CN109447107B - On-line detection method for daily energy consumption mode abnormality of air conditioner of office building based on information entropy - Google Patents

On-line detection method for daily energy consumption mode abnormality of air conditioner of office building based on information entropy Download PDF

Info

Publication number
CN109447107B
CN109447107B CN201811072397.XA CN201811072397A CN109447107B CN 109447107 B CN109447107 B CN 109447107B CN 201811072397 A CN201811072397 A CN 201811072397A CN 109447107 B CN109447107 B CN 109447107B
Authority
CN
China
Prior art keywords
daily
daily energy
air conditioner
energy
information entropy
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201811072397.XA
Other languages
Chinese (zh)
Other versions
CN109447107A (en
Inventor
周璇
字学辉
梁列全
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
South China University of Technology SCUT
Original Assignee
South China University of Technology SCUT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by South China University of Technology SCUT filed Critical South China University of Technology SCUT
Priority to CN201811072397.XA priority Critical patent/CN109447107B/en
Publication of CN109447107A publication Critical patent/CN109447107A/en
Application granted granted Critical
Publication of CN109447107B publication Critical patent/CN109447107B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Air Conditioning Control Device (AREA)

Abstract

The invention discloses an on-line detection method for abnormal daily energy mode of an air conditioner of an office building based on information entropy, which comprises the following steps: s1, establishing a normal daily energy consumption mode historical data set under different working conditions; s2, detecting the abnormal daily energy consumption mode of the building air conditioner on line; the method comprises the steps of firstly carrying out normal daily energy use mode extraction on historical energy use mode data of the building air conditioner, including working condition division, information entropy calculation of each daily energy use record under different working conditions, information entropy clustering, establishment of normal daily energy use modes under the working conditions and the like, then collecting daily energy use data of the air conditioner on line, judging the working condition to which the daily energy use data belong, calculating the information entropy of each daily energy use record under the working condition to which the daily energy use data belong, clustering, and judging the daily energy use data to be an abnormal energy use mode if the daily energy use records are singly clustered into one type. The invention effectively realizes the on-line detection of the abnormal daily energy consumption mode of the building air conditioner, and can be used for detecting the energy consumption of the air conditioner of a single building and also can be used for detecting the energy consumption of the air conditioner of a large building group.

