CN110570023B - Short-term commercial power load prediction method based on SARIMA-GRNN-SVM - Google Patents
Short-term commercial power load prediction method based on SARIMA-GRNN-SVM Download PDFInfo
- Publication number
- CN110570023B CN110570023B CN201910758297.0A CN201910758297A CN110570023B CN 110570023 B CN110570023 B CN 110570023B CN 201910758297 A CN201910758297 A CN 201910758297A CN 110570023 B CN110570023 B CN 110570023B
- Authority
- CN
- China
- Prior art keywords
- prediction
- model
- commercial power
- power load
- svm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000012549 training Methods 0.000 claims abstract description 28
- 241000728173 Sarima Species 0.000 claims abstract description 14
- 238000005457 optimization Methods 0.000 claims abstract description 9
- 238000012795 verification Methods 0.000 claims abstract description 4
- 230000001932 seasonal effect Effects 0.000 claims description 15
- 210000002569 neuron Anatomy 0.000 claims description 8
- 238000002790 cross-validation Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 238000012546 transfer Methods 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000000780 augmented Dickey–Fuller test Methods 0.000 claims description 2
- 125000004122 cyclic group Chemical group 0.000 claims description 2
- 230000001419 dependent effect Effects 0.000 claims description 2
- 238000009499 grossing Methods 0.000 claims description 2
- 238000012886 linear function Methods 0.000 claims description 2
- 238000012216 screening Methods 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 claims 1
- 230000000737 periodic effect Effects 0.000 abstract description 5
- 230000008859 change Effects 0.000 abstract description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Tourism & Hospitality (AREA)
- Mathematical Optimization (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Pure & Applied Mathematics (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Business, Economics & Management (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Entrepreneurship & Innovation (AREA)
- Water Supply & Treatment (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Public Health (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
Abstract
The invention relates to a short-term commercial power load prediction method based on SARIMA-GRNN-SVM, which comprises the following steps: obtaining influence factors of commercial power load fluctuation through analysis of a load curve; constructing a single prediction model for the commercial power load time sequence prediction and the multi-factor regression prediction; constructing an SVM model, carrying out parameter optimization and training on the SVM model by utilizing a training sample, and carrying out parameter optimization on the SVM model by a grid search and cross verification method; and inputting the predicted value of the predicted day obtained by the SARIMA model and the GRNN model into the trained SVM model to obtain the predicted value of the commercial power load of the predicted day. The method solves the problem that the single prediction model cannot comprehensively consider the periodic change of the commercial load and influence factors to cause the prediction result to be easy to generate larger errors, and improves the prediction accuracy and the robustness.
Description
Technical Field
The invention belongs to the technical field of electric vehicle parts, and particularly relates to a short-term commercial power load prediction method based on SARIMA-GRNN-SVM.
Background
Short-term power load prediction is the basis for power supply enterprises to make power generation and scheduling plans and ensure stable and reliable operation of a power system. Under the background of rapid development of the power spot market, the short-term power load prediction can enable a power supply enterprise to master the power consumption situation in time, reasonable power price is formulated, a power selling strategy is flexibly adjusted, and the trading efficiency and economic benefits of the power spot market are improved.
In recent years, commercial power loads have a remarkable rising trend due to the adjustment of industrial structures in China and the improvement of living standards of residents. Compared with industrial load and residential load, the commercial power load has stronger controllability and more obvious peak-valley difference, and is suitable for participating in the trade of the power spot market. It can be seen that the short-term prediction of commercial power load has important significance for the safe and economical operation of the power grid and the smooth development of the power spot market. The current research on the prediction of the commercial power load is less, the prediction of the medium-long-term commercial power load is mostly carried out, the influence of factors such as weather and temperature on the commercial power load is not considered, and the requirements of the power spot market on the short-term commercial power load prediction are difficult to meet.
Therefore, based on the problems, the short-term commercial power load prediction method based on the SARIMA-GRNN-SVM, which solves the problem that a single prediction model cannot comprehensively consider the periodic change of commercial load and influence factors to cause a large error in a prediction result, has important practical significance.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a short-term commercial power load prediction method based on SARIMA-GRNN-SVM, which solves the problem that a single prediction model cannot comprehensively consider the periodic change of commercial load and influence factors to cause a large error in a prediction result.
The invention solves the technical problems by adopting the following technical scheme:
a short-term commercial power load prediction method based on SARIMA-GRNN-SVM, comprising the steps of:
s1, collecting recent local commercial power load data, drawing an average load curve in the period of time or/and a load curve in a certain period of time, and analyzing the load curve to obtain influence factors of commercial power load fluctuation;
s2, constructing a single prediction model for the commercial power load time sequence prediction and the multi-factor regression prediction;
s201, establishing an SARIMA model to realize the prediction of a commercial power load time sequence;
and establishing an SARIMA model, wherein the formula is as follows:
wherein:
θ(B)＝1-θ 1 B-…-θ q B q
Φ(B s )＝1-Φ 1 B S -…-Φ P B S
Θ(B s )＝1-Θ 1 B S -…-Θ Q B S
where x (t) is the commercial electrical load time series,
is a smooth time sequence after difference; b represents a hysteresis operator, < >>
Representing a difference operator; />
For seasonal autoregressive model, < >>
Is an autoregressive polynomial of order p, +.>
Is a non-seasonal autoregressive parameter; phi (B) s ) Representing seasonal autoregressive polynomials, Φ 1 ，Φ 2 ，…Φ P The parameter is the P-order seasonal autoregressive parameter; Θ (B) S ) Theta (B) represents a season shiftA moving average model, wherein θ (B) represents a q-order moving average polynomial, θ 1 ，θ 2 ，…θ q For non-seasonal moving average parameter, Θ (B S ) Representing a season moving average polynomial, Θ 1 ，Θ 2 ，…Θ Q For the Q-order seasonal moving average parameter, ε (t) is Gaussian noise;
establishing an SARIMA model, taking a historical load and a time sequence of a continuous time before a predicted day as input, adopting an Augmented Dickey-Fuller test to stabilize an original time sequence, determining a differential order, judging possible model parameters through an autocorrelation coefficient and a partial autocorrelation coefficient graph, screening by using a red pool information amount criterion and a Bayesian information criterion, and outputting a predicted value of a predicted day;
s202, adopting GRNN model to conduct multi-factor regression prediction on commercial power load
Constructing a GRNN model, taking influence factors of commercial power load fluctuation determined in the step S1 of predicting a period of continuous time before the day and the day-to-day moment load of the previous same date type as inputs, and utilizing a training method of circulating cross verification, so that the output predicted value is optimal;
s3, constructing an SVM model, carrying out parameter optimization and training on the SVM model by using a training sample, and carrying out parameter optimization on the SVM model by using a grid search and cross validation method; the method comprises the steps of selecting a single prediction model predicted value and a load actual value of the first three working days as training samples for working day prediction; aiming at rest day prediction, a single prediction model prediction value and a load actual value of two days before the prediction day and the last same date type day are selected as training samples; the single prediction model prediction value adopts the prediction value obtained by utilizing the SARIMA model and the GRNN model in the steps S201 and S202;
s4, inputting the prediction value of the prediction day obtained by the SARIMA model and the GRNN model in the steps S201 and S202 into the SVM model trained in the step S3, and obtaining the commercial power load prediction value of the prediction day.
Further, in the step S3, before training the SVM model, the prediction date in the training sample is judged by using the rada criterion, and if the prediction result of the prediction date is an abnormal value, the prediction date is removed, and the training sample is forward and forward for one day.
Further, the local commercial power load data collected in step S1 is at least one year in the near future.
Further, the GRNN model in the step S202 is composed of an input layer, a mode layer, a summation layer and an output layer; each neuron of the input layer directly transmits the input data to the mode layer, the mode layer transmits the samples to the summation layer through a radial transfer function, the summation layer sums through two algorithms and transmits the data to the output layer, and the output layer outputs the result by adopting a linear function.
Further, the radial transfer function of the mode layer is:
wherein X is a network input variable; x is X i A learning sample corresponding to the ith neuron; sigma represents a smoothing factor, where the optimal sigma is found using a training method of cyclic cross validation.
Further, the summation layer utilizes two summation modes: one is to calculate a weighted sum of the outputs of the neurons of the pattern layer; the other is to calculate the sum of the outputs of the neurons of the pattern layer; the two formulas are respectively as follows:
wherein j=1, 2, …, m; h is a ij Is the j-th element in the dependent variable of the i-th training sample.
The invention has the advantages and positive effects that:
the SARIMA model adopted in the invention can describe the sequence showing the periodic characteristic time by seasonal variation, and can simultaneously consider the correlation between data and the interference of random factors when processing the time sequence problem, so that the prediction precision is higher; the GRNN model is an improved algorithm of the radial base network, and has high calculation accuracy and short required time; the SVM model adopted in the invention utilizes training sample data to construct an optimal hyperplane in a high-dimensional space, has stronger generalization capability and needs less samples; the method solves the problem that the single prediction model cannot comprehensively consider the periodic change of the commercial load and influence factors to cause the prediction result to be easy to generate larger errors, and improves the prediction accuracy and the robustness.
Drawings
The technical solution of the present invention will be described in further detail below with reference to the accompanying drawings and examples, but it should be understood that these drawings are designed for the purpose of illustration only and thus are not limiting the scope of the present invention. Moreover, unless specifically indicated otherwise, the drawings are intended to conceptually illustrate the structural configurations described herein and are not necessarily drawn to scale.
FIG. 1 is a graph of average daily load of a shopping mall 2017 provided in example 1 of the present invention;
FIG. 2 (a) is a plot of average load over time for business hours for a certain shopping mall 2017, month 1, provided in example 1 of the present invention;
FIG. 2 (b) is a plot of average load over time for business hours for a leisure center 2017, month 4, for a shopping mall according to example 1 of the present invention;
FIG. 2 (c) is a plot of average load over time for a business hour of 7 months in 2017 of a shopping mall provided by example 1 of the present invention;
FIG. 2 (d) is a plot of average load over time for a business hour of 10 months in 2017 of a shopping mall provided by example 1 of the present invention;
FIG. 3 is a predicted daily business hours load actual value and a single model predicted value provided in example 1 of the present invention;
FIG. 4 is a predicted daily business hour load actual value and SVM model predicted value provided in embodiment 1 of the present invention;
Detailed Description
It should be noted that, in the case of no conflict, the embodiments and features in the embodiments may be combined with each other.
Example 1
FIG. 1 is a graph of average daily load of a shopping mall 2017 provided in example 1 of the present invention; FIG. 2 (a) is a plot of average load over time for business hours for a certain shopping mall 2017, month 1, provided in example 1 of the present invention; FIG. 2 (b) is a plot of average load over time for business hours for a leisure center 2017, month 4, for a shopping mall according to example 1 of the present invention; FIG. 2 (c) is a plot of average load over time for a business hour of 7 months in 2017 of a shopping mall provided by example 1 of the present invention; FIG. 2 (d) is a plot of average load over time for a business hour of 10 months in 2017 of a shopping mall provided by example 1 of the present invention; FIG. 3 is a predicted daily business hours load actual value and a single model predicted value provided in example 1 of the present invention; FIG. 4 is a predicted daily business hour load actual value and SVM model predicted value provided in embodiment 1 of the present invention;
according to fig. 1,2 (a), 2 (b), 2 (c), 2 (d): the influence factors of commercial power load fluctuation are obtained through analysis of the load curve: weather, maximum temperature, minimum temperature, date type;
business power load prediction is carried out on business hours from 5 days of 7 months to 10 days of 7 months of 2017 of a certain large leisure shopping center. A SARIMA and GRNN single prediction model is built by taking the actual value of the commercial power load of each predicted business hours of 28 days before the day and the weather, the highest temperature, the lowest temperature, the date type and the day-time load of the previous co-date type of 14 days before the day as inputs, and the prediction results are shown in figure 3.
As can be seen from fig. 3, the predicted value of each single prediction model is substantially identical to the actual value of the commercial power load, but the accuracy of the predicted result of the single prediction model fluctuates greatly, so that the robustness is poor.
The method of the invention is used for carrying out parameter optimization and training on the SVM: parameter optimization of the SVM model is carried out through a grid search and cross verification method; the method comprises the steps of selecting a single prediction model predicted value and a load actual value of the first three working days as training samples for working day prediction; aiming at rest day prediction, a single prediction model prediction value and a load actual value of two days before the prediction day and the last same date type day are selected as training samples; the single prediction model prediction value adopts a prediction value obtained by utilizing a SARIMA model and a GRNN model;
after parameter optimization and training are carried out on the SVM, a single prediction model prediction result from 5 days of 7 months of 2017 to 10 days of 7 months is input into the SVM, and a combined prediction model prediction value is obtained, as shown in figure 4.
As can be obtained from fig. 4, the combined prediction model provided by the invention better fits the variation trend of commercial power load, the prediction result is similar to the actual value, the difference between the predicted value and the actual value is larger only in 7 months and 6 days, and after detection by the rada criterion, the abnormal values are 5, so that the data of 7 months and 6 days are not adopted in the subsequent prediction. Specifically, the time errors e of three prediction days in the training sample are used for 1 ，e 2 ，…，e 36 Calculating the average value E and the standard deviation sigma of the judgment standard; respectively calculating the error v of different times of the judged day 1 ，v 2 ，…v 12 Difference V from the judgment reference average E n (n=1, 2, … 12); if |V n The predicted value at that time is considered to be an outlier if 3 sigma; if the predicted daily outliers exceed 5, then the subsequent predictions will not take the daily data any more.
In order to measure the accuracy and robustness of the prediction model, the embodiment adopts Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) as error criteria, and the error criteria are respectively shown in the following formulas:
where n is the number of predicted times.
Average absolute percent error Em and root mean square error Er for single and combined models from 2017, 7, 5, and 7, 10 are obtained using the above formulas, as shown in table 1:
table 1 error comparison of combined prediction model and single prediction model
As can be seen from table 1, both MAPE and RMSE of the combined prediction model provided in this embodiment are smaller than those of the single prediction model, which indicates that the combined prediction model combines the advantages of the single prediction model, extracts more comprehensive information, and has better prediction accuracy and robustness than those of the single prediction model.
Example 2
Business power load predictions are made for business hours of 1 month 13 to 1 month 19 of another business complex 2017. The combined model and single model error pairs are shown in table 2.
Table 2 error comparison of combined prediction model and single prediction model
As can be seen from Table 2, since the winter heating load is smaller than the summer cooling load, the fluctuation of the winter commercial power load is smaller than the summer load, and the prediction accuracy of the single prediction model and the combined prediction model is better than that of the summer prediction. And the MAPE and the RMSE of the combined prediction model provided by the embodiment are smaller than those of each single prediction model, the prediction precision and the robustness are better than those of the single prediction model, and the requirements of short-term prediction precision and period of commercial power load can be met.
The foregoing examples illustrate the invention in detail, but are merely preferred embodiments of the invention and are not to be construed as limiting the scope of the invention. All equivalent changes and modifications within the scope of the present invention are intended to be covered by the present invention.
Claims (5)
1. A short-term commercial power load prediction method based on a SARIMA-GRNN-SVM, characterized by: the method comprises the following steps:
s1, collecting recent local commercial power load data, drawing an average load curve in the period of time or/and a load curve in a certain period of time, and analyzing the load curve to obtain influence factors of commercial power load fluctuation;
s2, constructing a single prediction model for the commercial power load time sequence prediction and the multi-factor regression prediction;
s201, establishing a seasonal autoregressive differential moving average SARIMA model to realize the prediction of a commercial power load time sequence;
and establishing an SARIMA model, wherein the formula is as follows:
wherein:
θ(B)=1-θ 1 B-…-θ q B q
Φ(B s )＝1-Φ 1 B S -…-Φ P B S
Θ(B s )＝1-Θ 1 B S -…-Θ Q B S
where x (t) is the commercial electrical load time series,
is a smooth time sequence after difference; b represents a hysteresis operator, < >>
Representing a difference operator; />
For seasonal autoregressive model, < >>
Is an autoregressive polynomial of the order p,
is a non-seasonal autoregressive parameter; phi (B) s ) Representing seasonal autoregressive polynomials, Φ 1 ，Φ 2 ，…Φ P The parameter is the P-order seasonal autoregressive parameter; Θ (B) S ) θ (B) represents a seasonal moving average model, where θ (B) represents a q-order moving average polynomial, θ 1 ，θ 2 ，…θ q For non-seasonal moving average parameter, Θ (B S ) Representing a season moving average polynomial, Θ 1 ，Θ 2 ，…Θ Q For the Q-order seasonal moving average parameter, ε (t) is Gaussian noise;
establishing an SARIMA model, taking a historical load and a time sequence of a continuous time before a predicted day as input, adopting an Augmented Dickey-Fuller test to stabilize an original time sequence, determining a differential order, judging possible model parameters through an autocorrelation coefficient and a partial autocorrelation coefficient graph, screening by using a red pool information amount criterion and a Bayesian information criterion, and outputting a predicted value of a predicted day;
s202, performing multi-factor regression prediction on commercial power load by adopting generalized regression neural network GRNN model
Constructing a GRNN model, taking influence factors of commercial power load fluctuation determined in the step S1 of predicting a period of continuous time before the day and the day-to-day moment load of the previous same date type as inputs, and utilizing a training method of circulating cross verification, so that the output predicted value is optimal;
the GRNN model consists of an input layer, a mode layer, a summation layer and an output layer; each neuron of the input layer directly transmits input data to a mode layer, the mode layer transmits samples to a summation layer through a radial transfer function, the summation layer sums through two algorithms and transmits the data to an output layer, and the output layer outputs a result by adopting a linear function;
s3, constructing an SVM model, carrying out parameter optimization and training on the SVM model by using a training sample, and carrying out parameter optimization on the SVM model by using a grid search and cross validation method; the method comprises the steps of selecting a single prediction model predicted value and a load actual value of the first three working days as training samples for working day prediction; aiming at rest day prediction, a single prediction model prediction value and a load actual value of two days before the prediction day and the last same date type day are selected as training samples; the single prediction model prediction value adopts the prediction value obtained by utilizing the SARIMA model and the GRNN model in the steps S201 and S202;
s4, inputting the prediction value of the prediction day obtained by the SARIMA model and the GRNN model in the steps S201 and S202 into the SVM model trained in the step S3, and obtaining the commercial power load prediction value of the prediction day.
2. A method of short term commercial power load prediction based on SARIMA-GRNN-SVM as defined in claim 1, wherein: in the step S3, before training the SVM model, the prediction date in the training sample is judged by using the rada criterion, and if the prediction result of the prediction date is an abnormal value, the prediction date is removed, and the training sample is forward and forward for one day.
3. A method of short term commercial power load prediction based on SARIMA-GRNN-SVM as defined in claim 1, wherein: the local commercial power load data collected in step S1 is at least one year in the near future.
4. A method of short term commercial power load prediction based on SARIMA-GRNN-SVM as defined in claim 1, wherein: the radial transfer function of the mode layer is:
wherein X is a network input variable; x is X i A learning sample corresponding to the ith neuron; sigma represents a smoothing factor, where the optimal sigma is found using a training method of cyclic cross validation.
5. A method for short-term commercial power load prediction based on a SARIMA-GRNN-SVM as defined in claim 4, wherein: the summation layer utilizes the following two summation modes: one is to calculate a weighted sum of the outputs of the neurons of the pattern layer; the other is to calculate the sum of the outputs of the neurons of the pattern layer; the two formulas are respectively as follows:
wherein j=1, 2, …, m; h is a ij Is the j-th element in the dependent variable of the i-th training sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910758297.0A CN110570023B (en) | 2019-08-16 | 2019-08-16 | Short-term commercial power load prediction method based on SARIMA-GRNN-SVM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910758297.0A CN110570023B (en) | 2019-08-16 | 2019-08-16 | Short-term commercial power load prediction method based on SARIMA-GRNN-SVM |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110570023A CN110570023A (en) | 2019-12-13 |
| CN110570023B true CN110570023B (en) | 2023-05-16 |
Family
ID=68775658
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910758297.0A Active CN110570023B (en) | 2019-08-16 | 2019-08-16 | Short-term commercial power load prediction method based on SARIMA-GRNN-SVM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110570023B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111382906B (en) * | 2020-03-06 | 2024-02-27 | 南京工程学院 | Power load prediction method, system, equipment and computer readable storage medium |
| CN111815051B (en) * | 2020-07-06 | 2022-09-23 | 安徽建筑大学 | GRNN photovoltaic power generation prediction method considering weather influence factors |
| CN112116144B (en) * | 2020-09-15 | 2021-04-13 | 山东科技大学 | Regional power distribution network short-term load prediction method |
| CN112465250B (en) * | 2020-12-08 | 2024-08-20 | 深圳供电局有限公司 | Power load prediction method, device, computer equipment and storage medium |
| CN117272121B (en) * | 2023-11-21 | 2024-03-12 | 江苏米特物联网科技有限公司 | Hotel load influence factor quantitative analysis method based on Deep SHAP |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015172560A1 (en) * | 2014-05-16 | 2015-11-19 | 华南理工大学 | Central air conditioner cooling load prediction method based on bp neural network |
| CN106779129A (en) * | 2015-11-19 | 2017-05-31 | 华北电力大学（保定） | A kind of Short-Term Load Forecasting Method for considering meteorologic factor |
| CN108229754A (en) * | 2018-01-31 | 2018-06-29 | 杭州电子科技大学 | Short-term load forecasting method based on similar day segmentation and LM-BP networks |
| CN109657844A (en) * | 2018-11-28 | 2019-04-19 | 国电南瑞科技股份有限公司 | Power Short-Term Load Forecasting method and apparatus |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030182250A1 (en) * | 2002-03-19 | 2003-09-25 | Mohammad Shihidehpour | Technique for forecasting market pricing of electricity |
-
2019
- 2019-08-16 CN CN201910758297.0A patent/CN110570023B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015172560A1 (en) * | 2014-05-16 | 2015-11-19 | 华南理工大学 | Central air conditioner cooling load prediction method based on bp neural network |
| CN106779129A (en) * | 2015-11-19 | 2017-05-31 | 华北电力大学（保定） | A kind of Short-Term Load Forecasting Method for considering meteorologic factor |
| CN108229754A (en) * | 2018-01-31 | 2018-06-29 | 杭州电子科技大学 | Short-term load forecasting method based on similar day segmentation and LM-BP networks |
| CN109657844A (en) * | 2018-11-28 | 2019-04-19 | 国电南瑞科技股份有限公司 | Power Short-Term Load Forecasting method and apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110570023A (en) | 2019-12-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110570023B (en) | Short-term commercial power load prediction method based on SARIMA-GRNN-SVM | |
| Al Mamun et al. | A comprehensive review of the load forecasting techniques using single and hybrid predictive models | |
| Deihimi et al. | Application of echo state networks in short-term electric load forecasting | |
| Chen et al. | Online 24-h solar power forecasting based on weather type classification using artificial neural network | |
| CN111260136A (en) | Building short-term load prediction method based on ARIMA-LSTM combined model | |
| Moradzadeh et al. | Heating and cooling loads forecasting for residential buildings based on hybrid machine learning applications: A comprehensive review and comparative analysis | |
| Dong et al. | Applying the ensemble artificial neural network-based hybrid data-driven model to daily total load forecasting | |
| Sengar et al. | Ensemble approach for short term load forecasting in wind energy system using hybrid algorithm | |
| Abdulrahman et al. | A review on deep learning with focus on deep recurrent neural network for electricity forecasting in residential building | |
| CN112418526A (en) | Comprehensive energy load control method and device based on improved deep belief network | |
| CN110380444B (en) | Capacity planning method for distributed wind power orderly access to power grid under multiple scenes based on variable structure Copula | |
| Al-Rousan et al. | Assessment of predicting hourly global solar radiation in Jordan based on Rules, Trees, Meta, Lazy and Function prediction methods | |
| Selvi et al. | Investigation of performance of electric load power forecasting in multiple time horizons with new architecture realized in multivariate linear regression and feed-forward neural network techniques | |
| Li et al. | Short-term electricity consumption prediction for buildings using data-driven swarm intelligence based ensemble model | |
| Lauret et al. | Nonlinear models for short-time load forecasting | |
| Tian et al. | An adaptive ensemble predictive strategy for multiple scale electrical energy usages forecasting | |
| Li et al. | Estimate and characterize PV power at demand-side hybrid system | |
| Safari et al. | Multi-term electrical load forecasting of smart cities using a new hybrid highly accurate neural network-based predictive model | |
| Kampouropoulos et al. | A combined methodology of adaptive neuro-fuzzy inference system and genetic algorithm for short-term energy forecasting | |
| CN111222700A (en) | Day-ahead electricity price probability prediction method based on dynamic network quantile model | |
| Akpinar et al. | Forecasting natural gas consumption with hybrid neural networks—Artificial bee colony | |
| Souhe et al. | A hybrid model for forecasting the consumption of electrical energy in a smart grid | |
| Zhang et al. | Load Prediction Based on Hybrid Model of VMD‐mRMR‐BPNN‐LSSVM | |
| Bhandari et al. | Short term electric load forecasting of kathmandu valley of nepal using artificial neural network | |
| Xie et al. | A multi-information fusion model for short term load forecasting of an architectural complex considering spatio-temporal characteristics |
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 |