Electric energy dynamic anomaly analysis and early warning system for urban buildings in severe cold areas

Abstract

The invention discloses an electric energy dynamic anomaly analysis and early warning system for urban buildings in severe cold areas, and relates to the technical field of electric energy consumption analysis. The objective of the invention is to solve the problem of low identification accuracy of an existing electricity consumption abnormity identification method. The method comprises the following steps: acquiring electric energy consumption of a building in each acquisition period, acquiring dynamic abnormal power consumption behaviors and static abnormal power consumption behaviors of the residential building by utilizing the electric energy consumption of the building in each acquisition period, and giving an alarm if the power consumption abnormal behaviors occur. And performing short-term prediction and long-term prediction on the high power utilization risk of the public building by using the electric energy consumption of the building in each acquisition period, and giving an alarm if the high power utilization risk occurs. The method is used for electric energy consumption abnormity analysis.

Description

Technical Field

[0001] This invention relates to the field of electrical energy analysis technology, and in particular to a dynamic anomaly analysis and early warning system for urban buildings in frigid regions. Background Technology

[0002] Dynamic anomaly analysis of building electrical energy is a crucial aspect of building energy management. By collecting, processing, and analyzing electrical energy data during building operation, patterns of electrical energy changes can be identified, and abnormal electricity consumption behaviors can be pinpointed, providing a basis for optimizing building electrical energy operation. Especially in urban clusters with a wide distribution of buildings, achieving dynamic analysis and anomaly identification of electrical energy across multiple buildings at the city scale will offer significant value for the coordinated regulation of urban electricity and the construction of low-carbon cities.

[0003] In existing technologies, building power analysis methods involve collecting power data at fixed time intervals and then simply comparing it with historical power data to determine if abnormal power consumption has occurred. While this method achieves basic power monitoring, its analytical precision is insufficient, resulting in low accuracy in anomaly identification. Existing technologies also utilize single statistical models for short-term predictions of abnormal power consumption. However, this method is slow to respond to sudden changes in power consumption and its prediction results are heavily influenced by single factors, leading to low accuracy in identifying power anomalies. Summary of the Invention

[0004] This invention addresses the problem of low accuracy in existing methods for identifying abnormal electricity consumption by proposing a dynamic analysis and early warning system for abnormal electricity consumption in urban buildings in frigid regions.

[0005] A city-level building power dynamic anomaly analysis and early warning system for severely cold regions includes: a building power monitoring module, a residential building power consumption early warning module, and a public building power consumption early warning module;

[0006] The building power monitoring module is used to obtain the power consumption of each building in each data collection cycle within the monitoring range. ,Will Send to the residential building electricity consumption early warning module and the public building electricity consumption early warning module;

[0007] The buildings include: residential buildings and public buildings;

[0008] The residential building electricity consumption early warning module includes: a residential building electricity consumption dynamic early warning unit and a residential building electricity consumption static early warning unit;

[0009] The residential building electricity consumption dynamic early warning unit uses the electricity consumption of the residential building in each collection cycle to obtain dynamic abnormal electricity consumption behavior of the residential building and issue early warnings.

[0010] The residential building electricity consumption static early warning unit uses the daily total electricity consumption data of the residential building to obtain the static abnormal electricity consumption behavior of the residential building and issue an early warning.

[0011] The public building electricity consumption early warning module includes: a public building short-term electricity consumption early warning unit and a public building long-term electricity consumption early warning unit;

[0012] The public building short-term power consumption early warning unit uses the power consumption of the public building in each collection cycle to determine whether the current public building has excessive power consumption. If excessive power consumption occurs, a high power consumption early warning is issued; otherwise, no high power consumption early warning is issued.

[0013] The long-term power consumption early warning unit for public buildings establishes a building power prediction model using the power consumption of public buildings in each collection cycle, and obtains the power consumption prediction results of public buildings using the building power prediction model. Then, based on the power consumption prediction results of public buildings, it determines whether there is a high power consumption risk in public buildings. If there is a high power consumption risk, a high power consumption risk warning is issued; otherwise, no high power consumption risk warning is issued.

[0014] Furthermore, the building energy monitoring module is used to acquire the power consumption of each building in each acquisition cycle within the monitoring range. And based on the power consumption of each building in each collection period, the total daily power consumption data of each building is obtained. Monthly total power consumption data and total annual power consumption data Simultaneously, the power consumption per unit area of ​​each building is obtained within each data collection period. Specifically:

[0015] First, set the preset duration as the data collection cycle, and obtain the power consumption of each building within the monitoring range for each data collection cycle. ;

[0016] The preset duration is less than or equal to 1 hour;

[0017] Then, by using the power consumption of each building in each data collection cycle, the total daily power consumption can be obtained. Total monthly power consumption and total annual power consumption Specifically:

[0018]

[0019]

[0020]

[0021] in, It is the collection cycle number. It is the total number of collection cycles within a day. It is a date label. It is the total number of days in the month. It's a month number. It is architecture No. Power consumption per data acquisition cycle It is architecture No. Total daily electricity consumption It is architecture No. Total monthly power consumption It is architecture No. Total annual electricity consumption;

[0022] Finally, the unit area power consumption index for each building in each data collection cycle was obtained. Specifically:

[0023]

[0024] in, It is the unit area electricity index of building i in the t-th data collection period. This represents the power consumption of building i during the t-th data collection cycle. It is the area of ​​building i.

[0025] Furthermore, the residential building electricity consumption dynamic early warning unit uses the electricity consumption of the residential building in each collection cycle to obtain dynamic abnormal electricity consumption behavior of the residential building and issue early warnings, specifically as follows:

[0026] A1. Utilizing residential buildings The power consumption of residential buildings in the t-th collection cycle is obtained. The month-on-month growth rate of electricity consumption is as follows:

[0027]

[0028] in, Residential buildings The year-on-year growth rate of electricity consumption in the t-th data collection period Residential buildings No. Power consumption per data acquisition cycle Residential buildings No. Power consumption per data collection cycle;

[0029] A2. Utilize the month-on-month growth rate of residential building electricity consumption to obtain dynamic abnormal electricity consumption behavior in residential buildings and issue early warnings, specifically:

[0030] A residential building electricity consumption rate that meets any of the following conditions will be issued a residential building electricity consumption rate report. Dynamic abnormal electricity consumption behavior warning; otherwise, no dynamic abnormal electricity consumption behavior warning will be issued.

