CN109035067A - Building energy consumption processing method and processing device based on RF and ARMA algorithm - Google Patents
Building energy consumption processing method and processing device based on RF and ARMA algorithm Download PDFInfo
- Publication number
- CN109035067A CN109035067A CN201810799398.8A CN201810799398A CN109035067A CN 109035067 A CN109035067 A CN 109035067A CN 201810799398 A CN201810799398 A CN 201810799398A CN 109035067 A CN109035067 A CN 109035067A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- model
- prediction model
- consumption prediction
- value
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/08—Construction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The invention discloses a kind of building energy consumption processing method and processing devices based on RF and ARMA algorithm.This method includes acquisition data source and obtains default engineering characteristics after passing through pretreatment；Energy consumption prediction model is established according to the default engineering characteristics；And it assesses the energy consumption prediction model and the energy consumption prediction model is corrected according to assessment result.The present invention solves shortage architecture, configurable, adaptive prediction model to meet the technical issues of the needs of actual prediction system complex is changeable, can with equipment using the variation of the factors such as, personal information and climate condition and dynamically adapting is reduced with improving energy consumption precision of prediction because of false alarm rate caused by temporarily working overtime/have a holiday or vacation.
Description
Technical field
This application involves the analysis of building trade data and excavation applications, are calculated in particular to one kind based on RF and ARMA
The building energy consumption processing method and processing device of method.
Background technique
As China's science and technology and economic are constantly progressive, intelligent building has also obtained rapid development and has carved machine, large-scale Super High
Public building springs up like bamboo shoots after a spring rain, and construction area exponential type increases, and the level of IT application is also constantly promoted.Meanwhile the use of people
Demand is also growing, and energy for building intensity shows the situation that increases substantially.According to the annual developmental research report of Chinese architecture energy conservation
2017. statistics are accused, ratio of the building energy consumption in China in social total energy consumption is rapidly increased to existing by the 10% of late nineteen seventies
30%.As an important indicator for measuring building energy consumption level, the amplification of building electricity consumption in the past decade can
Up to 10%~13%.Currently, China's building energy conservation level lags far behind developed country, and energy for building presence is very serious
Wasting phenomenon.Therefore, building energy utilization rate how is improved, reduction energy for building waste has become and implements China's sustainable development
In exhibition strategy the problem of urgent need to resolve.
It can obtain about 20% additional energy-saving potential Waide P using advanced control strategy, 2013.It sufficiently excavates and builds
The inherent value of data is built, fault diagnosis, control optimization and PREDICTIVE CONTROL etc. can be carried out.According to investigations, about 20% building
Energy consumption is that Katipamula S is caused due to equipment fault, sensor failure, control failure and unreasonable operation, 2005.Therefore,
The prediction of energy consumption and abnormal alarm on the one hand can allow manager learn in advance with can trend, formulate it is optimal with can strategy such as
It avoids the peak hour and uses energy；On the other hand energy consumption exception can be found in time, and personnel do not accommodate energy consumption wave caused by reducing because of equipment fault etc.
Take.
Prediction technique about building energy consumption emerges in multitude, and relates generally to decision tree, BP neural network algorithm, time series
Analysis, support vector machines and linear regression etc..However, existing energy consumption prediction technique often rests on theoretical research level, in reality
Rarely have application in the information system of border, tracing it to its cause mainly has: lacking architecture further investigation about characteristic factor selection；Prediction
For algorithm since model selection and parameter configuration are complicated, result precision is difficult to online evaluation and correction；Prediction model often relies on
In the active operation mode that date and hour divides, for temporarily work overtime or temporarily go out etc. energy consumption caused by scenes increase or
Reduction can generate energy consumption exception false-alarm.
Meet actual prediction system for shortage architecture in the related technology, configurable, adaptive prediction model
The problem of demand complicated and changeable of uniting, currently no effective solution has been proposed.
Summary of the invention
The main purpose of the application is to provide a kind of building energy consumption processing method and processing device based on RF and ARMA algorithm,
Lack architecture, configurable, adaptive prediction model with solution to meet the needs of actual prediction system complex is changeable
The problem of.
To achieve the goals above, according to the one aspect of the application, a kind of building based on RF and ARMA algorithm is provided
Build energy consumption processing method.
The data processing method for building energy consumption according to the application includes: acquisition data source and obtains after passing through pretreatment
To default engineering characteristics；Energy consumption prediction model is established according to the default engineering characteristics；And the assessment energy consumption prediction model
And the energy consumption prediction model is corrected according to assessment result.
Further, it assesses the energy consumption prediction model and is gone back after correcting the energy consumption prediction model according to assessment result
It include: the difference for analyzing energy consumption predicted value and energy consumption actual measurement value；If difference meets switching condition, energy consumption operation mould is matched
Formula；And if difference satisfaction is unsatisfactory for switching condition, determine exception.
Further, establishing energy consumption prediction model according to the default engineering characteristics includes: to return to calculate using random forest
Method is analyzed and establishes the energy consumption prediction model.
Further, it assesses the energy consumption prediction model and the energy consumption prediction model is corrected according to assessment result and include:
The energy consumption prediction model is corrected as benchmark model using autoregression sliding mean value model in model training.
Further, it assesses the energy consumption prediction model and the energy consumption prediction model is corrected according to assessment result and include:
Assess and switch the selection energy consumption prediction model as benchmark model using autoregression sliding mean value model in model application,
Wherein, the energy consumption prediction model includes operating mode and non-operating mode.
Further, it assesses the energy consumption prediction model and is gone back after correcting the energy consumption prediction model according to assessment result
It include: the processing result execution abnormal alarm operation to Exception Model or abnormal energy consumption is determined as.
To achieve the goals above, it according to the another aspect of the application, provides at a kind of data for building energy consumption
Manage device.
It include: acquisition module according to the data processing equipment for building energy consumption of the application, for acquiring data source simultaneously
By obtaining default engineering characteristics after pretreatment；Model building module, it is pre- for establishing energy consumption according to the default engineering characteristics
Survey model；And model evaluation module, it is pre- to assess the energy consumption prediction model and correct the energy consumption according to assessment result
Survey model.
Further, device further include: analysis module, the analysis module include: analytical unit, analyze energy consumption predicted value
With the difference of energy consumption actual measurement value；Matching unit, for matching energy consumption operational mode when difference meets switching condition；And it is different
Constant element, for when difference satisfaction is unsatisfactory for switching condition, then determining exception.
Further, the model building module includes: model foundation unit, and the model evaluation module includes: correction
Unit, assessment unit, the model foundation unit, for the energy consumption prediction to be analyzed and established using random forest regression algorithm
Model；The correction unit, for correcting the energy as benchmark model using autoregression sliding mean value model in model training
Consume prediction model；And the assessment unit, for sliding mean value model as benchmark mould using autoregression in model application
Type is assessed and switches the selection energy consumption prediction model, wherein the energy consumption prediction model includes: operating mode and inoperative mould
Formula.
Further, described device further include: alarm module, the alarm module include: abnormal alarm unit, for pair
It is determined as that the processing result of Exception Model or abnormal energy consumption executes abnormal alarm operation.
In the embodiment of the present application, by the way of acquisition data source and by obtaining default engineering characteristics after pretreatment,
Energy consumption prediction model is established according to the default engineering characteristics, by assessing the energy consumption prediction model and according to assessment result school
The just described energy consumption prediction model, reached used with equipment, the factors variation such as personal information and climate condition and dynamically adapting with
Energy consumption precision of prediction is improved, the purpose because of false alarm rate caused by temporarily working overtime/having a holiday or vacation is reduced, is configurable to realize foundation
And the adaptive technical effect for architectural energy consumption prediction model, and then solve lack architecture, it is can configure, adaptive
The prediction model answered meets the technical issues of the needs of actual prediction system complex is changeable.
Detailed description of the invention
The attached drawing constituted part of this application is used to provide further understanding of the present application, so that the application's is other
Feature, objects and advantages become more apparent upon.The illustrative examples attached drawing and its explanation of the application is for explaining the application, not
Constitute the improper restriction to the application.In the accompanying drawings:
Fig. 1 is the building energy consumption processing method schematic diagram according to the application first embodiment；
Fig. 2 is the building energy consumption processing method schematic diagram according to the application second embodiment；
Fig. 3 is the building energy consumption processing unit schematic diagram according to the application first embodiment；
Fig. 4 is the building energy consumption processing unit schematic diagram according to the application second embodiment；
Fig. 5 is for the core algorithm flow diagram in the application；
Fig. 6 is the architecture workflow diagrams for the application；
Fig. 7 is normal (operating mode) schematic diagram of one embodiment energy consumption of the application；
Fig. 8 is normal (non-operating mode) schematic diagram of one embodiment energy consumption of the application
Fig. 9 is abnormal (interim overtime work) schematic diagram of one embodiment energy consumption of the application；And
Figure 10 is abnormal (interim outgoing) schematic diagram of one embodiment energy consumption of the application.
Specific embodiment
In order to make those skilled in the art more fully understand application scheme, below in conjunction in the embodiment of the present application
Attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is only
The embodiment of the application a part, instead of all the embodiments.Based on the embodiment in the application, ordinary skill people
Member's every other embodiment obtained without making creative work, all should belong to the model of the application protection
It encloses.
It should be noted that the description and claims of this application and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that using in this way
Data be interchangeable under appropriate circumstances, so as to embodiments herein described herein.In addition, term " includes " and " tool
Have " and their any deformation, it is intended that cover it is non-exclusive include, for example, containing a series of steps or units
Process, method, system, product or equipment those of are not necessarily limited to be clearly listed step or unit, but may include without clear
Other step or units listing to Chu or intrinsic for these process, methods, product or equipment.
In this application, term " on ", "lower", "left", "right", "front", "rear", "top", "bottom", "inner", "outside",
" in ", "vertical", "horizontal", " transverse direction ", the orientation or positional relationship of the instructions such as " longitudinal direction " be orientation based on the figure or
Positional relationship.These terms are not intended to limit indicated dress primarily to better describe the application and embodiment
Set, element or component must have particular orientation, or constructed and operated with particular orientation.
Also, above-mentioned part term is other than it can be used to indicate that orientation or positional relationship, it is also possible to for indicating it
His meaning, such as term " on " also are likely used for indicating certain relations of dependence or connection relationship in some cases.For ability
For the those of ordinary skill of domain, the concrete meaning of these terms in this application can be understood as the case may be.
In addition, term " installation ", " setting ", " being equipped with ", " connection ", " connected ", " socket " shall be understood in a broad sense.For example,
It may be a fixed connection, be detachably connected or monolithic construction；It can be mechanical connection, or electrical connection；It can be direct phase
It even, or indirectly connected through an intermediary, or is two connections internal between device, element or component.
For those of ordinary skills, the concrete meaning of above-mentioned term in this application can be understood as the case may be.
The data processing method for building energy consumption in the application is returned based on random forest and the building energy consumption of ARMA
Prediction and abnormal alarm.Building energy consumption real-time prediction model is gone out using data random forests algorithm dynamic self-identifying, is returned using oneself
Return sliding mean value (ARMA) model as benchmark model and correct and assess, and according to energy consumption predicted value, measured value and a reference value
Deviation carries out abnormal alarm setting.
Fully consider first in this application building energy consumption feature used with equipment, personal information and climate condition etc. because
Element variation, the pivot factor (number, state of weather and outdoor temperature humidity) for influencing energy consumption is determined using signature analysis, wherein number
It can be characterized with festivals or holidays and operation time information indirect.Then data mining (random forest) regression algorithm is introduced to go through energy consumption
History data and corresponding pivot influence factor data carry out model training, learn energy consumption operational mode out and predict 24 hours following
Power consumption values.A reference value is predicted using autoregression sliding mean value as energy consumption simultaneously, is introduced power consumption mode handover mechanism, is finally based on
The difference of energy consumption predicted value, measured value and a reference value sets abnormal alarm threshold value.The characteristic of this data-driven method exists
In by identifying that the energy consumption prediction model established of data inherent law can be with dynamic self-adapting each influence factor (personnel, equipment
Configuration, season and weather etc.) variation, independent of the engineering experience of industry specialists and the concrete configuration of building；For
It practises model and is difficult to the problem of assessing and correcting, propose the appraisal procedure using sliding window historical data mean value as benchmark, drop
The interference to model accuracy is arranged in low algorithm parameter；Meanwhile introducing energy consumption operational mode handover mechanism facilitates system and moves
State identifies operating mode and non-operating mode, overcomes because of interim overtime work or interim outgoing bring energy consumption exception false-alarm problem.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.The application is described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
As shown in Figure 1, this method includes the following steps, namely S102 to step S106:
Step S102 acquires data source and by obtaining default engineering characteristics after pretreatment；
Acquisition data source refers to acquisition subitem energy consumption and related data, the acquisition modes of data source include sensor detection,
Database is read and website crawls, and involved initial data includes that energy consumption adds up reading, date and time stamp, weather conditions index
And temperature and humidity value, wherein the accumulative reading of energy consumption can periodically read directly from ammeter and save (containing date and time stamp), it is vaporous
Condition is read from the public data of local meteorological observatory website, and outdoor temperature humidity value is measured in real time using sensor and teletransmission is protected
It deposits.
Data prediction includes: singular value is rejected, missing values are filled up, normalized, format is converted and time series data is converted etc.,
Wherein singular value rejects the normal range (NR) for needing empirically determined data；Missing values are filled up using moving average method, i.e. dt=
(dt+1+dt-1)/2,dtFor t-th of data；Normalization is then that dimension disunity bring is avoided to influence, can be using maximum-most
Small value method, i.e. Id=(dmax-d)/(dmax-dmin)；Format conversion is then the format or type for converting the data into needs, such as will
The power consumption values of the available unit interval of energy consumption gauge outfit value difference point；Time series data conversion is the equivalent for taking certain unit interval,
Specific formula are as follows:E (T) is T time section energy consumption equivalent value, etIt is adopted for the t times energy consumption in T time section
Sample value, αtFor weight factor, default takes 1.
Specifically, acquisition subitem energy consumption and related data, line number of going forward side by side Data preprocess, including singular value rejecting, missing values
Fill up, normalize and time series data conversion etc., wherein time series data conversion be that equivalent is taken as unit of hour, specific formula are as follows:
Wherein, E (T) is T time section (hour) energy consumption equivalent value, etFor the t times energy consumption sampled value, α in T time sectiontFor power
Repeated factor, default take 1.
Step S104 establishes energy consumption prediction model according to the default engineering characteristics；
Using the pivot factor of Feature Engineering analyzing influence energy consumption, mode input/output variable is determined.
Specifically, using the pivot factor of Feature Engineering analyzing influence energy consumption, determine mode input/output variable, i.e., it is defeated
Entering variable includes energy consumption historical data, date, moment, weather conditions and temperature and humidity value, and output data is energy consumption forecasting sequence value
(24 hours following), wherein the corresponding date and hour of energy consumption historical data characterizes personal information, i.e. number overall state and work
Work and rest rule, if the date divides festivals or holidays and working day, the time divides the work and rest moment on and off duty, directly affects air-conditioning, illumination
Equal capital equipments energy distribution situation；Weather conditions and temperature and humidity value characterize equipment operation condition indirectly, and the air-conditioning that is such as warm is set
Standby operating load is big, and energy consumption is more.
Further, establishing energy consumption prediction model includes:
Model selection and parameter setting, parameter include decision tree quantity, maximum number depth, maximum branch mailbox number, character subset
Selection Strategy and random seed etc.；
Regressive prediction model training, is learnt and is trained with pivot factor data to energy consumption historical data, and keep instructing
Practice model；
Autoregressive moving-average model calculates, and carries out regression analysis to energy consumption unit time equivalent, obtains based on average meaning
The predicted value of justice；
Model evaluation and prediction, are assessed on the basis of mean prediction result, are modified to model.
Step S106 assesses the energy consumption prediction model and corrects the energy consumption prediction model according to assessment result.
Assessing the energy consumption prediction model and correcting the energy consumption prediction model according to assessment result mainly includes model choosing
Select with parameter setting, regressive prediction model, autoregressive moving-average model and model evaluation and prediction and etc..
Specifically, when carrying out model selection with parameter setting, random forest parameter includes decision tree quantity, maximal tree depth
It spends, maximum characteristic, internal node subdivided required smallest sample number and the minimum sample number of leaf node etc., wherein decision tree number
Amount is bigger can to allow model to have better performance, but will increase computational processing simultaneously, need to make choice according to the actual situation；
Maximal tree depth depends on data distribution, commonly uses value 10-100, the sample more than feature is needed to limit depth capacity；It is maximum
Characteristic generally takes the root of total characteristic number；The subtree that smallest sample number needed for internal node is subdivided limits continues the item divided
Part；The minimum sample number of leaf node limits the least sample number of leaf node.Autoregressive moving-average model parameter includes returning certainly
Return order, difference order and moving average order etc., wherein Autoregressive and moving average order are usually no more than 2.
Regressive prediction model learns energy consumption historical data and pivot factor data using random forest regression algorithm
And training, available following 24 hours energy consumptions predict value sequence { Et(T), T=1,2 ... 24 }；Random forest regression algorithm is
The integrated study mode being made of multiple Cart Tree Classifiers, wherein each Cart tree has the pumping put back at random inside sample set
A part is taken to be trained, multiple Tree Classifiers just constitute a training pattern matrix, and the sample that then will classify is brought into
Tree Classifier then with the principle that the minority is subordinate to the majority decides by vote the final classification type of this sample out one by one for this.
Autoregressive moving-average model is to carry out regression analysis to energy consumption unit time equivalent, obtains based on average
Predicted value as a reference value； For T moment energy consumption predicted value,To return certainly
Return coefficient, εtFor white noise, θj(j=1,2 ..., m) is sliding average coefficient.
Model evaluation is the evaluation situation to model selection and parameter setting, according to energy consumption predicted value Et(T) and a reference valueDetermined, ifThen model output reaches desired precision；IfThen model output cannot reach expectation quality, need to come back for model correction and parameter tune
It is whole.
It can be seen from the above description that the application realizes following technical effect:
In the embodiment of the present application, by the way of acquisition data source and by obtaining default engineering characteristics after pretreatment,
Energy consumption prediction model is established according to the default engineering characteristics, by assessing the energy consumption prediction model and according to assessment result school
The just described energy consumption prediction model, reached used with equipment, the factors variation such as personal information and climate condition and dynamically adapting with
Energy consumption precision of prediction is improved, the purpose because of false alarm rate caused by temporarily working overtime/having a holiday or vacation is reduced, is configurable to realize foundation
And the adaptive technical effect for architectural energy consumption prediction model, and then solve lack architecture, it is can configure, adaptive
The prediction model answered meets the technical issues of the needs of actual prediction system complex is changeable.
According to the embodiment of the present application, as preferred in the present embodiment, as shown in Fig. 2, assessing the energy consumption prediction model
And after according to assessment result correcting the energy consumption prediction model further include:
Step S202 analyzes the difference of energy consumption predicted value and energy consumption actual measurement value；
Predicted value and measured value difference analysis, analysis method include but is not limited to t inspection, variance analysis, the inspection of card side
It tests and non-parametric test (such as Kruskal-Wallis inspection)；
Step S204 matches energy consumption operational mode if difference meets switching condition；
If there is significant difference, then energy consumption operational mode matching is carried out, energy consumption operational mode is according to festivals or holidays and work
It is divided into operating mode and non-operating mode as the daily schedule, operational mode matching can have a holiday or vacation (outer to avoid interim overtime work or temporarily
Interference out)；
Step S206 determines exception if difference is unsatisfactory for switching condition.