Description

On-line detection method for daily energy consumption mode abnormality of air conditioner of office building based on information entropy
Technical Field
The invention relates to the research field of abnormal energy consumption detection of an air conditioner of an office building, in particular to an online detection method of an abnormal daily energy consumption mode of the air conditioner of the office building based on information entropy.
Background
In recent years, the problems of resource shortage, energy price rise, global warming and the like become serious, and the energy conservation and emission reduction work is urgent. At present, the energy consumption of buildings in China is about more than 33% of the total energy consumption of the whole country, the energy consumption of public buildings is huge, the energy consumption of air conditioners in hot summer and warm winter areas is more than 40% of the energy consumption of the public buildings, and the energy conservation of the air conditioners is imperative. The detection of the abnormal daily energy mode of the air conditioner is the premise of normal operation of the air conditioner and energy conservation of the air conditioner. Along with popularization and application of the fractional measurement of the energy consumption of the public buildings, a large amount of energy consumption data of the air conditioners are collected and stored, and a detection algorithm for an abnormal daily energy mode of the air conditioners of the public buildings is researched, so that the method has important significance for energy conservation of the public buildings. The detection of the abnormal daily energy mode of the air conditioner of the public building mainly aims to detect the abnormal daily energy mode caused by sensor faults, interference noise, communication interruption, abnormal operation faults of the air conditioner equipment, improper operation management and the like from energy consumption data, provide basis for energy-saving management and operation maintenance of the air conditioner of the building and achieve the aim of saving energy consumption. The energy consumption abnormity of the air conditioner is mainly divided into two types: single outlier anomalies due to noise interference or communication interruptions; an abnormal power consumption pattern due to a device operational failure or mismanagement of operation.
Different methods are needed to be adopted for detecting different types of abnormal daily energy modes; the single abnormal daily energy pattern data (outlier data) caused by noise interference or communication interruption is generally judged by adopting a 3 sigma principle, a graphical method, a threshold method, a cluster analysis method, an energy consumption prediction method and the like, and the method is relatively mature; however, for the daily energy consumption mode abnormality caused by equipment operation failure or improper operation management, most of the commonly used detection methods extract characteristic parameters of the daily energy consumption mode, such as maximum values, minimum values, root mean square and other conventional characteristic quantities, by calculating the statistical characteristics of the daily time-by-time energy consumption, convert high-dimensional data into low-dimensional data, and then judge the abnormal mode by using a threshold method. However, the setting of the threshold is difficult to judge abnormal data, the detection is easily missed when the threshold is large, and the erroneous judgment is easily caused when the threshold is too small.
Disclosure of Invention
The invention mainly aims to overcome the defects and shortcomings of the prior art, and provides an on-line detection method for an abnormal daily energy consumption mode of an air conditioner of an office building based on information entropy, which effectively realizes the on-line detection of the abnormal daily energy consumption mode of the air conditioner of the building, and can be used for detecting the energy consumption of the air conditioner of a single building and a large building group. The invention has the premise that the daily energy consumption data is cleaned and does not contain abnormal energy consumption data of the outlier mutation point.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention provides an on-line detection method for abnormal daily energy modes of an air conditioner of an office building based on information entropy, which comprises the following steps:
s1, establishing a normal daily energy consumption mode historical data set under different working conditions, specifically:
s1.1, obtaining historical data, wherein the historical data comprises outdoor dry bulb temperature t and building air conditioner energy consumption data Q;
s1.2, converting Data, namely converting the acquired historical Data into daily energy records of the air conditioners of the building, wherein the number of the daily energy records of the historical Data is counted by N, the date of the recorded day is Data, judging the attribute of the date corresponding to the daily energy record of each air conditioner, counting the corresponding actual starting time length of each air conditioner, and calculating the average value of the outdoor dry bulb temperature during actual starting;
s1.3, working condition division is carried out, the daily energy record of the building air conditioner is divided according to date attributes, actual starting time and the average value of the outdoor dry bulb temperature, and the discretization interval of the outdoor average dry bulb temperature is set as tcRecording the total number of different working conditions as M;
s1.4, calculating information entropy values recorded by daily energy of each air conditioner under the kth working condition;
s1.5, clustering information entropy values recorded by daily energy of each air conditioner under the same working condition by using a clustering algorithm, and clustering data into two types by using the clustering algorithm, namely normal and abnormal information entropy value clusters, wherein the information entropy value cluster with less data amount is considered as an abnormal information entropy value cluster, and the information entropy value is used as a characteristic parameter for reflecting the daily energy mode under the same working condition, so that the normal daily energy mode and the abnormal daily energy mode under the working condition are distinguished;
s1.6, outputting a clustering result, deleting an abnormal information entropy value cluster, establishing a normal daily use energy recording data set under the k working condition, namely a k normal daily use energy mode, and after deleting abnormal data, recording daily use energy under the working condition to obtain N 'in total'k(ii) a strip;
s1.7, repeating the processes of the steps S1.4 to S1.6 for all working conditions until the normal daily energy recording data sets under the M working conditions are established;