[0031] a1. Residential buildings in three consecutive data collection periods The month-on-month growth rate of electricity consumption was greater than 15%;

[0032] a2. Residential buildings for six consecutive data collection periods The month-on-month growth rate of electricity consumption is greater than 10%.

[0033] Furthermore, the residential building electricity consumption static early warning unit uses the daily total electricity consumption data of the residential building to obtain static abnormal electricity consumption behavior of the residential building and issue early warnings, specifically as follows:

[0034] B1. Obtain the information for each residential building within the monitoring range. The total daily electricity consumption for each day of the month, and the first electricity consumption for each residential building. The total daily electricity consumption for each day of the month is sorted in ascending order to obtain the sequence of total daily electricity consumption for each residential building;

[0035] B2. Obtain residential buildings based on the daily total power consumption sequence corresponding to residential buildings. Base threshold Specifically:

[0036]

[0037] in, It is the first The 90th percentile of the daily total electricity consumption sequence for each residential building. , It is the total number of residential buildings within the monitoring area. It refers to the residential building identification number within the monitoring area. It is the first The 90th percentile of the daily total electricity consumption sequence for each residential building. It is the first The 90th percentile value of the daily total electricity consumption sequence corresponding to each residential building;

[0038] B3. Obtain static abnormal electricity consumption behavior of residential buildings based on the basic threshold of residential buildings and issue early warnings.

[0039] Furthermore, in B3, static abnormal electricity consumption behavior of residential buildings is obtained based on basic thresholds for residential buildings, and early warnings are issued accordingly. Specifically:

[0040] Determine the current season. If the current season is the cooling season, execute b1; if the current season is the heating season, execute b2; if the current season is the transitional season, execute b3.

[0041] When the daily average temperature is below 5°C for more than a preset number of days, it indicates that the current period is the heating season; when the daily maximum temperature is above 26°C for more than a preset number of days, it indicates that the current period is the cooling season; other times are the transitional season.

[0042] b1. If it is currently the cooling season, obtain the current day's highest temperature. And the highest temperature in the current day's historical data A static abnormal power consumption behavior warning will be issued if any of the following conditions are met:

[0043] b1.1, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0044] b1.2, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0045] b1.3, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0046] b1.4, if Then judge Is it greater than or equal to? ;like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0047] b1.5, if Then judge Is it greater than or equal to? ;like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0048] b1.6, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0049] b2. If it is currently the heating season, obtain the current daily average temperature. and the current daily historical average temperature A static abnormal power consumption behavior warning will be issued if any of the following conditions are met:

[0050] b2.1, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0051] b2.2, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0052] b2.3, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0053] b2.4, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0054] b2.5, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0055] b3. If it is currently the transitional season, obtain the current daily average temperature. and the current daily historical average temperature A static abnormal power consumption behavior warning will be issued if any of the following conditions are met:

[0056] b3.1, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0057] b3.2, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0058] Furthermore, the short-term power consumption early warning unit for public buildings uses the power consumption of the public building in each collection cycle to determine whether the current power consumption of the public building is too high. If the power consumption is too high, a high power consumption early warning for the public building is issued; otherwise, no high power consumption early warning is issued. Specifically:

[0059] C1. Set 24 collection cycles as a time window, and obtain the average power consumption within each time window in the power consumption sequence corresponding to each public building, specifically:

[0060]

[0061] in, It is the first Time window of the power consumption sequence corresponding to each public building Average power consumption within, It is the first The first public building Power consumption per data collection cycle; C2, Utilization of public buildings The power consumption of public buildings in the t-th collection cycle is obtained. The month-on-month growth rate of electricity consumption is as follows:

[0062]

[0063] in, It is a public building The year-on-year growth rate of electricity consumption in the t-th data collection period It is a public building The Power consumption per data acquisition cycle It is a public building No. Power consumption per data collection cycle;

[0064] C3, using the first Time window of the power consumption sequence corresponding to each public building Average power consumption in public buildings The month-on-month growth rate of electricity consumption is used to determine whether there is excessive electricity consumption in public buildings. If excessive electricity consumption occurs, a high electricity consumption warning for public buildings is issued; otherwise, no high electricity consumption warning is issued.

[0065] Furthermore, the use of the first in C3 Time window of the power consumption sequence corresponding to each public building Average power consumption in public buildings The month-on-month growth rate of electricity consumption is used to determine whether there is excessive electricity consumption in public buildings. If excessive electricity consumption is found, a high electricity consumption warning for public buildings is issued; otherwise, no high electricity consumption warning is issued. Specifically:

[0066] C3.1, Setting up public buildings Base threshold ;

[0067] C3.2 Determine the current acquisition cycle Is it greater than or equal to? ,like Then issue public buildings Excessive power consumption warning; otherwise, determine the current data collection cycle. Is it during working hours? During working hours, execute C3.3; if During non-working hours, execute C3.4;

[0068] C3.3, Judgment Is it greater than or equal to 20%? Then output acquisition cycle For abnormal data collection cycles, public buildings are issued. A warning will be issued if electricity consumption is too high; otherwise, no warning will be issued.

[0069] C3.4, Judgment Is it greater than or equal to 5%? Then output acquisition cycle For abnormal data collection cycles, public buildings are issued. A warning will be issued if electricity consumption is too high; otherwise, no warning will be issued.

[0070] Furthermore, the long-term electricity consumption early warning unit for public buildings establishes a building electricity prediction model using the electricity consumption of the public building in each collection cycle, and uses the building electricity prediction model to obtain the electricity consumption prediction results of the public building. Then, based on the electricity consumption prediction results of the public building, it determines whether the public building has a high electricity consumption risk. If there is a high electricity consumption risk, a high electricity consumption risk warning is issued; otherwise, no high electricity consumption risk warning is issued. Specifically:

[0071] S1. Obtain the power consumption sequence of the public building for the week preceding the predicted collection period. Input the power consumption sequence of the public building for the week preceding the predicted date into the building power prediction model to obtain the predicted power consumption result of the public building for the predicted collection period. ;

[0072] The building power prediction model is a trained first LSTM model, a trained second LSTM model, or a trained third LSTM model.

[0073] If the month in which the data collection period to be predicted is located is a strongly correlated month, then the building power prediction model is the first LSTM model; if the month in which the data collection period to be predicted is located is a moderately correlated month, then the building power prediction model is the second LSTM model; if the month in which the data collection period to be predicted is located is a weakly correlated month, then the building power prediction model is the third LSTM model.

[0074] S2. Obtain the specific year corresponding to the data collection period to be predicted. ,moon ,day and collection cycle Then obtain Year moon, Daily collection cycle and Year moon, Daily collection cycle Public building power consumption and and obtain , , Average value:

[0075]

[0076] S3, Utilization Get the base threshold ,like Greater than If the power consumption risk warning is not issued, a high power consumption risk warning will be issued; otherwise, no high power consumption risk warning will be issued.

[0077] The basic threshold .

[0078] Furthermore, the months with strong correlation, moderate correlation, and weak correlation are determined as follows:

[0079] E1. Obtain the power consumption sequence of public buildings for each collection period within a year, and divide the power consumption sequence of public buildings for each collection period within a year into 12 groups according to the month. Then, obtain the outdoor average temperature for each month and each collection period to form an outdoor average temperature sequence, and obtain the Pearson correlation value between the outdoor average temperature of the k-th group and the power consumption of public buildings in the k-th group. :

[0080]

[0081] in, It is a variable and covariance, It is a variable standard deviation It is a variable standard deviation , It is the first Group of public buildings power consumption and the first Group average outdoor temperature;

[0082] E2, according to the first Group's average outdoor temperature and the first The Pearson correlation value for the power consumption of the group of public buildings is the first. The power consumption of public buildings is labeled with power consumption tags, specifically:

[0083] like Then the first The label for the electricity consumption of the group of public buildings is set to weak correlation;

[0084] like Then the first The label for the electricity consumption of public buildings is set to be relevant.

[0085] like Then the first The labels for the electricity consumption of public buildings are set to be strongly correlated;

[0086] E3. Based on the power consumption labels, the power consumption of public buildings is divided into strongly correlated, moderately correlated, and weakly correlated groups. The months corresponding to the data in the strongly correlated group are the strongly correlated months, the months corresponding to the data in the moderately correlated group are the moderately correlated months, and the months corresponding to the data in the weakly correlated group are the weakly correlated months.

[0087] Furthermore, the trained first LSTM model, the trained second LSTM model, or the trained third LSTM model is obtained through the following method:

[0088] D1. Obtain the power consumption sequence of public buildings for each collection period within a year, and divide the power consumption sequence of public buildings for each collection period within a year into 12 groups according to the month. Then, obtain the average outdoor temperature for each month and each collection period to form an average outdoor temperature sequence, and obtain the Pearson correlation value between the average outdoor temperature of the k-th group and the power consumption of public buildings in the k-th group. ;

[0089] D2, according to the first Group's average outdoor temperature and the first The Pearson correlation value for the power consumption of the group of public buildings is the first. Set power consumption labels for public buildings;

[0090] D3. Based on the power consumption labels, divide the power consumption of public buildings into strongly correlated, moderately correlated, and weakly correlated groups. Divide the strongly correlated, moderately correlated, and weakly correlated groups into multiple data segments according to a one-week time span. Use the power consumption of public buildings in week W and week W+1 corresponding to the strongly correlated group to form the first training set, use the power consumption of public buildings in week W and week W+1 corresponding to the moderately correlated group to form the second training set, and use the power consumption of public buildings in week W and week W+1 corresponding to the weakly correlated group to form the third training set.

[0091] D4. Train the first LSTM model using the first training set to obtain the trained first LSTM model; train the second LSTM model using the second training set to obtain the trained second LSTM model; train the third LSTM model using the third training set to obtain the trained third LSTM model.

[0092] The beneficial effects of this invention are as follows:

[0093] This invention designs a dynamic anomaly analysis and early warning system for building electricity consumption in urban areas of frigid regions. By considering the different characteristics of residential and public buildings, this invention specifically designs anomaly identification methods. By utilizing the power consumption in each data collection cycle combined with different abnormal power consumption standards for different buildings, it can accurately identify whether abnormal power consumption behavior is occurring in the current building, avoiding the problem of a single factor significantly affecting the identification results. For residential buildings, this invention employs two methods for identifying abnormal power consumption behavior: a dynamic early warning unit primarily based on dynamic identification and a static early warning unit primarily based on static identification. This allows the system to accurately detect abnormal power consumption behavior in residential buildings, improving the accuracy of anomaly identification. For public buildings, this invention uses a short-term power consumption early warning unit for short-term anomaly warnings, enabling the system to respond promptly to abnormal power consumption behavior; and a long-term power consumption early warning unit for long-term anomaly warnings, allowing for timely equipment adjustments based on long-term warning results to prevent potential abnormal power consumption behavior. This invention combines short-term and long-term power consumption early warning results to improve the accuracy of public building power consumption anomaly identification. Attached Figure Description

[0094] Figure 1 This is a block diagram of the present invention. Detailed Implementation

[0095] Specific implementation method one: as follows Figure 1 As shown in the figure, this embodiment of a city-level building power dynamic anomaly analysis and early warning system in a cold region includes: a building power monitoring module, a residential building power consumption early warning module, a public building power consumption early warning module, and an energy consumption data comparison module;

[0096] The building power monitoring module is used to obtain the power consumption of each building in each data collection cycle within the monitoring range. And based on the power consumption of each building in each collection period, the total daily power consumption data of each building is obtained. Monthly total power consumption data and total annual power consumption data Simultaneously, the power consumption per unit area of ​​each building is obtained within each data collection period. and will , , , and Send to the energy consumption data comparison module. Send to the residential building electricity consumption early warning module and the public building electricity consumption early warning module;

[0097] The buildings include: residential buildings and public buildings;

[0098] The residential building electricity consumption early warning module includes: a residential building electricity consumption dynamic early warning unit and a residential building electricity consumption static early warning unit;

[0099] The residential building electricity consumption dynamic early warning unit uses dynamic identification as the main method to enable the system to accurately detect abnormal electricity consumption behavior in residential buildings. It uses the electricity consumption of residential buildings in each collection cycle to obtain dynamic abnormal electricity consumption behavior and issue early warnings.

[0100] Residential buildings refer to buildings used for people's daily living, including: houses, villas, dormitories, and apartments;

[0101] Electricity consumption refers to the total electrical energy consumed by all electrical systems within a building, including heating, air conditioning, lighting, power, and office equipment, per unit of time during operation. The electricity consumption figure is calculated based on the difference in meter readings per unit of time, and the commonly used unit is kilowatt-hour (kWh).

