Gaussian probability is as follows:

mu and sigma²The expected value and variance calculated in the above steps, x is the coefficient to be determined, the expected value of normal distributionMu determines its position, its variance sigma²Determines the amplitude of the distribution;

step 2.3: selecting a judgment boundary, and carrying out abnormal detection judgment on the electricity consumption behavior of the user:

and f (x) is 3 σ as a determination boundary. When f (x) is less than 3 sigma, judging that the test sample is an abnormal sample; and when f (x) is larger than 3 sigma, judging the test sample as a normal sample.

3. The method for monitoring abnormal behavior of users as claimed in claim 1, wherein the step 3 comprises the following steps:

step 3.1: the data of the opening and closing state and the current power of the electric appliance are set as { x¹,x²,x³…x^mSorting according to the sequence from small to large;

step 3.2: and carrying out grade division on each group of data, wherein the division formula is as follows:

wherein x is⁽ⁱ⁾For a data set { x¹,x²,x³…x^mData in (j), m is the total size of data, rankⁱThen it is the rank of the set of data;

step 3.3: and judging whether the data belongs to the abnormal electricity utilization condition of the user according to the grade condition of the group of data, wherein the data with the highest grade or the lowest grade is regarded as abnormal electricity utilization data.

Technical Field

The invention relates to a monitoring method for user electricity consumption behavior abnormity in the technical field of electricity consumption behavior abnormity identification.

Background

At present, series problems of electricity stealing, electric leakage and the like exist in the electricity utilization process of a user, the problems can cause power Loss, which is collectively called as Non-Technical Loss NTL (Non-Technical Loss). the Non-Technical Loss not only can influence the normal operation of an electric power system, but also can obviously influence the scheduling of the whole power grid, and even can generate safety accidents caused by actions such as electricity stealing and the like, so that electric power enterprises and the electric power system can inspect the electricity utilization actions of the user.

Disclosure of Invention

The invention aims to overcome the defects of the prior art, and provides monitoring methods for user electricity consumption behavior abnormity.

The technical schemes for realizing the aim are that monitoring methods for abnormal electricity utilization behaviors of users comprise the following steps:

step 1: collecting the on-off state data and the electric appliance power data of the electric appliance switch of each user as a data set through user electricity consumption data collecting equipment, wherein the data set corresponding to the on-off state of the mth type of electric appliance switch and the current power of the electric appliance is x¹,x²,x³…x^m}；

Step 2: will data set { x¹,x²,x³…x^mInputting a Gaussian probability distribution function, and carrying out monitoring and judgment on the electricity utilization behavior of a user to obtain a Gaussian model abnormal sample set;

and step 3: will { x¹,x²,x³…x^mInputting a Percentile detection algorithm, and carrying out monitoring and judgment on the power utilization behavior of a user to obtain a Percentile detection algorithm abnormal sample set;

and 4, step 4: marking a union set of all abnormal samples in the abnormal sample set of the Gaussian model or the Percentile detection algorithm, then carrying out power consumption abnormal behavior subdivision on the abnormal samples, and outputting a subdivision behavior sample set.

, the step 2 comprises the following steps:

step 2.1: data using switch on and off states and appliance power { x¹,x²,x³…x^mCalculating the expected value mu and variance sigma of the Gaussian probability distribution function²The calculation formula is as follows:

wherein x is⁽ⁱ⁾For a data set { x¹,x²,x³…x^mData in (c);

step 2.2: constructing a distribution function of Gaussian probability, wherein the distribution function of the Gaussian probability is as follows:

mu and sigma²The expected value and variance calculated for the above steps, x is the coefficient to be determined, the expected value μ of the normal distribution determines its position, and the variance σ determines its position²Determines the amplitude of the distribution;

step 2.3: selecting a judgment boundary, and carrying out abnormal detection judgment on the electricity consumption behavior of the user:

and f (x) is 3 σ as a determination boundary. When f (x) is less than 3 sigma, judging that the test sample is an abnormal sample; and when f (x) is larger than 3 sigma, judging the test sample as a normal sample.

, the concrete steps of step 3 are:

step 3.1: the data of the opening and closing state and the current power of the electric appliance are set as { x¹,x²,x³…x^mSorting according to the sequence from small to large;

step 3.2: and carrying out grade division on each group of data, wherein the division formula is as follows:

wherein x is⁽ⁱ⁾For a data set { x¹,x²,x³…x^mData in (j), m is the total size of data, rankⁱThen it is the rank of the set of data;

step 3.3: and judging whether the data belongs to the abnormal electricity utilization condition of the user according to the grade condition of the group of data, wherein the data with the highest grade or the lowest grade is regarded as abnormal electricity utilization data.

The method for monitoring abnormal electricity consumption behavior of users firstly collects the on-off state data and the electric appliance power data of the electric appliance switch of each user as a data set { x }¹,x²,x³…x^mAnd respectively importing the data set into a Gaussian probability distribution function and a Percentile detection algorithm, so that two mathematical models for detecting the abnormal electricity consumption behavior of the user are constructed for verification, the efficiency and the accuracy of detecting the abnormal electricity consumption condition of the client are improved, and the inspection range is narrowed.

Detailed Description

In order to better understand the technical solution of the present invention, the following detailed description is made by specific examples:

the invention provides monitoring methods for user electricity consumption behavior abnormity, and constructs two mathematical models for detecting the user electricity consumption behavior abnormity, wherein the two mathematical models comprise a detection algorithm based on a Gaussian probability distribution function and a detection algorithm based on Percentile.

Taking detection cycles as an example, 60-day user data of 6 groups of families are collected, and electric appliance switch data and electric appliance power data of each group of families are collected as data sets, wherein the opening and closing states of a switch and the current power of an electric appliance are respectively used as data sets { x }¹,x²,x³…x^mThe input is to the model.

Firstly, the specific application of the power utilization abnormity analysis of the detection algorithm based on the Gaussian probability distribution function comprises the following specific steps:

1) the expected value and standard deviation of the gaussian probability distribution function are calculated.

Data using switch on and off states and appliance power { x¹,x²,x³…x^mCalculating the expected value mu and variance sigma of the Gaussian probability distribution function²The calculation formula is as follows:

wherein x is⁽ⁱ⁾For a data set { x¹,x²,x³…x^mThe data in (c).

2) A distribution function of gaussian probabilities is constructed.

The distribution function of gaussian probability is:

mu and sigma²The expected value and variance calculated for the above steps, x is the coefficient to be determined, the expected value μ of the normal distribution determines its position, and the variance σ determines its position²Determines the amplitude of the distribution.

3) And selecting a judgment boundary, and detecting and judging the abnormal electricity utilization behavior of the user.

And f (x) is 3 σ as a determination boundary. When f (x) is less than 3 sigma, judging that the test sample is an abnormal sample; and when f (x) is larger than 3 sigma, judging the test sample as a normal sample.

Then, carrying out specific application of power utilization abnormity analysis based on a Percentile detection algorithm, wherein the specific steps are as follows:

1) data of switch on-off state and electric appliance power { x¹,x²,x³…x^mThe sorting is performed from small to large.

2) And grading each group of data.

The division formula is as follows:

xⁱis the value of the set of data, m is the total size of the data, rankⁱIt is the rank of the set of data.

3) And judging whether the data belongs to the abnormal electricity utilization condition of the user according to the grade condition of the group of data, wherein the data with the highest grade or the lowest grade is regarded as abnormal electricity utilization data.

Through the detection of the two models, if the group of data is the abnormal data of the power utilization behavior, the abnormal data of the power utilization behavior is subdivided in a manual marking mode, such as electricity stealing and electricity leakage.

By adopting the method, the abnormal electricity consumption behavior of the user can be detected and verified through the dual mathematical model, so that the initial screening of the electricity consumption data sample of the user is realized, the workload of screening the abnormal electricity consumption behavior is greatly simplified, and the accuracy of screening the abnormal electricity consumption behavior is improved.

The higher the true-normal ratio, the lower the false-positive ratio, generally indicating the better this model. It should be understood by those skilled in the art that the above embodiments are only for illustrating the present invention and are not to be used as a limitation of the present invention, and that changes and modifications to the above described embodiments are within the scope of the claims of the present invention as long as they are within the spirit and scope of the present invention.