CN108830417A - A kind of residential energy consumption prediction technique and system based on ARMA and regression analysis - Google Patents
A kind of residential energy consumption prediction technique and system based on ARMA and regression analysis Download PDFInfo
- Publication number
- CN108830417A CN108830417A CN201810609136.0A CN201810609136A CN108830417A CN 108830417 A CN108830417 A CN 108830417A CN 201810609136 A CN201810609136 A CN 201810609136A CN 108830417 A CN108830417 A CN 108830417A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- regression analysis
- arma
- residential energy
- model
- 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.)
- Granted
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
- 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 residential energy consumption prediction technique and system based on ARMA and regression analysis obtain residential energy consumption project and its measured value per capita；First sample corresponding with residential energy consumption measured value per capita is established, time series is constructed；According to Bayesian Information amount criterion, the order of arma modeling is determined, construct arma modeling；It establishes with the impact factor of reality factor and the sample set of time series as the second sample；Regression analysis is carried out to the second sample, obtains combination forecasting；Prediction is combined to time series using combination forecasting.The present invention, which is used, learns prediction model based on ARMA and the ensemble machine of regression analysis, can better adapt to the characteristic and accurate description reality influence factor of time series, with the high beneficial effect of test accuracy.
Description
Technical field
The present invention relates to energy forecast the field of data mining, and in particular to a kind of life energy based on ARMA and regression analysis
Source consumption predictions method and system.
Background technique
The energy occupies an important position in economic development, is an important factor for influencing national strategy and policy.In recent years,
Flourishing for Chinese energy industry provides endlessly power for China's economic growth, but in energy industry development process
In, the problems such as per capita energy is insufficient, efficiency of energy utilization is low, environmental pollution is serious, is increasingly prominent, needs the energy to China thus
Source structure and energy-consuming are adjusted and control, and predict residential energy consumption per capita, help to formulate reasonable
Energy adjustment measure, to the sound development important in inhibiting of economy and environment.At present is mostly used for the prediction of the energy time
Serial method, time series method are predicted Future Data, but the single time by finding the potential rule in historical data
For series model when predicting nonlinear chaos sequence, prediction result often has biggish error.In addition, under physical condition
The value of time series at a time depends not only on the changing rule of itself, is also influenced by factors such as population, economy, and
Temporal model can not describe the characteristic information of real influence factor.
In conclusion big for prediction result error in the prior art, temporal model can not describe real influence factor
Problem, still shortage effective solution scheme.
Summary of the invention
In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a kind of life energy based on ARMA and regression analysis
Source consumption predictions method and system learn prediction model using the ensemble machine based on ARMA and regression analysis, can preferably fit
The characteristic and accurate description reality influence factor for answering time series, with the high beneficial effect of test accuracy.
The technical scheme adopted by the invention is that：
A kind of residential energy consumption prediction technique based on ARMA and regression analysis, this approach includes the following steps：
Obtain residential energy consumption project and its measured value per capita；
First sample corresponding with residential energy consumption measured value per capita is established, time series is constructed；
According to Bayesian Information amount criterion, the order of arma modeling is determined, construct arma modeling；
It establishes with the impact factor of reality factor and the sample set of time series as the second sample；
Regression analysis is carried out to the second sample, obtains combination forecasting；
Prediction is combined to first sample using combination forecasting.
Further, further include being screened to residential energy consumption measured value per capita, reject residential energy consumption per capita
Missing values in amount；Measurement missing values in residential energy consumption measured value per capita are fitted, to after screening and fitting
The measured value of residential energy consumption per capita the step of formatting.
Further, the construction method of the time series is：
First sample corresponding with residential energy consumption measured value per capita is established, using first sample as initiation sequence；
The operation of k order difference is carried out to initiation sequence, obtains being based on residential energy consumption project per capita and meet stationarity to want
The time series asked.
Further, if initiation sequence is not stationary sequence, need to do initiation sequence k order difference operation, k is to make sequence
Column meet the minimum differential operation times of stationarity requirement；If initiation sequence meets stationarity requirement, do not need to make the difference partite transport
It calculates, k value is 0 at this time.
Further, the method for carrying out regression analysis to the second sample includes：
Using the impact factor of reality factor and time series as the independent variable of regression analysis, by the time series a certain moment
Value as regression analysis dependent variable carry out Function Fitting, obtain several regression analysis samples, establish regression analysis sample
This collection；
3 samples are randomly selected from regression analysis sample set as verifying collection, remaining sample is as training set to recurrence
Analysis model is trained；
Compare relative error of each regression analysis model on verifying collection, takes first three the smallest model structure of relative error
Combination forecasting is built, the combination forecasting includes arma modeling, Support vector regression model and ridge regression model.
Further, intercepted length is the partial sequence of n as time series independent variable in time series in order
Value, step-length 1, n are the quantity of time series independent variable in regression analysis.
Further, the method for being combined prediction to first sample using combination forecasting includes：
First sample is predicted using arma modeling, Support vector regression model and ridge regression model, obtains three
The prediction result of a model is distributed the prediction result of three models to certain weight and is weighted and averaged, and obtains final pre-
Survey result.
Further, the calculation method of the weight is：
Wherein, τiIndicate the model compatible degree containing mean value and standard deviation, the value of i is 1,2,3, ωiIndicate the power of model
Value.
Further, the calculating function of the model compatible degree containing mean value and standard deviation is：
Wherein, σ and μ respectively indicates the standard deviation and mean value of verifying collection true value；ξiIndicate phase of the model i on verifying collection
To error, the value of i is 1,2,3；σiIndicate the standard deviation of predicted value obtained by model i predicts verifying collection, μiIndicate mould
Type i predict to verifying collection the mean value of obtained predicted value.
A kind of residential energy consumption forecasting system based on ARMA and regression analysis, the system include：
Energy detecting device, for obtaining residential energy consumption project and its measured value；And
Processor is connected with energy-consuming detection device, for realizing as described above based on ARMA and regression analysis
Residential energy consumption prediction technique；And
Display unit is connected with processor, the prediction result for output processor.
Compared with prior art, the beneficial effects of the invention are as follows：
(1) present invention learns prediction model using based on ARMA and the ensemble machine of regression analysis, when having better adapted to
Between sequence characteristic and can accurate description reality influence factor, future life energy-consuming can be changed over time situation progress
Prediction, and then reasonable energy adjustment measure is formulated, test accuracy is high；
(2) present invention carries out optimal models screening, by comparing relative error of each model on verifying collection, takes opposite
First three the smallest model construction combination forecasting of error, enables residential energy consumption combination forecasting to better adapt to
The characteristic of time series, last prediction result is by the prediction result of three single models by distributing certain weight and being weighted
It averagely obtains, is predicted by way of weighted array, with high, the highly reliable and more stable beneficial effect of test accuracy
Fruit, hence it is evident that better than the method for conventional construction built-up pattern.
Detailed description of the invention
The accompanying drawings constituting a part of this application is used to provide further understanding of the present application, and the application's shows
Meaning property embodiment and its explanation are not constituted an undue limitation on the present application for explaining the application.
Fig. 1 is the residential energy consumption prediction technique flow chart based on ARMA and regression analysis；
Fig. 2 is dependent variable of the present invention and time series independent variable figure；
Fig. 3 is regression analysis sample graph；
Fig. 4 is relative error figure of the model on verifying collection；
Fig. 5 is mean value and standard deviation figure；
Fig. 6 is predicted value of the present invention and true value figure.
Specific embodiment
The invention will be further described with embodiment with reference to the accompanying drawing.
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
Term explains part:ARMA is autoregressive moving-average model
As background technique is introduced, exist in the prior art that prediction result error is larger, and temporal model can not describe
The deficiency of real influence factor, in order to solve technical problem as above, present applicant proposes one kind to be based on ARMA and regression analysis
Residential energy consumption prediction technique and system, better adapted to the characteristic of time series and can accurate description reality influence because
Element can change over time situation to future life energy-consuming and predict, and then formulate reasonable energy adjustment measure, tool
There is the beneficial effect that test accuracy is high.
In a kind of typical embodiment of the application, as shown in Figure 1, providing a kind of based on ARMA and regression analysis
Residential energy consumption prediction technique, this approach includes the following steps：
Step 101：Residential energy consumption project and its measured value per capita are obtained, and to residential energy consumption measured value per capita
It screened, be fitted and converted.
Include to the step of residential energy consumption measured value screens per capita：
Residential energy consumption measured value per capita is screened, the missing values in household energy consumption per capita are rejected.
Include to the step of residential energy consumption measured value is fitted per capita：
Measurement missing values in residential energy consumption measured value per capita are fitted.
Include to the step of residential energy consumption measured value is converted per capita：
The measured value of residential energy consumption per capita after screening and fitting is formatted.
Step 102：Based on residential energy consumption measured value per capita, time series, the second sample are constructed.
The construction method of time series is：
First sample corresponding with residential energy consumption measured value per capita is established, using first sample as initiation sequence；
The operation of k order difference (k is the minimum differential operation times for making sequence meet stationarity requirement) is carried out to initiation sequence, is based on
Residential energy consumption project per capita meets the time series of stationarity requirement.
If initiation sequence is not stationary sequence, need to do initiation sequence k order difference operation, k is to meet sequence to put down
The minimum differential operation times that stability requires；If initiation sequence meets stationarity requirement, do not need to do calculus of differences, at this time k
Value is 0.
The k value is calculated as：
Z=S ' | S ' is stationary } (1)
K=min (Q) (3)
Wherein, Z indicates the set comprising all time serieses for meeting stationarity requirement；S indicates initial time sequence；D
(S, l) expression does the obtained new time series of l order difference operation to initial time sequence S；Q, which indicates all, keeps sequence satisfaction flat
The set for the calculus of differences number that stability requires.
After obtaining time series, establish with based on reality factors such as population, economy impact factor and the time sequence
The sample set of column is the second sample.
Step 103：According to Bayesian Information amount criterion, the order of arma modeling is determined, construct arma modeling.
According to Bayesian Information amount criterion, the order construction strongest arma modeling of generalization ability of arma modeling is determined, it is right
Future Data is predicted.
The expression formula of arma modeling is：
Wherein, YtFor the value for predicting object under t moment, etIt is the sequence of random variables that mean value is greater than 0 for 0, variance, p, q
For the order of arma modeling, it is to determine suitable p, q value that arma modeling, which determines rank,.
The order of arma modeling is by Bayesian Information amount criterion (Bayesian Information Criterion, BIC)
It determines, to overcome because poor fitting and over-fitting cause the defect of model generalization scarce capacity, the calculation of BIC index is：
BIC=-2lnL+KlnN (5)
Wherein, L is the maximum likelihood estimator of model, and K is the quantity that model uses variable, and N is that model uses number
According to quantity.
Step 104：Regression analysis constructs combination forecasting.
Regression analysis includes a variety of homing methods that linear and nonlinear returns, respectively multiple linear regression, ridge
Return, random forest return, decision tree return, extreme random tree return, gradient boosted tree return, Support vector regression and
LASSO (Least Absolute Shrinkage And Selection Operator, LASSO) is returned.
Using based on reality factors such as population, economy impact factor and time series as the independent variable of regression analysis, will
Dependent variable of the value at time series a certain moment as regression analysis, as follows：
Wherein, y indicates dependent variable, W1...Wm+nIndicate the coefficient of independent variable, m, n respectively indicate impact factor independent variable
(f1...fm) and time series independent variable (o1...on) quantity, in order in time series intercepted length be n partial order
Value of the column as time series independent variable, step-length 1, such as time series (s1,s2,s3,s4,s5,s6,s7,s8), if returning
The quantity for returning time series independent variable in analysis is 4 (n=4), i.e. (o1,o2,o3,o4), then dependent variable and time series independent variable
Value it is as shown in Figure 2.
Will affect the independent variable of the factor and time series as regression analysis, using the value at time series a certain moment as
The dependent variable of regression analysis carries out Function Fitting, obtains several regression analysis samples, constructs regression analysis sample set.
3 samples are randomly selected from regression analysis sample set as verifying collection, remaining sample is as training set to recurrence
Analysis model is trained；Relative error on training set by comparing each regression analysis model on verifying collection, takes phase
First three model construction combination forecasting the smallest to error, combination forecasting include arma modeling, support vector machines time
Return model and ridge regression model.
Step 105：Prediction is combined to first sample using combination forecasting.
Prediction is combined to first sample using combination forecasting, last prediction result is pre- by three single models
Result is surveyed by distributing certain weight and being weighted and averaged to obtain.And the prediction of per capita energy consumption in time series is presented
As a result.
The weight computing is：
Wherein, τiIndicate the compatible degree of model, the value of i is 1,2,3, ωiIndicating the weight of model, the value of i is 1,2,
3。
The model compatible degree of arma modeling is calculated as：
Wherein, ξiIndicate that relative error of the model on verifying collection, the value of i are 1,2,3.
The application introduces mean value and standard deviation, the mould for predicted value and true value with the mean value and standard deviation being closer to
Type distributes biggish weight.Compatible degree calculating function is replaced with and calculates function containing the compatible degree of mean value and standard deviation, is：
Wherein, ξiIndicate that relative error of the model on verifying collection, the value of i are 1,2,3；σ and μ respectively indicates verifying collection
The standard deviation and mean value of true value；σiIndicate the standard deviation of predicted value obtained by model i predicts verifying collection, μiIndicate mould
Type i predict to verifying collection the mean value of obtained predicted value.
The residential energy consumption prediction technique based on ARMA and regression analysis that the embodiment of the present invention proposes, so that life energy
Source consumption combination forecasting can better adapt to the characteristic of time series, and be predicted have by way of weighted array
There is the beneficial effect that test accuracy is high, highly reliable and more stable, hence it is evident that better than the method for conventional construction built-up pattern.
In order to make those skilled in the art be better understood by the present invention, a specific calculated examples are set forth below, originally
Inventive embodiments provide a kind of residential energy consumption prediction technique based on ARMA and regression analysis, including：
Step 201：Choose annual household energy consumption per capita.
The historical data that the present embodiment uses is household energy consumption per capita annual in nineteen eighty-three to 2015, unit
For kilogram standard coal and male demographic's proportion, as shown in Table 1 and Table 2；And to 2008 to 2015 years domestic energies per capita
Consumption figure carries out prediction verifying.The regression analysis model established is using male's ratio shared in Chinese total population as influence
Factor independent variable.
The equal household energy consumption of 1 people of table
2 male of table proportion in Chinese total population
| Year | Specific gravity/% | Year | Specific gravity/% | Year | Specific gravity/% |
| 1983 | 51.6 | 1994 | 51.1 | 2005 | 51.53 |
| 1984 | 51.6 | 1995 | 51.03 | 2006 | 51.52 |
| 1985 | 51.7 | 1996 | 50.82 | 2007 | 51.5 |
| 1986 | 51.7 | 1997 | 51.07 | 2008 | 51.47 |
| 1987 | 51.5 | 1998 | 51.25 | 2009 | 51.44 |
| 1988 | 51.52 | 1999 | 51.43 | 2010 | 51.27 |
| 1989 | 51.55 | 2000 | 51.63 | 2011 | 51.26 |
| 1990 | 51.52 | 2001 | 51.46 | 2012 | 51.25 |
| 1991 | 51.34 | 2002 | 51.47 | 2013 | 51.24 |
| 1992 | 51.05 | 2003 | 51.5 | 2014 | 51.23 |
| 1993 | 51.02 | 2004 | 51.52 | 2015 | 51.22 |
Step 202：Household energy consumption is screened per capita every year.
To residential energy consumption project and its measured value screen per capita every year, annual residential energy consumption per capita is rejected
Measurement missing values in amount；Measurement missing values every year per capita in household energy consumption are fitted；Turn to through screening and
Annual household energy consumption measured value per capita after fitting formats.
Firstly, carrying out missing values cleaning.Data are observed, its missing values ratio is calculated, determines the range of missing values.According to scarce
Mistake ratio and field importance, take different processing strategies.Feature high for importance, miss rate is low, by experience or
Professional knowledge estimation is filled；The feature high for importance, miss rate is high is calculated using other more complicated models and is mended
Entirely.Importance is high and miss rate is low, is supplemented by approximating method；Miss rate is high and importance is low, directly removes.
Secondly, carrying out Data Format Transform.The problem of not being aligned partially is arranged existing for data to importing, and has more column
The case where, carry out artificial treatment.
Step 202：Time series building.
First sample corresponding with residential energy consumption measured value per capita is established, k scale is carried out to the first sample
Partite transport calculates (k is the minimum differential operation times for making sequence meet stationarity requirement), obtains based on residential energy consumption item per capita
Purpose, meet stationarity requirement time series, establish with based on reality factors such as population, economy impact factor and it is described when
Between sequence sample set be the second sample；The order that model is determined according to Bayesian Information amount criterion is p=1, q=0.
Step 20:3：Regression analysis constructs combination forecasting, is predicted.
For regression analysis model, dependent variable is the Chinese per capita energy consumption in a certain year, and independent variable includes male
Property population proportion and time series variable, specifically, the present invention is using 1 as step-length, in order when per capita energy consumes
Between intercepted length is 11 in sequence part-time sequence, the sample of regression analysis is as shown in Figure 3.In the sample set of regression analysis
In select 3 samples at random as verifying collection, remaining sample is trained regression model as training set, arma modeling and respectively
Relative error of the regression analysis model on verifying collection is as shown in Figure 4.
Thus obtaining arma modeling, Support vector regression (linear kernel) and ridge regression is lesser three moulds of relative error
Type, relative error are respectively 1%, 4% and 4%, therefore by the prediction result weighted sum of above three model, are realized pair with this
The prediction of 2008 to 2015 household energy consumptions per capita.Fig. 5 indicates Support vector regression (linear kernel) and ridge regression
The mean value and standard deviation of mean value and standard deviation, verifying collection true value on verifying collection.
In order to further determine each Model Weight, arma modeling relative error, mean value, standard deviation are substituted into, are obtained by the present invention
Compatible degree to arma modeling is 97.1934463, obtains the compatible degree of Support vector regression (linear kernel function) and ridge regression
Respectively 4.22926306 and 3.291428455, the weight of final arma modeling is 0.9281788, Support vector regression (line
Property kernel function) weight be 0.04038865, the weight of ridge regression is 0.0314325.
Predicted value of the combination forecasting constructed by the present invention to 2008 to 2015 Chinese per capita energy consumptions
It is as shown in Figure 6 with true value.
If being chosen at verifying collects upper maximum three models of error, the relative error of predicted value is 50.3%；Divide if returning
The compatible degree of model is analysed by formulaIt calculates, the relative error of predicted value is 5.3%；If the weight of three models is 1/
3, the relative error of predicted value is 8.48%；Method proposed by the present invention, which can not only be screened effectively, is suitble to prediction domestic energy to disappear
Take the model of sequence, moreover it is possible to which the model to filter out distributes reasonable weight.It is the smallest that the present embodiment chooses the upper error of verifying collection
Three models are calculated the compatible degree of arma modeling and regression analysis model by formula 9 and 10 respectively, are obtained the opposite of predicted value and are missed
Difference is 4.25%, hence it is evident that better than the method for other building built-up patterns.
The residential energy consumption prediction technique based on ARMA and regression analysis that the embodiment of the present invention proposes, effectively reflects
Influence of the reality factor to time series, and combination forecasting can filter out properly according to the characteristics of time series itself
Single model, and reasonable weight is distributed for model, with high, the highly reliable and more stable beneficial effect of test accuracy.
Another exemplary embodiment of the application provides a kind of residential energy consumption based on ARMA and regression analysis
Forecasting system, the system include sequentially connected energy detecting device, processor and display unit.
Energy detecting device, for obtaining residential energy consumption project and its measured value.
Processor is for realizing the residential energy consumption prediction technique as described above based on ARMA and regression analysis；
Display unit, the prediction result for output processor.
What the embodiment of the present invention proposed is better adapted to based on ARMA and the residential energy consumption forecasting system of regression analysis
The characteristic of time series and can accurate description reality influence factor, situation can be changed over time to future life energy-consuming
It is predicted.
Above-mentioned, although the foregoing specific embodiments of the present invention is described with reference to the accompanying drawings, not protects model to the present invention
The limitation enclosed, those skilled in the art should understand that, based on the technical solutions of the present invention, those skilled in the art are not
Need to make the creative labor the various modifications or changes that can be made still within protection scope of the present invention.
Claims (10)
1. a kind of residential energy consumption prediction technique based on ARMA and regression analysis, characterized in that include the following steps：
Obtain residential energy consumption project and its measured value per capita；
First sample corresponding with residential energy consumption measured value per capita is established, time series is constructed；
According to Bayesian Information amount criterion, the order of arma modeling is determined, construct arma modeling；
It establishes with the impact factor of reality factor and the sample set of time series as the second sample；
Regression analysis is carried out to the second sample, obtains combination forecasting；
Prediction is combined to first sample using combination forecasting.
2. the residential energy consumption prediction technique according to claim 1 based on ARMA and regression analysis, characterized in that also
Including screening to residential energy consumption measured value per capita, the missing values in household energy consumption per capita are rejected；To per capita
Measurement missing values in residential energy consumption measured value are fitted, and are surveyed to the residential energy consumption per capita after screening and fitting
The step of magnitude formats.
3. the residential energy consumption prediction technique according to claim 1 based on ARMA and regression analysis, characterized in that institute
The construction method for stating time series is：
First sample corresponding with residential energy consumption measured value per capita is established, using first sample as initiation sequence；
The operation of k order difference is carried out to initiation sequence, obtain based on residential energy consumption project per capita and meets stationarity requirement
Time series.
4. the residential energy consumption prediction technique according to claim 3 based on ARMA and regression analysis, characterized in that if
Initiation sequence is not stationary sequence, then needs to do initiation sequence k order difference operation, and k makes sequence meet stationarity requirement
Minimum differential operation times；If initiation sequence meets stationarity requirement, do not need to do calculus of differences, k value is 0 at this time.
5. the residential energy consumption prediction technique according to claim 1 based on ARMA and regression analysis, characterized in that institute
Stating the method for carrying out regression analysis to the second sample includes：
Using the impact factor of reality factor and time series as the independent variable of regression analysis, by taking for time series a certain moment
It is worth and carries out Function Fitting as the dependent variable of regression analysis, obtains several regression analysis samples, establish regression analysis sample set；
3 samples are randomly selected from regression analysis sample set as verifying collection, remaining sample is as training set to regression analysis
Model is trained；
Compare relative error of each regression analysis model on verifying collection, takes first three the smallest model construction group of relative error
Prediction model is closed, the combination forecasting includes arma modeling, Support vector regression model and ridge regression model.
6. the residential energy consumption prediction technique according to claim 5 based on ARMA and regression analysis, characterized in that press
Sequence intercepted length in time series is value of the partial sequence of n as time series independent variable, and step-length 1, n is to return to divide
The quantity of time series independent variable in analysis.
7. the residential energy consumption prediction technique according to claim 1 based on ARMA and regression analysis, characterized in that institute
It states and includes using the method that combination forecasting is combined prediction to first sample：
First sample is predicted using arma modeling, Support vector regression model and ridge regression model, obtains three moulds
The prediction result of type is distributed the prediction result of three models to certain weight and is weighted and averaged, obtains final prediction knot
Fruit.
8. the residential energy consumption prediction technique according to claim 7 based on ARMA and regression analysis, characterized in that institute
The calculation method for stating weight is：
Wherein, τiIndicate the model compatible degree containing mean value and standard deviation, the value of i is 1,2,3, ωiIndicate the weight of model.
9. the residential energy consumption prediction technique according to claim 8 based on ARMA and regression analysis, characterized in that
The calculating function of the model compatible degree containing mean value and standard deviation is：
Wherein, σ and μ respectively indicates the standard deviation and mean value of verifying collection true value；ξiIndicate model i missing on verifying collection relatively
Difference, the value of i are 1,2,3；σiIndicate the standard deviation of predicted value obtained by model i predicts verifying collection, μiIndicate model i
Predict to verifying collection the mean value of obtained predicted value.
10. a kind of residential energy consumption forecasting system based on ARMA and regression analysis, characterized in that including：
Energy detecting device, for obtaining residential energy consumption project and its measured value；And
Processor is connected with energy-consuming detection device, for realizing it is of any of claims 1-9 based on ARMA and
The residential energy consumption prediction technique of regression analysis；And
Display unit is connected with processor, the prediction result for output processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810609136.0A CN108830417B (en) | 2018-06-13 | 2018-06-13 | ARMA (autoregressive moving average) and regression analysis based life energy consumption prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810609136.0A CN108830417B (en) | 2018-06-13 | 2018-06-13 | ARMA (autoregressive moving average) and regression analysis based life energy consumption prediction method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108830417A true CN108830417A (en) | 2018-11-16 |
| CN108830417B CN108830417B (en) | 2021-01-08 |
Family
ID=64141426
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810609136.0A Expired - Fee Related CN108830417B (en) | 2018-06-13 | 2018-06-13 | ARMA (autoregressive moving average) and regression analysis based life energy consumption prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108830417B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109854242A (en) * | 2019-01-08 | 2019-06-07 | 浙江大学 | A kind of coal mining machine roller automatic Prediction system based on chaology |
| CN110991696A (en) * | 2019-11-04 | 2020-04-10 | 广州丰石科技有限公司 | Method for filling missing of passenger flow data |
| CN111310960A (en) * | 2018-12-12 | 2020-06-19 | 华北电力大学扬中智能电气研究中心 | Energy consumption prediction method and device |
| CN115358157A (en) * | 2022-10-20 | 2022-11-18 | 正大农业科学研究有限公司 | Prediction analysis method and device for litter size of individual litters and electronic equipment |
| CN116976860A (en) * | 2023-08-14 | 2023-10-31 | 上海工业自动化仪表研究院有限公司 | Implementation method and application of algorithm for improving operation and maintenance capabilities of welding workstation |
-
2018
- 2018-06-13 CN CN201810609136.0A patent/CN108830417B/en not_active Expired - Fee Related
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111310960A (en) * | 2018-12-12 | 2020-06-19 | 华北电力大学扬中智能电气研究中心 | Energy consumption prediction method and device |
| CN109854242A (en) * | 2019-01-08 | 2019-06-07 | 浙江大学 | A kind of coal mining machine roller automatic Prediction system based on chaology |
| CN110991696A (en) * | 2019-11-04 | 2020-04-10 | 广州丰石科技有限公司 | Method for filling missing of passenger flow data |
| CN115358157A (en) * | 2022-10-20 | 2022-11-18 | 正大农业科学研究有限公司 | Prediction analysis method and device for litter size of individual litters and electronic equipment |
| CN116976860A (en) * | 2023-08-14 | 2023-10-31 | 上海工业自动化仪表研究院有限公司 | Implementation method and application of algorithm for improving operation and maintenance capabilities of welding workstation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108830417B (en) | 2021-01-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108830417A (en) | A kind of residential energy consumption prediction technique and system based on ARMA and regression analysis | |
| CN115276006B (en) | Load prediction method and system for power integration system | |
| CN105117602B (en) | A kind of metering device running status method for early warning | |
| CN107992968B (en) | Electric energy meter metering error prediction method based on integrated time series analysis technology | |
| CN109145949A (en) | Non-intrusive electrical load monitoring and decomposition method and system based on integrated study | |
| CN109858758A (en) | A kind of the combination weighting appraisal procedure and system of distribution network electric energy quality | |
| CN109472403B (en) | Medium-and-long-term runoff forecasting method integrating empirical mode decomposition and remote correlation | |
| CN106533750A (en) | System and method for predicting non-steady application user concurrency in cloud environment | |
| CN112149873A (en) | Low-voltage transformer area line loss reasonable interval prediction method based on deep learning | |
| CN110335168A (en) | Method and system based on GRU optimization power information acquisition terminal fault prediction model | |
| CN111008790A (en) | Hydropower station group power generation electric scheduling rule extraction method | |
| CN111178594A (en) | Thermal power generating unit peak regulation capability prediction method, device and system | |
| CN109754122A (en) | A kind of Numerical Predicting Method of the BP neural network based on random forest feature extraction | |
| CN112417764A (en) | K nearest neighbor regression prediction method for boiler special equipment steam flow prediction | |
| CN110321493A (en) | A kind of abnormality detection of social networks and optimization method, system and computer equipment | |
| CN111126865B (en) | Technology maturity judging method and system based on technology big data | |
| CN110751176A (en) | Lake water quality prediction method based on decision tree algorithm | |
| CN113554213A (en) | Natural gas demand prediction method, system, storage medium and equipment | |
| CN115438897A (en) | Industrial process product quality prediction method based on BLSTM neural network | |
| CN109523077B (en) | Wind power prediction method | |
| CN111027841A (en) | Low-voltage transformer area line loss calculation method based on gradient lifting decision tree | |
| CN109491709B (en) | Software code controllability comprehensive evaluation method based on AHP and neural network | |
| CN109116300B (en) | Extreme learning positioning method based on insufficient fingerprint information | |
| He et al. | A study on prediction of customer churn in fixed communication network based on data mining | |
| CN115310727B (en) | Building cooling, heating and power load prediction method and system based on transfer learning |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210108 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |