CN109978268A - A kind of short-term load forecasting method, system and relevant apparatus - Google Patents
A kind of short-term load forecasting method, system and relevant apparatus Download PDFInfo
- Publication number
- CN109978268A CN109978268A CN201910249984.XA CN201910249984A CN109978268A CN 109978268 A CN109978268 A CN 109978268A CN 201910249984 A CN201910249984 A CN 201910249984A CN 109978268 A CN109978268 A CN 109978268A
- Authority
- CN
- China
- Prior art keywords
- deviation
- neural network
- short
- network model
- index
- 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
- 238000013277 forecasting method Methods 0.000 title claims abstract description 24
- 238000003062 neural network model Methods 0.000 claims abstract description 63
- 238000012549 training Methods 0.000 claims abstract description 59
- 238000010606 normalization Methods 0.000 claims abstract description 41
- 238000004364 calculation method Methods 0.000 claims abstract description 37
- 210000002569 neuron Anatomy 0.000 claims description 42
- 238000013528 artificial neural network Methods 0.000 claims description 33
- 238000012937 correction Methods 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 13
- 238000012795 verification Methods 0.000 claims description 11
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 5
- 210000004218 nerve net Anatomy 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 240000002853 Nelumbo nucifera Species 0.000 claims 1
- 235000006508 Nelumbo nucifera Nutrition 0.000 claims 1
- 235000006510 Nelumbo pentapetala Nutrition 0.000 claims 1
- 238000000034 method Methods 0.000 abstract description 12
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 210000005036 nerve Anatomy 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 230000005611 electricity Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000005034 decoration Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000452 restraining effect Effects 0.000 description 1
Classifications
-
- 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
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Educational Administration (AREA)
- General Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Water Supply & Treatment (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of short-term load forecasting method provided herein, comprising: obtain the index parameter of each index in demand history data；The index parameter of each index is inputted into default BP neural network model, the short-term load forecasting result of each index after output normalization；Wherein, the deviation for presetting BP neural network model is to be obtained by the deviation calculation formula of increase complexity convergent, and deviation is not more than deviation convergence threshold.The deviation of default BP neural network model in this method is obtained by the deviation calculation formula of increase complexity convergent, and deviation is not more than deviation convergence threshold, it can guarantee that the complexity of the default BP neural network model is not too big, also it can guarantee that the default BP neural network model will not over-fitting, and then Error Margin of Short-Term Electric Load Forecasting can be reduced while guaranteeing model training precision.The application also provides a kind of Short Term Load Forecasting System, equipment and computer readable storage medium, all has above-mentioned beneficial effect.
Description
Technical field
This application involves electric power system dispatchings to run field, in particular to a kind of short-term load forecasting method, system, equipment
And computer readable storage medium.
Background technique
Artificial neural network (Artificial Neural Network, i.e. ANN) is the artificial intelligence being concerned in recent years
Energy algorithm, BP neural network becomes most commonly used people with its stronger learning ability in numerous branches of artificial neural network
Artificial neural networks algorithm, and be widely used in power-system short-term load forecasting.
So-called short-term load forecasting, which refers to, carries out the indexs such as the load curve of next day, peak load, electricity consumption a few days ago
The load prediction type of prediction.Pertinent literature research shows that the multinomial factor such as temperature, distributed generation resource, Demand Side Response will
Short term is had an impact.And when carrying out short-term load forecasting using BP neural network intelligent algorithm, often face
" over-fitting " problem as caused by correlative factor is more, historical data is complicated, although it is higher to cause the training stage that can obtain
Training precision, and the problem that forecast period test error is larger.
Therefore, how while guaranteeing model training precision, reducing Error Margin of Short-Term Electric Load Forecasting is those skilled in the art
The technical issues of member's urgent need to resolve.
Summary of the invention
The purpose of the application is to provide a kind of short-term load forecasting method, system, equipment and computer readable storage medium,
Error Margin of Short-Term Electric Load Forecasting can be reduced while guaranteeing model training precision.
In order to solve the above technical problems, the application provides a kind of short-term load forecasting method, comprising:
Obtain the index parameter of each index in demand history data；
The index parameter of each index is inputted into default BP neural network model, it is each after output normalization
The short-term load forecasting result of the index；Wherein, the deviation of the default BP neural network model is by increase complexity
What the deviation calculation formula of convergent obtained, and the deviation is not more than deviation convergence threshold.
Preferably, the index parameter by each index inputs default BP neural network model, and output is returned
The short-term load forecasting result of each index after one change, comprising:
After carrying out data normalization to original loads training data and handling to obtain weight training data, the load is utilized
Training data determines the connection weight and amount of bias of each neuron in BP neural network；
The connection weight of each neuron and the amount of bias are being substituted into corresponding numerical value conversion function, obtained
To after the first BP neural network model, described first will be inputted by the load verification data that the data normalization is handled
BP neural network model obtains output result；
In the connection weight for utilizing each neuron and preset weight coefficient, the complexity convergence is constructed
Xiang Hou determines the deviation calculation formula using the complexity convergent and original deflection calculation formula；
The load verification data and the output result are being substituted into the deviation calculation formula, are obtaining the deviation
Afterwards, judge whether the deviation is greater than the deviation convergence threshold；
If it is not, the first BP neural network model is then determined as the default BP neural network model；
The index parameter of each index is inputted into the default BP neural network model, after output normalization
The short-term load forecasting result of each index.
Preferably, if judging, the deviation is greater than the deviation convergence threshold, comprising:
According to correction formula to the connection weight of each neuron of the first BP neural network model and
The amount of bias is modified operation, obtains corresponding connection weight correction value and amount of bias correction value.
It is preferably, described after carrying out data normalization to original loads training data and handling to obtain weight training data,
The connection weight and amount of bias of each neuron in BP neural network are determined using the weight training data, comprising:
Data normalization processing is carried out to the original loads training data according to linear transformation formula, obtains the load
Training data；
The connection weight of each neuron in the BP neural network is determined using the weight training data
With the amount of bias.
The application also provides a kind of Short Term Load Forecasting System, comprising:
Index parameter obtains module, for obtaining the index parameter of each index in demand history data；
Short-term load forecasting result output module, for the index parameter of each index to be inputted default BP mind
Through network model, the short-term load forecasting result of each index after output normalization；Wherein, the default BP nerve net
The deviation of network model is to be obtained by the deviation calculation formula of increase complexity convergent, and the deviation is not more than deviation
Convergence threshold.
Preferably, the short-term load forecasting result output module, comprising:
Connection weight and amount of bias determination unit, for being handled to original loads training data progress data normalization
To after weight training data, the connection weight of each neuron and partially is determined in BP neural network using the weight training data
The amount of setting；
As a result output unit, for the connection weight of each neuron and the amount of bias to be substituted into correspondence
Numerical value conversion function, after obtaining the first BP neural network model, the load handled by the data normalization is tested
It demonstrate,proves data and inputs the first BP neural network model, obtain output result；
Deviation calculation formula determination unit, in the connection weight and preset power using each neuron
Weight coefficient, after constructing the complexity convergent, using the complexity convergent and original deflection calculation formula, determine described in
Deviation calculation formula；
Judging unit, for the load verification data and the output result to be substituted into the deviation calculation formula,
After obtaining the deviation, judge whether the deviation is greater than the deviation convergence threshold；
Default BP neural network model determination unit, if being not more than the deviation convergence threshold for the deviation,
The first BP neural network model is determined as the default BP neural network model；
Short-term load forecasting result output unit, for the index parameter input of each index is described default
BP neural network model, the short-term load forecasting result of each index after output normalization.
Preferably, the short-term load forecasting result output module, comprising:
Operating unit is corrected, if being greater than the deviation convergence threshold for the deviation, according to correction formula to institute
The connection weight and the amount of bias for stating each neuron of the first BP neural network model are modified operation, obtain
To corresponding connection weight correction value and amount of bias correction value.
Preferably, the connection weight and amount of bias determination unit, comprising:
Data normalization handles subelement, for being counted according to linear transformation formula to the original loads training data
According to normalized, the weight training data are obtained；
Connection weight and amount of bias determine subelement, for determining the BP neural network using the weight training data
In each neuron the connection weight and the amount of bias.
The application also provides a kind of equipment, comprising:
Memory and processor；Wherein, the memory is for storing computer program, and the processor is for executing institute
The step of short-term load forecasting method described above is realized when stating computer program.
The application also provides a kind of computer readable storage medium, and the computer-readable recording medium storage has computer
The step of program, the computer program realizes short-term load forecasting method described above when being executed by processor.
A kind of short-term load forecasting method provided herein, comprising: obtain each index in demand history data
Index parameter；The index parameter of each index is inputted into default BP neural network model, it is each after output normalization
The short-term load forecasting result of a index.Wherein, the deviation of the default BP neural network model is complicated by increasing
What the deviation calculation formula of degree convergent obtained, and the deviation is not more than deviation convergence threshold.
The deviation of default BP neural network model in this method is to calculate public affairs by the deviation of increase complexity convergent
What formula obtained, and the deviation is not more than deviation convergence threshold, can guarantee the complexity of the default BP neural network model
It is not too big, can also guarantee the default BP neural network model will not over-fitting, and then can guarantee model training essence
While spending, reduce Error Margin of Short-Term Electric Load Forecasting.The application also provides a kind of Short Term Load Forecasting System, equipment and computer can
Storage medium is read, all has above-mentioned beneficial effect, details are not described herein.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The embodiment of application for those of ordinary skill in the art without creative efforts, can also basis
The attached drawing of offer obtains other attached drawings.
Fig. 1 is a kind of flow chart of short-term load forecasting method provided by the embodiment of the present application；
Fig. 2 is a kind of BP neural network structural schematic diagram provided by the embodiment of the present application；
Fig. 3 is a kind of neuronal structure schematic diagram provided by the embodiment of the present application；
Fig. 4 is a kind of structural block diagram of Short Term Load Forecasting System provided by the embodiment of the present application.
Specific embodiment
The core of the application is to provide a kind of short-term load forecasting method, can while guaranteeing model training precision,
Reduce Error Margin of Short-Term Electric Load Forecasting.Another core of the application is to provide a kind of Short Term Load Forecasting System, equipment and computer
Readable storage medium storing program for executing.
To keep the purposes, technical schemes and advantages of the embodiment of the present application clearer, below in conjunction with the embodiment of the present application
In attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is
Some embodiments of the present application, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art
Every other embodiment obtained without making creative work, shall fall in the protection scope of this application.
Currently, often facing when carrying out short-term load forecasting using BP neural network intelligent algorithm due to correlation
" over-fitting " problem caused by factor is more, historical data is complicated, although causing the training stage that can obtain higher training essence
Degree, and the problem that forecast period test error is larger.A kind of short-term load forecasting method provided by the present application, can be in Lifting Modules
While type training precision, reduce Error Margin of Short-Term Electric Load Forecasting, it is specific referring to FIG. 1, Fig. 1 is provided by the embodiment of the present application
A kind of flow chart of short-term load forecasting method, the short-term load forecasting method specifically include:
S101, the index parameter for obtaining each index in demand history data；
The embodiment of the present application obtained the index parameter of each index in demand history data before this, at this to the demand history
The amount of data is not especially limited, and history duration corresponding for demand history data is also not especially limited, should be by this field
Technical staff makes corresponding setting according to the actual situation.The demand history data of the embodiment of the present application are the fingers by many indexes
Parameter composition is marked, so monitoring the index parameter of many indexes when obtaining the demand history data.At this to the quantity of index
It is not especially limited with type, corresponding setting should be made according to the actual situation by those skilled in the art, These parameters can
Think meteorology, distributed generation resource electricity etc..
S102, the index parameter of each index is inputted to default BP neural network model, each finger after output normalization
Target short-term load forecasting result.Wherein, the deviation for presetting BP neural network model is by the inclined of increase complexity convergent
What poor calculation formula obtained, and deviation is not more than deviation convergence threshold.
The embodiment of the present application obtain demand history data in each index index parameter after, by the index of each index
Parameter inputs default BP neural network model, the short-term load forecasting result of each index after output normalization.At this to pre-
If the acquisition modes of BP neural network model are not especially limited, usually by obtained from model training.The default BP mind
Deviation through network model is to be obtained by the deviation calculation formula of increase complexity convergent, and deviation is not more than deviation
Convergence threshold is not especially limited deviation convergence threshold at this, should make according to the actual situation phase by those skilled in the art
The setting answered.
Further, the above-mentioned index parameter by each index inputs default BP neural network model, after output normalization
Each index short-term load forecasting result, comprising: to original loads training data carry out data normalization handle to obtain
After weight training data, the connection weight and amount of bias of each neuron in BP neural network are determined using weight training data；
The connection weight of each neuron and amount of bias are being substituted into corresponding numerical value conversion function, obtaining the first BP neural network model
Afterwards, the load verification data handled by data normalization are inputted into the first BP neural network model, obtains output result；
It is restrained after constructing complexity convergent using complexity in the connection weight for utilizing each neuron and preset weight coefficient
Item and original deflection calculation formula, determination deviation calculation formula；It is calculated load verification data and output result are substituted into deviation
Formula, after obtaining deviation, whether judgment bias value is greater than deviation convergence threshold；If it is not, then by the first BP neural network model
It is determined as default BP neural network model；The index parameter of each index is inputted into default BP neural network model, exports normalizing
The short-term load forecasting result of each index after change.
Data normalization mentioned above is handled differences such as load, temperature involved in short-term load forecasting, wind-force
The data of dimension are converted to the dimensionless number between [0,1], to promote the convergence rate in neural network training process.Herein
Data normalization processing method is not especially limited, the methods of linear transformation, bipolar conversion can be used in data normalization, linearly
Conversion method uses more universal, and formula may be expressed as:
Wherein, x in formulaUNor、xNorRespectively indicate the numerical value of normalization front and back index, xUN,max、xUN,minRespectively non-normalizing
Change the value bound of the index.The basic structure of above-mentioned BP neural network is as shown in Fig. 2, Fig. 2 is mentioned by the embodiment of the present application
A kind of BP neural network structural schematic diagram supplied, as shown in Figure 2, BP neural network structure contains input layer, hidden layer, output
Layer, wherein input layer includesnA input parameter；Hidden layer includeshA neuron, it is correspondinghA hidden layer output；Output layer includes m
A neuron, corresponding m output result.x1、x2…xnRespectively indicate input item, y1、y2....yhRespectively indicate neuron, z1、
z2…zmRespectively indicate output item.The structure of each neuron is as shown in figure 3, Fig. 3 is a kind of mind provided by the embodiment of the present application
Through meta structure schematic diagram, each neuron include weighted sum, increase biasing, numerical value conversion three operation, input item with it is defeated
N-th-trem relation n may be expressed as: out
In formula,yiFor neuroniOutput item, x1、x2、……xnThe respectively input item of neuron i, wi1、wi2、……
winFor the corresponding connection weight of each input item of neuron i, θ is amount of bias, and f is numerical value conversion function, and threshold function table, S can be used
The expression-forms such as shape function are manually selected by load prediction personnel.
The indexs such as meteorology, distributed power generation amount are optional in historical data does input item, and short term to be predicted is pre-
Index is surveyed as output item, then output valve can successively be calculated according to above-mentioned BP neural network.
The output item of hidden layer may be expressed as:
In formula f IMThe respectively output item of hidden layer i and numerical value transfer function,Corresponding hidden layer nerve
The connection weight and amount of bias of member.
The output item of output layer may be expressed as:
In formula f OUThe respectively output item of output layer i and numerical value transfer function,Corresponding output layer nerve
The connection weight and amount of bias of member.
Further, above-mentioned to handle to obtain weight training data to original loads training data progress data normalization
Afterwards, the connection weight and amount of bias that each neuron in BP neural network is determined using weight training data, are generally included: according to
Linear transformation formula carries out data normalization processing to original loads training data, obtains weight training data；It is instructed using load
Practice connection weight and amount of bias that data determine each neuron in BP neural network.
From the foregoing, it can be understood that the embodiment of the present application is in the connection weight for utilizing each neuron and preset weight coefficient, structure
After building complexity convergent, complexity convergent and original deflection calculation formula, determination deviation calculation formula are utilized.Original deflection
Calculation formula is the deviation determined between the output valve and the corresponding numerical value of historical data of study stage output layer, can be indicated such as
Under:
ε ' is the calculating deviation under traditional mode in formula,For the historical values of output item,For corresponding output item
Calculated value.And when using deviation calculation method corresponding to the formula, due to input item, historical data is more and is easy to produce
" over-fitting " problem.Deviation calculation method is adjusted by the present invention thus, supplemented with the net being calculated by connection weight
Network complexity, revised deviation may be expressed as:
ε is deviation calculation method proposed by the invention in formula, and i ∈ BP indicates all nerves in the BP neural network
Member, j ∈ i indicate all connection weights for belonging to neuron in BP neural network, wijFor the numerical value of the connection weight, λ is to introduce
Weight coefficient, meet 0 < λ < 1, by manually giving specific numerical value, when level off to 1 when, then will be promoted the study stage effect
Rate, and when level off to 0 when, complexity will be reduced, to weaken the possibility occurrence of " over-fitting ".
If meetingε≤ε0, then show to restrain, namely indicate that above-mentioned first BP neural network model has reached the use of the application
Condition can go to working stage；Otherwise, show not restrain, BP neural network is modified according to deviation.Wherein,ε0For people
The given deviation convergence threshold of work.Further, if judging, deviation is greater than deviation convergence threshold, generally includes: according to repairing
Positive formula is modified operation to the connection weight and amount of bias of each neuron of the first BP neural network model, is corresponded to
Connection weight correction value and amount of bias correction value.
The correction formula of each layer neuron can be calculated according to steepest descent algorithm.It is not repeated in the present invention to push away
Process is led, correction result is directly given:
Formula (1) is into formula (4)Respectively represent hidden layer, the connection weight of output layer is repaired
Positive value and amount of bias correction value, η are correction factor, and value meets0 < η < 1, for specific value by manually giving, numerical value is bigger, amendment
Speed is faster, but stability is lower, the preferred value 0.4 of the present invention.
From the foregoing, it can be understood that the index parameter of each index is inputted default BP neural network model by the application, normalizing is exported
The short-term load forecasting of each index after change is as a result, be set as xF,Nor, then the resulting result of data convert meets:
xF,UNor=(xUN,max-xUN,min)xF,Nor+xUN,min
In formula, xF,UNorFor load prediction numerical result resulting after data convert.
The deviation of default BP neural network model in the application is to calculate public affairs by the deviation of increase complexity convergent
What formula obtained, and deviation is not more than deviation convergence threshold, can guarantee that the complexity of the default BP neural network model will not
It is excessive, can also guarantee the default BP neural network model will not over-fitting, and then can guarantee model training precision
Meanwhile reducing Error Margin of Short-Term Electric Load Forecasting.The application is increased again by restraining decision stage in traditional BP neural network model
Miscellaneous degree convergent, a possibility that realizing the coordination between training effectiveness and training pattern complexity, reduce over-fitting.
Below to a kind of Short Term Load Forecasting System provided by the embodiments of the present application, equipment and computer readable storage medium
It is introduced, Short Term Load Forecasting System, equipment and computer readable storage medium described below and above-described short-term
Load forecasting method can correspond to each other reference.
Referring to FIG. 4, Fig. 4 is a kind of structural block diagram of Short Term Load Forecasting System provided by the embodiment of the present application；It should
Short Term Load Forecasting System includes:
Index parameter obtains module 401, for obtaining the index parameter of each index in demand history data；
Short-term load forecasting result output module 402, for the index parameter of each index to be inputted default BP nerve net
Network model, the short-term load forecasting result of each index after output normalization.Wherein, the deviation of BP neural network model is preset
Value is to be obtained by the deviation calculation formula of increase complexity convergent, and deviation is not more than deviation convergence threshold.
Based on the above embodiment, the present embodiment middle or short term load prediction results output module 402, generally includes:
Connection weight and amount of bias determination unit, for being handled to original loads training data progress data normalization
To after weight training data, the connection weight and biasing of each neuron in BP neural network are determined using weight training data
Amount；
As a result output unit, for the connection weight of each neuron and amount of bias to be substituted into corresponding numerical value conversion letter
The load verification data handled by data normalization after obtaining the first BP neural network model, are inputted the first BP by number
Neural network model obtains output result；
Deviation calculation formula determination unit, for using each neuron connection weight and preset weight coefficient,
After constructing complexity convergent, complexity convergent and original deflection calculation formula, determination deviation calculation formula are utilized；
Judging unit, for load verification data and output result to be substituted into deviation calculation formula, after obtaining deviation,
Whether judgment bias value is greater than deviation convergence threshold；
Default BP neural network model determination unit, if being not more than deviation convergence threshold for deviation, by the first BP
Neural network model is determined as default BP neural network model；
Short-term load forecasting result output unit, for the index parameter of each index to be inputted default BP neural network mould
Type, the short-term load forecasting result of each index after output normalization.
Based on the above embodiment, the present embodiment middle or short term load prediction results output module 402, generally includes:
Operating unit is corrected, if being greater than deviation convergence threshold for deviation, according to correction formula to the first BP nerve
The connection weight and amount of bias of each neuron of network model are modified operation, obtain corresponding connection weight correction value and partially
The amount of setting correction value.
Based on the above embodiment, connection weight and amount of bias determination unit in the present embodiment, generally include:
Data normalization handles subelement, returns for carrying out data to original loads training data according to linear transformation formula
One change processing, obtains weight training data；
Connection weight and amount of bias determine subelement, for determining each mind in BP neural network using weight training data
Connection weight and amount of bias through member.
The application also provides a kind of equipment, comprising: memory and processor；Wherein, memory is for storing computer journey
The step of sequence, processor is for realizing the short-term load forecasting method of above-mentioned any embodiment when executing computer program.
The application also provides a kind of computer readable storage medium, and computer-readable recording medium storage has computer journey
Sequence, the step of short-term load forecasting method of above-mentioned any embodiment is realized when computer program is executed by processor.
The computer readable storage medium may include: USB flash disk, mobile hard disk, read-only memory (Read-Only
Memory, ROM), random access memory (Random Access Memory, RAM), magnetic or disk etc. is various to deposit
Store up the medium of program code.
Each embodiment is described in a progressive manner in specification, the highlights of each of the examples are with other realities
The difference of example is applied, the same or similar parts in each embodiment may refer to each other.For embodiment provide system and
Speech, since it is corresponding with the method that embodiment provides, so being described relatively simple, related place is referring to method part illustration
?.
Professional further appreciates that, unit described in conjunction with the examples disclosed in the embodiments of the present disclosure
And algorithm steps, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These
Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession
Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered
Think beyond the scope of this invention.
The step of method described in conjunction with the examples disclosed in this document or algorithm, can directly be held with hardware, processor
The combination of capable software module or the two is implemented.Software module can be placed in random access memory (RAM), memory, read-only deposit
Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology
In any other form of storage medium well known in field.
Above to a kind of short-term load forecasting method, system, equipment and computer-readable storage medium provided herein
Matter is described in detail.Specific examples are used herein to illustrate the principle and implementation manner of the present application, above
The explanation of embodiment is merely used to help understand the present processes and its core concept.It should be pointed out that for the art
Those of ordinary skill for, under the premise of not departing from the application principle, can also to the application carry out it is several improvement and repair
Decorations, these improvement and modification are also fallen into the protection scope of the claim of this application.
Claims (10)
1. a kind of short-term load forecasting method characterized by comprising
Obtain the index parameter of each index in demand history data；
The index parameter of each index is inputted into default BP neural network model, it is each described after output normalization
The short-term load forecasting result of index；Wherein, the deviation of the default BP neural network model is restrained by increase complexity
What the deviation calculation formula of item obtained, and the deviation is not more than deviation convergence threshold.
2. short-term load forecasting method according to claim 1, which is characterized in that it is described will be described in each index
Index parameter inputs default BP neural network model, the short-term load forecasting of each index after output normalization as a result,
Include:
After carrying out data normalization to original loads training data and handling to obtain weight training data, the weight training is utilized
Data determine the connection weight and amount of bias of each neuron in BP neural network；
The connection weight of each neuron and the amount of bias are being substituted into corresponding numerical value conversion function, obtaining the
After one BP neural network model, the load verification data handled by the data normalization are inputted into the first BP mind
Through network model, output result is obtained；
In the connection weight for utilizing each neuron and preset weight coefficient, the complexity convergent is constructed
Afterwards, using the complexity convergent and original deflection calculation formula, the deviation calculation formula is determined；
The load verification data and the output result are being substituted into the deviation calculation formula, after obtaining the deviation,
Judge whether the deviation is greater than the deviation convergence threshold；
If it is not, the first BP neural network model is then determined as the default BP neural network model；
The index parameter of each index is inputted into the default BP neural network model, it is each after output normalization
The short-term load forecasting result of the index.
3. short-term load forecasting method according to claim 2, which is characterized in that if judging, the deviation is greater than institute
State deviation convergence threshold, comprising:
According to correction formula to the connection weight of each neuron of the first BP neural network model and described
Amount of bias is modified operation, obtains corresponding connection weight correction value and amount of bias correction value.
4. short-term load forecasting method according to claim 2, which is characterized in that described to original loads training data
It carries out data normalization to handle after obtaining weight training data, be determined using the weight training data each in BP neural network
The connection weight and amount of bias of neuron, comprising:
Data normalization processing is carried out to the original loads training data according to linear transformation formula, obtains the weight training
Data；
The connection weight of each neuron and institute in the BP neural network are determined using the weight training data
State amount of bias.
5. a kind of Short Term Load Forecasting System characterized by comprising
Index parameter obtains module, for obtaining the index parameter of each index in demand history data；
Short-term load forecasting result output module, for the index parameter of each index to be inputted default BP nerve net
Network model, the short-term load forecasting result of each index after output normalization；Wherein, the default BP neural network mould
The deviation of type is to be obtained by the deviation calculation formula of increase complexity convergent, and the deviation is restrained no more than deviation
Threshold value.
6. Short Term Load Forecasting System according to claim 5, which is characterized in that the short-term load forecasting result output
Module, comprising:
Connection weight and amount of bias determination unit, for handling and being born carrying out data normalization to original loads training data
After lotus training data, the connection weight and biasing of each neuron in BP neural network are determined using the weight training data
Amount；
As a result output unit, for the connection weight of each neuron and the amount of bias to be substituted into corresponding number
It is worth transfer function, after obtaining the first BP neural network model, the load verification number that the data normalization is handled will be passed through
According to the first BP neural network model is inputted, output result is obtained；
Deviation calculation formula determination unit, in the connection weight for utilizing each neuron and preset weight system
Number, after constructing the complexity convergent, using the complexity convergent and original deflection calculation formula, determines the deviation
Calculation formula；
Judging unit is obtained for the load verification data and the output result to be substituted into the deviation calculation formula
After the deviation, judge whether the deviation is greater than the deviation convergence threshold；
Default BP neural network model determination unit, if being not more than the deviation convergence threshold for the deviation, by institute
It states the first BP neural network model and is determined as the default BP neural network model；
Short-term load forecasting result output unit, for the index parameter input default BP of each index is refreshing
Through network model, the short-term load forecasting result of each index after output normalization.
7. Short Term Load Forecasting System according to claim 6, which is characterized in that the short-term load forecasting result output
Module, comprising:
Operating unit is corrected, if being greater than the deviation convergence threshold for the deviation, according to correction formula to described the
The connection weight and the amount of bias of each neuron of one BP neural network model are modified operation, obtain pair
The connection weight correction value and amount of bias correction value answered.
8. Short Term Load Forecasting System according to claim 6, which is characterized in that the connection weight and amount of bias determine
Unit, comprising:
Data normalization handles subelement, returns for carrying out data to the original loads training data according to linear transformation formula
One change processing, obtains the weight training data；
Connection weight and amount of bias determine subelement, each in the BP neural network for being determined using the weight training data
The connection weight and the amount of bias of a neuron.
9. a kind of equipment characterized by comprising
Memory and processor；Wherein, the memory is for storing computer program, the processor by execute it is described based on
The step of short-term load forecasting methods as described in any item such as Claims 1-4 are realized when calculation machine program.
10. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer journey
Sequence realizes such as Claims 1-4 described in any item short-term load forecasting methods when the computer program is executed by processor
The step of.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910249984.XA CN109978268A (en) | 2019-03-29 | 2019-03-29 | A kind of short-term load forecasting method, system and relevant apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910249984.XA CN109978268A (en) | 2019-03-29 | 2019-03-29 | A kind of short-term load forecasting method, system and relevant apparatus |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109978268A true CN109978268A (en) | 2019-07-05 |
Family
ID=67081685
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910249984.XA Pending CN109978268A (en) | 2019-03-29 | 2019-03-29 | A kind of short-term load forecasting method, system and relevant apparatus |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109978268A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110308658A (en) * | 2019-07-24 | 2019-10-08 | 广东工业大学 | A kind of pid parameter setting method, device, system and readable storage medium storing program for executing |
| CN110516889A (en) * | 2019-09-03 | 2019-11-29 | 广东电网有限责任公司 | A kind of load Comprehensive Prediction Method and relevant device based on Q-learning |
| CN111047191A (en) * | 2019-12-12 | 2020-04-21 | 广东电网有限责任公司 | Medium-and-long-term load prediction method and device for power market |
| CN111126565A (en) * | 2019-11-28 | 2020-05-08 | 广东电网有限责任公司 | Method and device for predicting block load density index based on deep learning |
| CN112531683A (en) * | 2020-11-19 | 2021-03-19 | 国网湖北省电力有限公司电力科学研究院 | Distribution network line load prediction method based on solution of Ornstein-Urnbek process |
| CN112561180A (en) * | 2020-12-21 | 2021-03-26 | 深圳大学 | Short-term wind speed prediction method and device, computer equipment and storage medium |
-
2019
- 2019-03-29 CN CN201910249984.XA patent/CN109978268A/en active Pending
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110308658A (en) * | 2019-07-24 | 2019-10-08 | 广东工业大学 | A kind of pid parameter setting method, device, system and readable storage medium storing program for executing |
| CN110308658B (en) * | 2019-07-24 | 2023-09-26 | 广东阿达智能装备有限公司 | PID parameter setting method, device and system and readable storage medium |
| CN110516889A (en) * | 2019-09-03 | 2019-11-29 | 广东电网有限责任公司 | A kind of load Comprehensive Prediction Method and relevant device based on Q-learning |
| CN110516889B (en) * | 2019-09-03 | 2023-07-07 | 广东电网有限责任公司 | Load comprehensive prediction method based on Q-learning and related equipment |
| CN111126565A (en) * | 2019-11-28 | 2020-05-08 | 广东电网有限责任公司 | Method and device for predicting block load density index based on deep learning |
| CN111047191A (en) * | 2019-12-12 | 2020-04-21 | 广东电网有限责任公司 | Medium-and-long-term load prediction method and device for power market |
| CN112531683A (en) * | 2020-11-19 | 2021-03-19 | 国网湖北省电力有限公司电力科学研究院 | Distribution network line load prediction method based on solution of Ornstein-Urnbek process |
| CN112561180A (en) * | 2020-12-21 | 2021-03-26 | 深圳大学 | Short-term wind speed prediction method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109978268A (en) | A kind of short-term load forecasting method, system and relevant apparatus | |
| CN111353656B (en) | Steel enterprise oxygen load prediction method based on production plan | |
| CN111027775A (en) | Step hydropower station generating capacity prediction method based on long-term and short-term memory network | |
| CN111460001B (en) | Power distribution network theoretical line loss rate evaluation method and system | |
| CN112183849A (en) | Short-term electric power quantity prediction method based on artificial neural network | |
| CN106127303A (en) | A kind of short-term load forecasting method towards multi-source data | |
| CN107563451A (en) | Running rate recognizing method under a kind of pumping plant steady state condition | |
| CN108921411A (en) | A kind of electric power accident emergency aid decision-making method of case-based reasioning | |
| CN104573857A (en) | Power grid load rate prediction method based on intelligent algorithm optimization and combination | |
| CN112819225A (en) | Carbon market price prediction method based on BP neural network and ARIMA model | |
| CN110866633A (en) | Micro-grid ultra-short term load prediction method based on SVR support vector regression | |
| CN110796293A (en) | Power load prediction method | |
| CN112803398A (en) | Load prediction method and system based on empirical mode decomposition and deep neural network | |
| CN112686693A (en) | Method, system, equipment and storage medium for predicting marginal electricity price of electric power spot market | |
| CN113762591A (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM counterstudy | |
| CN113361776A (en) | Power load probability prediction method based on user power consumption behavior clustering | |
| CN107679651A (en) | A kind of moon electricity demand forecasting method based on grey forecasting model Yu BP artificial nerve network models | |
| CN111563652A (en) | Method for evaluating power grid enterprise comprehensive plan execution condition | |
| CN112785022B (en) | Method and system for excavating electric energy substitution potential | |
| Sun et al. | Forecasting and identification of stock market based on modified RBF neural network | |
| CN113076696A (en) | Load short-term prediction method and prediction system based on IPSO-chaotic BP network | |
| Wang et al. | Short term load forecasting: A dynamic neural network based genetic algorithm optimization | |
| CN110956372A (en) | Power grid input-output marginal benefit analysis method and system | |
| Franco et al. | Very short-term bus reactive load forecasting models based on KDD approach | |
| Zhao et al. | Short-term load forecasting considering meteorological factors and electric vehicles |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190705 |
|
| RJ01 | Rejection of invention patent application after publication |