[0102] The residential building electricity consumption static early warning unit uses static identification as the main method to enable the system to accurately detect abnormal electricity consumption behavior in residential buildings. It uses the daily total electricity consumption data of residential buildings to obtain static abnormal electricity consumption behavior and issue early warnings.

[0103] The public building electricity consumption early warning module includes: a public building short-term electricity consumption early warning unit and a public building long-term electricity consumption early warning unit;

[0104] The aforementioned short-term power consumption early warning unit for public buildings enables the system to respond promptly to abnormal power consumption behavior. It uses the power consumption data of the public building in each collection cycle to determine whether the current power consumption of the public building is too high. If the power consumption is too high, a high power consumption early warning for the public building is issued; otherwise, no high power consumption early warning is issued.

[0105] Public buildings refer to buildings used for various public activities, including office buildings, commercial buildings, tourism buildings, science, education, culture and health buildings, communication buildings, and transportation buildings.

[0106] The long-term power consumption early warning unit for public buildings can adjust the equipment in a timely manner based on the long-term power consumption early warning results to prevent abnormal power consumption behaviors that have not yet occurred. It establishes a building power consumption prediction model using the power consumption of public buildings in each collection cycle, and uses the building power consumption prediction model to obtain the power consumption prediction results of public buildings. Then, based on the power consumption prediction results of public buildings, it determines whether there is a high power consumption risk in public buildings. If there is a high power consumption risk, a high power consumption risk warning is issued; otherwise, no high power consumption risk warning is issued.

[0107] The energy consumption data comparison module is used to compare the power consumption of each building within the monitoring range for each data collection cycle. Daily total power consumption data for each building Monthly total power consumption data Annual total power consumption data and the electricity index per unit area of ​​each building within each data collection period Create and output line charts, bar charts, and comparison charts respectively.

[0108] Specific Implementation Method Two: The building energy monitoring module is used to acquire the power consumption of each building within the monitoring range for each data acquisition cycle. And based on the power consumption of each building in each collection period, the total daily power consumption data of each building is obtained. Monthly total power consumption data Total annual electricity consumption data Simultaneously, the power consumption per unit area of ​​each building is obtained within each data collection period. Specifically:

[0109] First, a 1-hour data collection period was set. The meter data for each building was marked at the beginning and end of each collection period. The energy consumption within each collection period was calculated using the difference between the two values, and this energy consumption was taken as the power consumption for that collection period. The unit is kilowatt-hour (kWh);

[0110] In this step, the data collection period is recorded in the format of "year-month-day-hour", for example... This represents the power consumption of building i in the t-th data collection period. If the t-th data collection period is from 14:00 to 15:00 on September 2, 2025, then... This indicates the electricity consumption of building i from 14:00 to 15:00 on September 2, 2025;

[0111] Then, using the power consumption of each building in each collection cycle, the daily (from 1:00 AM to 12:00 AM on the current day) building power consumption data is obtained and marked as the total daily power consumption, denoted as . Specifically:

[0112]

[0113] in, It is the collection cycle number. It is the total number of collection cycles within a day. It is architecture No. Power consumption per data acquisition cycle It is architecture No. Total daily electricity consumption;

[0114] Then, using the daily total power consumption, the building's electricity consumption data for each month (from 1:00 AM on the 1st of the current month to 12:00 PM on the last day of the month) is obtained and marked as the monthly total power consumption, denoted as . Specifically:

[0115]

[0116] in, It is a date label. It is the total number of days in the month. It is architecture No. Total monthly power consumption;

[0117] Then, using the total monthly electricity consumption, the building's electricity consumption data for the entire year (from 1:00 AM on January 1st to 12:00 PM on December 31st) is obtained and marked as the total annual electricity consumption, denoted as . Specifically:

[0118]

[0119] in, It's a month number. It is architecture No. Total annual electricity consumption;

[0120] Finally, the unit area power consumption index for each building in each data collection cycle was obtained. Specifically:

[0121]

[0122] in, It is the unit area electricity index of building i in the t-th data collection period. This represents the power consumption of building i during the t-th data collection cycle. It is the area of ​​building i.

[0123] Specific Implementation Method 3: The residential building electricity consumption dynamic early warning unit, primarily using dynamic identification, enables the system to accurately detect abnormal electricity consumption behavior in residential buildings. It acquires dynamic abnormal electricity consumption behavior of residential buildings by utilizing the electricity consumption of residential buildings within each collection cycle and issues early warnings accordingly. Specifically:

[0124] A1. Utilizing residential buildings The power consumption of residential buildings in the t-th collection cycle is obtained. The month-on-month growth rate of electricity consumption is as follows:

[0125]

[0126] in, Residential buildings The year-on-year growth rate of electricity consumption in the t-th data collection period Residential buildings No. Power consumption per data acquisition cycle Residential buildings No. Power consumption per data collection cycle;

[0127] A2. Utilize the month-on-month growth rate of residential building electricity consumption to obtain dynamic abnormal electricity consumption behavior in residential buildings and issue early warnings, specifically:

[0128] A residential building electricity consumption rate that meets any of the following conditions will be issued a residential building electricity consumption rate report. Dynamic abnormal electricity consumption behavior warning; otherwise, no dynamic abnormal electricity consumption behavior warning will be issued.

[0129] a1. Residential buildings in three consecutive data collection periods The month-on-month growth rate of electricity consumption was greater than 15%;

[0130] a2. Residential buildings for six consecutive data collection periods The month-on-month growth rate of electricity consumption is greater than 10%.

[0131] Specific Implementation Method Four: The residential building electricity consumption static early warning unit, primarily using static identification, enables the system to accurately detect abnormal electricity consumption behavior in residential buildings. It utilizes the daily total electricity consumption data of the residential building to obtain and issue early warnings for static abnormal electricity consumption behavior. Specifically:

[0132] B1. Obtain the information for each residential building within the monitoring range. The total daily electricity consumption for each day of the month, and the first electricity consumption for each residential building. The total daily electricity consumption for each day of the month is sorted in ascending order to obtain the sequence of total daily electricity consumption for each residential building;

[0133] B2. Obtain residential buildings based on the daily total power consumption sequence corresponding to residential buildings. Base threshold Specifically:

[0134]

[0135] in, It is the first The 90th percentile of the daily total electricity consumption sequence for each residential building. , It is the total number of residential buildings within the monitoring area. It refers to the residential building identification number within the monitoring area;

[0136] B3. Obtain static abnormal electricity consumption behavior of residential buildings based on basic thresholds and issue early warnings, specifically:

[0137] Determine the current season. If the current season is the cooling season, execute b1; if the current season is the heating season, execute b2; if the current season is the transitional season, execute b3.

