Power consumption anomaly detection method and system based on multi-dimensional feature multi-check

Abstract

The invention discloses an electricity consumption anomaly detection method based on multi-dimensional feature multi-check, and the method comprises the following steps: carrying out the feature extraction processing of user basic data, user historical electrical parameters and historical electricity consumption data; according to the user industry type characteristics and the user electricity consumption behavior characteristics, user type electricity consumption grading processing is carried out; based on the weekly electricity consumption behavior characteristics of the users and the daily electricity consumption behavior characteristics of the users, screening out electricity consumption users with electricity consumption grade changes and electricity consumption viscosity demand degree changes to judge whether the electricity consumption behaviors of the users are abnormal or not; performing electricity consumption behavior abnormity check processing and electricity consumption behavior normal check processing according to the electricity consumption behavior judgment result of the user, and generating an electricity consumption behavior abnormity report; accuracy and reliability of user electricity consumption behavior anomaly detection are greatly improved, and safe and stable operation of an electric power system is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power consumption state detection, and in particular to a power consumption anomaly detection method and system based on multi-dimensional feature multi-checking. BACKGROUND

[0002] At present, the detection method of user power consumption anomaly is realized by using single user power consumption data or simple rule judgment. This method can only capture some obvious abnormal conditions, and the accuracy of power consumption anomaly detection is not ideal for complex and changeable power consumption behavior and power consumption anomaly caused by the joint action of multiple factors. For example, the power consumption of some users may fluctuate greatly due to weather changes, holidays and other external factors, and may be misjudged as power consumption anomaly by traditional power consumption anomaly detection method. Moreover, since the traditional method does not analyze the user power consumption behavior in depth and check the normal judgment result and the abnormal judgment result of power consumption behavior, it cannot accurately distinguish between real abnormal types and pseudo normal types of power consumption, affecting the accuracy and reliability of power consumption detection, and cannot accurately count the real power consumption demand of users for regional power distribution, thereby causing safety hazards to the safe and stable operation of the power system.

[0003] In order to overcome the above-mentioned deficiencies of the prior art, the present application provides a power consumption anomaly detection method and system based on multi-dimensional feature multi-checking, which comprehensively considers user basic data, historical electrical parameters, power consumption data and external influencing factors and other multi-dimensional features, analyzes the user power consumption behavior in depth, checks the normal judgment result and the abnormal judgment result of power consumption behavior, accurately distinguishes between real abnormal types and pseudo normal types of power consumption, greatly improves the accuracy and reliability of user power consumption behavior anomaly detection, and thus guarantees the safe and stable operation of the power system. SUMMARY

[0004] In order to overcome the deficiencies of the prior art, the present application provides a power consumption anomaly detection method and system based on multi-dimensional feature multi-checking, which realizes the improvement of the accuracy and reliability of user power consumption behavior anomaly detection through power consumption behavior anomaly checking processing and power consumption behavior normal checking processing, and thus guarantees the safe and stable operation of the power system.

[0005] In order to achieve the above-mentioned application purposes, the present application adopts the following technical solutions: The first aspect of the present application provides a power consumption anomaly detection method based on multi-dimensional feature multi-checking, comprising the following steps: S101, feature extraction processing is performed on user basic data and user historical electrical parameter and historical power consumption data, and user type power consumption level classification processing is performed according to user industry type features and user power consumption behavior features; S102, based on user weekly power consumption behavior features and user daily power consumption behavior features, power consumption users with power consumption level changes and power consumption stickiness demand degree changes are screened out to determine whether the user power consumption behavior is abnormal; S103, calling the influence factor features in the external influence factor database, the user power consumption behavior abnormality recognition processing result is processed to perform power consumption behavior abnormality checking processing; S104, based on the electrical features, the user power consumption behavior normal result in step S102 is processed to perform power consumption behavior normal checking processing; S105, based on the power consumption behavior abnormality checking processing result and the power consumption behavior normal checking processing result, power consumption behavior abnormality type classification processing is performed, and a power consumption behavior abnormality report is generated.

[0006] Further, the feature extraction processing is performed on the user basic data and the user historical electrical parameter and historical power consumption data, and the user type power consumption level classification processing is performed according to the user industry type features and the user power consumption behavior features, including the following steps: The user basic data is processed to obtain user ID features, user industry type features and user geographic location features; The user historical electrical parameter data is processed to obtain voltage parameter features, current parameter features and power parameter features, and the power parameter features include active power parameter features, reactive power parameter features and power factor parameter features; The user historical power consumption data is processed to obtain user daily power consumption behavior features and user weekly power consumption behavior features; Based on the user industry type features, the user weekly power consumption behavior features and the user daily power consumption behavior features, the user power consumption type classification processing is performed.

[0007] Further, based on the user industry type features, the user weekly power consumption behavior features and the user daily power consumption behavior features, the user power consumption type classification processing includes the following steps: Based on the user industry type features, the power consumption users are classified by industry type; Based on the user weekly power consumption behavior features, the power consumption users classified by industry type are classified by power consumption level; Based on the user daily power consumption behavior features, the power consumption users classified by power consumption level are classified by power consumption stickiness demand degree.

