Statistics-based periodic building energy consumption abnormity diagnosis method
By classifying and abnormal diagnosis of power energy consumption based on statistics, combining multiple algorithms to analyze periodic abnormalities in building power energy consumption, the problem of insufficient scientific diagnosis of power energy consumption in the existing technology is solved, and office buildings are realized In-depth analysis of power consumption and effective identification of abnormal causes.
Patent Information
- Application Number
- CN202411189450.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-28
- Publication Date
- 2024-12-03
AI Technical Summary
The existing research technology for power consumption diagnosis is difficult to achieve in-depth analysis and scientific diagnosis of building power consumption, especially when dealing with periodic total energy consumption abnormalities.
The cyclical building energy consumption abnormality diagnosis method based on statistics is adopted, and the daily energy consumption method and benchmark are constructed by classifying the historical data of electricity, combining the optimal similar daily and same working conditions sorting and short-term month-on-month algorithm for multiple energy consumption. Line abnormality diagnosis, and the Otsu algorithm and absolute distance algorithm are used to analyze the causes of abnormalities.
The interpretability diagnosis of periodic total energy consumption abnormalities in office buildings is achieved, the scientificity and accuracy of the diagnosis results are improved, and the cause analysis of abnormalities can be effectively combined with the historical data of the power consumption line.
Smart Images

Figure CN119066442A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of building energy consumption diagnosis, and in particular to a method for diagnosing abnormal periodic building energy consumption based on statistics. Background Art
[0002] The dynamic energy consumption monitoring platform of buildings is in a rapid development stage, among which the power energy consumption monitoring platform is developing the fastest. The emergence of the power energy consumption monitoring platform has changed the traditional way of manually recording the energy consumption of buildings, improved the real-time and accuracy of building energy consumption, and realized the in-depth analysis and application of building power energy consumption. Relying on the power energy consumption monitoring platform and other related power energy consumption monitoring equipment, the power energy consumption diagnosis research technology has flourished. However, the existing power energy consumption diagnosis research technology needs to be further deepened to make the diagnosis results more scientific and in line with the needs of development. Summary of the invention
[0003] The purpose of the present invention is to overcome the shortcomings and defects of the prior art and provide a statistically based method for diagnosing abnormal periodic building energy consumption. The method of the present invention, based on the classification and feature analysis of power energy consumption lines, classifies and analyzes the abnormality and causes of the periodic total energy consumption of office building power lines, and realizes the explainability of the abnormal periodic total energy consumption.
[0004] A statistically based method for diagnosing abnormal periodic building energy consumption comprises the following steps:
[0005] Based on the historical power data of the power database, the power consumption of office buildings is classified to obtain multiple types of energy consumption lines:
[0006] According to the historical power energy consumption characteristics, a daily line energy consumption method and a daily line energy consumption benchmark are constructed for lines with different energy consumption;
[0007] Based on the daily line energy consumption method and the daily line energy consumption benchmark, a diagnosis is made as to whether multiple types of energy consumption lines are abnormal; when energy consumption is diagnosed as abnormal, an analysis is performed on the causes of abnormal power energy consumption in office buildings.
[0008] Among them, the classification of office building power energy consumption based on historical data of the power database and multiple types of energy consumption lines are obtained by judging the correlation between the time series of daily and hourly energy consumption in the power historical data and the corresponding influencing factors, and then classifying the office building power energy consumption based on the energy consumption line data characteristics to obtain multiple types of energy consumption lines.
[0009] Among them, the multiple types of energy consumption lines include: Class I, Class II, Class III, Class IV, and Class V;
[0010] Among them, the data of Class I energy consumption lines are characterized by stable daily energy consumption and small discreteness, the data of Class II energy consumption lines are characterized by unstable daily energy consumption and stable hourly energy consumption, the data of Class III energy consumption lines are characterized by non-stable daily energy consumption and strong correlation between hourly energy consumption and climate conditions, the data of Class IV energy consumption lines are characterized by non-stable daily energy consumption and weak correlation with outdoor climate and stable hourly energy consumption during daily operation, and the data of Class V energy consumption lines are characterized by non-stable daily energy consumption and weak correlation with outdoor climate and non-stable hourly energy consumption during daily operation.
[0011] Among them, the daily line energy consumption method and daily line energy consumption benchmark of different energy consumption lines include the optimal similar day algorithm and its energy consumption benchmark, the same working condition sorting algorithm and its energy consumption benchmark and the short-term year-on-year algorithm and its energy consumption benchmark.
[0012] The optimal similar day algorithm is used to diagnose whether multiple types of energy consumption lines are abnormal, including:
[0013] Based on the test day, select N days of electricity historical data, and select the same type of historical data as the test day from the N days of electricity historical data;
[0014] After normalizing the selected historical data, the Mahalanobis distance is calculated to calculate the distance between the data and the test day data. The data with a Mahalanobis distance less than the set threshold are sorted in descending order. According to the sorted distance, the energy consumption data of m similar days are selected to take the average value, and it is determined whether the average value is less than the abnormal threshold. Whether the energy consumption is abnormal or not is diagnosed based on the judgment result.
[0015] The optimal similar day algorithm uses five parameters, namely, day type, weather type, maximum temperature, minimum temperature, and average temperature, as input parameters for determining the optimal similar day.
[0016] The same operating condition sorting algorithm is used to diagnose whether multiple types of energy consumption lines are abnormal, including:
[0017] Select the same working condition based on the day to be tested, and select N days of power historical data from the historical data based on the same working condition;
[0018] Arrange the data of the N days in ascending order according to energy consumption values;
[0019] Determine whether the energy consumption value of the day to be tested is higher than the sorting threshold in the ranking of the energy consumption values of the data of N days, and give a diagnosis result of whether the energy is abnormal or not based on the judgment result.
[0020] The short-term month-on-month algorithm is used to diagnose whether multiple types of energy consumption lines are abnormal, and the data trend within the month-on-month sliding T is used to perform abnormal analysis, including:
[0021]
[0022] Where t is the month-on-month abnormal threshold, count_num is the month-on-month abnormal count value, T is the month-on-month sliding window, i is the energy consumption value of the comparison day, and nowvalue is the energy consumption value of the day to be tested.
[0023] The analysis of the causes of abnormal power consumption in office buildings includes analyzing the causes of abnormal line energy consumption from the aspect of operating time using the Otsu algorithm, and analyzing the causes of abnormal line energy consumption from the aspect of hourly energy consumption level using the absolute distance algorithm to obtain analysis results.
[0024] Among them, the Otsu algorithm automatically identifies the daily operation start and stop times and calculates the line operation time of the day through image analysis of the operation interface of the operation equipment.
[0025] The present invention takes the energy consumption data of the office building sub-item power energy consumption monitoring platform as the research object, and through the characteristic analysis of the sub-item line energy consumption data of the office building sub-item power energy consumption monitoring platform, proposes a daily energy consumption abnormality diagnosis and cause analysis method and a real-time energy consumption diagnosis method for sub-lines with different data characteristics. It can effectively combine the historical data of the power energy consumption line, use the similar day, sorting and short-term year-on-year methods to analyze the causes of energy consumption abnormality of the power line, and provide corresponding energy consumption abnormal causes. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a flow chart of the statistical-based method for diagnosing abnormal periodic building energy consumption of the present invention.
[0027] Figure 2 It is a schematic diagram of the optimal similarity algorithm of the present invention for judging whether the energy consumption is abnormal or not.
[0028] Figure 3 It is a schematic diagram of the statistical-based periodic building energy consumption anomaly diagnosis method of the present invention.
[0029] Figure 4 It is a schematic diagram of judging whether energy consumption is abnormal based on the same working condition algorithm of the present invention. DETAILED DESCRIPTION
[0030] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0031] The present invention adopts the optimal similar day algorithm, sorting algorithm and short-term comparative algorithm to construct the periodic total energy consumption benchmark of each line in the office building, realizes the diagnosis of abnormal periodic total energy consumption of five types of lines, and analyzes the causes of abnormal line energy consumption from two aspects of operating time and hourly energy consumption level through Otsu algorithm and absolute distance calculation, realizes the diagnosis of abnormal periodic energy consumption of office building power sub-lines and the analysis of abnormal causes.
[0032] See also Figure 1 As shown, the statistically based method for diagnosing abnormal periodic building energy consumption includes the following steps:
[0033] Based on the historical power data of the power database, the power consumption of office buildings is classified to obtain multiple types of energy consumption lines:
[0034] According to the historical power energy consumption characteristics, a daily line energy consumption method and a daily line energy consumption benchmark are constructed for lines with different energy consumption;
[0035] Based on the daily line energy consumption method and the daily line energy consumption benchmark, a diagnosis is made as to whether multiple types of energy consumption lines are abnormal; when energy consumption is diagnosed as abnormal, an analysis is performed on the causes of abnormal power energy consumption in office buildings.
[0036] Among them, the classification of office building power energy consumption based on historical data of the power database and multiple types of energy consumption lines are obtained by judging the correlation between the time series of daily and hourly energy consumption in the power historical data and the corresponding influencing factors, and then classifying the office building power energy consumption based on the energy consumption line data characteristics to obtain multiple types of energy consumption lines.
[0037] Among them, the multiple types of energy consumption lines include: Class I, Class II, Class III, Class IV, and Class V;
[0038] Among them, the data of Class I energy consumption lines are characterized by stable daily energy consumption and small discreteness, the data of Class II energy consumption lines are characterized by unstable daily energy consumption and stable hourly energy consumption, the data of Class III energy consumption lines are characterized by non-stable daily energy consumption and strong correlation between hourly energy consumption and climate conditions, the data of Class IV energy consumption lines are characterized by non-stable daily energy consumption and weak correlation with outdoor climate and stable hourly energy consumption during daily operation, and the data of Class V energy consumption lines are characterized by non-stable daily energy consumption and weak correlation with outdoor climate and non-stable hourly energy consumption during daily operation.
[0039] Specifically, the energy consumption of Class I and Class II energy consumption lines is stable under known operating conditions, so the energy consumption under the same operating conditions (such as monthly type or daily type) is selected to build the building energy consumption benchmark for analysis. Class III energy consumption lines are strongly correlated with outdoor weather temperature, and similar days are used to build the building energy consumption benchmark; Class V energy consumption lines have weak correlation with weather and do not have month-on-month similarity. Line energy consumption is mainly affected by manual operation, and its energy consumption change law is not easy to obtain directly, but it can be analyzed indirectly through similar days. Therefore, Class III and Class V lines use similar days to build the building energy consumption benchmark for analysis. Class IV line energy consumption is weakly correlated with weather and the energy consumption is stable hour by hour during daily operation, which is not suitable for the similar day method. Therefore, energy consumption anomaly diagnosis is performed through a short-term month-on-month algorithm.
[0040] Based on the analysis of the time series characteristics of office building power consumption and the research on the correlation of influencing factors, the present invention proposes an office building energy consumption classification based on historical data. On the basis of understanding the office building equipment operation strategy and the basic composition of the line, the historical energy consumption characteristics are explored to lay the foundation for realizing the abnormal diagnosis of power consumption of multi-mode sub-item energy consumption. At the same time, it effectively broadens the application scope of the office building power consumption diagnosis method, and specifically solves the problem that the energy consumption anomaly cannot be effectively judged directly according to the equipment characteristics and indicators due to the impure composition of the office building energy consumption line. The data-based office building power consumption classification algorithm performs time series and influencing factor correlation judgment on the daily and hourly energy consumption, and finally obtains five types of energy consumption lines according to the energy consumption line data characteristics, laying a data foundation for the next step of establishing multi-mode energy consumption anomaly diagnosis using data characteristics.
[0041] The daily line energy consumption method and daily line energy consumption benchmark of different energy consumption lines include the best similar day algorithm and its energy consumption benchmark, the same working condition sorting algorithm and its energy consumption benchmark, and the short-term comparison algorithm and its energy consumption benchmark. The same working condition sorting algorithm and the short-term comparison algorithm use the absolute size of the data as the judgment condition.
[0042] The optimal similar day algorithm is used to diagnose whether multiple types of energy consumption lines are abnormal, including:
[0043] Based on the test day, select N days of electricity historical data, and select the same type of historical data as the test day from the N days of electricity historical data;
[0044] After normalizing the selected historical data, the Mahalanobis distance is calculated to calculate the distance between the data and the test day data. The data with a Mahalanobis distance less than the set threshold are sorted in descending order. According to the sorted distance, the energy consumption data of m similar days are selected to take the average value, and it is determined whether the average value is less than the abnormal threshold α. Whether the energy consumption is abnormal or not is diagnosed based on the judgment result.
[0045] According to different needs and application scenarios, the present invention has different methods and evaluation criteria of the similar day principle. The energy consumption diagnosis standard is established and diagnosed for the daily energy consumption of office building electricity energy consumption by similarity. The energy consumption of the same office building is complicated, and the daily energy consumption is mainly affected by two aspects: on the one hand, the daily energy consumption is related to the day type of the office building (such as the difference between working days and non-working days, the classification or type between working hours on working days and non-working days). The day type affects the behavior of personnel and the operation status of equipment in the office building, which is one of the main factors affecting energy consumption; on the other hand, the daily energy consumption is also affected by the outdoor climate, especially the temperature and humidity, among which the humidity distribution and change law are difficult to analyze, and the outdoor weather temperature and humidity are related to the outdoor weather type. Therefore, the outdoor weather type is used instead of humidity as one of the input conditions. Therefore, the optimal similar day algorithm uses the five parameters of day type, weather type, maximum temperature, minimum temperature, and average temperature as the judgment input parameters of the optimal similar day. Specifically, the Mahalanobis distance of the corresponding data is calculated for these parameters in the historical data, and then the average Mahalanobis distance of multiple data is taken, and the data are sorted according to the average Mahalanobis distance, so as to realize the selection of similar days.
[0046] Among them, the average value method is used to construct the diagnostic benchmark for daily abnormal energy consumption of the line, and the average of a set of m similar days is used to form the diagnostic benchmark for abnormal energy consumption. There are two main reasons for adopting this method: on the one hand, since the building automation level needs to be improved, there may be errors in the daily start and stop time of the equipment, and the average value method can reduce the impact of manual operation on the data; on the other hand, since the object is a real line, the voltage of the real power line is unstable. Combined with the characteristics of the line's own equipment, the average value method can reduce the impact of power fluctuations on hourly energy consumption and improve the accuracy of the energy consumption benchmark. At the same time, according to the abnormal sensitivity of each line to the power energy consumption of the office building, the abnormal energy consumption judgment threshold of each line is determined according to the actual building situation and management and operation requirements.
[0047] The Mahalanobis distance is calculated by first calculating the covariance matrix and then calculating based on the covariance matrix, wherein the matrix algorithm expression of the covariance is as follows:
[0048]
[0049] Where S represents the covariance matrix, x i ,y i Respectively represent the Dth and jth elements of the two groups of samples, μ i and μ j Respectively represent x i ,y i The mean of the corresponding dimension, E[·] represents the mathematical expectation;
[0050] After the covariance matrix S is known, the Mahalanobis distance D between two sample sets is N The calculation formula for (x, y) is:
[0051]
[0052] Among them, x and y are two sample vectors, S -1 is the inverse matrix of the covariance matrix S.
[0053] This Mahalanobis distance formula represents the Mahalanobis distance between samples x and y. It takes into account the correlation between dimensions, unlike the Euclidean distance which assumes that the dimensions are independent of each other.
[0054] The Mahalanobis distance algorithm is based on the Euclidean distance and takes into account the similarities between multiple factors to calculate the distance, so it has higher accuracy.
[0055] The same operating condition sorting algorithm is used to diagnose whether multiple types of energy consumption lines are abnormal, including:
[0056] Select the same working condition based on the day to be tested, and select N days of power historical data from the historical data based on the same working condition;
[0057] Arrange the data of the N days in ascending order according to energy consumption values;
[0058] Determine whether the energy consumption value of the day to be tested is higher than the sorting threshold β in the ranking of the energy consumption values of the data of N days, and give a diagnosis result of whether the energy is abnormal or not based on the judgment result.
[0059] The short-term comparison algorithm refers to comparing the current time range of the time series with the adjacent previous time range, and is suitable for setting the indicators of the current time range according to the numbers of the adjacent time range. Specifically, in this application, the short-term comparison algorithm is used to diagnose whether multiple types of energy consumption lines are abnormal, and the data trend within the comparison sliding T is used for abnormal analysis, including:
[0060]
[0061] Where t is the month-on-month abnormal threshold, coyunt_num is the month-on-month abnormal count value, T is the month-on-month sliding window, i is the energy consumption value of the comparison day, and nowvalue is the energy consumption value of the day to be tested.
[0062] The same working condition sorting algorithm and the short-term year-on-year algorithm are both common statistical analysis methods. The same working condition sorting algorithm sorts one or more groups of data according to a predetermined pattern through a specific algorithm and method, and the new sequence after sorting follows a specified rule. The same working condition sorting algorithm of the present invention is mainly used in Class I and Class II lines. The daily energy consumption lines are stable under the same working conditions. Therefore, the ranking comparison method is used to diagnose energy consumption anomalies, and the abnormal ranking range is set according to the sensitivity of the line energy consumption anomaly.
[0063] The analysis of the causes of abnormal power consumption in office buildings includes analyzing the causes of abnormal line energy consumption from the aspect of operating time using the Otsu algorithm, and analyzing the causes of abnormal line energy consumption from the aspect of hourly energy consumption level using the absolute distance algorithm to obtain analysis results.
[0064] Among them, the Otsu algorithm automatically identifies the daily operation start and stop times and calculates the line operation time of the day through image analysis of the monitoring operation interface of the operating equipment.
[0065] The Otsu method is an algorithm for determining the image binarization segmentation threshold, which assumes that the image pixels can be divided into two parts, background and target, according to the global threshold. Then, the optimal threshold for distinguishing the background from the image is obtained by calculating the following formula.
[0066] g=w0w1(u0-u1) 2 ;
[0067] Among them, w0 is the ratio of target points to the total image, w1 is the ratio of background points to the image, u0 is the average gray value of w0, and u1 is the average gray value of w1.
[0068] The present invention takes the energy consumption data of the office building sub-item power energy consumption monitoring platform as the research object, and proposes a daily energy consumption abnormality diagnosis and cause analysis method and a real-time energy consumption diagnosis method for sub-lines with different data characteristics through characteristic analysis of the sub-line energy consumption data of the office building sub-item power energy consumption monitoring platform. It can effectively combine the historical data of the power energy consumption line, use the similar day, sorting and short-term year-on-year comparison methods to analyze the causes of energy consumption abnormality of the power line, and provide corresponding abnormal causes.
[0069] The basic principles and main features of the present invention and the advantages of the present invention are shown and described above. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention.
[0070] Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the claims are included in the present invention.
[0071] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.
Claims
1. A statistically based method for diagnosing abnormal periodic building energy consumption, characterized in that: Including steps: Based on the historical power data in the power database, the power consumption of office buildings is classified to obtain multiple types of energy consumption lines: Based on the historical power consumption characteristics, a daily line energy consumption method and a daily line energy consumption benchmark are established for lines with different energy consumption; Based on the daily line energy consumption method and the daily line energy consumption benchmark, diagnose whether multiple types of energy consumption lines are abnormal; When abnormal energy consumption is diagnosed, the cause of abnormal power consumption in office buildings will be analyzed.
2. The statistically based method for diagnosing abnormal periodic building energy consumption according to claim 1, characterized in that: The classification of office building power energy consumption based on historical data of the power database and multiple types of energy consumption lines are obtained by judging the correlation between the time series of daily and hourly energy consumption in the power historical data and the corresponding influencing factors, and then classifying the office building power energy consumption based on the energy consumption line data characteristics to obtain multiple types of energy consumption lines.
3. The statistically based method for diagnosing abnormal periodic building energy consumption according to claim 1, characterized in that: The multiple types of energy consumption lines include: Class I, Class II, Class III, Class IV, and Class V; Among them, the data of Class I energy consumption lines are characterized by stable daily energy consumption and small discreteness, the data of Class II energy consumption lines are characterized by unstable daily energy consumption and stable hourly energy consumption, the data of Class III energy consumption lines are characterized by non-stationary daily energy consumption and strong correlation between hourly energy consumption and climate conditions, the data of Class IV energy consumption lines are characterized by non-stationary daily energy consumption and weak correlation with outdoor climate and stable hourly energy consumption during daily operation, and the data of Class V energy consumption lines are characterized by non-stationary daily energy consumption and weak correlation with outdoor climate and non-stationary hourly energy consumption during daily operation.
4. The statistically based method for diagnosing abnormal periodic building energy consumption according to claim 1, characterized in that: The daily line energy consumption method and daily line energy consumption benchmark for lines with different energy consumption include an optimal similar day algorithm and its energy consumption benchmark, a same operating condition sorting algorithm and its energy consumption benchmark, and a short-term comparison algorithm and its energy consumption benchmark.
5. The statistically based method for diagnosing abnormal periodic building energy consumption according to claim 4, characterized in that: The optimal similar day algorithm is used to diagnose whether multiple types of energy consumption lines are abnormal, including: Based on the test day, select N days of electricity historical data, and select historical data of the same type as the test day from the N days of electricity historical data; After normalizing the selected historical data, the Mahalanobis distance is calculated to calculate the distance between the data and the test day data. The data with a Mahalanobis distance less than the set threshold are sorted in descending order. Based on the sorted distance, the energy consumption data of m similar days are selected and the average value is taken. It is judged whether the average value is less than the abnormal threshold, and the energy consumption abnormality is diagnosed based on the judgment result.
6. The statistically based method for diagnosing abnormal periodic building energy consumption according to claim 4, characterized in that: The optimal similar day algorithm uses five parameters, namely day type, weather type, maximum temperature, minimum temperature and average temperature, as input parameters for determining the optimal similar day.
7. The statistically based method for diagnosing abnormal periodic building energy consumption according to claim 4, characterized in that: The same operating condition sorting algorithm is used to diagnose whether multiple types of energy consumption circuits are abnormal, including: Select the same operating conditions based on the day to be tested, and select N days of historical power data based on the same operating conditions; Arrange the data for the N days in ascending order according to energy consumption value; Determine whether the energy consumption value of the test day is higher than the sorting threshold in the ranking of the energy consumption value of the data of N days, and give a diagnosis result of whether the energy is abnormal or not based on the judgment result.
8. The statistically based method for diagnosing abnormal periodic building energy consumption according to claim 4, characterized in that: The short-term comparison algorithm is used to diagnose whether multiple types of energy consumption lines are abnormal. The data trend within the sliding T of the comparison is used to perform abnormal analysis, including: Where t is the month-on-month anomaly threshold, count_num is the month-on-month anomaly count value, T is the month-on-month sliding window, i is the energy consumption value of the comparison day, and nowvalue is the energy consumption value of the test day.
9. The statistically based method for diagnosing abnormal periodic building energy consumption according to claim 1, characterized in that: The analysis of the causes of abnormal power energy consumption in office buildings includes analyzing the causes of abnormal line energy consumption from the perspective of operating time using the Otsu algorithm, and analyzing the causes of abnormal line energy consumption from the perspective of hourly energy consumption levels using the absolute distance algorithm to obtain analysis results.
10. The statistically based method for diagnosing abnormal periodic building energy consumption according to claim 9, characterized in that: The Otsu algorithm automatically identifies the daily operation start and stop times and calculates the daily line operation time by analyzing the image of the operation interface of the operation equipment.
Citation Information
Patent Citations
Dynamic detection method for building energy consumption abnormity
CN112381181A
Building operation energy consumption abnormity identification method and system, and computer readable storage medium
CN112686286A