s2, detecting abnormal daily energy modes of the building on line, and specifically comprising the following steps:
s2.1, collecting energy consumption data of the building air conditioner on line;
s2.2, judging the working condition of the newly collected daily energy record of the air conditioner, wherein the newly collected daily energy record of the air conditioner is called a new daily energy record for short, namely judging the date attribute of the new daily energy record, counting the starting time, calculating the average value of the outdoor dry bulb temperature during starting, and judging the working condition of the new daily energy record according to the conditions;
s2.3, establishing a distance matrix by combining the new daily energy record and the historical daily energy record data under the same working condition, and outputting an information entropy value;
s2.4, clustering the information entropy values by using a clustering algorithm, and judging the number of objects in a cluster where the new daily performance record is located after clustering is finished;
s2.5, when the number of the objects in the cluster is 1, judging that the new daily energy consumption record is an abnormal energy consumption mode, and recording and alarming; when the number of the objects in the cluster is not 1, judging that the new daily energy record is a normal energy record mode, adding a normal daily energy record data set of the working condition of the new daily energy record to detect the abnormal daily energy record mode in the next detection period;
and S2.6, finishing the online detection of the abnormal daily energy mode.
As a preferred technical solution, step S1.2 is specifically as follows:
the ith record comprises date, date attribute, outdoor dry bulb temperature ti,jEnergy consumption Q of air conditioneri,jWherein j is delta, 2 delta, …, W delta, delta is energy consumption data acquisition step length, W is 24 h/delta, h is hour, the acquisition step length is 15 minutes, half an hour, 1 hour or 2 hours, and each data time span capable of being recorded is a natural day; the air conditioner is started up in the same day at the initial starting-up time niDelta, the step length of energy consumption data acquisition is delta, and the actual starting time is piDelta, calculating the average outdoor dry bulb temperature during the starting time of the air conditioner on the day
Figure GDA0003107647330000041
As a preferred technical solution, step S1.4 specifically is: constructing an input energy consumption matrix Q under the k working conditionkEstablishing a distance matrix Dk；
Assuming that the number of the energy-use records under the k-th working condition is NkThe step length of energy consumption data acquisition is delta;
Figure GDA0003107647330000042
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000043
the energy consumption of the ith day and the jth moment under the kth working condition is shown;
Figure GDA0003107647330000044
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000045
the Euclidean distance between the daily energy records of the ith and ith 'days under the k working conditions, i, i' is belonged to [1,2, …, Nk]；
DkThe calculation formula of (2):
Figure GDA0003107647330000046
when i ═ i',
Figure GDA0003107647330000047
calculating a distance contribution matrix:
Figure GDA0003107647330000048
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000049
under the k operating condition
Figure GDA0003107647330000051
To overall distance
Figure GDA0003107647330000052
Degree of contribution of
Figure GDA0003107647330000053
i,i′∈[1,2,…Nk]；
Calculating N under the k working conditionkThe formula for calculating the information entropy recorded by daily use is as follows:
Figure GDA0003107647330000054
i∈[1,2,…,Nk]wherein, CkIs a constant number, Ck＝1/ln(Nk)，NkRecording the number of the energy consumption data under the kth working condition;
outputting all daily recordable information entropy column vectors under the kth working condition:
Figure GDA0003107647330000055
as a preferred technical scheme, in step S1.5, the clustering algorithm can cluster similar data into one class, and cluster normal and abnormal information entropy value clusters by the clustering algorithm; the information entropy value is used as a characteristic parameter reflecting the daily performance mode, so that the normal daily performance mode and the abnormal daily performance mode are distinguished.
As a preferred technical solution, in step S2.1, daily energy record of the day is recorded as:
Figure GDA0003107647330000056
the date is recorded as Data (N + 1).
As a preferred technical scheme, in step S2.2, when the date attribute is judged, the data attribute may be divided into a Workday and a holiday Restday, the boot time length is counted, the average value of the outdoor dry bulb temperature of the boot time is calculated, and then the working condition of the new day energy-consumption record is judged according to the above conditions.
As a preferred technical solution, step S2.3 specifically is:
inputting an energy consumption matrix:
Figure GDA0003107647330000057
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000058
is the energy consumption at the j time of the ith day under the k working conditions, N'kRecording the data volume of the data set for normal daily use energy under the kth working condition;
outputting a distance matrix:
Figure GDA0003107647330000061
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000062
is the distance between the energy consumption data records of the ith day and the ith ' day under the k working condition, i, i ' belongs to [1,2, …, N 'k,N′k+1]。
As a preferable technical solution, the step S2.3 further includes the steps of:
according to the distance matrix, a distance contribution degree matrix is established, and the distance contribution degree calculation method is as follows:
Figure GDA0003107647330000063
under the k operating condition
Figure GDA0003107647330000064
To overall distance
Figure GDA0003107647330000065
Degree of contribution of
Figure GDA0003107647330000066
i,i′∈[1,2,…N′k+1]。
Calculating the information entropy of each piece of data according to the established distance contribution matrix, wherein the information entropy calculation method is the same as S1.5;
output information entropy column vector:
Figure GDA0003107647330000067
preferably, in step S2.4, if the number of objects in the cluster where the new daily energy recording is located is greater than 1, the new daily energy recording is in the normal daily energy mode, and if the number is equal to 1, the new daily energy recording is in the abnormal daily energy mode.
As a preferable technical solution, before the step S1, the method further includes the steps of:
the daily energy consumption data is cleaned and does not contain the outlier mutation point abnormal energy consumption data.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1. the invention provides an on-line detection method for an abnormal daily energy consumption mode of an air conditioner of an office building based on information entropy, aiming at identifying an unreasonable energy consumption mode instead of identifying a single energy consumption abnormal point;
2. according to the invention, the working conditions are divided according to the date attribute, the starting time of the air conditioner and the average temperature of the outdoor dry bulb in the starting time of the air conditioner, so that the interference of various influencing factors on the identification of the abnormal daily energy pattern is reduced, and the possibility of misjudgment is reduced;
3. the method comprises the steps of describing distribution characteristics of all-day energy consumption data of the air conditioner by using information entropy values, and converting high-dimensional data into 1-dimensional data; the method has the advantages that the influence of dimension disaster on the abnormal detection can be eliminated, and the detection precision is improved;
4. the method utilizes a clustering algorithm to detect abnormal daily energy consumption mode data; the method has the advantages that the abnormal daily energy pattern data are detected by utilizing a mode of judging the cluster size without setting a threshold value;
5. the invention can detect the abnormal daily energy consumption mode of the building air conditioner under various energy consumption working conditions on line;
6. the invention provides an on-line detection method for abnormal daily energy consumption mode of a central air conditioner of a building, which can be used for detecting the energy consumption of the air conditioner of a single building and can also be used for detecting the energy consumption of the air conditioner of a large-scale building group.
7. The model data of the invention is continuously optimized along with the updating of the on-line data, thereby greatly improving the detection precision.
Drawings
FIG. 1 is a flow chart of the detection method of the present invention.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but the present invention is not limited thereto.
Examples of the embodiments
As shown in fig. 1, the method for detecting abnormal daily energy pattern of air conditioner in office building based on information entropy in the embodiment includes the following steps:
s1, establishing a normal daily energy consumption mode historical data set under different working conditions, specifically;
s1.1, obtaining historical data, wherein the historical data comprises outdoor dry bulb temperature t and building air conditioner energy consumption data Q;
s1.2, converting Data, namely converting the acquired historical Data into daily energy records of the air conditioners of the building, wherein the number of the daily energy records of the historical Data is counted by N, the date of the recorded day is Data, judging the attribute of the date corresponding to the daily energy record of each air conditioner, counting the corresponding actual starting time length of each air conditioner, and calculating the average value of the outdoor dry bulb temperature during actual starting;
s1.3, working condition division is carried out, wherein the working condition division is carried out on daily energy records of the building air conditioner according to date attributes, actual starting time and an average value of outdoor dry bulb temperature, a discretization interval of the outdoor average dry bulb temperature is set to be 2 ℃, and the total number of different working conditions is recorded as M;
s1.4, calculating information entropy values recorded by daily energy of each air conditioner under the kth working condition;
s1.5, clustering information entropy values recorded by daily energy of each air conditioner under the same working condition by using a clustering algorithm, and clustering data into two types by using the clustering algorithm, namely normal and abnormal information entropy value clusters, wherein the information entropy value cluster with less data amount is considered as an abnormal information entropy value cluster, and the information entropy value is used as a characteristic parameter for reflecting the daily energy mode under the same working condition, so that the normal daily energy mode and the abnormal daily energy mode under the working condition are distinguished;
s1.6, outputting a clustering result, deleting an abnormal information entropy value cluster, establishing a normal daily use energy recording data set under the k working condition, namely a k normal daily use energy mode, and after deleting abnormal data, recording daily use energy under the working condition to obtain N 'in total'k(ii) a strip;
s1.7, repeating the processes of the steps S1.4 to S1.6 for all working conditions until the normal daily energy recording data sets under the M working conditions are established;
s2, detecting abnormal daily energy modes of the building on line, and specifically comprising the following steps:
s2.1, collecting energy consumption data of the building air conditioner on line;
s2.2, judging the working condition of the newly-collected daily energy record of the air conditioner, wherein the newly-collected daily energy record of the air conditioner is called a new daily energy record for short, namely judging the date attribute of the new daily energy record, counting the starting time, calculating the average value of the outdoor dry bulb temperature during starting, and judging the working condition of the newly-collected daily energy record of the air conditioner according to the conditions;
s2.3, establishing a distance matrix by combining the new daily energy record and the historical daily energy record data under the same working condition, and outputting an information entropy value;
s2.4, clustering the information entropy values by using a clustering algorithm, and judging the number of objects in a cluster where the new daily performance record is located after clustering is finished;
s2.5, when the number of the objects in the cluster is 1, judging that the new daily energy consumption record is an abnormal energy consumption mode, and recording and alarming; when the number of the objects in the cluster is not 1, judging that the new daily energy record is a normal energy record mode, adding a normal daily energy record data set of the working condition of the new daily energy record to detect the abnormal daily energy record mode in the next detection period;
and S2.6, finishing the online detection of the abnormal daily energy mode.
Step S1.2 is specifically as follows:
the ith record comprises date, date attribute, outdoor dry bulb temperature ti,jEnergy consumption Q of air conditioneri,jWherein j is 1,2, …,24, and the step size of energy consumption data acquisition is 1 hour; the air conditioner is started up in the same day at the initial starting-up time niΔ, actual boot time is piDelta, calculating the average outdoor dry bulb temperature during the starting time of the air conditioner on the day
