CN108808657B - Short-term prediction method for power load - Google Patents
Short-term prediction method for power load Download PDFInfo
- Publication number
- CN108808657B CN108808657B CN201810542186.1A CN201810542186A CN108808657B CN 108808657 B CN108808657 B CN 108808657B CN 201810542186 A CN201810542186 A CN 201810542186A CN 108808657 B CN108808657 B CN 108808657B
- Authority
- CN
- China
- Prior art keywords
- power load
- load data
- brownian motion
- fractional brownian
- historical
- 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
Images
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
Abstract
The invention relates to a short-term prediction method of a power load, which comprises the following steps: (1) obtaining historical electricityA force load data sequence, calculating the Hurst index H of the historical power load data sequence; (2) establishing a fractional Brownian motion model for predicting the power load based on the Hurst index H; (3) carrying out global optimization on the Hurst index H in the fractional Brownian motion model to obtain the optimal value H of the Hurst indexgbestFurther obtaining a fractional Brownian motion optimization model; (4) and predicting the power load data by using a fractional Brownian motion optimization model. Compared with the prior art, the method can predict the short-term non-steady power load data with high precision.
Description
Technical Field
The invention relates to a power load prediction method, in particular to a power load short-term prediction method.
Background
The power load prediction is an important link of the operation of a power system and is an important content of power dispatching. According to power scheduling, an electric power system operator can determine the operation time of the power grid and reduce potential loss, so that accurate power load prediction is helpful for the operator to grasp future power development trend and better schedule the power grid.
At present, many traditional power load methods exist, a Grey Model (GM) is widely applied to power load prediction, however, the accuracy of power load prediction is often influenced by various factors, and GM exponential growth rules cannot process the factors, so that a reasonable prediction effect is obtained. Synthetic autoregressive moving average (ARIMA) was successfully used for estimated power demand, but fails to address multivariate ARIMA and heteroscedasticity related problems. In recent years, an error Back Propagation (BP) algorithm is widely applied to power prediction, but has the defects of low convergence rate, easy falling into a local space and the like. The use of artificial neural networks for power load prediction ensures accuracy, however, artificial neural networks also have drawbacks, such as problems with overfitting and large training examples required. Support vector regression (SVM) machines have better prediction results than neural networks, but have a complex computational process.
At present, a Particle Swarm Optimization (PSO) algorithm is often adopted to solve the optimization problem, the PSO algorithm has an inherent defect, and because the PSO algorithm and the moving speed of each particle in a search space search the optimal and global optimal of each particle, the PSO algorithm is easy to fall into a local optimal solution. Due to the fact that the quantum theory of the particles is considered, a QPSO algorithm is utilized, and the velocity vector is not needed to obtain the global optimal solution. Therefore, one can avoid falling into a local optimal solution, and therefore, the optimization effect of the QPSO algorithm is better than that of the particle swarm optimization algorithm based on Newton motion. Herein, we predict power demand using QPSO optimization model parameters.
Disclosure of Invention
The present invention is directed to a method for short-term prediction of electrical load, which overcomes the above-mentioned drawbacks of the prior art.
The purpose of the invention can be realized by the following technical scheme:
a method for short-term prediction of an electrical load, the method comprising the steps of:
(1) acquiring a historical power load data sequence, and calculating a Hurst index H of the historical power load data sequence;
(2) establishing a fractional Brownian motion model for predicting the power load based on the Hurst index H;
(3) carrying out global optimization on the Hurst index H in the fractional Brownian motion model to obtain the optimal value H of the Hurst indexgbestFurther obtaining a fractional Brownian motion optimization model;
(4) and predicting the power load data by using a fractional Brownian motion optimization model.
And (3) acquiring the hestert index H of the power load data sequence in the step (1) by a re-standard range analysis method.
The fractional brownian motion model in the step (2) is specifically as follows:
for historical power load data series yt,t=0,1,2...n},ytRepresenting historical power load data at time t, the fractional brownian motion model is as follows:
yt+1=yt+uytΔt+σytw1(t)(Δt)H+λytw2(t)(Δt)2H,
wherein, yt+1Representing power load data at time t +1, Δ t representing the time interval between two adjacent historical power load data, w1(t) and w2(t) is obedience independentNormal parameters of normal distribution;
wherein, E represents the mathematical expectation,T=[0,1,2,…,n]y1 'is the transpose of y1, and T' is the transpose of T;
in the step (3), a quantum behavior particle swarm optimization method is adopted to carry out global optimization on the Hurst index H to obtain the optimal value H of the Hurst indexgbest。
Before the step (4), the method further comprises the following steps: changing the number of historical data in the historical power load data sequence and respectively using the historical data as a training sample, repeatedly executing the steps (1) to (3) on each training sample to respectively obtain corresponding fractional Brownian motion optimization models, comparing the prediction result errors of the fractional Brownian motion optimization models, and selecting the fractional Brownian motion optimization model with the minimum prediction result error to complete the prediction of the power load data in the step (4).
And (3) judging the Hurst index H, if the H is more than 0.5 and less than 1, the historical power load data sequence has long correlation, continuing to execute the step (2), and if not, ending.
Compared with the prior art, the invention has the following advantages:
(1) according to the method, the historical power load data can be used for predicting the future power load data, the scheduling workload of scheduling personnel is effectively reduced, the prediction effect is more accurate than the result of manual estimation, and large-scale enterprises and power departments can reduce the gateway flow and reduce the power consumption cost on the basis of the short-term prediction result according to the load tracking capability of the self-contained generator through the research of the method, so that greater economic benefits are brought to the enterprises and the society.
(2) The prediction result of the fractional Brown motion model is accurate and reliable;
(3) according to the method, a plurality of training samples are adopted for training, and then the fractional Brownian motion optimization model with the minimum error of the prediction result is selected for power load prediction, so that the accuracy of the model is effectively improved, and the prediction result is more accurate.
Drawings
Fig. 1 is a flow chart of a short-term prediction method of an electric load according to the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. Note that the following description of the embodiments is merely a substantial example, and the present invention is not intended to be limited to the application or the use thereof, and is not limited to the following embodiments.
Examples
As shown in fig. 1, a method for short-term prediction of an electrical load includes the following steps:
(1) acquiring a historical power load data sequence, and calculating a Hurst index H of the historical power load data sequence;
(2) establishing a fractional Brownian motion model for predicting the power load based on the Hurst index H;
(3) carrying out global optimization on the Hurst index H in the fractional Brownian motion model to obtain the optimal value H of the Hurst indexgbestFurther obtaining a fractional Brownian motion optimization model;
(4) and predicting the power load data by using a fractional Brownian motion optimization model.
And (2) acquiring the Herster index H of the power load data sequence in the step (1) by a re-standard polar difference analysis method (R/S method).
For historical power load data series yt,t=0,1,2...n},ytHistorical power load data representing time t, the partial sum of which is:
the sample variance is:
then the statistical formula for R/S is:
R/S is the range of data reforming, R/S values are calculated for different n values, and the Hurst index H of the group of power load data sequences can be obtained by drawing R/S and n curves on a logarithmic graph and performing least square fitting.
The fractional brownian motion model in the step (2) is specifically as follows:
for historical power load data series yt,t=0,1,2...n},ytRepresenting historical power load data at time t, the fractional brownian motion model is as follows:
yt+1=yt+uytΔt+σytw1(t)(Δt)H+λytw2(t)(Δt)2H,
wherein, yt+1Representing power load data at time t +1, Δ t representing the time interval between two adjacent historical power load data, w1(t) and w2(t) is a constant parameter that follows an independent normal distribution;
wherein, E represents the mathematical expectation,T=[0,1,2,…,n]y1 'is the transpose of y1, and T' is the transpose of T;
in the step (3), a quantum behavior particle swarm optimization (QPSO) method is adopted to carry out global optimization on the heuster index H to obtain the optimal value H of the heuster indexgbest。
In the above optimization process, the initialized population N is 20, and the population of the hurst index H is shown in table 1.
TABLE 1 20 particles of the Herster index H
1 | 3 | 1 | 3 | 1 |
3 | 3 | 3 | 3 | 2 |
3 | 1 | 2 | 1 | 1 |
1 | 2 | 3 | 1 | 1 |
Before the step (4), the method further comprises the following steps: changing the number of historical data in the historical power load data sequence and respectively using the historical data as a training sample, repeatedly executing the steps (1) to (3) on each training sample to respectively obtain corresponding fractional Brownian motion optimization models, comparing the prediction result errors of the fractional Brownian motion optimization models, and selecting the fractional Brownian motion optimization model with the minimum prediction result error to complete the prediction of the power load data in the step (4). The step aims to obtain the optimal prediction step length, historical power load data of the previous 24 hours, the previous 36 hours and the historical power load data of the previous 48 hours are respectively selected as three different training samples to establish a fractional Brown motion optimization model, and then the data of the next 24 hours are predicted. By comparing the relative errors of the predictions, a smaller error is selected to determine the appropriate prediction step size. And continuously adjusting the H value in the training process by using a QPSO algorithm to achieve the optimal approximation effect simulation. The parameters of the model are trained and then used for prediction respectively, and according to different step sizes, the prediction result error is best selected in 48 hours as shown in table 2.
TABLE 2 prediction error for different step sizes
Step size (hours) | Maximum error | Mean error | Median number | Standard deviation of |
24 | 6.212871 | 1.814809 | 1.652649 | 1.364628 |
36 | 3.516927 | 1.214638 | 0.975981 | 0.924972 |
48 | 2.588324 | 0.991533 | 0.93945 | 0.692363 |
And (3) judging the Hurst index H, if the H is more than 0.5 and less than 1, the historical power load data sequence has long correlation, continuing to execute the step (2), and if not, ending.
The above embodiments are merely examples and do not limit the scope of the present invention. These embodiments may be implemented in other various manners, and various omissions, substitutions, and changes may be made without departing from the technical spirit of the present invention.
Claims (3)
1. A method for short-term prediction of an electrical load, comprising the steps of:
(1) acquiring a historical power load data sequence, and calculating a Hurst index H of the historical power load data sequence;
(2) establishing a fractional Brownian motion model for predicting the power load based on the Hurst index H;
(3) carrying out global optimization on the Hurst index H in the fractional Brownian motion model to obtain the optimal value H of the Hurst indexgbestFurther obtaining a fractional Brownian motion optimization model;
(4) predicting power load data by using a fractional Brownian motion optimization model;
before the step (4), the method further comprises the following steps: changing the number of historical data in the historical power load data sequence and respectively using the historical data as a training sample, repeatedly executing the steps (1) to (3) on each training sample to respectively obtain corresponding fractional Brownian motion optimization models, comparing the prediction result errors of the fractional Brownian motion optimization models, and selecting the fractional Brownian motion optimization model with the minimum prediction result error to complete the prediction of the power load data in the step (4);
the fractional brownian motion model in the step (2) is specifically as follows:
for historical power load data series yt,t=0,1,2...n},ytRepresenting historical power load data at time t, the fractional brownian motion model is as follows:
yt+1=yt+uytΔt+σytw1(t)(Δt)H+λytw2(t)(Δt)2H,
wherein, yt+1Representing power load data at time t +1, Δ t representing the time interval between two adjacent historical power load data, w1(t) and w2(t) is a constant parameter that follows an independent normal distribution;
wherein, E represents the mathematical expectation,T=[0,1,2,…,n]y1 'is the transpose of y1, and T' is the transpose of T;
in the step (3), a quantum behavior particle swarm optimization method is adopted to carry out global optimization on the Hurst index H to obtain the optimal value H of the Hurst indexgbest。
2. The method as claimed in claim 1, wherein the hester exponent H of the power load data sequence in step (1) is obtained by a re-standard range analysis.
3. The method as claimed in claim 1, wherein the step (1) further comprises a step of judging a hester index H, if 0.5 < H < 1, the historical power load data sequence has a long correlation, and the step (2) is continued, otherwise, the method is ended.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810542186.1A CN108808657B (en) | 2018-05-30 | 2018-05-30 | Short-term prediction method for power load |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810542186.1A CN108808657B (en) | 2018-05-30 | 2018-05-30 | Short-term prediction method for power load |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108808657A CN108808657A (en) | 2018-11-13 |
CN108808657B true CN108808657B (en) | 2020-07-10 |
Family
ID=64089569
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810542186.1A Active CN108808657B (en) | 2018-05-30 | 2018-05-30 | Short-term prediction method for power load |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108808657B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110398368B (en) * | 2019-07-26 | 2021-05-11 | 上海工程技术大学 | FBM-based bearing inner ring fault residual life prediction method based on long correlation model |
CN112801363B (en) * | 2021-01-26 | 2024-01-23 | 国网上海市电力公司 | Power load loss prediction method, medium, electronic equipment and system |
CN113887812B (en) * | 2021-10-14 | 2023-07-07 | 广东电网有限责任公司 | Clustering-based small sample load prediction method, device, equipment and storage medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104318334A (en) * | 2014-10-29 | 2015-01-28 | 上海工程技术大学 | Short-time power load forecasting method based on long-range dependence FARIMA model |
CN107590562A (en) * | 2017-09-05 | 2018-01-16 | 西安交通大学 | A kind of Short-Term Load Forecasting of Electric Power System based on changeable weight combination predicted method |
-
2018
- 2018-05-30 CN CN201810542186.1A patent/CN108808657B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104318334A (en) * | 2014-10-29 | 2015-01-28 | 上海工程技术大学 | Short-time power load forecasting method based on long-range dependence FARIMA model |
CN107590562A (en) * | 2017-09-05 | 2018-01-16 | 西安交通大学 | A kind of Short-Term Load Forecasting of Electric Power System based on changeable weight combination predicted method |
Non-Patent Citations (1)
Title |
---|
美国加州电力市场电价的分形特征;孙可等;《能源工程》;20060531(第5期);第1-3页 * |
Also Published As
Publication number | Publication date |
---|---|
CN108808657A (en) | 2018-11-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Sun et al. | Using Bayesian deep learning to capture uncertainty for residential net load forecasting | |
CN110135612B (en) | Method for monitoring capacity of material supply Shang Sheng and early warning abnormality based on electricity consumption analysis | |
CN111199016A (en) | DTW-based improved K-means daily load curve clustering method | |
CN108808657B (en) | Short-term prediction method for power load | |
CN111783953A (en) | 24-point power load value 7-day prediction method based on optimized LSTM network | |
CN104504475A (en) | AR*-SVM (support vector machine) hybrid modeling based haze time series prediction method | |
CN110837915B (en) | Low-voltage load point prediction and probability prediction method for power system based on hybrid integrated deep learning | |
CN108717579B (en) | Short-term wind power interval prediction method | |
CN104331635A (en) | Predication method for electric power optical fiber communication optical power | |
CN108667069B (en) | Short-term wind power prediction method based on partial least squares regression | |
CN105205326B (en) | A kind of power plant's Real-time Load on-line prediction method | |
CN109919421A (en) | Short-term power load prediction model establishment method based on VMD-PSO-BPNN | |
CN112257928A (en) | Short-term power load probability prediction method based on CNN and quantile regression | |
CN110807490A (en) | Intelligent prediction method for construction cost of power transmission line based on single-base tower | |
CN109754122A (en) | A kind of Numerical Predicting Method of the BP neural network based on random forest feature extraction | |
CN115545333A (en) | Method for predicting load curve of multi-load daily-type power distribution network | |
CN109388863A (en) | A kind of distributed photovoltaic power output power forecasting method based on ARIMA model | |
CN110533249B (en) | Metallurgical enterprise energy consumption prediction method based on integrated long-term and short-term memory network | |
Wang et al. | Cloud computing and extreme learning machine for a distributed energy consumption forecasting in equipment-manufacturing enterprises | |
CN103605493A (en) | Parallel sorting learning method and system based on graphics processing unit | |
CN109767037A (en) | A kind of prediction technique of summer Daily treatment cost | |
CN112836876B (en) | Power distribution network line load prediction method based on deep learning | |
CN115907000A (en) | Small sample learning method for optimal power flow prediction of power system | |
Browell et al. | Cluster-based regime-switching AR for the EEM 2017 Wind Power Forecasting Competition | |
CN108985563A (en) | A kind of dynamically labeled method of Mechatronic Systems military service based on self-organizing feature map |
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 |