[0138] When the daily average temperature is below 5°C for more than a preset number of days, it indicates that the current period is the heating season; when the daily maximum temperature is above 26°C for more than a preset number of days, it indicates that the current period is the cooling season; other times are the transitional season.

[0139] b1. If it is currently the cooling season, obtain the current day's highest temperature. And the highest temperature in the current day's historical data A static abnormal power consumption behavior warning will be issued if any of the following conditions are met:

[0140] b1.1, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0141] Among them, the highest temperature in the current day's historical data Obtained through the following methods:

[0142] Get the current date (e.g., 2025.5.4); then get the temperature data for all dates with the same date as the current date from the historical monitoring data (temperature monitoring data for the 4th of that month in the historical data, such as 2024.5.4, 2023.5.4), and get the maximum value among all temperature data as the highest temperature in the historical data for the current day. ;

[0143] b1.2, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0144] b1.3, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0145] b1.4, if Then judge Is it greater than or equal to? ;like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0146] b1.5, if Then judge Is it greater than or equal to? ;like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0147] b1.6, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0148] b2. If it is currently the heating season, obtain the current daily average temperature. and the current daily historical average temperature A static abnormal power consumption behavior warning will be issued if any of the following conditions are met:

[0149] b2.1, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0150] The average temperature of the current day's historical data is obtained as follows: obtain the date of the current day (e.g., 2025.5.4); then obtain the temperature data of all the same dates as the current day in the historical monitoring data (temperature monitoring data of the 4th of the month in the historical data, such as 2024.5.4, 2023.5.4), and obtain the average of all temperature data as the average temperature of the previous day's historical data;

[0151] b2.2, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0152] b2.3, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0153] b2.4, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0154] b2.5, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0155] b3. If it is currently the transitional season, obtain the current daily average temperature. and the current daily historical average temperature A static abnormal power consumption behavior warning will be issued if any of the following conditions are met:

[0156] b3.1, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0157] b3.2, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

[0158] In this step, the cooling season refers to the period of the year when outdoor temperatures are relatively high and residential buildings generally require the use of cooling systems. It is determined by a sustained daily maximum temperature above 26°C, or by the air conditioning usage period published by the local energy management department.

[0159] The heating season refers to the period of the year when outdoor temperatures are relatively low and residential buildings generally require the use of heating systems. It is determined based on a continuous daily average temperature below 5°C, or according to the centralized heating period stipulated by the local government.

[0160] The transition season refers to the time of year excluding the heating and cooling seasons.

[0161] Specific Implementation Method Five: The aforementioned short-term power consumption early warning unit for public buildings enables the system to respond promptly to abnormal power consumption behavior. It uses the power consumption of public buildings within each collection cycle to determine whether the current power consumption of the public building is excessive. If excessive power consumption occurs, a high power consumption early warning for the public building is issued; otherwise, no high power consumption early warning is issued. Specifically:

[0162] C1. Set 24 collection cycles as a time window, and obtain the average power consumption within each time window in the power consumption sequence corresponding to each public building, specifically:

[0163]

[0164] in, It is the first Time window of the power consumption sequence corresponding to each public building Average power consumption within;

[0165] In this step, the time window The data in is the first One collection cycle to the previous 24 collection cycles;

[0166] C2. Public buildings using the t-th collection period The power consumption of public buildings The month-on-month growth rate of electricity consumption is as follows:

[0167]

[0168] in, It is the public building in the t-th collection cycle. The month-on-month growth rate of electricity consumption It is the first Public buildings in each collection cycle Power consumption It is the first Public buildings in each collection cycle Power consumption;

[0169] C3, using the first Time window of the power consumption sequence corresponding to each public building Average power consumption in public buildings The month-on-month growth rate of electricity consumption is used to determine whether there is excessive electricity consumption in public buildings. If excessive electricity consumption is found, a high electricity consumption warning for public buildings is issued; otherwise, no high electricity consumption warning is issued. Specifically:

[0170] C3.1, Setting up public buildings Base threshold ;

[0171] C3.2 Determine the current acquisition cycle Is it greater than or equal to? ,like Then issue public buildings Excessive power consumption warning; otherwise, determine the current data collection cycle. Is it during working hours? During working hours, execute C3.3; if During non-working hours, execute C3.4;

[0172] C3.3, Judgment Is it greater than or equal to 20%? Then output acquisition cycle For abnormal data collection cycles, public buildings are issued. A warning will be issued if electricity consumption is too high; otherwise, no warning will be issued.

[0173] C3.4, Judgment Is it greater than or equal to 5%? Then output acquisition cycle For abnormal data collection cycles, public buildings are issued. A warning will be issued if electricity consumption is too high; otherwise, no warning will be issued.

[0174] In this step, working hours refer to the time period during which a public building provides services or performs its main functions under normal operating conditions. The specific start and end times are determined based on the building type and management regulations. For example, office buildings are generally open from 8:00 AM to 6:00 PM on weekdays, schools during class teaching hours, and shopping malls during business hours. Non-working hours refer to all other times besides the aforementioned working hours, including nighttime, holidays, or times when the building is not operating its main functions. This invention also counts the number of abnormal electricity consumption collection cycles for the current month. If the number of abnormal collection cycles exceeds 20% of the total number of collection cycles for the current month, it is reported to the relevant management department.

[0175] Specific Implementation Method Six: The long-term electricity consumption early warning unit for public buildings can adjust its equipment in a timely manner based on the long-term electricity consumption early warning results to prevent abnormal electricity consumption behaviors that have not yet occurred. It establishes a building energy prediction model using the electricity consumption of public buildings within each collection cycle, and obtains the predicted electricity consumption results of public buildings using this model. Then, based on the predicted electricity consumption results, it determines whether the public building has a high electricity consumption risk. If there is a high electricity consumption risk, a high electricity consumption risk warning is issued; otherwise, no high electricity consumption risk warning is issued. Specifically:

[0176] S1. Obtain the power consumption sequence of the public building for the week preceding the predicted collection period. Input the power consumption sequence of the public building for the week preceding the predicted date into the building power prediction model to obtain the predicted power consumption result of the public building for the predicted collection period. ;

[0177] The building power prediction model is a trained first LSTM model, a trained second LSTM model, or a trained third LSTM model.

[0178] If the month in which the data collection period to be predicted is located is a strongly correlated month, then the building power prediction model is the first LSTM model; if the month in which the data collection period to be predicted is located is a moderately correlated month, then the building power prediction model is the second LSTM model; if the month in which the data collection period to be predicted is located is a weakly correlated month, then the building power prediction model is the third LSTM model.

[0179] The trained first LSTM model, the trained second LSTM model, and the trained third LSTM model are obtained through the following methods:

[0180] D1. Obtain the power consumption sequence of public buildings for each collection period within a year, and divide the power consumption sequence of public buildings for each collection period within a year into 12 groups according to the month. Then, obtain the average outdoor temperature for each month and each collection period to form an average outdoor temperature sequence, and obtain the Pearson correlation value between the average outdoor temperature of the k-th group and the power consumption of public buildings in the k-th group. ,as follows:

[0181]

[0182] in, It is a variable and covariance, It is a variable standard deviation It is a variable standard deviation , It is the first Group of public buildings power consumption and the first Group average outdoor temperature;

[0183] In this step, the Pearson correlation coefficient is used to measure the degree of linear correlation between two continuous variables. Its value ranges from -1 to 1. The closer the absolute value is to 1, the stronger the correlation, and the closer it is to 0, the weaker the correlation.

[0184] D2, according to the first Group's average outdoor temperature and the first The Pearson correlation value for the power consumption of the group of public buildings is the first. The power consumption of public buildings is labeled with power consumption tags, specifically:

[0185] like Then the first The label for the electricity consumption of the group of public buildings is set to weak correlation;

[0186] like Then the first The label for the electricity consumption of public buildings is set to be relevant.

[0187] like Then the first The labels for the electricity consumption of public buildings are set to be strongly correlated;

[0188] D3. Based on the power consumption labels, divide the power consumption of public buildings into strongly correlated, moderately correlated, and weakly correlated groups. Divide the strongly correlated, moderately correlated, and weakly correlated groups into multiple data segments according to a one-week time span. Use the power consumption of public buildings in week W and week W+1 corresponding to the strongly correlated group to form the first training set, use the power consumption of public buildings in week W and week W+1 corresponding to the moderately correlated group to form the second training set, and use the power consumption of public buildings in week W and week W+1 corresponding to the weakly correlated group to form the third training set.

[0189] D4. Train the first LSTM model using the first training set to obtain the trained first LSTM model; train the second LSTM model using the second training set to obtain the trained second LSTM model; train the third LSTM model using the third training set to obtain the trained third LSTM model.

[0190] The loss function for training the first LSTM model, the second LSTM model, and the third LSTM model is the error between the predicted result and the true value.

[0191] This step involves iteratively optimizing the parameters to minimize the prediction error until the model converges and training is complete.

[0192] S2. Obtain the specific year corresponding to the data collection period to be predicted. ,moon ,day and collection cycle Then obtain Year moon, Daily collection cycle and Year moon, Daily collection cycle Public building power consumption and and obtain , , Average value:

[0193]

[0194] S3, Utilization Get the base threshold ,like Greater than If the power consumption risk warning is not issued, a high power consumption risk warning will be issued; otherwise, no high power consumption risk warning will be issued.

[0195] The basic threshold .

[0196] Specific Implementation Method Seven: Strongly correlated months, moderately correlated months, and weakly correlated months are determined using the following method:

[0197] E1. Obtain the power consumption sequence of public buildings for each collection period within a year, and divide the power consumption sequence of public buildings for each collection period within a year into 12 groups according to the month. Then, obtain the outdoor average temperature for each month and each collection period to form an outdoor average temperature sequence, and obtain the Pearson correlation value between the outdoor average temperature of the k-th group and the power consumption of public buildings in the k-th group. :

[0198]

[0199] in, It is a variable and covariance, It is a variable standard deviation It is a variable standard deviation , It is the first Group of public buildings power consumption and the first Group average outdoor temperature;

[0200] E2, according to the first Group's average outdoor temperature and the first The Pearson correlation value for the power consumption of the group of public buildings is the first. The power consumption of public buildings is labeled with power consumption tags, specifically:

[0201] like Then the first The label for the electricity consumption of the group of public buildings is set to weak correlation;

[0202] like Then the first The label for the electricity consumption of public buildings is set to be relevant.

[0203] like Then the first The labels for the electricity consumption of public buildings are set to be strongly correlated;

[0204] E3. Based on the power consumption labels, the power consumption of public buildings is divided into strongly correlated, moderately correlated, and weakly correlated groups. The months corresponding to the data in the strongly correlated group are the strongly correlated months, the months corresponding to the data in the moderately correlated group are the moderately correlated months, and the months corresponding to the data in the weakly correlated group are the weakly correlated months.

Claims

1. A dynamic anomaly analysis and early warning system for building electrical power in urban areas of frigid regions, characterized in that... The system includes: a building power monitoring module, a residential building power consumption early warning module, and a public building power consumption early warning module; The building power monitoring module is used to obtain the power consumption of each building in each data collection cycle within the monitoring range. ,Will Send to the residential building electricity consumption early warning module and the public building electricity consumption early warning module; The buildings include: residential buildings and public buildings; The residential building electricity consumption early warning module includes: a residential building electricity consumption dynamic early warning unit and a residential building electricity consumption static early warning unit; The residential building electricity consumption dynamic early warning unit uses the electricity consumption of the residential building in each collection cycle to obtain dynamic abnormal electricity consumption behavior of the residential building and issue early warnings. The residential building electricity consumption static early warning unit uses the daily total electricity consumption data of the residential building to obtain the static abnormal electricity consumption behavior of the residential building and issue an early warning. The public building electricity consumption early warning module includes: a public building short-term electricity consumption early warning unit and a public building long-term electricity consumption early warning unit; The public building short-term power consumption early warning unit uses the power consumption of the public building in each collection cycle to determine whether the current public building has excessive power consumption. If excessive power consumption occurs, a high power consumption early warning is issued; otherwise, no high power consumption early warning is issued. The long-term power consumption early warning unit for public buildings establishes a building power prediction model using the power consumption of public buildings in each collection cycle, and obtains the power consumption prediction results of public buildings using the building power prediction model. Then, based on the power consumption prediction results of public buildings, it determines whether there is a high power consumption risk in public buildings. If there is a high power consumption risk, a high power consumption risk warning is issued; otherwise, no high power consumption risk warning is issued.