Specifically, abnormal determination obtains new predicted value and corresponding sliding average base after referring to switching energy consumption operational mode
Quasi- value, carries out energy consumption abnormal determination and model abnormal determination；
It is abnormal to illustrate that energy consumption occurs, is reported if measured value deviates predicted value and a reference value for energy consumption abnormal alarm
It is alert；
Model abnormal alarm illustrates model misalignment if predicted value deviates measured value and a reference value, carries out alarm and school
Just.
Preferably, assess the energy consumption prediction model and according to assessment result correct the energy consumption prediction model include:
The energy consumption prediction model is corrected as benchmark model using autoregression sliding mean value model in model training.
Preferably, assess the energy consumption prediction model and according to assessment result correct the energy consumption prediction model include:
Assess and switch the selection energy consumption prediction model in model application as benchmark model using autoregression sliding mean value model,
In, the energy consumption prediction model includes: operating mode and non-operating mode.
Preferably, it assesses the energy consumption prediction model and is also wrapped after correcting the energy consumption prediction model according to assessment result
It includes: abnormal alarm operation is executed to the processing result for being determined as Exception Model or abnormal energy consumption.
It should be noted that step shown in the flowchart of the accompanying drawings can be in such as a group of computer-executable instructions
It is executed in computer system, although also, logical order is shown in flow charts, and it in some cases, can be with not
The sequence being same as herein executes shown or described step.
According to the embodiment of the present application, additionally provide a kind of for implementing the above-mentioned data processing method for building energy consumption
Device, as shown in figure 3, the device includes: acquisition module 10, for acquiring data source and by obtaining default engineering after pretreatment
Feature；Model building module 20, for establishing energy consumption prediction model according to the default engineering characteristics；And model evaluation module
30, to assess the energy consumption prediction model and correct the energy consumption prediction model according to assessment result.
Data source is acquired in the acquisition module 10 of the embodiment of the present application refers to acquisition subitem energy consumption and related data, data source
Acquisition modes include sensor detection, database is read and website crawls, involved initial data, which includes that energy consumption is accumulative, to be read
Number, date and time stamp, weather conditions index and temperature and humidity value, wherein the accumulative reading of energy consumption can periodically be read simultaneously directly from ammeter
It saves and (contains date and time stamp), weather conditions are read from the public data of local meteorological observatory website, and outdoor temperature humidity value utilizes biography
Sensor is measured in real time and teletransmission saves.
Data prediction includes: singular value is rejected, missing values are filled up, normalized, format is converted and time series data is converted etc.,
Wherein singular value rejects the normal range (NR) for needing empirically determined data；Missing values are filled up using moving average method, i.e. dt=
(dt+1+dt-1)/2,dtFor t-th of data；Normalization is then that dimension disunity bring is avoided to influence, can be using maximum-most
Small value method, i.e. Id=(dmax-d)/(dmax-dmin)；Format conversion is then the format or type for converting the data into needs, such as will
The power consumption values of the available unit interval of energy consumption gauge outfit value difference point；Time series data conversion is the equivalent for taking certain unit interval,
Specific formula are as follows:E (T) is T time section energy consumption equivalent value, etIt is adopted for the t times energy consumption in T time section
Sample value, αtFor weight factor, default takes 1.
Specifically, acquisition subitem energy consumption and related data, line number of going forward side by side Data preprocess, including singular value rejecting, missing values
Fill up, normalize and time series data conversion etc., wherein time series data conversion be that equivalent is taken as unit of hour, specific formula are as follows:
Wherein, E (T) is T time section (hour) energy consumption equivalent value, etFor the t times energy consumption sampled value, α in T time sectiontFor power
Repeated factor, default take 1.
The pivot factor that Feature Engineering analyzing influence energy consumption is utilized in the model building module 20 of the embodiment of the present application, determines
Mode input/output variable.
Specifically, using the pivot factor of Feature Engineering analyzing influence energy consumption, determine mode input/output variable, i.e., it is defeated
Entering variable includes energy consumption historical data, date, moment, weather conditions and temperature and humidity value, and output data is energy consumption forecasting sequence value
(24 hours following), wherein the corresponding date and hour of energy consumption historical data characterizes personal information, i.e. number overall state and work
Work and rest rule, if the date divides festivals or holidays and working day, the time divides the work and rest moment on and off duty, directly affects air-conditioning, illumination
Equal capital equipments energy distribution situation；Weather conditions and temperature and humidity value characterize equipment operation condition indirectly, and the air-conditioning that is such as warm is set
Standby operating load is big, and energy consumption is more.
Further, establishing energy consumption prediction model includes:
Model selection and parameter setting, parameter include decision tree quantity, maximum number depth, maximum branch mailbox number, character subset
Selection Strategy and random seed etc.；
Regressive prediction model training, is learnt and is trained with pivot factor data to energy consumption historical data, and keep instructing
Practice model；
Autoregressive moving-average model calculates, and carries out regression analysis to energy consumption unit time equivalent, obtains based on average meaning
The predicted value of justice；
Model evaluation and prediction, are assessed on the basis of mean prediction result, are modified to model.
The energy consumption prediction model is assessed in the model evaluation module 30 of the embodiment of the present application and is corrected according to assessment result
The energy consumption prediction model mainly include model selection with parameter setting, regressive prediction model, autoregressive moving-average model and
Model evaluation and prediction.
Specifically, when carrying out model selection with parameter setting, random forest parameter includes decision tree quantity, maximal tree depth
It spends, maximum characteristic, internal node subdivided required smallest sample number and the minimum sample number of leaf node etc., wherein decision tree number
Amount is bigger can to allow model to have better performance, but will increase computational processing simultaneously, need to make choice according to the actual situation；
Maximal tree depth depends on data distribution, commonly uses value 10-100, the sample more than feature is needed to limit depth capacity；It is maximum
Characteristic generally takes the root of total characteristic number；The subtree that smallest sample number needed for internal node is subdivided limits continues the item divided
Part；The minimum sample number of leaf node limits the least sample number of leaf node.Autoregressive moving-average model parameter includes returning certainly
Return order, difference order and moving average order etc., wherein Autoregressive and moving average order are usually no more than 2.