[0008] Further, based on the user weekly power consumption behavior characteristics and the user daily power consumption behavior characteristics, the power consumption users with power consumption level changes and power consumption stickiness demand degree changes are screened out to determine whether the user power consumption behavior is abnormal, including the following steps: The power consumption users with power consumption level changes are screened out based on the user weekly power consumption behavior characteristics; The power consumption users with power consumption stickiness demand degree changes are screened out based on the user daily power consumption behavior characteristics; If the power consumption users with power consumption level changes do not have power consumption stickiness demand degree changes, it is determined that the user power consumption behavior is normal; if the power consumption users with power consumption level changes have power consumption stickiness demand degree changes, it is determined that the user power consumption behavior is abnormal.

[0009] Further, the influence factor characteristics in the external influence factor database are called to process the user power consumption behavior abnormality recognition result, and the power consumption behavior abnormality checking processing includes the following steps: Based on the user geographic location characteristics, the corresponding meteorological characteristics and holiday characteristics are extracted from the external influence factor database; Based on the meteorological characteristics and holiday characteristics, it is determined whether they are the reasons for causing the user power consumption level and power consumption stickiness demand degree to change; If the meteorological characteristics or / and holiday characteristics are the reasons for causing the user power consumption behavior to be abnormal, the abnormality recognition processing result is corrected, and it is determined that the user power consumption behavior is normal; If the meteorological characteristics or / and holiday characteristics are not the reasons for causing the user power consumption behavior to be abnormal, the original abnormality recognition processing result is maintained, and it is determined that the user power consumption behavior is abnormal.

[0010] Further, based on the electrical characteristics, the user power consumption behavior normal in step S102 is checked, and the power consumption behavior normal checking processing includes the following steps: If the current parameter characteristic fluctuation is within the normal range, the active power parameter characteristic and the power factor reduction fluctuation are within the abnormal range, or the voltage parameter characteristic fluctuation is within the normal range, the current parameter characteristic increase fluctuation and the active power parameter characteristic reduction fluctuation are within the abnormal range, it is determined that the user power consumption behavior is abnormal or pseudo-normal; If the voltage parameter characteristic, the current parameter characteristic, the active power parameter characteristic, the reactive power parameter characteristic and the power factor parameter characteristic fluctuation are within the normal range, it is determined that the user power consumption behavior is normal.

[0011] Further, based on the power consumption behavior abnormality checking processing result and the power consumption behavior normal checking processing result, the power consumption behavior abnormality type classification processing is performed, and a power consumption behavior abnormality report is generated, including the following steps: The user whose power consumption behavior is corrected to be normal in the abnormal comparison processing result is marked as a misjudgment type caused by external factors; The user whose power consumption behavior is abnormal in the abnormal comparison processing result is marked as an abnormal prominent type; The user whose power consumption behavior is abnormal in the normal comparison processing result is marked as an abnormal concealed type; The user whose power consumption behavior is pseudo-normal in the normal comparison processing result is marked as a pseudo-normal type, and a power consumption behavior abnormality report is generated.

[0012] The second aspect of the present application provides a power consumption abnormality detection system based on multi-dimensional feature multi-comparison, comprising: A first data processing unit is configured to perform feature extraction processing on user basic data, user historical electrical parameter data and user historical power consumption data, and perform user type power consumption level division processing according to user industry type features and user power consumption behavior features; A second data processing unit is configured to filter out power consumption users with power consumption level changes and power consumption stickiness demand degree changes based on user weekly power consumption behavior features and user daily power consumption behavior features, so as to determine whether the user power consumption behavior is abnormal; A third data processing unit is configured to call influence factor features in an external influence factor database to perform power consumption behavior abnormality comparison processing on the user power consumption behavior abnormality identification processing result; A fourth data processing unit is configured to perform power consumption behavior normal comparison processing on power consumption users with power consumption level changes and without power consumption stickiness demand degree changes based on electrical features; A fifth data processing unit is configured to perform power consumption behavior abnormality type division processing based on the power consumption behavior abnormality comparison processing result and the power consumption behavior normal comparison processing result, and generate a power consumption behavior abnormality report.

[0013] Further, the first data processing unit performs feature extraction processing on user basic data, user historical electrical parameter data and user historical power consumption data, and performs user type power consumption level division processing according to user industry type features and user power consumption behavior features, which comprises: User basic feature extraction processing is performed on the user basic data to obtain user ID features, user industry type features and user geographic location features; Electrical feature extraction processing is performed on the user historical electrical parameter data to obtain voltage parameter features, current parameter features and power parameter features, wherein the power parameter features include active power parameter features, reactive power parameter features and power factor parameter features; Power consumption behavior feature extraction processing is performed on the user historical power consumption data to obtain user daily power consumption behavior features and user weekly power consumption behavior features; The user electricity type classification processing is performed based on the user industry type feature, the user weekly electricity consumption behavior feature, and the user daily electricity consumption behavior feature.

[0014] Further, the first data processing unit performs the user electricity type classification processing based on the user industry type feature, the user weekly electricity consumption behavior feature, and the user daily electricity consumption behavior feature, which includes: performing industry type classification processing on the electricity user based on the user industry type feature; performing electricity consumption level classification processing on the electricity user classified by the industry type based on the user weekly electricity consumption behavior feature; performing electricity stickiness demand degree classification processing on the electricity user classified by the electricity consumption level based on the user daily electricity consumption behavior feature.

[0015] Further, the second data processing unit filters out the electricity user with changes in electricity consumption level and changes in electricity stickiness demand degree based on the user weekly electricity consumption behavior feature and the user daily electricity consumption behavior feature, to determine whether the user electricity consumption behavior is abnormal, which includes: filtering out the electricity user with changes in electricity consumption level based on the user weekly electricity consumption behavior feature; filtering out the electricity user with changes in electricity stickiness demand degree based on the user daily electricity consumption behavior feature; if the electricity user with changes in electricity consumption level does not have changes in electricity stickiness demand degree, it is determined that the user electricity consumption behavior is normal; if the electricity user with changes in electricity consumption level has changes in electricity stickiness demand degree, it is determined that the user electricity consumption behavior is abnormal.

[0016] Further, the third data processing unit performs electricity consumption behavior abnormality checking processing on the user electricity consumption behavior abnormality identification processing result by calling the influence factor features in the external influence factor database, which includes: extracting corresponding meteorological features and holiday features from the external influence factor database based on the user geographic location feature; judging whether the meteorological features and holiday features are the reasons for changes in the user electricity consumption level and the electricity stickiness demand degree; if the meteorological features or / and the holiday features are the reasons for the user electricity consumption behavior abnormality, the abnormality identification processing result is corrected, and it is determined that the user electricity consumption behavior is normal; if the meteorological features or / and the holiday features are not the reasons for the user electricity consumption behavior abnormality, the original abnormality identification processing result is maintained, and it is determined that the user electricity consumption behavior is abnormal.

[0017] Further, the fourth data processing unit performs normal use electricity amount behavior check processing on the use electricity user whose use electricity amount level changes and whose use electricity stickiness demand degree does not change based on the electrical characteristics, including: If the current parameter characteristic fluctuation is within the normal range, the active power parameter characteristic and the power factor decrease fluctuation are within the abnormal range, or the voltage parameter characteristic fluctuation is within the normal range, the current parameter characteristic increase fluctuation and the active power parameter characteristic decrease fluctuation are within the abnormal range, it is determined that the use electricity amount behavior of the user is abnormal or pseudo-normal; If the voltage parameter characteristic, the current parameter characteristic, the active power parameter characteristic, the reactive power parameter characteristic, and the power factor parameter characteristic fluctuation are within the normal range, it is determined that the use electricity amount behavior of the user is normal.

[0018] Further, the fifth data processing unit performs use electricity amount behavior abnormal type division processing based on the use electricity amount behavior abnormal check processing result and the use electricity amount behavior normal check processing result, and generates a use electricity amount behavior abnormal report, including: For the user in the abnormal check processing result who is corrected to be normal use electricity amount behavior, it is marked as an external factor misjudgment type; For the user in the abnormal check processing result whose use electricity amount behavior is abnormal, it is marked as an abnormal prominent type; For the user in the use electricity amount behavior normal check processing result whose use electricity amount behavior is abnormal, it is marked as an abnormal hidden type; For the user in the use electricity amount behavior normal check processing result whose use electricity amount behavior is pseudo-normal, it is marked as a pseudo-normal type, and a use electricity amount behavior abnormal report is generated.

[0019] The application has the following beneficial effects: By dividing the use electricity user into user industry type, use electricity amount level, and use electricity stickiness demand degree in detail, and screening the use electricity user whose use electricity amount level changes and whose use electricity stickiness demand degree changes, it is determined whether the use electricity amount behavior of the user is abnormal according to the use electricity amount level change and the use electricity stickiness demand degree change, thereby improving the accuracy and reliability of the use electricity amount behavior abnormal detection of the user. By performing use electricity amount behavior abnormal check processing and use electricity amount behavior normal check processing on the use electricity amount behavior judgment result, it is detected that the misjudgment of the use electricity amount behavior caused by external influencing factors, and the abnormal hidden type and the pseudo-normal type of the use electricity amount behavior abnormality are detected, thereby further greatly improving the accuracy and reliability of the use electricity amount behavior abnormal detection of the user. Through the generated use electricity amount behavior abnormal report, the specific situation of the use electricity amount behavior abnormality of the user of different industry types can be clearly understood, and through the use electricity amount behavior abnormal report, the basic information (user ID, user industry type, user geographic location, etc.) of the user can be understood, which is convenient for accurate positioning and classified management of the use electricity amount behavior abnormal user, thereby ensuring the safe and stable operation of the power system. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description only represent some embodiments of the present application, and for those skilled in the art, other drawings can be obtained from these drawings without creative labor.

[0021] Figure 1 is a step schematic diagram of a power consumption anomaly detection method based on multi-dimensional feature multi-checking according to the present application. DETAILED DESCRIPTION

[0022] The embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0023] The embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0024] Embodiment one A power consumption anomaly detection method based on multi-dimensional feature multi-checking, comprising the following steps: S101, performing feature extraction processing on user basic data and user historical electrical parameter and historical power consumption data, and performing user type power consumption level division processing according to user industry type features and user power consumption behavior features; The user basic data, user historical electrical parameter data and user historical power consumption data are processed by feature extraction, and the user type power consumption level is divided based on the user industry type feature and the user power consumption behavior feature, which includes the following steps:

[0025] The user basic data, user historical electrical parameter data and user historical power consumption data are processed by feature extraction, and the user type power consumption level is divided based on the user industry type feature and the user power consumption behavior feature, which includes the following steps: The user basic data is processed by user basic feature extraction to obtain the user ID feature, the user industry type feature and the user geographic location feature; The user historical electrical parameter data is processed by electrical feature extraction to obtain the voltage parameter feature, the current parameter feature and the power parameter feature, and the power parameter feature includes the active power parameter feature, the reactive power parameter feature and the power factor parameter feature; The user historical power consumption data is processed by power consumption behavior feature extraction to obtain the user daily power consumption behavior feature and the user weekly power consumption behavior feature; The user power consumption type is divided based on the user industry type feature, the user weekly power consumption behavior feature and the user daily power consumption behavior feature.

[0026] The user power consumption type is divided based on the user industry type feature, the user weekly power consumption behavior feature and the user daily power consumption behavior feature, which includes the following steps: The power consumption users are divided by industry type based on the user industry type feature; The power consumption users divided by industry type are divided by power consumption level based on the user weekly power consumption behavior feature; The power consumption users divided by power consumption level are divided by power consumption stickiness demand degree based on the user daily power consumption behavior feature.

[0027] For example, by classifying the electricity users according to the user industry type characteristics, the electricity users are classified into residential type electricity users, industrial type electricity users, commercial type electricity users and agricultural type users. Based on the user weekly electricity consumption behavior characteristics, the user weekly total electricity consumption is obtained, and the electricity consumption level can be set as high electricity consumption level, medium electricity consumption level and low electricity consumption level. According to the user weekly total electricity consumption, the residential type electricity users, industrial type electricity users, commercial type electricity users and agricultural type users are classified into high electricity consumption level, medium electricity consumption level and low electricity consumption level. Based on the comparison between the user daily electricity consumption behavior characteristics and the user high electricity consumption threshold, the user high electricity consumption period is screened out, and the proportion of the user high electricity consumption period is obtained. The electricity stickiness demand degree can be set as heavy electricity stickiness demand degree, medium electricity stickiness demand degree and light electricity stickiness demand degree. According to the proportion of the user high electricity consumption period, the electricity stickiness demand degree of the electricity users classified according to the electricity consumption level is classified, and the residential type electricity users, industrial type electricity users, commercial type electricity users and agricultural type users with different electricity stickiness demand degrees in high electricity consumption level, medium electricity consumption level and low electricity consumption level are obtained.

[0028] S102, based on the user weekly electricity consumption behavior characteristics and the user daily electricity consumption behavior characteristics, the electricity users with changes in electricity consumption level and electricity stickiness demand degree are screened out to determine whether the user electricity consumption behavior is abnormal; Based on the user weekly electricity consumption behavior characteristics, the user weekly total electricity consumption is obtained, and the user electricity consumption level is determined according to the user weekly total electricity consumption. If the user electricity consumption level of the i-th week is different from that of the i-1-th week, it is determined that the user electricity consumption level has changed. Otherwise, it is determined that the user electricity consumption level has not changed. Based on the user daily electricity consumption behavior characteristics, the proportion of the user high electricity consumption period is obtained, and the user electricity stickiness demand degree is determined according to the proportion of the user high electricity consumption period. If the user electricity stickiness demand degree of the j-th day is different from that of the j-1-th day, it is determined that the user electricity stickiness demand degree has changed. Otherwise, it is determined that the user electricity stickiness demand degree has not changed. By determining whether the user electricity consumption behavior is abnormal according to the changes in electricity consumption level and electricity stickiness demand degree, the credibility of the user electricity consumption behavior determination result is improved, and the traditional single comparison between the j-th day and the j-1-th day electricity consumption difference and the fixed electricity consumption threshold is avoided, which is easy to cause the user electricity consumption behavior misjudgment, resulting in low credibility of the user electricity consumption behavior determination result.

[0029] Based on the user weekly electricity consumption behavior characteristics and the user daily electricity consumption behavior characteristics, electricity users with electricity consumption level changes and electricity stickiness demand degree changes are screened out to determine whether the user electricity consumption behavior is abnormal, including the following steps: Based on the user weekly electricity consumption behavior characteristics, electricity users with electricity consumption level changes are screened out; Based on the user daily electricity consumption behavior characteristics, electricity users with electricity stickiness demand degree changes are screened out; If the electricity users with electricity consumption level changes do not have electricity stickiness demand degree changes, it is determined that the user electricity consumption behavior is normal; if the electricity users with electricity consumption level changes have electricity stickiness demand degree changes, it is determined that the user electricity consumption behavior is abnormal.

[0030] For example, based on the user weekly electricity consumption behavior characteristics, electricity users A and B with electricity consumption level changes are screened out. The user industry types of electricity users A and B are both industrial types. The electricity consumption level of electricity user A changes from high electricity consumption level to medium electricity consumption level, and the electricity stickiness demand degree of electricity user A changes from heavy electricity stickiness demand degree to medium electricity stickiness demand degree, so it is determined that the electricity consumption behavior of user A is abnormal. The electricity consumption level of electricity user B changes from high electricity consumption level to medium electricity consumption level, and the electricity stickiness demand degree of electricity user B does not change, so it is determined that the electricity consumption behavior of user B is normal.