Figure GDA0003107647330000091
Step S1.4 specifically includes: constructing an input energy consumption matrix Q under the k working conditionkEstablishing a distance matrix Dk；
Assuming that the number of the energy-use records under the k-th working condition is NkThe step length of energy consumption data acquisition is 1 hour;
Figure GDA0003107647330000092
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000093
the energy consumption of the ith day and the jth moment under the kth working condition is shown;
Figure GDA0003107647330000094
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000095
the Euclidean distance between the daily energy records of the ith and ith 'days under the k working conditions, i, i' is belonged to [1,2, …, Nk]；
DkThe calculation formula of (2):
Figure GDA0003107647330000101
when i ═ i',
Figure GDA0003107647330000102
calculating a distance contribution matrix:
Figure GDA0003107647330000103
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000104
under the k operating condition
Figure GDA0003107647330000105
To overall distance
Figure GDA0003107647330000106
Degree of contribution of
Figure GDA0003107647330000107
i,i′∈[1,2,…Nk]。
Calculating N under the k working conditionkThe formula for calculating the information entropy recorded by daily use is as follows:
Figure GDA0003107647330000108
i∈[1,2,…,Nk]wherein, CkIs a constant number, Ck＝1/ln(Nk)，NkRecording the number of the energy consumption data under the kth working condition;
outputting all daily recordable information entropy column vectors under the kth working condition:
Figure GDA0003107647330000109
in the step S1.5, the hierarchical clustering algorithm can cluster similar data into one class, and normal and abnormal information entropy value clusters are clustered through the hierarchical clustering algorithm; the information entropy value is used as a characteristic parameter reflecting the daily performance mode, so that the normal daily performance mode and the abnormal daily performance mode are distinguished.
Clustering the information entropy values recorded by daily energy by using a hierarchical clustering algorithm, wherein the specific method comprises the following steps:
(1) dividing elements of each information entropy vector into a cluster, NkThe elements of the information entropy value vector correspond to NkThe number of the clusters is small,
Figure GDA00031076473300001010
calculating Euclidean distance values between clusters, and establishing a distance value row vector D 'of information entropy'k(i,i′)：
Figure GDA0003107647330000111
Wherein i, i' e [1, …, (N)k-1)*Nk]When i is equal to i', the reaction proceeds,
Figure GDA0003107647330000112
and D'k(i,i′)＝D′k(i′,i)
(2) Find D'kMinimum value Min [ D ' of non-zero (i, i ') vector elements 'k(i,i′)]The minimum value corresponds to the original two clusters
Figure GDA0003107647330000113
And
Figure GDA0003107647330000114
(3) number of original clusters is NkThe number of the main components is one,
Figure GDA0003107647330000115
merging two information entropy clusters with the nearest distance in the previous step into a new cluster
Figure GDA0003107647330000116
The number of clusters becomes Nk-1 of the number of the cells to be tested,
Figure GDA0003107647330000117
(4) calculating merged Nk-distance between 1 cluster, using Average Linkage calculation method:
Figure GDA0003107647330000118
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000119
to be a new cluster
Figure GDA00031076473300001110
Two objects in (1)
Figure GDA00031076473300001111
And
Figure GDA00031076473300001112
j∈[1,2,…Nk-2]。
(5) repeating the calculation of (1) to (4) until the number of the clustering clusters is 2, wherein the distance between any two merging clusters containing a plurality of objects in the clustering process is shown as the following formula:
Figure GDA00031076473300001113
in the formula
Figure GDA00031076473300001114
And
Figure GDA00031076473300001115
new clusters containing n and m objects respectively,
Figure GDA00031076473300001116
is composed of
Figure GDA00031076473300001117
F-th in a cluster1The object is a person to be examined,
Figure GDA00031076473300001118
is composed of
Figure GDA00031076473300001119
F-th in a cluster2An object;
in step S2.1, daily performance records of the same day are recorded as:
Figure GDA00031076473300001120
the date is recorded as Data (N + 1).
Step 2.2, when judging the date attribute, the data attribute can be divided into working day Workday and rest day Restday, the starting time length is counted, the average value of the outdoor dry bulb temperature of the starting time is calculated, and then the newly collected daily energy of the air conditioner is judged according to the conditions to record the working condition of the newly collected daily energy of the air conditioner;
Figure GDA00031076473300001211
step S2.3 specifically includes:
inputting an energy consumption matrix:
Figure GDA0003107647330000123
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000124
is the ith day under the k working conditionEnergy consumption at time j, N'kRecording the data volume of the data set for normal daily use energy under the kth working condition;
outputting a distance matrix:
Figure GDA0003107647330000125
in the formula (I), the compound is shown in the specification,
Figure GDA0003107647330000126
is the distance between the energy consumption data records of the ith day and the ith ' day under the k working condition, i, i ' belongs to [1,2, …, N 'k,N′k+1]。
In step S2.3, the following steps are also included:
according to the distance matrix, a distance contribution degree matrix is established, and the distance contribution degree calculation method is as follows:
Figure GDA0003107647330000127
under the k operating condition
Figure GDA0003107647330000128
To overall distance
Figure GDA0003107647330000129
Degree of contribution of
Figure GDA00031076473300001210
i,i′∈[1,2,…N′k+1]。
Calculating the information entropy of each piece of data according to the established distance contribution matrix, wherein the information entropy calculation method is the same as S1.5;
output information entropy column vector:
Figure GDA0003107647330000131
in step S2.4, the cluster where the new daily energy record is located is recorded as a, if the number of objects in the cluster is greater than 1, the new daily energy record is the normal energy use mode, and if the number of objects in the cluster is equal to 1, the new daily energy record is the abnormal daily energy use mode.
Figure GDA0003107647330000132
Before step S1, the method further includes the following steps:
the daily energy consumption data is cleaned and does not contain the outlier mutation point abnormal energy consumption data.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.