Regressive prediction model learns energy consumption historical data and pivot factor data using random forest regression algorithm
And training, available following 24 hours energy consumptions predict value sequence { Et(T), T=1,2 ... 24 }；Random forest regression algorithm is
The integrated study mode being made of multiple Cart Tree Classifiers, wherein each Cart tree has the pumping put back at random inside sample set
A part is taken to be trained, multiple Tree Classifiers just constitute a training pattern matrix, and the sample that then will classify is brought into
Tree Classifier then with the principle that the minority is subordinate to the majority decides by vote the final classification type of this sample out one by one for this.
Autoregressive moving-average model is to carry out regression analysis to energy consumption unit time equivalent, obtains based on average
Predicted value as a reference value； For T moment energy consumption predicted value,To return certainly
Return coefficient, εtFor white noise, θj(j=1,2 ..., m) is sliding average coefficient.
Model evaluation is the evaluation situation to model selection and parameter setting, according to energy consumption predicted value Et (T) and a reference valueDetermined, ifThen model output reaches desired precision；IfThen model output cannot reach expectation quality, need to come back for model correction and parameter tune
It is whole.
According to the embodiment of the present application, as preferred in the present embodiment, as shown in figure 4, device further include: analysis module
40, the analysis module includes: analytical unit, analyzes the difference of energy consumption predicted value and energy consumption actual measurement value；Matching unit is used for
When difference meets switching condition, energy consumption operational mode is matched；And anomaly unit, for when difference is unsatisfactory for switching condition,
Then determine exception.
According to the embodiment of the present application, as preferred in the present embodiment, as shown in figure 4, device further include: alarm module
50, the alarm module includes: abnormal alarm unit, for the processing result execution to Exception Model or abnormal energy consumption is determined as
Abnormal alarm operation.
According to the embodiment of the present application, as preferred in the present embodiment, the model building module includes: model foundation list
Member, the model evaluation module include: correction unit, assessment unit, the model foundation unit, for being returned using random forest
Reduction method is analyzed and establishes the energy consumption prediction model；The correction unit, for being slided in model training using autoregression
Mean value model corrects the energy consumption prediction model as benchmark model；And the assessment unit, for being used in model application
Autoregression sliding mean value model is assessed as benchmark model and switches the selection energy consumption prediction model, wherein the energy consumption is pre-
Surveying model includes: operating mode and non-operating mode.
The energy consumption prediction model is analyzed and established using random forest regression algorithm, and mean value is slided using autoregression
Model corrects the energy consumption prediction model as benchmark model, assesses institute as benchmark model using autoregression sliding mean value model
When stating energy consumption prediction model, description below is carried out to the realization principle of the application:
Referring to figure 5., based on the prediction of the building energy consumption of random forest and ARMA and abnormal alarm method, including following step
Suddenly,
1) data source is acquired, i.e., subitem energy consumption and related data, line number of going forward side by side Data preprocess, including singular value are rejected, lacked
Mistake value fills up, normalizes and time series data conversion etc., and wherein time series data conversion is that equivalent is taken as unit of hour, specific formula
Are as follows:
Wherein, E (T) is T time section (hour) energy consumption equivalent value, etFor the t times energy consumption sampled value, α in T time sectiontFor power
Repeated factor, default take 1.
2) the pivot factor for utilizing Feature Engineering analyzing influence energy consumption, determines mode input/output variable
3) it is analyzed using random forests algorithm and establishes energy consumption prediction model, key step includes:
A. model selection and parameter setting, parameter include decision tree quantity, maximum number depth, maximum branch mailbox number, feature
Collect Selection Strategy and random seed etc.；
B. regressive prediction model training, is learnt and is trained with pivot factor data to energy consumption historical data, and kept
Training pattern；
C. autoregressive moving-average model calculates, and carries out regression analysis to energy consumption unit time equivalent, obtains based on average
The predicted value of meaning；
D. model evaluation and prediction, are assessed on the basis of mean prediction result, are modified to model.
4) energy consumption abnormal alarm mechanism is designed based on prediction result, key step includes:
A. predicted value and measured value difference analysis, analysis method includes but is not limited to t inspection, variance analysis, card side
It examines and non-parametric test (such as Kruskal-Wallis inspection)；
B. if there is significant difference, then carry out energy consumption operational mode matching, energy consumption operational mode according to festivals or holidays and
The work scheduling time is divided into operating mode and non-operating mode, and operational mode matching can have a holiday or vacation to avoid interim overtime work or temporarily
The interference of (outgoing)；
C. abnormal determination obtains new predicted value and corresponding sliding average a reference value after switching energy consumption operational mode, into
The abnormal determination of row energy consumption and model abnormal determination；
D. it is abnormal to illustrate that energy consumption occurs, is reported if measured value deviates predicted value and a reference value for energy consumption abnormal alarm
It is alert；
F. model abnormal alarm illustrates model misalignment if predicted value deviates measured value and a reference value, carry out alarm and
Correction.
Preferably, according to internal association rule, mechanism knowledge and engineering experience between data, the main shadow of energy consumption analysis is determined
The factor of sound: personal information, facility information and climate condition, personal information include number information and work scheduling feature, are generally led to
It crosses sensor to be difficult to accurately obtain, but the places such as office building metastable for flow of personnel, can be read by access control system
It takes, number overall situation (more/few) can also be characterized indirectly according to national festivals or holidays and operation time；Facility information refers mainly to
Device type, quantity and operating status (ON/OFF) can be obtained by information systems such as equipment operation management platforms；Weather
Situation mainly includes that season situation (cooling supply, heat supply and conditioning in Transition Season), weather conditions (fine day, polynary, rain, snow etc.) and outdoor are warm and humid
Detected value is spent, season situation can generally be provided by building control system, and weather conditions can be obtained in weather site, and temperature and humidity value can benefit
It is surveyed, can also be obtained in weather site with sensor.This Feature Engineering gives the pivot with building energy consumption strong correlation more comprehensively
Factor and data acquiring mode can take the circumstances into consideration to select according to the actual situation and adjustment, Feature Engineering directly influence analysis result
Precision, however existing energy consumption prediction technique about characteristic factor selection lack architecture further investigation.
Preferably, it according to the historical data of pivot factor, introduces in data mining regression algorithm study energy for building and is advising
Rule, to provide following 24 hours energy consumption predicted values.Regression algorithm includes but is not limited to random forest, decision tree and GBDT
Deng the forecast analysis that relationship between dependent variable (target) and independent variable (fallout predictor) is used for continuous variable can be learnt.We
Method can be built with dynamically adapting difference, Various Seasonal and weather feature learning go out prediction model, the work independent of industry specialists
The concrete configuration of journey experience and building, this is the application and other traditional energy consumption prediction technique differences.Regression algorithm
Model selection is directly related to prediction result precision with parameter setting, and the application is proposed using autoregression sliding mean value as model
The benchmark of assessment solves the problems, such as to be difficult to assess and correct in algorithm model application, this is the application and other data-drivens
Prediction technique difference.