[0031] S103, calling the influence factor characteristics in the external influence factor database to process the user electricity consumption behavior abnormality recognition result, and performing electricity consumption behavior abnormality checking processing; Based on the user geographic location characteristics, meteorological data and statutory holiday table data are obtained, meteorological characteristics extraction processing is performed on the meteorological data, and temperature characteristics, rainfall level characteristics and snowfall level characteristics at the user geographic location are obtained. The statutory holiday table data is subjected to holiday marking processing to obtain holiday characteristics. Based on the meteorological characteristics extraction processing result and the holiday marking processing result, an external influence factor database is constructed. By calling the influence factor characteristics in the external influence factor database, the user electricity consumption abnormality recognition processing result is checked to determine whether the user electricity consumption behavior abnormality is caused by external influence factors, to avoid misjudgment of users with normal electricity consumption behavior due to external influence factors as users with abnormal electricity consumption behavior, thereby improving the accuracy of user electricity consumption behavior abnormality detection.

[0032] Calling the influence factor characteristics in the external influence factor database to process the user electricity consumption behavior abnormality recognition result, and performing electricity consumption behavior abnormality checking processing includes the following steps: Based on the user geographic location characteristics, corresponding meteorological characteristics and holiday characteristics are extracted from the external influence factor database; determine whether the meteorological feature and the holiday feature are the cause of the change in the user electricity consumption level and the electricity stickiness demand degree; If the meteorological feature or / and the holiday feature is not the cause of the abnormal user electricity consumption behavior, the original abnormality identification processing result is maintained, and it is determined that the user electricity consumption behavior is abnormal. If the meteorological feature or / and the holiday feature is not the cause of the abnormal user electricity consumption behavior, the original abnormality identification processing result is maintained, and it is determined that the user electricity consumption behavior is abnormal.

[0033] For example, after the electricity user A is initially determined to have abnormal electricity consumption behavior, the external influencing factor database is called, and if the user's region has low temperature weather, heavy rainfall weather, heavy snow weather, or a holiday during the time period, the user's electricity consumption is correspondingly reduced, causing the user's electricity consumption level and electricity stickiness demand degree to change, it is determined that the user's electricity consumption behavior abnormality identification processing result is misjudged, and the misjudged user's electricity consumption behavior is corrected to be normal; if the electricity user A does not have the above external influencing factors, it is determined that the user's electricity consumption behavior abnormality identification processing result is not misjudged. The external influencing factor feature is used to check the user's electricity consumption behavior abnormality identification processing result, which can determine that the user's electricity consumption behavior abnormality is not caused by the user itself, but is affected by external factors such as weather and holidays, thereby avoiding misjudgment.

[0034] S104, based on the electrical feature, the user electricity consumption behavior normal result determined in step S102 is subjected to normality check processing; The voltage parameter feature, the current parameter feature, the active power parameter feature, the reactive power parameter feature, and the power factor parameter feature in the user historical electrical parameter data are extracted, and the user electricity consumption behavior normal result determined in step S102 is subjected to normality check processing. If the voltage parameter feature, the current parameter feature, the active power parameter feature, the reactive power parameter feature, and the power factor parameter feature in the electrical parameter fluctuate within a normal range, it is determined that the user's electricity consumption behavior is normal; if the voltage parameter feature, the current parameter feature, the active power parameter feature, the reactive power parameter feature, and the power factor parameter feature in the electrical parameter fluctuate within an abnormal range, it is determined that the user's electricity consumption behavior is abnormal. The user electricity consumption behavior normality check processing is performed on the user whose electricity consumption level changes and whose electricity stickiness demand degree does not change, so as to accurately determine the user whose electricity consumption behavior is abnormal from the user whose electricity consumption behavior is normal, thereby further improving the accuracy of the user electricity consumption behavior abnormality detection.

[0035] The normality checking process of the user's power consumption behavior in step S102 based on the electrical characteristics includes the following steps: If the current parameter characteristic fluctuation is within the normal range, the active power parameter characteristic and the power factor decrease fluctuation are within the abnormal range, or the voltage parameter characteristic fluctuation is within the normal range, the current parameter characteristic increase fluctuation and the active power parameter characteristic decrease fluctuation are within the abnormal range, it is determined that the user's power consumption behavior is abnormal or pseudo-normal; If the voltage parameter characteristic, current parameter characteristic, active power parameter characteristic, reactive power parameter characteristic and power factor parameter characteristic fluctuation are within the normal range, it is determined that the user's power consumption behavior is normal.

[0036] For example, the power consumption behavior of the power user B is normal, the power consumption level of the power user B is changed from the high power consumption level to the medium power consumption level, and the power consumption stickiness demand degree of the power user B does not change. If the current parameter characteristic fluctuation is within the normal range, the active power parameter characteristic and the power factor decrease fluctuation are within the abnormal range, it is determined that the power consumption behavior of the power user B is abnormal. If the voltage parameter characteristic fluctuation is within the normal range, the current parameter characteristic increase fluctuation and the active power parameter characteristic decrease fluctuation are within the abnormal range, it is determined that the power consumption behavior of the power user B is pseudo-normal. If the voltage parameter characteristic, current parameter characteristic, active power parameter characteristic, reactive power parameter characteristic and power factor parameter characteristic fluctuation are within the normal range, it is determined that the power consumption behavior of the power user B is normal.

[0037] S105, based on the abnormality checking process result and the normality checking process result of the power consumption behavior, the abnormal type of the power consumption behavior is divided, and the power consumption behavior abnormal report is generated; The user whose power consumption behavior is corrected to be normal in the abnormality checking processing result is marked as a misjudgment type caused by external factors, and it is specified which meteorological feature or holiday feature causes the misjudgment. The user whose power consumption behavior is abnormal in the abnormality checking processing result is marked as an abnormal prominent type, and it is specified the specific situation of the power consumption level and the change of the power consumption stickiness demand degree of the user. The user whose power consumption behavior is abnormal in the normality checking processing result is marked as an abnormal hidden type, and it is specified the abnormal fluctuation of the electrical parameter feature. The user whose power consumption behavior is pseudo-normal in the normality checking processing result is marked as a pseudo-normal type, and it is specified the abnormal fluctuation of the electrical parameter feature. Through the power consumption behavior abnormality checking processing result and the power consumption behavior normality checking processing result, the power consumption behavior abnormality type classification processing is performed, and the power consumption behavior abnormality report is generated, so that the specific situation of the power consumption behavior abnormality of the user of different types can be clearly understood, and through the power consumption behavior abnormality report, the basic information (user ID, user industry type, user geographical location, etc.) of the user can be understood, so that the user with power consumption behavior abnormality can be accurately positioned and classified and managed, thereby ensuring the safe and stable operation of the power system.

[0038] Based on the power consumption behavior abnormality checking processing result and the power consumption behavior normality checking processing result, the power consumption behavior abnormality type classification processing is performed, and the power consumption behavior abnormality report is generated, including the following steps: The user whose power consumption behavior is corrected to be normal in the abnormality checking processing result is marked as a misjudgment type caused by external factors; The user whose power consumption behavior is abnormal in the abnormality checking processing result is marked as an abnormal prominent type; The user whose power consumption behavior is abnormal in the normality checking processing result is marked as an abnormal hidden type; The user whose power consumption behavior is pseudo-normal in the normality checking processing result is marked as a pseudo-normal type, and the power consumption behavior abnormality report is generated.

[0039] Embodiment Two The above is a power consumption abnormality detection method based on multi-dimensional feature multi-time checking provided in the embodiments of the present application, and the following is a power consumption abnormality detection system based on multi-dimensional feature multi-time checking provided in the embodiments of the present application.

[0040] A power consumption abnormality detection system based on multi-dimensional feature multi-time checking, comprising: A first data processing unit is configured to perform feature extraction processing on user basic data and user historical electrical parameters and historical power consumption data, and perform user type power consumption level classification processing according to user industry type features and user power consumption behavior features. The second data processing unit is configured to filter out power users with changes in power consumption level and changes in power stickiness demand degree based on the user weekly power consumption behavior characteristics and the user daily power consumption behavior characteristics, so as to determine whether the power consumption behavior of the user is abnormal. The third data processing unit is configured to call the influence factor characteristics in the external influence factor database to perform power consumption behavior abnormality identification processing on the user, and perform power consumption behavior abnormality checking processing on the result of the power consumption behavior abnormality identification processing. The fourth data processing unit is configured to perform power consumption behavior normality checking processing on the power users with changes in power consumption level and no changes in power stickiness demand degree based on the electrical characteristics. The fifth data processing unit is configured to perform power consumption behavior abnormality type classification processing based on the result of the power consumption behavior abnormality checking processing and the result of the power consumption behavior normality checking processing, and generate a power consumption behavior abnormality report.

[0041] The first data processing unit is configured to perform feature extraction processing on the user basic data, the user historical electrical parameters and the historical power consumption data, and perform user type power consumption level classification processing according to the user industry type characteristics and the user power consumption behavior characteristics, including: The user basic data is subjected to user basic feature extraction processing to obtain user ID characteristics, user industry type characteristics and user geographic location characteristics. The user historical electrical parameter data is subjected to electrical feature extraction processing to obtain voltage parameter characteristics, current parameter characteristics and power parameter characteristics, wherein the power parameter characteristics include active power parameter characteristics, reactive power parameter characteristics and power factor parameter characteristics. The user historical power consumption data is subjected to power consumption behavior feature extraction processing to obtain user daily power consumption behavior characteristics and user weekly power consumption behavior characteristics. The user power consumption type classification processing is performed based on the user industry type characteristics, the user weekly power consumption behavior characteristics and the user daily power consumption behavior characteristics.

[0042] The first data processing unit is configured to perform user power consumption type classification processing based on the user industry type characteristics, the user weekly power consumption behavior characteristics and the user daily power consumption behavior characteristics, including: The power users are subjected to industry type classification processing based on the user industry type characteristics. The power users subjected to the industry type classification processing are subjected to power consumption level classification processing based on the user weekly power consumption behavior characteristics. The power users subjected to the power consumption level classification processing are subjected to power stickiness demand degree classification processing based on the user daily power consumption behavior characteristics.

[0043] The second data processing unit is configured to filter out the electricity users with electricity consumption level change and electricity stickiness demand degree change based on the user weekly electricity consumption behavior characteristics and the user daily electricity consumption behavior characteristics, so as to determine whether the electricity consumption behavior of the user is abnormal, including: filtering out the electricity users with electricity consumption level change based on the user weekly electricity consumption behavior characteristics; filtering out the electricity users with electricity stickiness demand degree change based on the user daily electricity consumption behavior characteristics; if the electricity users with electricity consumption level change do not have electricity stickiness demand degree change, it is determined that the electricity consumption behavior of the user is normal; if the electricity users with electricity consumption level change have electricity stickiness demand degree change, it is determined that the electricity consumption behavior of the user is abnormal.

[0044] The third data processing unit is configured to call the influence factor characteristics in the external influence factor database to process the user electricity consumption behavior abnormality identification result, and perform electricity consumption behavior abnormality checking processing, including: extracting the corresponding meteorological characteristics and holiday characteristics from the external influence factor database based on the user geographic location characteristics; judging whether the meteorological characteristics and holiday characteristics are the reasons for causing the change of the user electricity consumption level and the electricity stickiness demand degree; if the meteorological characteristics or / and the holiday characteristics are the reasons for causing the abnormality of the user electricity consumption behavior, the abnormality identification processing result is corrected, and it is determined that the user electricity consumption behavior is normal; if the meteorological characteristics or / and the holiday characteristics are not the reasons for causing the abnormality of the user electricity consumption behavior, the original abnormality identification processing result is maintained, and it is determined that the user electricity consumption behavior is abnormal.

[0045] The fourth data processing unit is configured to perform electricity consumption behavior normality checking processing on the electricity users with electricity consumption level change and electricity stickiness demand degree change based on the electrical characteristics, including: if the current parameter characteristic fluctuation is within a normal range, the active power parameter characteristic and the power factor reduction fluctuation are within an abnormal range, or the voltage parameter characteristic fluctuation is within a normal range, the current parameter characteristic increase fluctuation and the active power parameter characteristic reduction fluctuation are within an abnormal range, it is determined that the electricity consumption behavior of the user is abnormal or pseudo-normal; if the voltage parameter characteristic, the current parameter characteristic, the active power parameter characteristic, the reactive power parameter characteristic and the power factor parameter characteristic fluctuation are within a normal range, it is determined that the electricity consumption behavior of the user is normal.

[0046] The fifth data processing unit is configured to perform electricity consumption behavior abnormality type classification processing based on the electricity consumption behavior abnormality checking processing result and the electricity consumption behavior normality checking processing result, and generate an electricity consumption behavior abnormality report, including: The user whose power consumption behavior is corrected to normal in the abnormality checking processing result is marked as a misjudgment type caused by external factors; The user whose power consumption behavior is abnormal in the abnormality checking processing result is marked as an abnormality prominent type; The user whose power consumption behavior is abnormal in the power consumption behavior normal checking processing result is marked as an abnormality concealed type; The user whose power consumption behavior is pseudo-normal in the power consumption behavior normal checking processing result is marked as a pseudo-normal type, and a power consumption behavior abnormality report is generated.

[0047] The above-described embodiments are merely used to illustrate the technical solutions of the present application, but not to limit the present application; although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting abnormal electricity consumption based on multiple verifications of multi-dimensional features, characterized in that, Includes the following steps: S101. Perform feature extraction processing on user basic data, user historical electrical parameters and historical electricity consumption data, and classify user type electricity consumption levels according to user industry type characteristics and user electricity consumption behavior characteristics. S102. Based on the weekly and daily electricity consumption behavior characteristics of users, select electricity users whose electricity consumption levels and electricity stickiness needs change, in order to determine whether there are any abnormalities in the users' electricity consumption behavior. S103. Call the characteristics of influencing factors in the external influencing factors database to identify and process the abnormal user electricity consumption behavior, and perform an abnormal electricity consumption behavior verification process. S104. Based on the electrical characteristics, perform a normal power consumption behavior verification process on the result of determining that the user's power consumption behavior is normal in step S102. S105. Based on the results of the verification and processing of abnormal electricity consumption behavior and the results of the verification and processing of normal electricity consumption behavior, perform classification processing of abnormal electricity consumption behavior types and generate an abnormal electricity consumption behavior report.

2. The method for detecting abnormal electricity consumption based on multiple verifications of multi-dimensional features according to claim 1, characterized in that, Step S101 includes the following steps: User basic features are extracted from the user basic data to obtain user ID features, user industry type features, and user geographical location features. Electrical feature extraction processing is performed on the user's historical electrical parameter data to obtain voltage parameter features, current parameter features, and power parameter features. The power parameter features include active power parameter features, reactive power parameter features, and power factor parameter features. The user's historical electricity consumption data is processed to extract electricity consumption behavior characteristics, resulting in the user's daily electricity consumption behavior characteristics and the user's weekly electricity consumption behavior characteristics. User electricity consumption type is classified based on user industry type characteristics, user weekly electricity consumption behavior characteristics, and user daily electricity consumption behavior characteristics.

3. The method for detecting abnormal electricity consumption based on multiple verifications of multi-dimensional features according to claim 2, characterized in that, The process of classifying user electricity consumption types based on user industry type characteristics, user weekly electricity consumption behavior characteristics, and user daily electricity consumption behavior characteristics includes the following steps: Electricity users are categorized by industry type based on their industry characteristics. Based on the weekly electricity consumption behavior characteristics of users, electricity users who have been classified by industry type are classified into different electricity consumption levels. Based on the characteristics of users' daily electricity consumption behavior, electricity users who have been classified into electricity consumption levels are further classified according to their degree of electricity stickiness.

4. The method for detecting abnormal electricity consumption based on multiple verifications of multi-dimensional features according to claim 1, characterized in that, Step S102 includes the following steps: Users whose electricity consumption levels change are identified based on their weekly electricity consumption behavior characteristics. Based on the characteristics of users' daily electricity consumption behavior, electricity users with changing electricity stickiness and demand were screened. If a user whose electricity consumption level changes does not show a change in their electricity consumption stickiness, then the user's electricity consumption behavior is considered normal; if a user whose electricity consumption level changes shows a change in their electricity consumption stickiness, then the user's electricity consumption behavior is considered abnormal.

5. The method for detecting abnormal electricity consumption based on multiple verifications of multi-dimensional features according to claim 1, characterized in that, Step S103 includes the following steps: Based on user geographic location characteristics, corresponding meteorological and holiday characteristics are extracted from the database of external influencing factors; Based on meteorological and holiday characteristics, determine whether these are the reasons for changes in users' electricity consumption levels and electricity stickiness. If meteorological characteristics and / or holiday characteristics are the cause of abnormal user electricity consumption behavior, the abnormal identification and processing result will be corrected and the user electricity consumption behavior will be judged to be normal. If meteorological characteristics and / or holiday characteristics are not the cause of abnormal user electricity consumption behavior, the original abnormal identification and processing result shall be maintained and the user electricity consumption behavior shall be judged as abnormal.

6. The method for detecting abnormal electricity consumption based on multiple verifications of multi-dimensional features according to claim 1, characterized in that, Step S104 includes the following steps: If the fluctuation of current parameter characteristics is within the normal range, while the fluctuation of active power parameter characteristics and power factor decrease is within the abnormal range, or if the fluctuation of voltage parameter characteristics is within the normal range, while the fluctuation of current parameter characteristics increases and the fluctuation of active power parameter characteristics decreases is within the abnormal range, then the user's electricity consumption behavior is determined to be abnormal or pseudo-normal. If the fluctuations in voltage, current, active power, reactive power, and power factor parameters are within the normal range, then the user's electricity consumption behavior is considered normal.

7. The method for detecting abnormal electricity consumption based on multiple verifications of multi-dimensional features according to claim 1, characterized in that, Step S105 includes the following steps: Users whose electricity consumption behavior is corrected to be normal in the abnormal verification and processing results are marked as misjudged by external factors. Users whose electricity consumption behavior is found to be abnormal in the abnormality verification and processing results are marked as the abnormality prominence type; Users whose electricity consumption behavior is found to be abnormal in the normal electricity consumption behavior verification process are marked as having hidden abnormalities. Users whose electricity consumption behavior is found to be pseudo-normal in the normal electricity consumption behavior verification results are marked as pseudo-normal and an abnormal electricity consumption behavior report is generated.

8. A power consumption anomaly detection system based on multiple verifications of multi-dimensional features, used to implement the power consumption anomaly detection method based on multiple verifications of multi-dimensional features as described in any one of claims 1-7, characterized in that, include: The first data processing unit is used to perform feature extraction processing on user basic data, user historical electrical parameters and historical electricity consumption data, and to classify user type electricity consumption levels according to user industry type characteristics and user electricity consumption behavior characteristics. The second data processing unit is used to filter out electricity users whose electricity consumption levels and electricity stickiness changes based on the user's weekly and daily electricity consumption behavior characteristics, so as to determine whether there are any abnormalities in the user's electricity consumption behavior. The third data processing unit is used to call the characteristics of influencing factors in the external influencing factors database to identify and process the abnormal user electricity consumption behavior, and to perform abnormal electricity consumption behavior verification. The fourth data processing unit is used to perform normal verification of electricity consumption behavior for electricity users whose electricity consumption level has changed but whose electricity stickiness demand has not changed, based on electrical characteristics. The fifth data processing unit is used to classify abnormal electricity consumption behavior types based on the results of abnormal electricity consumption behavior verification and processing and the results of normal electricity consumption behavior verification and processing, and to generate an abnormal electricity consumption behavior report.

9. The power consumption anomaly detection system based on multiple verifications of multi-dimensional features according to claim 8, characterized in that, The second data processing unit, based on the user's weekly and daily electricity consumption behavior characteristics, filters out electricity users whose electricity consumption levels and electricity stickiness changes, in order to determine whether there are any abnormalities in the user's electricity consumption behavior, including: Users whose electricity consumption levels change are identified based on their weekly electricity consumption behavior characteristics. Based on the characteristics of users' daily electricity consumption behavior, electricity users with changing electricity stickiness and demand were screened. If a user whose electricity consumption level changes does not show a change in their electricity consumption stickiness, then the user's electricity consumption behavior is considered normal; if a user whose electricity consumption level changes shows a change in their electricity consumption stickiness, then the user's electricity consumption behavior is considered abnormal.

10. The power consumption anomaly detection system based on multiple verifications of multi-dimensional features according to claim 8, characterized in that, The third data processing unit calls upon the characteristics of influencing factors in the external influencing factors database to process the abnormal user electricity consumption behavior identification results, and performs abnormal electricity consumption behavior verification processing, including: Based on user geographic location characteristics, corresponding meteorological and holiday characteristics are extracted from the database of external influencing factors; Based on meteorological and holiday characteristics, determine whether these are the reasons for changes in users' electricity consumption levels and electricity stickiness. If meteorological characteristics and / or holiday characteristics are the cause of abnormal user electricity consumption behavior, the abnormal identification and processing result will be corrected and the user electricity consumption behavior will be judged to be normal. If meteorological characteristics and / or holiday characteristics are not the cause of abnormal user electricity consumption behavior, the original abnormal identification and processing result shall be maintained and the user electricity consumption behavior shall be judged as abnormal.

Images