CN117010553A - Method and device for improving daily load prediction accuracy of power grid - Google Patents
Method and device for improving daily load prediction accuracy of power grid Download PDFInfo
- Publication number
- CN117010553A CN117010553A CN202310806537.6A CN202310806537A CN117010553A CN 117010553 A CN117010553 A CN 117010553A CN 202310806537 A CN202310806537 A CN 202310806537A CN 117010553 A CN117010553 A CN 117010553A
- Authority
- CN
- China
- Prior art keywords
- load
- load prediction
- data
- neural network
- factors
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000003062 neural network model Methods 0.000 claims abstract description 72
- 238000012549 training Methods 0.000 claims abstract description 38
- 238000012937 correction Methods 0.000 claims abstract description 23
- 238000013528 artificial neural network Methods 0.000 claims description 20
- 230000001537 neural effect Effects 0.000 claims description 16
- 230000035945 sensitivity Effects 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 14
- 230000001133 acceleration Effects 0.000 claims description 9
- 238000005286 illumination Methods 0.000 claims description 7
- 210000005036 nerve Anatomy 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 210000002364 input neuron Anatomy 0.000 claims description 4
- 230000010354 integration Effects 0.000 claims description 3
- 230000036316 preload Effects 0.000 claims description 3
- 238000010183 spectrum analysis Methods 0.000 claims description 3
- 230000001737 promoting effect Effects 0.000 claims 1
- 238000001556 precipitation Methods 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000010438 heat treatment Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 238000005057 refrigeration Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
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
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- 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
- G06N3/0499—Feedforward networks
-
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Human Resources & Organizations (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Operations Research (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Power Engineering (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The application provides a method and a device for improving the daily load prediction accuracy of a power grid, which belong to the technical field of daily load prediction of the power grid, and the method comprises the following steps: correcting weather forecast information according to the weather forecast change trend, establishing a weather factor load prediction model, training the weather factor load prediction model by using the weather forecast information after correction, and predicting daily load; correcting and normalizing the historical load data, establishing a BP neural network model for load prediction, training the BP neural network model by using the processed historical load data, and predicting daily load; and integrating daily load data predicted by the meteorological factor load prediction model and the BP neural network model. According to the application, the load prediction accuracy of daily load prediction is improved by introducing the load prediction model of the meteorological factors and the neural network model for correcting the historical load data, so that the adjustment of the power grid operation mode is facilitated, and the power grid load control capability is improved.
Description
Technical Field
The application belongs to the technical field of daily load prediction of a power grid, and particularly relates to a method and a device for improving daily load prediction accuracy of the power grid.
Background
The daily load prediction of the power system is an important basis for scheduling plans, operation mode arrangement and scheduling operation, and is an important guarantee for safe and stable operation of the power system. The existing daily load prediction accuracy rate calculation formula is as follows:
wherein n is the number of daily load prediction points, E i The relative error of the i-point load prediction is expressed as:
the power dispatching department has a requirement on the daily load prediction accuracy of the power system, the current daily load prediction rate mainly comprises four parts of system load prediction accuracy, bus load prediction accuracy, highest load prediction accuracy and lowest load prediction accuracy, which are multiplied by the weight coefficients of the four parts and summed up, and the problem that the daily load prediction rate is low is mainly caused by the fact that the system load prediction rate is low through statistical analysis.
The main factor causing the low system load prediction rate is as follows: firstly, weather factors, temperature, precipitation and other factors have larger influence on daily system loads, temperature change influences refrigeration load and heating load in domestic electricity and commercial load, precipitation change has larger influence on agricultural load, and current daily load accuracy prediction of the system is not combined with the weather factors. Secondly, the load prediction algorithm does not correct the correct historical data, the historical load data is an important basis for load prediction work, and the accuracy of the data has great influence on the accuracy of subsequent load prediction. The jump, the deletion and the like of the table code of the power dispatching automation system can cause inaccurate data, abnormal fluctuation of the load can also form abnormal data, a large amount of abnormal data exist in the historical data of the current daily load forecast use, and correction is not carried out.
In summary, factors that contribute to low accuracy of grid daily load forecast are unbound weather factors and using grid history data without correction.
This is a deficiency of the prior art, and therefore, it is necessary to provide a method and a device for improving the daily load prediction accuracy of the power grid in order to address the above-mentioned drawbacks of the prior art.
Disclosure of Invention
Aiming at the defects that the factors causing low daily load prediction accuracy of the power grid in the prior art are unconjugated weather factors and uncorrected power grid historical data, the application provides a method and a device for improving the daily load prediction accuracy of the power grid, and aims to solve the technical problems.
In a first aspect, the present application provides a method for improving the daily load prediction accuracy of a power grid, including the following steps:
s1, correcting weather forecast information according to a weather forecast change trend, establishing a weather factor load prediction model, training the weather factor load prediction model by using the weather forecast information after correction, and carrying out daily load prediction by using the trained weather factor load prediction model;
s2, correcting and normalizing the historical load data, establishing a BP neural network model for load prediction, training the BP neural network model by using the processed historical load data, and performing daily load prediction by using the trained BP neural network model;
and S3, integrating daily load data predicted by the meteorological factor load prediction model and the BP neural network model.
Further, the specific steps of step S1 are as follows:
s11, acquiring next-day weather forecast information in an area range, and correcting the next-day weather forecast information by combining a weather forecast change trend in a set time period to obtain weather forecast sensitivity data;
s12, selecting load factors and corresponding data samples of meteorological factors, establishing a meteorological factor load prediction model by using an Elman neural network, and training the meteorological factor load prediction model by using the data samples of the load factors;
and S13, establishing a weather information input module, and inputting weather prediction sensitivity data into a trained weather factor load prediction model to obtain first day pre-load prediction data.
Further, the specific steps of step S11 are as follows:
s111, referring to specific weather information of each channel in a set area range, and integrating the specific weather information to obtain detailed weather forecast information;
s112, obtaining the deviation between the weather forecast and the actual weather in the current set time period to predict the change trend of the weather forecast, and generating the deviation degree of the weather forecast;
and S113, correcting the detailed weather forecast information according to the weather forecast deviation degree to obtain weather forecast sensitivity data. The detailed weather forecast information comprises the next day temperature, precipitation, wind power, illumination intensity, weather early warning information and weather forecast information of a preset set period, wherein the wind power comprises wind speed and wind direction.
Further, the specific steps of step S12 are as follows:
s121, selecting a load factor of a meteorological factor, introducing the meteorological load factor, and establishing an associated polynomial of the meteorological load factor and the load factor of the meteorological factor;
s122, selecting a load which is influenced by the weather factors and has temperature change exceeding a threshold value, establishing a quantitative relation between the load and the load factors of the weather factors, obtaining a data sample of the load factors of the weather factors by combining historical data of a set time period, and setting the data sample as a first data sample;
s123, establishing a polynomial regression model to fit a change rule of load factors of meteorological factors to the load, and solving polynomial coefficients and orders;
s124, designing an E lman neural network model, and carrying out spectrum analysis on a load curve by adopting fast Fourier change;
s125, determining historical data required by an E lman neural network model input neuron, and selecting from a first data sample of a load factor of a meteorological factor;
and S126, training the Elman neural network model by using the data selected by the first data sample to obtain an E lman neural network load model based on the meteorological load factor.
Further, the load factor of the meteorological factors is selected from air temperature factors, illumination factors, rainfall factors, wind power factors and humidity factors.
Further, the specific steps of step S2 are as follows:
s21, carrying out correction processing on the historical load data to generate historical load sample data, and carrying out normalization processing on the historical load sample data;
s22, establishing a high-order load prediction BP neural network, and training a load prediction BP neural network model by using the normalized historical load sample data;
s23, burning the trained load prediction BP neural network model to a neural computing rod;
s24, debugging a nerve computation stick for completing the burning of the load prediction BP nerve network model;
s25, predicting the daily load by using a load prediction BP neural network model after acceleration of the neural computing rod, and obtaining second daily load prediction data.
Further, the specific steps of step S21 are as follows:
s211, acquiring historical load data;
s212, judging whether historical load data is missing or not;
if yes, go to step S213;
if not, go to step S214;
s213, complementing the missing historical load data;
s214, correcting the historical load data from the vertical dimension and the horizontal dimension respectively;
s215, normalizing the corrected historical load data.
In a second aspect, the present application provides a device for improving the daily load prediction accuracy of a power grid, including:
the load prediction weather factor introducing unit is used for correcting weather forecast information according to the change trend of the weather forecast, establishing a weather factor load prediction model, training the weather factor load prediction model by using the corrected weather forecast information, and carrying out daily load prediction by using the trained weather factor load prediction model;
the load prediction historical load correction unit is used for correcting and normalizing historical load data, establishing a load prediction BP neural network model, training the BP neural network model by using the processed historical load data, and carrying out daily load prediction by using the trained BP neural network model;
and the load data integration unit is used for integrating daily load data predicted by the meteorological factor load prediction model and the BP neural network model.
Further, the load prediction weather factor introduction unit includes:
the weather forecast correction subunit is used for acquiring the next-day weather forecast information in the area range, and correcting the next-day weather forecast information by combining the weather forecast change trend in the set time period to obtain weather forecast sensitivity data;
the meteorological factor load prediction model training subunit is used for selecting load factors and corresponding data samples of meteorological factors, establishing a meteorological factor load prediction model by using an E lman neural network, and training the meteorological factor load prediction model by using the data samples of the load factors;
and the daily load prediction data output subunit is used for establishing a meteorological information input module, inputting weather prediction sensitivity data into a trained meteorological factor load prediction model and obtaining first daily load prediction data.
Further, the load prediction history load correction unit includes:
the historical load data processing subunit is used for carrying out correction processing on the historical load data to generate historical load sample data and carrying out normalization processing on the historical load sample data;
the load prediction neural network model training subunit is used for establishing a high-order load prediction BP neural network and training a load prediction BP neural network model by using the normalized historical load sample data;
the neural network model hardware burning subunit is used for burning the trained load prediction BP neural network model to a neural computing rod;
the neural network model acceleration hardware debugging subunit is used for debugging the neural computing rod burnt by the load prediction BP neural network model;
and the daily load prediction data output subunit is used for predicting the daily load by using a load prediction BP neural network model after acceleration of the neural computing rod to obtain second daily load prediction data.
The application has the beneficial effects that:
according to the method and the device for improving the daily load prediction accuracy of the power grid, the daily load prediction accuracy is improved by introducing the load prediction model of the meteorological factors and the neural network model for correcting the historical load data, so that the adjustment of the power grid operation mode is facilitated, the power grid load control capability is improved, the real-time operation risk of the power grid is reduced, and the dispatching pressure of the power grid is reduced.
In addition, the application has reliable design principle, simple structure and very wide application prospect.
It can be seen that the present application has outstanding substantial features and significant advances over the prior art, as well as the benefits of its implementation.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the description of the embodiments or the prior art will be briefly described below, and it will be obvious to those skilled in the art that other drawings can be obtained from these drawings without inventive effort.
Fig. 1 is a schematic flow chart of an embodiment 1 of a method for improving the daily load prediction accuracy of a power grid.
Fig. 2 is a schematic flow chart of an embodiment 2 of a method for improving the daily load prediction accuracy of a power grid.
FIG. 3 is a flow chart of the application for correcting weather forecast information.
FIG. 4 is a flow chart of the present application for modifying historical load data.
Fig. 5 is a schematic diagram of a device for improving the daily load prediction accuracy of a power grid.
Detailed Description
In order to make the technical solution of the present application better understood by those skilled in the art, the technical solution of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present application without making any inventive effort, shall fall within the scope of the present application.
Example 1:
as shown in fig. 1, the application provides a method for improving the daily load prediction accuracy of a power grid, which comprises the following steps:
s1, correcting weather forecast information according to a weather forecast change trend, establishing a weather factor load prediction model, training the weather factor load prediction model by using the weather forecast information after correction, and carrying out daily load prediction by using the trained weather factor load prediction model;
s2, correcting and normalizing the historical load data, establishing a BP neural network model for load prediction, training the BP neural network model by using the processed historical load data, and performing daily load prediction by using the trained BP neural network model;
and S3, integrating daily load data predicted by the meteorological factor load prediction model and the BP neural network model.
Example 2:
as shown in fig. 2, the application provides a method for improving the daily load prediction accuracy of a power grid, which comprises the following steps:
s1, correcting weather forecast information according to a weather forecast change trend, establishing a weather factor load prediction model, training the weather factor load prediction model by using the weather forecast information after correction, and carrying out daily load prediction by using the trained weather factor load prediction model; statistical analysis shows that the difference of the system load prediction accuracy rates under different weather is larger, and the system load prediction accuracy rate is higher when the weather is clearer. The accuracy is highest in sunny days and lowest in rainy and snowy days; the specific steps of the step S1 are as follows:
s11, acquiring next-day weather forecast information in an area range, and correcting the next-day weather forecast information by combining a weather forecast change trend in a set time period to obtain weather forecast sensitivity data; as shown in fig. 3, the specific steps of step S11 are as follows:
s111, referring to specific weather information of each channel in a set area range, and integrating the specific weather information to obtain detailed weather forecast information; for example, specific weather information of a set area range can be obtained by consulting multiple channels such as a weather website, an APP, a WeChat public number and the like; the detailed weather forecast information comprises the next day temperature, precipitation, wind power, illumination intensity, weather early warning information and week weather forecast information, wherein the wind power comprises wind speed and wind direction;
s112, obtaining the deviation between the weather forecast and the actual weather in the current set time period to predict the change trend of the weather forecast, and generating the deviation degree of the weather forecast;
s113, correcting detailed weather forecast information according to the weather forecast deviation degree to obtain weather forecast sensitivity data;
the accuracy of the weather forecast of the next day is improved by correcting the detailed weather forecast information;
s12, selecting load factors and corresponding data samples of meteorological factors, establishing a meteorological factor load prediction model by using an Elman neural network, and training the meteorological factor load prediction model by using the data samples of the load factors; the specific steps of step S12 are as follows:
s121, selecting a load factor of a meteorological factor, introducing the meteorological load factor, and establishing an associated polynomial of the meteorological load factor and the load factor of the meteorological factor; the load factors of the meteorological factors are selected from air temperature factors, illumination factors, rainfall factors, wind power factors and humidity factors;
for example, identifying a load susceptible to weather as lw=f (T, V, R, H, W), where T is an air temperature factor, V is an illumination factor, R is a rainfall factor, H is a humidity factor, and W is a wind power factor;
introduction of a meteorological load factor m=k 1 ΔT+K 2 ΔV+K 3 ΔR+K 4 ΔH+K 5 ΔW；
S122, selecting a load which is influenced by the weather factors and has temperature change exceeding a threshold value, establishing a quantitative relation between the load and the load factors of the weather factors, obtaining a data sample of the load factors of the weather factors by combining historical data of a set time period, and setting the data sample as a first data sample; for example, a quantitative relation is established between summer cooling load, winter heating load and other weather factors such as temperature, precipitation, wind power and the like which are easily affected by weather is selected, historical data from 2015 to 2020 are processed, and a large number of data samples are formed;
s123, establishing a polynomial regression model to fit a change rule of load factors of meteorological factors to the load, and solving polynomial coefficients and orders;
for example, a polynomial regression model is established to simulate the change rule of the temperature T, the illumination V, the rainfall R, the humidity H and the wind power W to the load:
L＝a 0 +a 1 T+a 2 T 2 +…+a s1 T s1 +
b 0 +b 1 V+b 2 V 2 +…+b s2 V s2 +
c 0 +c 1 T+c 2 T 2 +…+c s3 R s3 +
d 0 +d 1 T+d 2 T 2 +…+d s4 H s4 +
e 0 +e 1 W+e 2 W 2 +…+e s5 W s5 +Y
wherein the coefficients are obtained by a least square method, and the polynomial order is obtained by an AI C criterion
S124, designing an E lman neural network model, and carrying out spectrum analysis on a load curve by adopting fast Fourier change;
s125, determining historical data required by an E lman neural network model input neuron, and selecting from a first data sample of a load factor of a meteorological factor; for example, the input neurons include actual load data that predicts the first 1 hour and the previous day:
l (n-1), L (n-24), weather load factor M (n) of the predicted time, type of day before the predicted time and time of the predicted time;
s126, training an Elman neural network model by using data selected by the first data sample to obtain an E lman neural network load model based on a meteorological load factor;
s13, establishing a weather information input module, and inputting weather prediction sensitivity data into a trained weather factor load prediction model to obtain first day pre-load prediction data;
in actual use, detailed weather forecast information of the next day is applied to 1 to 30 days of 4 months of a certain year, and an E lman neural network load prediction model based on meteorological factors is combined, so that the system load prediction rate reaches 97.96% in cloudy and overcast weather, the system load prediction rate reaches 97.78% in rainy and snowy weather, and the system load prediction accuracy rate reaches 98.03%;
s2, correcting and normalizing the historical load data, establishing a BP neural network model for load prediction, training the BP neural network model by using the processed historical load data, and performing daily load prediction by using the trained BP neural network model; the specific steps of the step S2 are as follows:
s21, carrying out correction processing on the historical load data to generate historical load sample data, and carrying out normalization processing on the historical load sample data; as shown in fig. 4, the specific steps of step S21 are as follows:
s211, acquiring historical load data;
s212, judging whether historical load data is missing or not;
if yes, go to step S213;
if not, go to step S214;
s213, complementing the missing historical load data;
s214, correcting the historical load data from the vertical dimension and the horizontal dimension respectively;
s215, carrying out normalization processing on the corrected historical load data;
s22, establishing a high-order load prediction BP neural network, and training a load prediction BP neural network model by using the normalized historical load sample data;
s23, burning the trained load prediction BP neural network model to a neural computing rod; the high-order neural network operation has high requirement on the performance of a computer, a neural computing rod, namely NCS is a deep learning reasoning tool and an artificial intelligent accelerator based on a USB mode recently introduced by Int e l company, and the deep neural network can be directly and rapidly operated in real time on the premise of no networking and no cloud;
s24, debugging a nerve computation stick for completing the burning of the load prediction BP nerve network model;
s25, predicting daily load by using a load prediction BP model after acceleration of the nerve computation stick to obtain second daily load prediction data;
the current load prediction method only considers single influencing factors, has no generality, establishes a correction history load data link through the methods of supplementing missing data, correcting vertical data, correcting horizontal data and the like, combines the correction history load data link with a high-order BP neural network, establishes a load prediction algorithm considering multiple factors through training an improved high-order BP neural network model, and is suitable for load prediction calculation under the influence of multiple factors;
after the BP neural network system load prediction method based on the neural computing rod is introduced, the system load prediction accuracy rate of the 1 log 30 days of 4 months of a certain year is counted to reach 98.04%.
And S3, integrating daily load data predicted by the meteorological factor load prediction model and the BP neural network model.
Example 3:
as shown in FIG. 5, the application provides a device for improving the daily load prediction accuracy of a power grid, which comprises the following steps:
the load prediction weather factor introducing unit is used for correcting weather forecast information according to the change trend of the weather forecast, establishing a weather factor load prediction model, training the weather factor load prediction model by using the corrected weather forecast information, and carrying out daily load prediction by using the trained weather factor load prediction model; the load prediction meteorological factor introducing unit comprises:
the weather forecast correction subunit is used for acquiring the next-day weather forecast information in the area range, and correcting the next-day weather forecast information by combining the weather forecast change trend in the set time period to obtain weather forecast sensitivity data;
the meteorological factor load prediction model training subunit is used for selecting load factors and corresponding data samples of meteorological factors, establishing a meteorological factor load prediction model by using an E lman neural network, and training the meteorological factor load prediction model by using the data samples of the load factors;
the daily load prediction data output subunit based on the weather information is used for establishing a weather information input module, inputting weather prediction sensitivity data into a trained weather factor load prediction model and obtaining first daily load prediction data;
the load prediction historical load correction unit is used for correcting and normalizing historical load data, establishing a load prediction BP neural network model, training the BP neural network model by using the processed historical load data, and carrying out daily load prediction by using the trained BP neural network model; the load prediction history load correction unit includes:
the historical load data processing subunit is used for carrying out correction processing on the historical load data to generate historical load sample data and carrying out normalization processing on the historical load sample data;
the load prediction neural network model training subunit is used for establishing a high-order load prediction BP neural network and training a load prediction BP neural network model by using the normalized historical load sample data;
the neural network model hardware burning subunit is used for burning the trained load prediction BP neural network model to a neural computing rod;
the neural network model acceleration hardware debugging subunit is used for debugging the neural computing rod burnt by the load prediction BP neural network model;
the daily load prediction data output subunit is used for predicting daily load by using a load prediction BP neural network model after acceleration of the neural computing rod to obtain second daily load prediction data;
and the load data integration unit is used for integrating daily load data predicted by the meteorological factor load prediction model and the BP neural network model.
Although the present application has been described in detail by way of preferred embodiments with reference to the accompanying drawings, the present application is not limited thereto. Various equivalent modifications and substitutions may be made in the embodiments of the present application by those skilled in the art without departing from the spirit and scope of the present application, and it is intended that all such modifications and substitutions be within the scope of the present application/be within the scope of the present application as defined by the appended claims. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. The method for improving the daily load prediction accuracy of the power grid is characterized by comprising the following steps of:
s1, correcting weather forecast information according to a weather forecast change trend, establishing a weather factor load prediction model, training the weather factor load prediction model by using the weather forecast information after correction, and carrying out daily load prediction by using the trained weather factor load prediction model;
s2, correcting and normalizing the historical load data, establishing a BP neural network model for load prediction, training the BP neural network model by using the processed historical load data, and performing daily load prediction by using the trained BP neural network model;
s3, integrating daily load data predicted by the meteorological factor load prediction model and the BP neural network model.
2. The method for improving the daily load prediction accuracy of the power grid according to claim 1, wherein the step S1 specifically comprises the following steps:
s11, acquiring next-day weather forecast information in an area range, and correcting the next-day weather forecast information by combining a weather forecast change trend in a set time period to obtain weather forecast sensitivity data;
s12, selecting load factors and corresponding data samples of meteorological factors, establishing a meteorological factor load prediction model by using an Elman neural network, and training the meteorological factor load prediction model by using the data samples of the load factors;
s13, a weather information input module is established, weather prediction sensitivity data are input into a trained weather factor load prediction model, and first day pre-load prediction data are obtained.
3. The method for improving the daily load prediction accuracy of the power grid according to claim 2, wherein the step S11 specifically comprises the following steps:
s111, referring to specific weather information of each channel in a set area range, and integrating the specific weather information to obtain detailed weather forecast information;
s112, obtaining the deviation between the weather forecast and the actual weather in the current set time period to predict the change trend of the weather forecast, and generating the deviation degree of the weather forecast;
s113, correcting the detailed weather forecast information according to the weather forecast deviation degree to obtain weather forecast sensitivity data.
4. The method for improving the daily load prediction accuracy of the power grid according to claim 2, wherein the step S12 specifically comprises the following steps:
s121, selecting load factors of meteorological factors, introducing the meteorological load factors, and establishing an associated polynomial of the meteorological load factors and the load factors of the meteorological factors;
s122, selecting a load which is influenced by the meteorological factors and has temperature change exceeding a threshold value, establishing a quantitative relation between the load and the load factors of the meteorological factors, obtaining a data sample of the load factors of the meteorological factors by combining historical data of a set time period, and setting the data sample as a first data sample;
s123, establishing a polynomial regression model to fit a change rule of load factors of meteorological factors to the load, and solving polynomial coefficients and orders;
s124, designing an Elman neural network model, and carrying out spectrum analysis on a load curve by adopting fast Fourier change;
s125, determining historical data required by an Elman neural network model input neuron, and selecting from a first data sample of a load factor of a meteorological factor;
s126, training the Elman neural network model by using data selected by the first data sample to obtain an Elman neural network load model based on the meteorological load factor.
5. The method for improving the daily load prediction accuracy of a power grid according to claim 4, wherein the load factors of the meteorological factors are selected from the group consisting of air temperature factors, illumination factors, rainfall factors, wind power factors and humidity factors.
6. The method for improving the daily load prediction accuracy of the power grid according to claim 1, wherein the step S2 specifically comprises the following steps:
s21, correcting the historical load data to generate historical load sample data, and normalizing the historical load sample data;
s22, establishing a high-order load prediction BP neural network, and training a load prediction BP neural network model by using the normalized historical load sample data;
s23, burning the trained load prediction BP neural network model to a neural computing rod;
s24, debugging a nerve computation stick for completing the burning of the load prediction BP nerve network model;
s25, predicting daily load by using a load prediction BP neural network model after acceleration of the neural computing rod, and obtaining second daily load prediction data.
7. The method for improving the daily load prediction accuracy of the power grid according to claim 6, wherein the step S21 specifically comprises the following steps:
s211, acquiring historical load data;
s212, judging whether historical load data are missing or not;
if yes, go to step S213;
if not, go to step S214;
s213, complementing the missing historical load data;
s214, correcting historical load data from vertical and horizontal dimensions respectively;
s215, normalizing the corrected historical load data.
8. The utility model provides a device for promoting power grid daily load prediction rate of accuracy, which is characterized in that includes:
the load prediction weather factor introducing unit is used for correcting weather forecast information according to the change trend of the weather forecast, establishing a weather factor load prediction model, training the weather factor load prediction model by using the corrected weather forecast information, and carrying out daily load prediction by using the trained weather factor load prediction model;
the load prediction historical load correction unit is used for correcting and normalizing historical load data, establishing a load prediction BP neural network model, training the BP neural network model by using the processed historical load data, and carrying out daily load prediction by using the trained BP neural network model;
and the load data integration unit is used for integrating daily load data predicted by the meteorological factor load prediction model and the BP neural network model.
9. The apparatus for improving the daily load prediction accuracy of a power grid according to claim 8, wherein the load prediction meteorological factor introduction unit comprises:
the weather forecast correction subunit is used for acquiring the next-day weather forecast information in the area range, and correcting the next-day weather forecast information by combining the weather forecast change trend in the set time period to obtain weather forecast sensitivity data;
the meteorological factor load prediction model training subunit is used for selecting load factors and corresponding data samples of meteorological factors, establishing a meteorological factor load prediction model by using an Elman neural network, and training the meteorological factor load prediction model by using the data samples of the load factors;
and the daily load prediction data output subunit is used for establishing a meteorological information input module, inputting weather prediction sensitivity data into a trained meteorological factor load prediction model and obtaining first daily load prediction data.
10. The apparatus for improving the daily load prediction accuracy of a power grid according to claim 8, wherein the load prediction history load correction unit comprises:
the historical load data processing subunit is used for carrying out correction processing on the historical load data to generate historical load sample data and carrying out normalization processing on the historical load sample data;
the load prediction neural network model training subunit is used for establishing a high-order load prediction BP neural network and training a load prediction BP neural network model by using the normalized historical load sample data;
the neural network model hardware burning subunit is used for burning the trained load prediction BP neural network model to a neural computing rod;
the neural network model acceleration hardware debugging subunit is used for debugging the neural computing rod burnt by the load prediction BP neural network model;
and the daily load prediction data output subunit is used for predicting the daily load by using a load prediction BP neural network model after acceleration of the neural computing rod to obtain second daily load prediction data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310806537.6A CN117010553A (en) | 2023-07-03 | 2023-07-03 | Method and device for improving daily load prediction accuracy of power grid |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310806537.6A CN117010553A (en) | 2023-07-03 | 2023-07-03 | Method and device for improving daily load prediction accuracy of power grid |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117010553A true CN117010553A (en) | 2023-11-07 |
Family
ID=88570071
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310806537.6A Pending CN117010553A (en) | 2023-07-03 | 2023-07-03 | Method and device for improving daily load prediction accuracy of power grid |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117010553A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118036816A (en) * | 2024-03-04 | 2024-05-14 | 国网山东省电力公司青岛供电公司 | Novel power system load online prediction method |
| CN118316127A (en) * | 2024-06-06 | 2024-07-09 | 国网浙江省电力有限公司慈溪市供电公司 | Park power resource scheduling method and device |
| CN118536841A (en) * | 2024-07-26 | 2024-08-23 | 国网山东省电力公司无棣县供电公司 | Power grid load prediction method, system, equipment and medium |
-
2023
- 2023-07-03 CN CN202310806537.6A patent/CN117010553A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118036816A (en) * | 2024-03-04 | 2024-05-14 | 国网山东省电力公司青岛供电公司 | Novel power system load online prediction method |
| CN118316127A (en) * | 2024-06-06 | 2024-07-09 | 国网浙江省电力有限公司慈溪市供电公司 | Park power resource scheduling method and device |
| CN118536841A (en) * | 2024-07-26 | 2024-08-23 | 国网山东省电力公司无棣县供电公司 | Power grid load prediction method, system, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109002915B (en) | Photovoltaic power station short-term power prediction method based on Kmeans-GRA-Elman model | |
| CN112257941B (en) | Photovoltaic power station short-term power prediction method based on improved Bi-LSTM | |
| CN114004139B (en) | Photovoltaic power generation power prediction method | |
| CN112749904B (en) | Power distribution network fault risk early warning method and system based on deep learning | |
| CN109086928B (en) | Photovoltaic power station real-time power prediction method based on SAGA-FCM-LSSVM model | |
| CN103106544B (en) | A kind of photovoltaic generation prognoses system based on T-S Fuzzy neutral net | |
| WO2023216576A1 (en) | Photovoltaic power generation short-term power prediction method and system | |
| CN114048896B (en) | Method, system, equipment and medium for predicting photovoltaic power generation data | |
| CN114792156A (en) | Photovoltaic output power prediction method and system based on curve characteristic index clustering | |
| CN114021420A (en) | Distributed photovoltaic ultra-short-term power generation prediction method and system | |
| CN114648176B (en) | Wind-solar power supply consumption optimization method based on data driving | |
| CN112285807A (en) | Meteorological information prediction method and device | |
| CN115796393B (en) | Energy management optimization method, system and storage medium based on multi-energy interaction | |
| CN112101669A (en) | Photovoltaic power interval prediction method based on improved extreme learning machine and quantile regression | |
| CN117010553A (en) | Method and device for improving daily load prediction accuracy of power grid | |
| CN111626473A (en) | Two-stage photovoltaic power prediction method considering error correction | |
| CN118040681B (en) | Method and system for predicting and correcting photovoltaic power generation power based on LSTM model | |
| CN115965132A (en) | Distributed photovoltaic digital twin system power prediction method based on GA-BP neural network | |
| CN116090678B (en) | Data processing method, device and equipment | |
| CN116402203A (en) | Method, system and medium for predicting short-time photovoltaic power generation capacity considering weather conditions | |
| CN112633565A (en) | Photovoltaic power aggregation interval prediction method | |
| CN117767446A (en) | Multi-source power scheduling method and system based on big data | |
| CN111697560A (en) | Method and system for predicting load of power system based on LSTM | |
| CN113887833A (en) | Distributed energy user side time-by-time load prediction method and system | |
| CN113112085A (en) | New energy station power generation load prediction method based on BP neural network |
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 |