Preferably, abnormal alarm comprehensively considers energy consumption predicted value, measured value and a reference value and is set, and mainly includes
Energy consumption abnormal alarm and model abnormal alarm, wherein energy consumption predicted value is provided by regression model, and measured value is sensor (ammeter)
Detection obtains, and a reference value is that arma modeling provides；If energy consumption actual measurement value deviates energy consumption predicted value and a reference value, illustrate energy consumption
Occur abnormal；If energy consumption predicted value deviates measured value and a reference value, illustrates model misalignment, need to correct.In addition, introducing work
Operation mode and non-operating mode energy consumption model handover mechanism overcome because energy consumption caused by temporarily working overtime or being temporarily outgoing is extremely empty
Alert problem, this is the difference of the application Yu other energy consumption prediction techniques.
Fig. 6 and Fig. 7 are please referred to, energy consumption of building prediction and abnormal alarm method based on data mining, by identifying data
Inherent law and the energy consumption prediction model established can be with dynamic self-adapting each influence factor (personnel, device configuration, season and weather
Deng) variation, provide energy consumption predicted value and energy consumption abnormal alarm.Energy consumption prediction and abnormal alarm architecture based on data mining
Workflow is as shown in fig. 6, mainly include data input step 201, data prediction step 202, Feature Engineering step 203, mould
Type establishment step 204, model evaluation step 205 and application output step 206 and etc., every part individual packages are at standardization mould
Block, can be with its parameter of flexible configuration, then each module compositional system workflow, under Various Seasonal Different climate feature
Different buildings, the workflow can be applicable in, it is only necessary to carry out parametrization configuration to corresponding module.Implement in the application
It is illustrated by taking certain the subitem energy consumption prediction of 11 layers of XX intelligent building Building A, Beijing and abnormity early warning as an example in example.
Data input the process that 201 steps are data acquisitions, including sensor detection, database are read and website crawls
Several mode, input data includes energy consumption historical data, date, moment, weather conditions and warm and humid in the embodiment of the present application
Angle value, wherein energy consumption historical data can periodically read directly from ammeter and save (containing date and time stamp), and the sampling period is
15min, weather conditions are read from the public data of local meteorological observatory website, sampling period 1h, and outdoor temperature humidity value utilizes biography
Sensor is measured in real time and teletransmission saves, sampling period 15min.
202 step of data prediction is will to collect initial data to be cleaned, converted and normalized, it is made to meet mould
The input requirements of type algorithm, need to carry out singular value rejecting in the embodiment of the present application, missing values are filled up, normalize and when ordinal number
According to conversion etc., wherein time series data conversion is that equivalent is taken as unit of hour, specific formula are as follows:
Wherein, E (T) is T time section (hour) energy consumption equivalent value, etFor the t times energy consumption sampled value, α in T time sectiontFor weight factor,
Default chooses 1.
203 step of Feature Engineering is the pivot factor of analyzing influence energy consumption, determines mode input/output variable, in this Shen
It please input data include energy consumption historical data, date, moment, weather conditions and temperature and humidity value in embodiment, output data is energy
Forecasting sequence value (24 hours following) is consumed, wherein the corresponding date and hour of energy consumption historical data characterizes personal information, i.e. number
Overall state and work scheduling rule, if the date divides festivals or holidays and working day, the time divides work and rest moment on and off duty, direct shadow
Ring the main energy distribution situation such as air-conditioning, illumination；Weather conditions and temperature and humidity value characterize equipment operation condition indirectly, are such as warm
Air-conditioning equipment operating load is big, and energy consumption is more.
204 step of model foundation is the core of data mining, is carried out in analysis searching using inputoutput data
In the process of rule.In the embodiment of the present application, random forests algorithm is chosen to analyze and establish energy consumption prediction model, key step
It is calculated including model selection and parameter setting, regressive prediction model training and autoregressive moving-average model.Random forests algorithm
Parameter includes decision tree quantity, maximal tree depth, maximum characteristic, the subdivided required smallest sample number of internal node and leaf section
Minimum sample number of point etc., wherein decision tree quantity is selected as 300, and maximal tree depth is 10, and maximum characteristic is 3, and internal node is again
Smallest sample number needed for dividing is defaulted as 2, and the minimum sample number of leaf node is defaulted as 1；Autoregressive moving-average model parameter packet
Autoregressive, difference order and moving average order etc. are included, wherein Autoregressive p=2, difference order d=1 is mobile flat
Equal order q=1.
205 step of model evaluation is that the evaluation situation to model selection and parameter setting calculates in the embodiment of the present application
A reference value of the mean value as prediction is slided in autoregressionSpecific formula are as follows:
For T moment energy consumption predicted value,For autoregressive coefficient, εtFor white noise, θj(j=1,2 ..., m) is sliding
Mean coefficient.Then energy consumption predicted value E is obtained using random forest regression algorithmt(T), if
Then model output reaches desired precision；If Then model output cannot reach expectation quality,
Need to come back for model correction and parameter adjustment.
It include energy consumption prediction and abnormal alarm using 206 steps of output, after the former reaches target after model training
Following 24 hours energy consumption predicted values are directly given, in the embodiment of the present application, Fig. 7 and Fig. 8 describe certain subitem energy consumption prediction
Result schematic diagram, energy consumption are divided into operating mode and two kinds of non-operating mode according to energy rule, and wherein curve 1 characterizes energy consumption actual measurement
Value, directly reads acquisition by ammeter；Curve 2 characterizes energy consumption a reference value, is obtained by autoregression sliding mean value computation；Curve 3 characterizes
Energy consumption predicted value is obtained by random forest regression forecasting.The latter carries out difference analysis according to energy consumption predicted value and measured value, such as
There are significant differences then to attempt running mode switching for fruit, then carries out abnormal determination, provides energy consumption abnormal alarm or model loses
Quasi- alarm.In the embodiment of the present application, difference analysis uses nonparametric Kruskal-Wallis method of inspection, if there is significant
Sex differernce then switches energy consumption operational mode, i.e., operating mode switches to non-operating mode or non-operating mode switches to Working mould
Formula.Existing energy consumption prediction model is to directly determine mode according to festivals or holidays and work scheduling time, it is difficult to which processing is because of interim overtime work
Or the energy consumption exception false-alarm predicament of (outgoing) of temporarily having a holiday or vacation.As Fig. 9 characterizes the energy consumption that certain weekend collective of company temporarily trains overtime work
Prediction case, curve 1 characterize energy consumption actual measurement value, and curve 2 characterizes energy consumption a reference value, and it is pre- that curve 3 characterizes energy consumption under non-operating mode
Measured value, there are significant differences between system detection energy consumption actual measurement value and predicted value, then switch to energy consumption predicted value under operating mode,
As shown in curve 4.Then whether detection measured value deviates predicted value and a reference value, if measured value still deviates predicted value,
It is abnormal to illustrate that energy consumption occurs, alarms；If predicted value deviates measured value and a reference value, illustrate model misalignment, needs
It is alarmed and is corrected.Equally, because energy consumption of temporarily having a holiday or vacation or go out is as shown in Figure 10, models switching and abnormality detection mistake
Journey is similar, repeats no more.The application can independently switch energy consumption operational mode, overcome because of energy caused by temporarily working overtime or have a holiday or vacation
Consume the predicament of abnormal false-alarm.
Obviously, those skilled in the art should be understood that each module of above-mentioned the application or each step can be with general
Computing device realize that they can be concentrated on a single computing device, or be distributed in multiple computing devices and formed
Network on, optionally, they can be realized with the program code that computing device can perform, it is thus possible to which they are stored
Be performed by computing device in the storage device, perhaps they are fabricated to each integrated circuit modules or by they
In multiple modules or step be fabricated to single integrated circuit module to realize.In this way, the application be not limited to it is any specific
Hardware and software combines.
The foregoing is merely preferred embodiment of the present application, are not intended to limit this application, for the skill of this field
For art personnel, various changes and changes are possible in this application.Within the spirit and principles of this application, made any to repair
Change, equivalent replacement, improvement etc., should be included within the scope of protection of this application.
Claims (10)
1. a kind of building energy consumption processing method based on RF and ARMA algorithm characterized by comprising
Acquire data source and by obtaining default engineering characteristics after pretreatment；
Energy consumption prediction model is established according to the default engineering characteristics；And
It assesses the energy consumption prediction model and the energy consumption prediction model is corrected according to assessment result.
2. the method according to claim 1, wherein assessing the energy consumption prediction model and according to assessment result school
After the just described energy consumption prediction model further include:
Analyze the difference of energy consumption predicted value and energy consumption actual measurement value；
If difference meets switching condition, energy consumption operational mode is matched；And
If difference is unsatisfactory for switching condition, exception is determined.
3. the method according to claim 1, wherein establishing energy consumption prediction model according to the default engineering characteristics
Include:
It is analyzed using random forest regression algorithm and establishes the energy consumption prediction model.
4. the method according to claim 1, wherein assessing the energy consumption prediction model and according to assessment result school
The just described energy consumption prediction model includes:
The energy consumption prediction model is corrected as benchmark model using autoregression sliding mean value model in model training.
5. the method according to claim 1, wherein assessing the energy consumption prediction model and according to assessment result school
The just described energy consumption prediction model includes:
It is assessed using autoregression sliding mean value model as benchmark model in model application and switches the selection energy consumption prediction
Model operating mode and non-operating mode.
6. the method according to claim 1, wherein assessing the energy consumption prediction model and according to assessment result school
After the just described energy consumption prediction model further include:
Abnormal alarm operation is executed to the processing result for being determined as Exception Model or abnormal energy consumption.
7. a kind of building energy consumption processing unit based on RF and ARMA algorithm characterized by comprising
Acquisition module, for acquiring data source and by obtaining default engineering characteristics after pretreatment；
Model building module, for establishing energy consumption prediction model according to the default engineering characteristics；And
Model evaluation module, for assessing the energy consumption prediction model and correcting the energy consumption prediction model according to assessment result.
8. device according to claim 7, which is characterized in that the device further include: analysis module, the analysis module packet
It includes:
Analytical unit analyzes the difference of energy consumption predicted value and energy consumption actual measurement value；
Matching unit, for matching energy consumption operational mode when difference meets switching condition；And
Anomaly unit, for when difference is unsatisfactory for switching condition, then determining exception.
9. device according to claim 7, which is characterized in that the model building module includes: model foundation unit, institute
Stating model evaluation module includes: correction unit, assessment unit,
The model foundation unit, for being analyzed using random forest regression algorithm and establishing the energy consumption prediction model；
The correction unit, for correcting the energy as benchmark model using autoregression sliding mean value model in model training
Consume prediction model；And
The assessment unit, for using autoregression sliding mean value model to assess and switch as benchmark model in model is applied
Select the energy consumption prediction model, wherein the energy consumption prediction model includes operating mode and non-operating mode.
10. device according to claim 7, which is characterized in that further include: alarm module, the alarm module include:
Abnormal alarm unit, for the processing result execution abnormal alarm operation to Exception Model or abnormal energy consumption is determined as.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810799398.8A CN109035067A (en) | 2018-07-19 | 2018-07-19 | Building energy consumption processing method and processing device based on RF and ARMA algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810799398.8A CN109035067A (en) | 2018-07-19 | 2018-07-19 | Building energy consumption processing method and processing device based on RF and ARMA algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109035067A true CN109035067A (en) | 2018-12-18 |
Family
ID=64644485
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810799398.8A Pending CN109035067A (en) | 2018-07-19 | 2018-07-19 | Building energy consumption processing method and processing device based on RF and ARMA algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109035067A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109741213A (en) * | 2018-12-24 | 2019-05-10 | 珠海格力电器股份有限公司 | Zero energy consumption architectural energy consumption control method of one kind and system |
| CN112836396A (en) * | 2021-03-10 | 2021-05-25 | 同济大学 | Building real-time energy consumption abnormity diagnosis system |
| CN112836720A (en) * | 2020-12-16 | 2021-05-25 | 博锐尚格科技股份有限公司 | Building operation and maintenance equipment abnormity diagnosis method and system and computer readable storage medium |
| CN113115311A (en) * | 2021-04-12 | 2021-07-13 | 江苏通付盾科技有限公司 | Support vector machine model-based fraud behavior identification method and system |
| CN113660335A (en) * | 2021-08-17 | 2021-11-16 | 北京计算机技术及应用研究所 | Equipment fine management method and system based on Internet of things |
| CN114741905A (en) * | 2022-06-13 | 2022-07-12 | 广东电网有限责任公司佛山供电局 | Actually measured energy consumption data correction method and device, electronic equipment and storage medium |
| CN115049083A (en) * | 2022-08-15 | 2022-09-13 | 天津理工大学 | Electromechanical equipment operation management method, device and system |
| CN116720634A (en) * | 2023-08-11 | 2023-09-08 | 北京泰豪智能工程有限公司 | Park operation data processing method and system |
| CN116805275A (en) * | 2023-08-21 | 2023-09-26 | 中国标准化研究院 | Public institution energy consumption intelligent supervision system based on data analysis |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102004859A (en) * | 2010-12-01 | 2011-04-06 | 南京工业大学 | On-line diagnostic method for abnormal energy consumption branch of building |
| US20130325377A1 (en) * | 2012-05-31 | 2013-12-05 | Johnson Controls Technology Company | Systems and methods for measuring and verifying energy usage in a building |
| CN105205558A (en) * | 2015-09-23 | 2015-12-30 | 南京磐能电力科技股份有限公司 | BP neural network model configuration method oriented to building energy consumption prediction |
| CN105631539A (en) * | 2015-12-25 | 2016-06-01 | 上海建坤信息技术有限责任公司 | Intelligent building energy consumption prediction method based on support vector machine |
-
2018
- 2018-07-19 CN CN201810799398.8A patent/CN109035067A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102004859A (en) * | 2010-12-01 | 2011-04-06 | 南京工业大学 | On-line diagnostic method for abnormal energy consumption branch of building |
| US20130325377A1 (en) * | 2012-05-31 | 2013-12-05 | Johnson Controls Technology Company | Systems and methods for measuring and verifying energy usage in a building |
| CN105205558A (en) * | 2015-09-23 | 2015-12-30 | 南京磐能电力科技股份有限公司 | BP neural network model configuration method oriented to building energy consumption prediction |
| CN105631539A (en) * | 2015-12-25 | 2016-06-01 | 上海建坤信息技术有限责任公司 | Intelligent building energy consumption prediction method based on support vector machine |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109741213A (en) * | 2018-12-24 | 2019-05-10 | 珠海格力电器股份有限公司 | Zero energy consumption architectural energy consumption control method of one kind and system |
| CN112836720A (en) * | 2020-12-16 | 2021-05-25 | 博锐尚格科技股份有限公司 | Building operation and maintenance equipment abnormity diagnosis method and system and computer readable storage medium |
| CN112836720B (en) * | 2020-12-16 | 2024-03-29 | 博锐尚格科技股份有限公司 | Building operation and maintenance equipment abnormality diagnosis method, system and computer readable storage medium |
| CN112836396A (en) * | 2021-03-10 | 2021-05-25 | 同济大学 | Building real-time energy consumption abnormity diagnosis system |
| CN113115311A (en) * | 2021-04-12 | 2021-07-13 | 江苏通付盾科技有限公司 | Support vector machine model-based fraud behavior identification method and system |
| CN113660335A (en) * | 2021-08-17 | 2021-11-16 | 北京计算机技术及应用研究所 | Equipment fine management method and system based on Internet of things |
| CN114741905B (en) * | 2022-06-13 | 2022-09-13 | 广东电网有限责任公司佛山供电局 | Actually measured energy consumption data correction method and device, electronic equipment and storage medium |
| CN114741905A (en) * | 2022-06-13 | 2022-07-12 | 广东电网有限责任公司佛山供电局 | Actually measured energy consumption data correction method and device, electronic equipment and storage medium |
| CN115049083A (en) * | 2022-08-15 | 2022-09-13 | 天津理工大学 | Electromechanical equipment operation management method, device and system |
| CN116720634A (en) * | 2023-08-11 | 2023-09-08 | 北京泰豪智能工程有限公司 | Park operation data processing method and system |
| CN116720634B (en) * | 2023-08-11 | 2023-10-13 | 北京泰豪智能工程有限公司 | Park operation data processing method and system |
| CN116805275A (en) * | 2023-08-21 | 2023-09-26 | 中国标准化研究院 | Public institution energy consumption intelligent supervision system based on data analysis |
| CN116805275B (en) * | 2023-08-21 | 2023-11-21 | 中国标准化研究院 | Public institution energy consumption intelligent supervision system based on data analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109035067A (en) | Building energy consumption processing method and processing device based on RF and ARMA algorithm | |
| Dong et al. | Wind power day-ahead prediction with cluster analysis of NWP | |
| WO2022135265A1 (en) | Failure warning and analysis method for reservoir dispatching rules under effects of climate change | |
| CN110097297A (en) | A kind of various dimensions stealing situation Intellisense method, system, equipment and medium | |
| Yalcintas | Energy-savings predictions for building-equipment retrofits | |
| Yildiz et al. | Household electricity load forecasting using historical smart meter data with clustering and classification techniques | |
| Serrano-Guerrero et al. | A new interval prediction methodology for short-term electric load forecasting based on pattern recognition | |
| Hadri et al. | A comparative study of predictive approaches for load forecasting in smart buildings | |
| Zhu et al. | Data-driven building load profiling and energy management | |
| Serrano-Guerrero et al. | Statistical methodology to assess changes in the electrical consumption profile of buildings | |
| CN106779129A (en) | A kind of Short-Term Load Forecasting Method for considering meteorologic factor | |
| JP2003242212A (en) | Apparatus and method for determining day of the week with similar utility consumption profile | |
| CN111680841B (en) | Short-term load prediction method, system and terminal equipment based on principal component analysis | |
| CN115796393B (en) | Energy management optimization method, system and storage medium based on multi-energy interaction | |
| CN114372360A (en) | Method, terminal and storage medium for power load prediction | |
| KR102495494B1 (en) | Method and apparatus for predicting home energy usage | |
| CN111161095A (en) | Method for detecting abnormal consumption of building energy | |
| CN108346009A (en) | A kind of power generation configuration method and device based on user model self study | |
| Yu et al. | A model integrating ARIMA and ANN with seasonal and periodic characteristics for forecasting electricity load dynamics in a state | |
| CN108154259B (en) | Load prediction method and device for heat pump, storage medium, and processor | |
| CN116633016B (en) | Power management system and method based on high-frequency data acquisition technology | |
| KR102332248B1 (en) | Performance evaluation system for Non-intrusive load monitoring according to multi-state energy classification and method thereof | |
| Ma et al. | A synchronous prediction method for hourly energy consumption of abnormal monitoring branch based on the data-driven | |
| CN116308883A (en) | Regional power grid data overall management system based on big data | |
| CN115796390A (en) | Data-driven building energy consumption abnormity diagnosis method and system |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181218 |