Claims (10)

1. An on-line detection method for an abnormal daily energy mode of an air conditioner of an office building based on information entropy is characterized by comprising the following steps:
s1, establishing a normal daily energy consumption mode historical data set under different working conditions, specifically:
s1.1, obtaining historical data, wherein the historical data comprises outdoor dry bulb temperature t and building air conditioner energy consumption data Q;
s1.2, converting Data, namely converting the acquired historical Data into daily energy records of the air conditioners of the building, wherein the number of the daily energy records of the historical Data is counted by N, the date of the recorded day is Data, judging the attribute of the date corresponding to the daily energy record of each air conditioner, counting the corresponding actual starting time length of each air conditioner, and calculating the average value of the outdoor dry bulb temperature during actual starting;
s1.3, working condition division, namely dividing the daily energy record of the building air conditioner according to date attributes, actual starting time and the average value of the outdoor dry bulb temperature during starting, and setting the discretization interval of the outdoor average dry bulb temperature as tcRecording the total number of different working conditions as M;
s1.4, calculating information entropy values recorded by daily energy of each air conditioner under the kth working condition;
s1.5, clustering information entropy values recorded by daily energy of each air conditioner under the same working condition by using a clustering algorithm, and clustering data into two types by using the clustering algorithm, namely normal and abnormal information entropy value clusters, wherein the information entropy value cluster with less data amount is considered as an abnormal information entropy value cluster, and the information entropy value is used as a characteristic parameter for reflecting the daily energy mode under the same working condition, so that the normal daily energy mode and the abnormal daily energy mode under the working condition are distinguished;
s1.6, outputting a clustering result, deleting an abnormal information entropy value cluster, establishing a normal daily use energy recording data set under the k working condition, namely a k normal daily use energy mode, and after deleting abnormal data, recording daily use energy under the working condition to obtain N 'in total'k(ii) a strip;
s1.7, repeating the processes of the steps S1.4 to S1.6 for all working conditions until the normal daily energy recording data sets under the M working conditions are established;
s2, detecting abnormal daily energy modes of the building on line, and specifically comprising the following steps:
s2.1, collecting energy consumption data of the building air conditioner on line;
s2.2, judging the working condition of the newly collected daily energy record of the air conditioner, wherein the newly collected daily energy record of the air conditioner is called a new daily energy record for short, namely judging the date attribute of the new daily energy record, counting the starting time, calculating the average value of the outdoor dry bulb temperature during starting, and judging the working condition of the new daily energy record according to the conditions;
s2.3, establishing a distance matrix by combining the new daily energy record and the historical daily energy record data under the same working condition, and outputting an information entropy value;
s2.4, clustering the information entropy values by using a clustering algorithm, and judging the number of objects in a cluster where the new daily performance record is located after clustering is finished;
s2.5, when the number of the objects in the cluster is 1, judging that the new daily energy consumption record is an abnormal energy consumption mode, and recording and alarming; when the number of the objects in the cluster is not 1, judging that the new daily energy record is a normal energy record mode, adding a normal daily energy record data set of the energy record working condition to which the new daily energy record belongs, and detecting the abnormal daily energy record mode in the next detection period;
and S2.6, finishing the online detection of the abnormal daily energy mode.
2. The on-line detection method for the abnormal daily energy pattern of the air conditioner of the office building based on the information entropy as claimed in claim 1, wherein the step S1.2 is as follows:
the ith record comprises date, date attribute, outdoor dry bulb temperature ti,jEnergy consumption Q of air conditioneri,jWherein j is delta, 2 delta, …, W delta, delta is energy consumption data acquisition step length, W is 24 h/delta, h is hour, the acquisition step length is 15 minutes, half an hour, 1 hour or 2 hours, and each data time span capable of being recorded is a natural day; the air conditioner is started up in the same day at the initial starting-up time niDelta, the step length of energy consumption data acquisition is delta, and the actual starting time is piDelta, calculating the average outdoor dry bulb temperature during the starting time of the air conditioner on the day
Figure FDA0003107647320000021
3. The on-line detection method for the abnormal daily energy pattern of the air conditioner of the office building based on the information entropy as claimed in claim 1, wherein the step S1.4 is specifically as follows: constructing an input energy consumption matrix Q under the k working conditionkEstablishing a distance matrix Dk；
Assuming that the number of the energy-use records under the k-th working condition is NkThe step length of energy consumption data acquisition is delta;