2. The urban-level building power dynamic anomaly analysis and early warning system for severely cold regions according to claim 1, characterized in that: The building power monitoring module is used to obtain the power consumption of each building in each data collection cycle within the monitoring range. And based on the power consumption of each building in each collection period, the total daily power consumption data of each building is obtained. Monthly total power consumption data and total annual power consumption data Simultaneously, the power consumption per unit area of ​​each building is obtained within each data collection period. Specifically: First, set the preset duration as the data collection cycle, and obtain the power consumption of each building within the monitoring range for each data collection cycle. ; The preset duration is less than or equal to 1 hour; Then, by using the power consumption of each building in each data collection cycle, the total daily power consumption can be obtained. Total monthly power consumption and total annual power consumption Specifically: in, It is the collection cycle number. It is the total number of collection cycles within a day. It is a date label. It is the total number of days in the month. It's a month number. It is architecture No. Power consumption per data acquisition cycle It is architecture No. Total daily electricity consumption It is architecture No. Total monthly power consumption It is architecture No. Total annual electricity consumption; Finally, the unit area power consumption index for each building in each data collection cycle was obtained. Specifically: in, It is the unit area electricity index of building i in the t-th data collection period. This represents the power consumption of building i during the t-th data collection cycle. It is the area of ​​building i.

3. The urban-level building power dynamic anomaly analysis and early warning system for severely cold regions according to claim 2, characterized in that: The residential building electricity consumption dynamic early warning unit uses the electricity consumption of the residential building in each collection cycle to obtain dynamic abnormal electricity consumption behavior of the residential building and issue early warnings, specifically: A1. Utilizing residential buildings The power consumption of residential buildings in the t-th collection cycle is obtained. The month-on-month growth rate of electricity consumption is as follows: in, Residential buildings The year-on-year growth rate of electricity consumption in the t-th data collection period Residential buildings No. Power consumption per data acquisition cycle Residential buildings No. Power consumption per data collection cycle; A2. Utilize the month-on-month growth rate of residential building electricity consumption to obtain dynamic abnormal electricity consumption behavior in residential buildings and issue early warnings, specifically: A residential building electricity consumption rate that meets any of the following conditions will be issued a residential building electricity consumption rate report. Dynamic abnormal electricity consumption behavior warning; otherwise, no dynamic abnormal electricity consumption behavior warning will be issued. a1. Residential buildings in three consecutive data collection periods The month-on-month growth rate of electricity consumption was greater than 15%; a2. Residential buildings for six consecutive data collection periods The month-on-month growth rate of electricity consumption is greater than 10%.

4. The urban-level building power dynamic anomaly analysis and early warning system for severely cold regions according to claim 3, characterized in that: The residential building electricity consumption static early warning unit uses the daily total electricity consumption data of the residential building to obtain static abnormal electricity consumption behavior of the residential building and issue early warnings, specifically: B1. Obtain the information for each residential building within the monitoring range. The total daily electricity consumption for each day of the month, and the first electricity consumption for each residential building. The total daily electricity consumption for each day of the month is sorted in ascending order to obtain the sequence of total daily electricity consumption for each residential building; B2. Obtain residential buildings based on the daily total power consumption sequence corresponding to residential buildings. Base threshold Specifically: in, It is the first The 90th percentile of the daily total electricity consumption sequence for each residential building. , It is the total number of residential buildings within the monitoring area. It refers to the residential building identification number within the monitoring area. It is the first The 90th percentile of the daily total electricity consumption sequence for each residential building. It is the first The 90th percentile value of the daily total electricity consumption sequence corresponding to each residential building; B3. Obtain static abnormal electricity consumption behavior of residential buildings based on the basic threshold of residential buildings and issue early warnings.

5. The urban-level building power dynamic anomaly analysis and early warning system for severely cold regions according to claim 4, characterized in that: In B3, static abnormal electricity consumption behavior of residential buildings is obtained based on the basic threshold of residential buildings, and early warning is issued. Specifically: Determine the current season. If the current season is the cooling season, execute b1; if the current season is the heating season, execute b2; if the current season is the transitional season, execute b3. When the daily average temperature is below 5°C for more than a preset number of days, it indicates that the current period is the heating season; when the daily maximum temperature is above 26°C for more than a preset number of days, it indicates that the current period is the cooling season; other times are the transitional season. b1. If it is currently the cooling season, obtain the current day's highest temperature. And the highest temperature in the current day's historical data A static abnormal power consumption behavior warning will be issued if any of the following conditions are met: b1.1, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b1.2, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b1.3, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b1.4, if Then judge Is it greater than or equal to? ;like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b1.5, if Then judge Is it greater than or equal to? ;like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b1.6, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b2. If it is currently the heating season, obtain the current daily average temperature. and the current daily historical average temperature A static abnormal power consumption behavior warning will be issued if any of the following conditions are met: b2.1, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b2.2, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b2.3, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b2.4, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b2.5, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b3. If it is currently the transitional season, obtain the current daily average temperature. and the current daily historical average temperature A static abnormal power consumption behavior warning will be issued if any of the following conditions are met: b3.1, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued. b3.2, if Then judge Is it greater than or equal to? ,like If the system detects abnormal static electricity consumption behavior in a residential building, an early warning will be issued; otherwise, no early warning will be issued.

6. The urban-level building power dynamic anomaly analysis and early warning system for severely cold regions according to claim 5, characterized in that: The short-term power consumption early warning unit for public buildings uses the power consumption data of each data collection cycle to determine whether the current power consumption of the public building is too high. If excessive power consumption occurs, a high power consumption early warning is issued; otherwise, no high power consumption early warning is issued. Specifically: C1. Set 24 collection cycles as a time window, and obtain the average power consumption within each time window in the power consumption sequence corresponding to each public building, specifically: in, It is the first Time window of the power consumption sequence corresponding to each public building Average power consumption within, It is the first The first public building Power consumption per data collection cycle; C2, Utilization of public buildings The power consumption of public buildings in the t-th collection cycle is obtained. The month-on-month growth rate of electricity consumption is as follows: in, It is a public building The year-on-year growth rate of electricity consumption in the t-th data collection period It is a public building The Power consumption per data acquisition cycle It is a public building No. Power consumption per data collection cycle; C3, using the first Time window of the power consumption sequence corresponding to each public building Average power consumption in public buildings The month-on-month growth rate of electricity consumption is used to determine whether there is excessive electricity consumption in public buildings. If excessive electricity consumption occurs, a high electricity consumption warning for public buildings is issued; otherwise, no high electricity consumption warning is issued.

7. A dynamic anomaly analysis and early warning system for building electrical power in urban areas of severely cold regions according to claim 6, characterized in that: The use of the first in C3 Time window of the power consumption sequence corresponding to each public building Average power consumption in public buildings The month-on-month growth rate of electricity consumption is used to determine whether there is excessive electricity consumption in public buildings. If excessive electricity consumption is found, a high electricity consumption warning for public buildings is issued; otherwise, no high electricity consumption warning is issued. Specifically: C3.1, Setting up public buildings Base threshold ; C3.2 Determine the current acquisition cycle Is it greater than or equal to? ,like Then issue public buildings Excessive power consumption warning; otherwise, determine the current data collection cycle. Is it during working hours? During working hours, execute C3.3; if During non-working hours, execute C3.4; C3.3, Judgment Is it greater than or equal to 20%? Then output acquisition cycle For abnormal data collection cycles, public buildings are issued. A warning will be issued if electricity consumption is too high; otherwise, no warning will be issued. C3.4, Judgment Is it greater than or equal to 5%? Then output acquisition cycle For abnormal data collection cycles, public buildings are issued. A warning will be issued if electricity consumption is too high; otherwise, no warning will be issued.

8. A dynamic anomaly analysis and early warning system for building electrical power in urban areas of frigid regions according to claim 7, characterized in that: The long-term electricity consumption early warning unit for public buildings establishes a building electricity prediction model using the electricity consumption data of each collection cycle. It then uses this model to obtain the predicted electricity consumption results for the public buildings. Based on these results, it determines whether the public buildings have a high electricity consumption risk. If a high electricity consumption risk is detected, a high electricity consumption risk warning is issued; otherwise, no warning is issued. Specifically: S1. Obtain the power consumption sequence of the public building for the week preceding the predicted collection period. Input the power consumption sequence of the public building for the week preceding the predicted date into the building power prediction model to obtain the predicted power consumption result of the public building for the predicted collection period. ; The building power prediction model is a trained first LSTM model, a trained second LSTM model, or a trained third LSTM model. If the month in which the data collection period to be predicted is located is a strongly correlated month, then the building power prediction model is the first LSTM model; if the month in which the data collection period to be predicted is located is a moderately correlated month, then the building power prediction model is the second LSTM model; if the month in which the data collection period to be predicted is located is a weakly correlated month, then the building power prediction model is the third LSTM model. S2. Obtain the specific year corresponding to the data collection period to be predicted. ,moon ,day and collection cycle Then obtain Year moon, Daily collection cycle and Year moon, Daily collection cycle Public building power consumption and and obtain , , Average value: S3, Utilization Get the base threshold ,like Greater than Then a high electricity consumption risk warning will be issued; Otherwise, no high electricity consumption risk warning will be issued; The basic threshold .

9. A dynamic anomaly analysis and early warning system for building electrical power in urban areas of frigid regions according to claim 8, characterized in that: Strongly correlated months, moderately correlated months, and weakly correlated months are determined using the following method: E1. Obtain the power consumption sequence of public buildings for each collection period within a year, and divide the power consumption sequence of public buildings for each collection period within a year into 12 groups according to the month. Then, obtain the outdoor average temperature for each month and each collection period to form an outdoor average temperature sequence, and obtain the Pearson correlation value between the outdoor average temperature of the k-th group and the power consumption of public buildings in the k-th group. : in, It is a variable and covariance, It is a variable standard deviation It is a variable standard deviation , It is the first Group of public buildings power consumption and the first Group average outdoor temperature; E2, according to the first Group's average outdoor temperature and the first The Pearson correlation value for the power consumption of the group of public buildings is the first. The power consumption of public buildings is labeled with power consumption tags, specifically: like Then the first The label for the electricity consumption of the group of public buildings is set to weak correlation; like Then the first The label for the electricity consumption of public buildings is set to be relevant. like Then the first The labels for the electricity consumption of public buildings are set to be strongly correlated; E3. Based on the power consumption labels, the power consumption of public buildings is divided into strongly correlated, moderately correlated, and weakly correlated groups. The months corresponding to the data in the strongly correlated group are the strongly correlated months, the months corresponding to the data in the moderately correlated group are the moderately correlated months, and the months corresponding to the data in the weakly correlated group are the weakly correlated months.

10. A dynamic anomaly analysis and early warning system for building electrical power in urban areas of frigid regions according to claim 9, characterized in that: The trained first LSTM model, the trained second LSTM model, or the trained third LSTM model are obtained through the following methods: D1. Obtain the power consumption sequence of public buildings for each collection period within a year, and divide the power consumption sequence of public buildings for each collection period within a year into 12 groups according to the month. Then, obtain the average outdoor temperature for each month and each collection period to form an average outdoor temperature sequence, and obtain the Pearson correlation value between the average outdoor temperature of the k-th group and the power consumption of public buildings in the k-th group. ; D2, according to the first Group's average outdoor temperature and the first The Pearson correlation value for the power consumption of the group of public buildings is the first. Set power consumption labels for public buildings; D3. Based on the power consumption labels, divide the power consumption of public buildings into strongly correlated, moderately correlated, and weakly correlated groups. Divide the strongly correlated, moderately correlated, and weakly correlated groups into multiple data segments according to a one-week time span. Use the power consumption of public buildings in week W and week W+1 corresponding to the strongly correlated group to form the first training set, use the power consumption of public buildings in week W and week W+1 corresponding to the moderately correlated group to form the second training set, and use the power consumption of public buildings in week W and week W+1 corresponding to the weakly correlated group to form the third training set. D4. Train the first LSTM model using the first training set to obtain the trained first LSTM model; train the second LSTM model using the second training set to obtain the trained second LSTM model; train the third LSTM model using the third training set to obtain the trained third LSTM model.

Images