Figure FDA0003107647320000022
in the formula (I), the compound is shown in the specification,
Figure FDA0003107647320000023
the energy consumption of the ith day and the jth moment under the kth working condition is shown;
Figure FDA0003107647320000031
in the formula (I), the compound is shown in the specification,
Figure FDA0003107647320000032
is as followsThe Euclidean distance between the daily use energy records of the ith day and the ith 'day under k working conditions, i, i' belongs to [1,2, …, Nk]；
DkThe calculation formula of (2):
Figure FDA0003107647320000033
when i ═ i',
Figure FDA0003107647320000034
calculating a distance contribution matrix:
Figure FDA0003107647320000035
in the formula (I), the compound is shown in the specification,
Figure FDA0003107647320000036
under the k operating condition
Figure FDA0003107647320000037
To overall distance
Figure FDA0003107647320000038
Degree of contribution of
Figure FDA0003107647320000039
Calculating N under the k working conditionkThe formula for calculating the information entropy recorded by daily use is as follows:
Figure FDA00031076473200000310
wherein, CkIs a constant number, Ck＝1/ln(Nk)，NkRecording the number of the energy consumption data under the kth working condition;
outputting all daily recordable information entropy column vectors under the kth working condition:
Figure FDA00031076473200000311
4. the information entropy-based on-line detection method for the abnormal daily energy pattern of the air conditioner of the office building is characterized in that in the step S1.5, the clustering algorithm can cluster similar data into one class, and normal and abnormal information entropy value clusters are clustered through the clustering algorithm; the information entropy value is used as a characteristic parameter reflecting the daily performance mode, so that the normal daily performance mode and the abnormal daily performance mode are distinguished.
5. The on-line detection method for the abnormal daily energy pattern of the air conditioner of the office building based on the information entropy as claimed in claim 1, wherein in step S2.1, daily energy records of the day are recorded as:
Figure FDA00031076473200000312
the date is recorded as Data (N + 1).
6. The on-line detection method for the abnormal daily energy mode of the air conditioner of the office building based on the information entropy as claimed in claim 1, wherein in the step S2.2, when the date attribute is judged, the date attribute can be divided into a working day Workday and a rest day Restday, the starting time is counted, the average value of the outdoor dry bulb temperature of the starting time is calculated, and then the working condition of the new daily energy record is judged according to the conditions.
7. The on-line detection method for the abnormal daily energy pattern of the air conditioner of the office building based on the information entropy as claimed in claim 1, wherein the step S2.3 is specifically as follows:
inputting an energy consumption matrix:
Figure FDA0003107647320000041
in the formula (I), the compound is shown in the specification,
Figure FDA0003107647320000042
is the energy consumption at the j time of the ith day under the k working conditions, N'kRecording the data volume of the data set for normal daily use energy under the kth working condition;
output distanceMatrix:
Figure FDA0003107647320000043
in the formula (I), the compound is shown in the specification,
Figure FDA0003107647320000044
is the distance between the energy consumption data records of the ith day and the ith ' day under the k working condition, i, i ' belongs to [1,2, …, N 'k,N′k+1]。
8. The on-line detection method for the abnormal daily energy pattern of the air conditioner of the office building based on the information entropy as claimed in claim 1, wherein in the step S2.3, the method further comprises the following steps:
according to the distance matrix, a distance contribution degree matrix is established, and the distance contribution degree calculation method is as follows:
Figure FDA0003107647320000045
under the k operating condition
Figure FDA0003107647320000046
To overall distance
Figure FDA0003107647320000047
Degree of contribution of
Figure FDA0003107647320000048
Figure FDA0003107647320000049
Calculating the information entropy of each piece of data according to the established distance contribution matrix, wherein the information entropy calculation method is the same as S1.5;
output information entropy column vector:
Figure DEST_PATH_IMAGE002
。
9. the on-line detection method for the abnormal daily energy mode of the air conditioner of the office building based on the information entropy of claim 1, wherein in the step S2.4, if the number of the objects in the cluster where the new daily energy record is located is more than 1, the new daily energy record is the normal energy mode, and if the number is equal to 1, the new daily energy record is the abnormal daily energy mode.
10. The on-line detection method for the abnormal daily energy pattern of the air conditioner of the office building based on the information entropy of claim 1, wherein before the step S1, the method further comprises the following steps:
the daily energy consumption data is cleaned and does not contain the outlier mutation point abnormal energy consumption data.
CN201811072397.XA 2018-09-14 2018-09-14 On-line detection method for daily energy consumption mode abnormality of air conditioner of office building based on information entropy Active CN109447107B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811072397.XA CN109447107B (en) 2018-09-14 2018-09-14 On-line detection method for daily energy consumption mode abnormality of air conditioner of office building based on information entropy

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811072397.XA CN109447107B (en) 2018-09-14 2018-09-14 On-line detection method for daily energy consumption mode abnormality of air conditioner of office building based on information entropy

Publications (2)

Publication Number Publication Date
CN109447107A CN109447107A (en) 2019-03-08
CN109447107B true CN109447107B (en) 2021-08-10

Family

ID=65532717

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811072397.XA Active CN109447107B (en) 2018-09-14 2018-09-14 On-line detection method for daily energy consumption mode abnormality of air conditioner of office building based on information entropy

Country Status (1)

Country Link
CN (1) CN109447107B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110210525B (en) * 2019-05-14 2023-07-04 天津大学 K-Means clustering-based design day meteorological element gradual change feature extraction method
CN110648250A (en) * 2019-09-25 2020-01-03 珠海格力电器股份有限公司 Energy consumption distribution calculation method and device and data server
CN112215474A (en) * 2020-09-18 2021-01-12 上海市建筑科学研究院有限公司 Running characteristic model for water chilling unit
CN112559501B (en) * 2020-12-01 2021-11-05 大连理工大学 Method for splitting power consumption of data-driven building air conditioner and lighting socket
CN113883698B (en) * 2021-09-23 2022-11-29 深圳达实智能股份有限公司 Air conditioning system refrigeration station starting strategy optimization method and system and electronic equipment
CN117633573B (en) * 2024-01-26 2024-04-16 南京群顶科技股份有限公司 Electric quantity data anomaly detection method and system based on air conditioner operation working condition

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104517041A (en) * 2015-01-26 2015-04-15 北京化工大学 Maximum information entropy-based chemical data stream real-time abnormity detection method
CN106910144A (en) * 2017-01-11 2017-06-30 华南理工大学 Based on timesharing carve it is actual with can coefficient heavy construction by when energy consumption on-line prediction method
CN107590567A (en) * 2017-09-13 2018-01-16 南京航空航天大学 A kind of Recognition with Recurrent Neural Network short-term load forecasting method based on comentropy cluster and notice mechanism
CN107633444A (en) * 2017-08-29 2018-01-26 南京理工大学紫金学院 Commending system noise filtering methods based on comentropy and fuzzy C-means clustering
CN107944464A (en) * 2017-10-12 2018-04-20 华南理工大学 A kind of office building by when energy consumption abnormal data online recognition and complementing method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2873199A1 (en) * 2012-09-03 2015-05-20 Siemens Aktiengesellschaft Interference-immune transmission of data telegrams in a communications network

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104517041A (en) * 2015-01-26 2015-04-15 北京化工大学 Maximum information entropy-based chemical data stream real-time abnormity detection method
CN106910144A (en) * 2017-01-11 2017-06-30 华南理工大学 Based on timesharing carve it is actual with can coefficient heavy construction by when energy consumption on-line prediction method
CN107633444A (en) * 2017-08-29 2018-01-26 南京理工大学紫金学院 Commending system noise filtering methods based on comentropy and fuzzy C-means clustering
CN107590567A (en) * 2017-09-13 2018-01-16 南京航空航天大学 A kind of Recognition with Recurrent Neural Network short-term load forecasting method based on comentropy cluster and notice mechanism
CN107944464A (en) * 2017-10-12 2018-04-20 华南理工大学 A kind of office building by when energy consumption abnormal data online recognition and complementing method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
Sampling for building energy consumption with fuzzy theory;Jinghong Qin.et al;《Energy and Buildings》;20170918;第78-84页 *

Also Published As

Publication number Publication date
CN109447107A (en) 2019-03-08

Similar Documents

Publication Publication Date Title
CN109447107B (en) On-line detection method for daily energy consumption mode abnormality of air conditioner of office building based on information entropy
CN109858522A (en) A kind of management line loss abnormality recognition method based on data mining
CN110264107B (en) Large data technology-based abnormal diagnosis method for line loss rate of transformer area
CN107944464A (en) A kind of office building by when energy consumption abnormal data online recognition and complementing method
CN111486552B (en) Method for identifying water supply temperature strategy of chilled water of air conditioner based on subentry metering data
CN103631681A (en) Method for online restoring abnormal data of wind power plant
CN105654238B (en) Data management method and device for public institution energy consumption monitoring platform
WO2021120829A1 (en) Method for diagnosing energy consumption of lighting socket branches of buildings during non-operational periods
CN106845748A (en) A kind of INTELLIGENT IDENTIFICATION method of low-voltage collecting meter reading system data deviation reason
CN116796403A (en) Building energy saving method based on comprehensive energy consumption prediction of commercial building
CN116148753A (en) Intelligent electric energy meter operation error monitoring system
CN106651093A (en) Smart re-check management and control method used for low voltage set copy system
CN114113773A (en) Non-invasive load monitoring method based on zero sample learning
CN116385210B (en) Power supply energy consumption monitoring system based on Internet of things
CN117559443A (en) Ordered power utilization control method for large industrial user cluster under peak load
CN115733258A (en) Control method of all-indoor intelligent substation system based on Internet of things technology
CN112113316B (en) Method for extracting air conditioner load
CN115392663A (en) Data acquisition and processing method based on big data
CN114048200A (en) User electricity consumption behavior analysis method considering missing data completion
CN109884426B (en) Method for obtaining environmental control system cooling tower energy efficiency evolution characteristic along with operation age
CN113128756A (en) User behavior analysis algorithm for water, electricity, gas and heat system
CN112381130A (en) Cluster analysis-based power distribution room multivariate data anomaly detection method
CN111814829A (en) Power generation abnormity identification method and system for photovoltaic power station
Li et al. [Retracted] Application of Internet of Things Based on Big Data Ecosystem in Factory Energy Consumption Analysis Model
CN109270898B (en) Building energy consumption data acquisition unit with data quality diagnosis and restoration functions

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant