CN107239859A - The heating load forecasting method of Recognition with Recurrent Neural Network is remembered based on series connection shot and long term - Google Patents
The heating load forecasting method of Recognition with Recurrent Neural Network is remembered based on series connection shot and long term Download PDFInfo
- Publication number
- CN107239859A CN107239859A CN201710414757.9A CN201710414757A CN107239859A CN 107239859 A CN107239859 A CN 107239859A CN 201710414757 A CN201710414757 A CN 201710414757A CN 107239859 A CN107239859 A CN 107239859A
- Authority
- CN
- China
- Prior art keywords
- long term
- data
- shot
- series connection
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000010438 heat treatment Methods 0.000 title claims abstract description 46
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 26
- 230000000306 recurrent effect Effects 0.000 title claims abstract description 23
- 238000013277 forecasting method Methods 0.000 title claims abstract description 18
- 230000007774 longterm Effects 0.000 title claims abstract description 14
- 230000007787 long-term memory Effects 0.000 claims abstract description 20
- 230000003044 adaptive effect Effects 0.000 claims abstract description 7
- 230000003134 recirculating effect Effects 0.000 claims abstract description 7
- 238000005457 optimization Methods 0.000 claims abstract description 6
- 238000000034 method Methods 0.000 claims description 24
- 230000006870 function Effects 0.000 claims description 22
- 238000012549 training Methods 0.000 claims description 15
- 238000010606 normalization Methods 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 5
- 210000002569 neuron Anatomy 0.000 claims description 5
- 230000006403 short-term memory Effects 0.000 claims description 2
- 210000004218 nerve net Anatomy 0.000 claims 1
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 11
- 230000005284 excitation Effects 0.000 description 6
- 230000008034 disappearance Effects 0.000 description 4
- 230000007935 neutral effect Effects 0.000 description 4
- 241001269238 Data Species 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 230000005611 electricity Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012804 iterative process Methods 0.000 description 3
- 210000005036 nerve Anatomy 0.000 description 3
- 238000005295 random walk Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 101100001674 Emericella variicolor andI gene Proteins 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 230000006386 memory function Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000013519 translation Methods 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Operations Research (AREA)
- Primary Health Care (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Biomedical Technology (AREA)
- Entrepreneurship & Innovation (AREA)
- Computational Linguistics (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
The invention discloses a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term, choose temperature weather, the heating load data of preset time section, sample data set is built, the input data and output data concentrated to sample data are standardized respectively；Input data is divided into two parts, it is set to respectively enter two independent shot and long term memory Recognition with Recurrent Neural Network, then two shot and long term memory Recognition with Recurrent Neural Network are merged, output data enters back into next layer of shot and long term memory Recognition with Recurrent Neural Network, finally enters two full articulamentums；The series connection shot and long term memory recirculating network of structure is trained, optimized using the adaptive moments estimation algorithm of parameter optimization；By data input to be predicted series connection shot and long term memory recirculating network, heating load forecasting result is calculated.The present invention can effectively screen input data, accelerate pace of learning, improve learning efficiency, and improve precision of prediction.
Description
Technical field
The invention belongs to heating load prediction field, and in particular to one kind is based on series connection shot and long term memory circulation nerve
The heating load forecasting method of network.
Background technology
The generating set that is mainly characterized by of coal-fired thermal power coproduction unit both produces electric energy, and does work(using steam turbine generator
Steam to user carry out heat supply.Therefore, in northerly Heating Season, coal-fired thermal power coproduction unit is mainly carry to specifically
The groundwork of resident's central heating in area.At present, the operating principle of the cogeneration power plant of national regulation is " electricity determining by heat ",
I.e.：The need for steam power plant should be according to thermic load, determine optimal operation scheme, and to meet thermic load the need for for main target.
Regional electric administrative department formulate steam power plant's power scheduling curve when, it is necessary to take into full account heating demand curve and energy-conservation because
Element, the external heat supply of electricity index limitation steam power plant is not able to.Therefore, regional heating load (referred to as " heating demand ") is entered
Row Accurate Prediction, had both contributed to grid company reasonable arrangement power plants generating electricity, and the local power resources of optimization distribution, elevator net is coordinated
Level, it helps power plant optimizes electric generation management, improves generating efficiency.
Heating demand is mainly by when inside even from weather such as ground temperature, weather, wind speed, and with data volume is big, randomness
The characteristics of height, fast change.At present, the method for existing heating load forecasting mainly has recurrence, time series, wavelet analysis, grey
Model, BP neural network predicted method and SVMs method etc..These methods belong to shallow-layer learning method, it is impossible to deeply dig
Dig the randomness and nonlinear characteristic of heating demand data.
Shot and long term memory Recognition with Recurrent Neural Network (hereinafter referred to as " LSTM ") is a weight in recent years in deep learning field
Want achievement in research.LSTM hidden layer output is not only relevant with the first output of current terminal nerve, at the same also as it is next when
Between the cycle hidden layer input so that influence following output result.In addition, different from traditional RNN, LSTM is remembered by introducing
Recall neuron (being mainly characterized by being provided with input gate, forgeing three Rule of judgment of door and out gate for it) and solve the training time
The gradient disappearance problem of backpropagation during excessive cycle.Therefore, LSTM can be goed deep into macrocyclic time series data
Excavate.At present, this neutral net shows good effect in terms of natural language processing and machine translation.
The adaptive moments estimation algorithm (Adam) of parameter optimization can optimize to traditional gradient descent algorithm, and it leads to
The desired value for crossing estimation gradient avoids the random walk of gradient in itself, also reduces the wind that model converges to locally optimal solution
Danger.Meanwhile, Adam sets the upper limit by the Learning Step to iteration each time, makes the value ratio of parameter in each step iterative process
Relatively stablize.
The content of the invention
The present invention is in order to solve the above problems, it is proposed that one kind is based on series connection shot and long term memory Recognition with Recurrent Neural Network
The heating load forecasting method of (Concatenated LSTMs), the present invention can solve the problem that traditional heat supply Forecasting Methodology is difficult to deeply
Mining data makes the problem of precision of prediction is not high.
To achieve these goals, the present invention is adopted the following technical scheme that：
A kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term, is comprised the following steps：
(1) temperature weather, the heating load data of preset time section are chosen, sample data set is built, sample data is concentrated
Input data and output data be standardized respectively；
(2) input data is divided into two parts, it is respectively enterd two independent shot and long term memory Recognition with Recurrent Neural Network
(LSTMs), then it is merged, output data enters back into next layer of shot and long term memory Recognition with Recurrent Neural Network, finally enters
Two full articulamentums；
(3) the series connection shot and long term memory recirculating network of structure is trained, utilizes the adaptive moments estimation of parameter optimization
Algorithm is optimized；
(4) by data input to be predicted series connection shot and long term memory recirculating network, heating load forecasting result is calculated.
In the step (1), the sample data of the preset time section of selection is arranged, extracted, form input, output
Data set, wherein input data include：Temperature, air pressure, wind speed, humidity data and heating demand data under each time point；It is defeated
It is to lag behind the output data after input data time in several cycles to go out data.
In the step (1), input data and output data in data set take the form of matrix in structure.
In the step (1), input and output data are normalized using minimax value method for normalizing, simultaneously
The result after normalization is set to be mapped within [0,1].
In the step (2), the data after normalization are divided into two groups, temperature, air pressure, wind speed and humidity data are one
Group, heating demand is another group, builds Concatenated LSTMs.
In the step (2), LSTM introduces input gate to neuron, forgets three Rule of judgment of door and out gate, input
Door represents to allow information to be added to the ratio in mnemon；Forget door and represent to retain the history letter stored in current state node
The ratio of breath；Out gate represent using the information of current state node as output ratio.
In the step (2), the activation primitive selection Relu functions of LSTM compacted zone.
In the step (3), in training process, selection mean square deviation is target loss function, chooses Adam as under gradient
The optimized algorithm of drop.
In the step (3), Adam detailed processes include：
The loss function that (3-1) calculates each moment calculates the gradient at the moment；
(3-2) calculates single order moments estimation and second order moments estimation, and carries out unbiased amendment to it；
(3-3) updates corresponding parameter.
In the step (4), data to be predicted are normalized, and be entered into the series connection short-term memory nerve that training is drawn
Calculated in Internet it is normalized predict the outcome, calculate normalized predict the outcome.
Compared with prior art, beneficial effects of the present invention are：
1st, LSTM is analyzed applied to heating demand, helps to realize to big number using many hidden neurons of neutral net
According to depth excavate；Simultaneously because LSTM can control the choice of historical information by input gate, forgetting door and Memory-Gate, solve
Gradient disappearance problem in the back-propagation process for parameter optimization of having determined, promotes the convergence of parameter, improves the standard of data prediction
True property and the speed of prediction, also optimize the use of system resource.Finally, the out gate expression formula in this LSTM models is used
Status information c under tt, so can further be carried when judging output valve using newest status information
High-accuracy.
2nd, traditional gradient descent algorithm can be optimized with Adam, it is by estimating that the desired value of gradient is avoided
The random walk of gradient in itself, also reduces the risk that model converges to locally optimal solution.Meanwhile, Adam passes through to each time
The Learning Step of iteration sets the upper limit, makes the value of parameter in each step iterative process more stable, it is to avoid gradient blast
Problem.
3rd, the Concatenated LSTMs of design can effectively screen input data, accelerate pace of learning, improve and learn
Efficiency is practised, precision of prediction is improved.
Brief description of the drawings
The Figure of description for constituting the part of the application is used for providing further understanding of the present application, and the application's shows
Meaning property embodiment and its illustrate be used for explain the application, do not constitute the improper restriction to the application.
Fig. 1 is the Concatenated LSTMs schematic diagrames of the present invention；
Fig. 2 is the heating load forecasting schematic flow sheet of the present invention；
Fig. 3 is the LSTM neuronal structure figures of the present invention.
Embodiment：
The invention will be further described with embodiment below in conjunction with the accompanying drawings.
It is noted that described further below is all exemplary, it is intended to provide further instruction to the application.Unless another
Indicate, all technologies used herein and scientific terminology are with usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singulative
It is also intended to include plural form, additionally, it should be understood that, when in this manual using term "comprising" and/or " bag
Include " when, it indicates existing characteristics, step, operation, device, component and/or combinations thereof.
As background technology is introduced, it is difficult to go deep into mining data in the presence of traditional heat supply Forecasting Methodology in the prior art to make
The not high deficiency of precision of prediction, in order to solve technical problem as above, Concatenated is based on present applicant proposes one kind
LSTMs heating load forecasting method.
A kind of heating load forecasting method optimized based on Concatenated LSTM and Adam, it is characterised in that including
Following steps：
Step 1: choosing given area, temperature weather, heating load data are collected, sample data set is built；
Step 2: being standardized respectively to the input data and output data in sample；
Step 3: input data is divided into two parts, it is set to respectively enter two independent LSTM neutral nets, it is then right
Two LSTM are merged, and output data enters back into next layer of LSTM, finally enter two full articulamentums；
Step 4：By data input ConcatenatedLSTMs to be predicted, heating load forecasting result is calculated.
Step one is specifically that the sample data of the preset time section of selection is arranged, extracted, and forms input, output number
According to collection.Wherein input data includes：Temperature, air pressure, wind speed, humidity data and heating demand data under some time point；Output
Data are to lag behind the output data after the n cycle of input data time.Inputoutput data takes the shape of matrix in structure
Formula.
Step 2 is specifically that input and output data are normalized using minimax value method for normalizing, is made simultaneously
Result after normalization is mapped to[0,1]Within.Normalized purpose is to eliminate input data dimension difference to the generation that predicts the outcome
Influence.
The function of minimax value method for normalizing can be expressed as：
Step 3 be specifically by data after the normalization in step 2 according to (temperature, air pressure, wind speed, humidity data) and
(heating demand) is divided into two groups of input datas, builds Concatenated LSTMs.
Wherein：LSTM solves the time by introducing input gate to neuron, forgeing three Rule of judgment of door and out gate
The gradient disappearance problem of backpropagation when step-length is long.Wherein：Input gate represents to allow information to be added to the ratio in mnemon
Example；Forget door to represent to retain the ratio of the historical information stored in current state node；Out gate is represented current state node
Information as output ratio.The excitation function of three doors represented using sigmoid functions, span 0 to 1 it
Between, expression formula is as follows：
In LSTM structures, it is x=[x to make list entries1,x2,x3,.....,xT], wherein T represents the length of time series
Degree, xtRepresent the input value of t；Output sequence is y=[y1,y2,y3,.....,yT], ytRepresent the output valve of t；Shape
State sequence is c=[c1,c2,c3,.....,cT], ctRepresent the state value of t.So t, out gate it, forget door
ftWith out gate otExpression formula be respectively：
it=sig (Wi·[yt-1,ct-1,xt]+bi) (3)
ft=sig (Wf·[yt-1,ct-1,xt]+bf) (4)
ot=sig (Wo·[yt-1,ct,xt]+bo) (5)
Here, Wi、Wf、WoInput gate is represented respectively, forget the weight matrix of door and Memory-Gate；bi、bf、boRepresent respectively defeated
Introduction, the constant for forgeing door and Memory-Gate；Sig table shows the excitation function of three doors, sees formula (2).In addition, out gate table here
Status information c under having used t up to formulat, so can when judging output valve using newest status information,
Improve accuracy rate.
According to the expression formula of input gate, forgetting door and Memory-Gate, it can be deduced that ctAnd ytExpression formula, it is as follows：
Wherein：The pointwise of element is multiplied in representing matrix, WcAnd bcThe state weight and state of neutral net are represented respectively
Constant；Tanh represents the excitation function from state node to output node, and its expression formula is：
Tanh (x)=2sig (x) -1 (8)
The activation primitive of compacted zone (Dense) selects Relu functions, and expression formula is：
F (x)=max (x, 0) (9)
Here, the advantage using fine and close layer functions is to accelerate the gradient descent method convergence rate in training process.
, it is necessary to define the optimized algorithm that target loss function and gradient decline in the training process of step 3.Here, select
It is target loss function to take mean square deviation (Meansquareerror), choose adaptive moments estimation Adam as gradient decline it is excellent
Change algorithm.Adam dynamically adjusts the learning rate of each parameter, its advantage master using the single order moments estimation and second order moments estimation of gradient
It is, by estimating that the desired value of gradient avoids the random walk of gradient in itself, to also reduce model and converge to local optimum
The risk of solution；Meanwhile, the upper limit is set by the Learning Step to iteration each time, makes the value of parameter in each step iterative process
It is more stable, it is to avoid the problem of gradient is exploded.
Adam algorithm main process is：
The first step：Calculate the gradient of t：
Wherein ftFor the loss function of t.
Second step：Calculate single order moments estimation：
mt=β1mt-1+(1-β1)gt (11)
The purpose of this step is to accelerate learning process while breaking away from locally optimal solution.
3rd step：Calculate second order moments estimation：
Purpose of this step is to estimate E [g2], provide the Learning Step upper limit, it is to avoid excessive Learning Step occur.By
InTherefore, when parameter be withWhen mode updates, the upper limit of Learning Step is exactly α.
4th step：Calculate the single order moments estimation of unbiased amendment：
5th step：Calculate the second order moments estimation of unbiased amendment：
6th step：Undated parameter：
Step 4 is specifically normalization data to be predicted, and is entered into the LSTM that training is drawn and calculates normalization
Predict the outcome, calculate normalized predict the outcome.Finally, inverse normalization is carried out to predicting the outcome, prediction conclusion is drawn.
In a kind of typical embodiment of the application, step one：Temperature, air pressure, wind speed, the humidity number of selection 3 months
According to heat supply data, data cycle is takes a numerical value in every 10 minutes；Output data is to lag behind the input data time 144
The heating load in individual cycle.This is in order that model can disposably export following 24 hours heating demand data.
Step 2：Specifically input and output data are normalized using minimax value method for normalizing, simultaneously
The result after normalization is set to be mapped within [0,1].Normalize formula selection as follows：
Especially, it is necessary to which whole training datas are normalized, normalized dimension is every for input data respectively
What the numerical value of one row was set.Because, each column data represents an index, only for belonging in this indication range
Data be normalized just it is meaningful.
Step 3：Data after the completion of being normalized in step 2 are divided into 2 classes, are respectivelyI.e.：Temperature, air pressure,
Wind speed, humidity data andI.e.：Heating demand.
LSTM_1, LSTM_2 layers of output dimension are set to be 200 for two class input datas respectively；By the god of two above layer
It is 100 to be merged through network and be input to LSTM_3 layer networks output dimension；LSTM_3 and first layer compacted zone Dense_1 is connected
Connect, while activation primitive Relu is set for the compacted zone, and output dimension is 32；By Dense_1 and second layer compacted zone
Dense_2 connections, are not provided with activation primitive, but output dimension is consistent with time step (seeing below section).
The batch processing quantity (batch_size) of input data is set to 5.Set batch processing quantity purpose be in order to point
Incoming training data is criticized to improve the service efficiency of internal memory.Time step is set to 6, and the purpose that step-length is set is to utilize
LSTM memory function, extracts the characteristic information of regularity from historical data.In order to avoid the gradient in back-propagation process
Disappearance problem, using LSTM network structure, wherein：The excitation function of input gate, forgetting door and out gate three is used
Sigmoid functions represent that span is between 0 to 1, expression formula is as follows：
In LSTM structures, it is x=[x to make list entries1,x2,x3,.....,xT], wherein T represents the length of time series
Degree, xtRepresent the input value of t；Output sequence is y=[y1,y2,y3,.....,yT], ytRepresent the output valve of t；Shape
State sequence is c=[c1,c2,c3,.....,cT], ctRepresent the state value of t.So t, out gate it, forget door
ftWith out gate otExpression formula be respectively：
it=sig (Wi·[yt-1,ct-1,xt]+bi) (18)
ft=sig (Wf·[yt-1,ct-1,xt]+bf) (19)
ot=sig (Wo·[yt-1,ct,xt]+bo) (20)
Here, Wi、Wf、WoInput gate is represented respectively, forget the weight matrix of door and Memory-Gate；bi、bf、boRepresent respectively defeated
The constant of the weight of introduction, forgetting door and Memory-Gate；Sig table shows the excitation function of three doors.In addition, output gate expression here
Formula has used the status information c under tt, so can be when judging output valve using newest status information,
Improve accuracy rate.
According to the expression formula of input gate, forgetting door and Memory-Gate, it can be deduced that ctAnd ytExpression formula, it is as follows：
Wherein：The pointwise of element is multiplied in representing matrix, WcAnd bcGod used when representing to calculate current state node respectively
State weight and state constant through network；Tanh represents the excitation function from state node to output node, and its expression formula is：
Tanh (x)=2sig (x) -1 (23)
LSTM training algebraically (epoch) is set as 150 times, while all to last time pair after every generation terminates
Status information cepochRecorded, and be transferred to training of future generation until completing all training.
In the training process of step 3, it is target loss function to choose mean square deviation (Meansquareerror), is selected from
The optimized algorithm that moments estimation (Adam) algorithm declines as gradient is adapted to, is comprised the concrete steps that：
Initial value is set：α=0.001, β1=0.9, β2=0.999, ε=10-8
The first step：Calculate the gradient of t：
Wherein ftFor the loss function of t.
Second step：Calculate single order moments estimation：
mt=β1mt-1+(1-β1)gt (25)
3rd step：Calculate second order moments estimation：
4th step：Calculate the single order moments estimation of unbiased amendment：
5th step：Calculate the second order moments estimation of unbiased amendment：
6th step：Undated parameter：
Step 4：Data to be predicted are normalized, and are entered into calculating in the ConcatenatedLSTMs that training is drawn
Go out normalized predict the outcome.Finally, inverse normalization is carried out to predicting the outcome, prediction conclusion is drawn.
The preferred embodiment of the application is the foregoing is only, the application is not limited to, for the skill of this area
For art personnel, the application can have various modifications and variations.It is all within spirit herein and principle, made any repair
Change, equivalent substitution, improvement etc., should be included within the protection domain of the application.
Although above-mentioned the embodiment of the present invention is described with reference to accompanying drawing, not to present invention protection model
The limitation enclosed, one of ordinary skill in the art should be understood that on the basis of technical scheme those skilled in the art are not
Need to pay various modifications or deform still within protection scope of the present invention that creative work can make.
Claims (10)
1. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term, it is characterized in that：Including following
Step：
(1) temperature weather, the heating load data of preset time section are chosen, sample data set is built, what sample data was concentrated is defeated
Enter data and output data is standardized respectively；
(2) input data is divided into two parts, it is respectively enterd two independent shot and long term memory Recognition with Recurrent Neural Network, then
Two shot and long term memory Recognition with Recurrent Neural Network are merged, output data enters back into next layer of shot and long term memory circulation nerve net
Network, finally enters two full articulamentums；
(3) the series connection shot and long term memory recirculating network of structure is trained, utilizes the adaptive moments estimation algorithm of parameter optimization
Optimize；
(4) by data input to be predicted series connection shot and long term memory recirculating network, heating load forecasting result is calculated.
2. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term as claimed in claim 1,
It is characterized in that：In the step (1), the sample data of the preset time section of selection is arranged, extracted, form input, defeated
Go out data set, wherein input data includes：Temperature, air pressure, wind speed, humidity data and heating demand data under each time point；
Output data is to lag behind the output data after input data time in several cycles.
3. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term as claimed in claim 1,
It is characterized in that：In the step (1), input data and output data in data set take the form of matrix in structure.
4. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term as claimed in claim 1,
It is characterized in that：In the step (1), input and output data are normalized using minimax value method for normalizing, together
When the result after normalization is mapped within [0,1].
5. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term as claimed in claim 1,
It is characterized in that：In the step (2), the data after normalization are divided into two groups, temperature, air pressure, wind speed and humidity data are one
Group, heating demand is another group, builds series connection shot and long term memory recirculating network.
6. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term as claimed in claim 1,
It is characterized in that：In the step (2), shot and long term Memory Neural Networks layer introduces input gate to neuron, forgets door and out gate
Three Rule of judgment, input gate represents to allow information to be added to the ratio in mnemon；Forget door to represent to retain current state
The ratio of the historical information stored in node；Out gate represent using the information of current state node as output ratio.
7. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term as claimed in claim 1,
It is characterized in that：In the step (2), the activation primitive selection Relu functions of the compacted zone of shot and long term Memory Neural Networks layer.
8. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term as claimed in claim 1,
It is characterized in that：In the step (3), in training process, selection mean square deviation is target loss function, chooses adaptive moments estimation and calculates
The optimized algorithm that method declines as gradient.
9. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term as claimed in claim 1,
It is characterized in that：In the step (3), adaptive moments estimation algorithm detailed process includes：
The loss function that (3-1) calculates each moment calculates the gradient at the moment；
(3-2) calculates single order moments estimation and second order moments estimation, and carries out unbiased amendment to it；
(3-3) updates corresponding parameter.
10. a kind of heating load forecasting method that Recognition with Recurrent Neural Network is remembered based on series connection shot and long term as claimed in claim 1,
It is characterized in that：In the step (4), data to be predicted are normalized, and be entered into the series connection short-term memory god that training is drawn
Through calculated in Internet it is normalized predict the outcome, calculate normalized predict the outcome.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710414757.9A CN107239859B (en) | 2017-06-05 | 2017-06-05 | Heating load forecasting method based on series connection shot and long term memory Recognition with Recurrent Neural Network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710414757.9A CN107239859B (en) | 2017-06-05 | 2017-06-05 | Heating load forecasting method based on series connection shot and long term memory Recognition with Recurrent Neural Network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107239859A true CN107239859A (en) | 2017-10-10 |
| CN107239859B CN107239859B (en) | 2018-05-08 |
Family
ID=59984938
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710414757.9A Active CN107239859B (en) | 2017-06-05 | 2017-06-05 | Heating load forecasting method based on series connection shot and long term memory Recognition with Recurrent Neural Network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107239859B (en) |
Cited By (47)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107909206A (en) * | 2017-11-15 | 2018-04-13 | 电子科技大学 | A kind of PM2.5 Forecasting Methodologies based on deep structure Recognition with Recurrent Neural Network |
| CN108051035A (en) * | 2017-10-24 | 2018-05-18 | 清华大学 | The pipe network model recognition methods of neural network model based on gating cycle unit |
| CN108319132A (en) * | 2018-01-11 | 2018-07-24 | 合肥工业大学 | Decision system and method for unmanned plane aerial opposition |
| CN108416690A (en) * | 2018-01-19 | 2018-08-17 | 中国矿业大学 | Load Forecasting based on depth LSTM neural networks |
| CN108648829A (en) * | 2018-04-11 | 2018-10-12 | 平安科技（深圳）有限公司 | Disease forecasting method and device, computer installation and readable storage medium storing program for executing |
| CN108711847A (en) * | 2018-05-07 | 2018-10-26 | 国网山东省电力公司电力科学研究院 | A kind of short-term wind power forecast method based on coding and decoding shot and long term memory network |
| CN108921341A (en) * | 2018-06-26 | 2018-11-30 | 国网山东省电力公司电力科学研究院 | A kind of steam power plant's short term thermal load forecasting method encoded certainly based on gate |
| CN108920888A (en) * | 2018-04-26 | 2018-11-30 | 浙江工业大学 | It is a kind of that autoclave reaction process discrimination method is continuously stirred based on deep neural network |
| CN109002942A (en) * | 2018-09-28 | 2018-12-14 | 河南理工大学 | A kind of short-term load forecasting method based on stochastic neural net |
| CN109063939A (en) * | 2018-11-01 | 2018-12-21 | 华中科技大学 | A kind of wind speed forecasting method and system based on neighborhood door shot and long term memory network |
| CN109146162A (en) * | 2018-08-07 | 2019-01-04 | 河海大学 | A kind of probability wind speed forecasting method based on integrated Recognition with Recurrent Neural Network |
| CN109165727A (en) * | 2018-09-04 | 2019-01-08 | 成都品果科技有限公司 | A kind of data predication method based on Recognition with Recurrent Neural Network |
| CN109242236A (en) * | 2018-07-05 | 2019-01-18 | 西安图迹信息科技有限公司 | Based on the PMU primary frequency modulation load forecasting method for being associated with full Connection Neural Network and LSTM |
| CN109253985A (en) * | 2018-11-28 | 2019-01-22 | 东北林业大学 | The method of near infrared light spectrum discrimination Chinese zither panel grading of timber neural network based |
| CN109325624A (en) * | 2018-09-28 | 2019-02-12 | 国网福建省电力有限公司 | A kind of monthly electric power demand forecasting method based on deep learning |
| CN109492839A (en) * | 2019-01-17 | 2019-03-19 | 东华大学 | A kind of mineral hot furnace operating condition prediction technique based on RNN-LSTM network |
| CN109583565A (en) * | 2018-11-07 | 2019-04-05 | 河海大学 | Forecasting Flood method based on the long memory network in short-term of attention model |
| CN109670629A (en) * | 2018-11-16 | 2019-04-23 | 浙江蓝卓工业互联网信息技术有限公司 | Coal-burning boiler thermal efficiency forecast method based on shot and long term Memory Neural Networks |
| CN109936464A (en) * | 2017-12-15 | 2019-06-25 | 中国移动通信集团浙江有限公司 | The open load forecasting method of communication capacity based on multilayer LSTMRNN and device |
| CN110007613A (en) * | 2019-04-10 | 2019-07-12 | 北方工业大学 | Warming prediction method and system for heat storage type electric heater and storage medium |
| CN110084357A (en) * | 2018-01-25 | 2019-08-02 | 三星电子株式会社 | Accelerate shot and long term memory network via selectivity trimming |
| CN110135634A (en) * | 2019-04-29 | 2019-08-16 | 广东电网有限责任公司电网规划研究中心 | Long-medium term power load forecasting device |
| CN110363289A (en) * | 2019-07-17 | 2019-10-22 | 山东浪潮人工智能研究院有限公司 | A kind of industrial steam amount prediction technique and device based on machine learning |
| CN110381524A (en) * | 2019-07-15 | 2019-10-25 | 安徽理工大学 | The mobile flow on-line prediction method of large scene based on Bi-LSTM, system and storage medium |
| CN110428053A (en) * | 2019-08-08 | 2019-11-08 | 东北大学 | A kind of steam produces the dynamic prediction method of consumption |
| CN110619389A (en) * | 2019-09-23 | 2019-12-27 | 山东大学 | Load prediction method and system of combined cooling heating and power system based on LSTM-RNN |
| CN110705743A (en) * | 2019-08-23 | 2020-01-17 | 国网浙江省电力有限公司 | New energy consumption electric quantity prediction method based on long-term and short-term memory neural network |
| CN110889494A (en) * | 2019-11-29 | 2020-03-17 | 北京龙田华远科技有限公司 | Coal mining machine positioning method and device |
| CN110991761A (en) * | 2019-12-13 | 2020-04-10 | 新奥数能科技有限公司 | Heat supply load prediction method and device |
| CN111160659A (en) * | 2019-12-31 | 2020-05-15 | 国家电网公司西南分部 | Power load prediction method considering temperature fuzzification |
| CN111222677A (en) * | 2019-10-22 | 2020-06-02 | 浙江运达风电股份有限公司 | Wind speed prediction method and system based on long-short term memory time neural network |
| CN111275169A (en) * | 2020-01-17 | 2020-06-12 | 北京石油化工学院 | Method for predicting building thermal load in short time |
| CN111353631A (en) * | 2019-11-26 | 2020-06-30 | 国网山东省电力公司电力科学研究院 | Thermal power plant condenser vacuum degree prediction method based on multilayer LSTM |
| CN111613338A (en) * | 2020-05-19 | 2020-09-01 | 中国电子科技集团公司第三十八研究所 | Method and system for constructing spike-slow complex wave detection model |
| CN111738520A (en) * | 2020-06-24 | 2020-10-02 | 中国电子科技集团公司第二十八研究所 | System load prediction method fusing isolated forest and long-short term memory network |
| CN111861013A (en) * | 2020-07-23 | 2020-10-30 | 长沙理工大学 | Power load prediction method and device |
| CN111895986A (en) * | 2020-06-30 | 2020-11-06 | 西安建筑科技大学 | MEMS gyroscope original output signal noise reduction method based on LSTM neural network |
| CN111932015A (en) * | 2020-08-12 | 2020-11-13 | 武汉中电节能有限公司 | Method and system for predicting cold and heat load of regional cooling and heating |
| CN112598222A (en) * | 2020-11-30 | 2021-04-02 | 河南汇祥通信设备有限公司 | Comprehensive pipe rack operation and maintenance prediction method based on long-time and short-time memory cyclic neural network |
| CN112712189A (en) * | 2019-10-25 | 2021-04-27 | 北京市热力集团有限责任公司 | Heat supply demand load prediction method |
| CN112733457A (en) * | 2021-01-18 | 2021-04-30 | 武汉大学 | Load electricity utilization characteristic modeling method for improving double-layer long-short term memory network |
| CN112733452A (en) * | 2021-01-13 | 2021-04-30 | 北京嘀嘀无限科技发展有限公司 | Trajectory prediction method and apparatus, electronic device and readable storage medium |
| CN112832744A (en) * | 2021-01-07 | 2021-05-25 | 中国石油大学(华东) | Pumping unit well pump detection period prediction method based on LSTM neural network |
| CN113420508A (en) * | 2021-07-07 | 2021-09-21 | 华北电力大学 | Unit combination calculation method based on LSTM |
| CN113623719A (en) * | 2021-06-23 | 2021-11-09 | 国家电投集团东北电力有限公司大连开热分公司 | Heat exchange station prediction control method based on effective room temperature detection |
| CN116362139A (en) * | 2023-04-14 | 2023-06-30 | 应急管理部沈阳消防研究所 | Multi-parameter fire detection method based on hierarchical long-short-time memory network |
| CN116361709A (en) * | 2023-03-31 | 2023-06-30 | 山东省计算中心（国家超级计算济南中心） | Self-adaptive power load identification method and device |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4975287B2 (en) * | 2005-08-08 | 2012-07-11 | パナソニック株式会社 | Prediction device |
| KR101295098B1 (en) * | 2011-08-10 | 2013-08-09 | 아주대학교산학협력단 | Heating supply determination method for district heating network |
| CN103955768A (en) * | 2014-04-30 | 2014-07-30 | 河北省电力勘测设计研究院 | CSP radiation and heat energy predication method based on BP neural network model |
| CN104504451A (en) * | 2014-11-04 | 2015-04-08 | 国家电网公司 | Optimization method of day-ahead unit commitment of thermal power system of thermal power plant capable of operating under multiple states |
| WO2015138535A1 (en) * | 2014-03-11 | 2015-09-17 | General Electric Company | Systems and methods for utility crew forecasting |
| CN105389980A (en) * | 2015-11-09 | 2016-03-09 | 上海交通大学 | Short-time traffic flow prediction method based on long-time and short-time memory recurrent neural network |
| CN105528466A (en) * | 2014-09-28 | 2016-04-27 | 国家电网公司 | Wind power optimal planning modeling method considering adaptability and economy of power system |
| CN105805822A (en) * | 2016-03-24 | 2016-07-27 | 常州英集动力科技有限公司 | Heat supply energy saving control method and system based on neural network prediction |
| CN105913141A (en) * | 2016-04-06 | 2016-08-31 | 上海电器科学研究院 | Load prediction method supporting big data |
| CN106127303A (en) * | 2016-06-15 | 2016-11-16 | 国网山东省电力公司菏泽供电公司 | A kind of short-term load forecasting method towards multi-source data |
| CN106126492A (en) * | 2016-06-07 | 2016-11-16 | 北京高地信息技术有限公司 | Statement recognition methods based on two-way LSTM neutral net and device |
| CN106339772A (en) * | 2016-08-11 | 2017-01-18 | 清华大学 | Heat supply pipe network thermal storage efficiency-based thermal-electric joint optimal dispatching method |
| CN106447152A (en) * | 2016-06-21 | 2017-02-22 | 中国南方电网有限责任公司电网技术研究中心 | Energy center-based power-gas-heat system co-scheduling method and system |
-
2017
- 2017-06-05 CN CN201710414757.9A patent/CN107239859B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4975287B2 (en) * | 2005-08-08 | 2012-07-11 | パナソニック株式会社 | Prediction device |
| KR101295098B1 (en) * | 2011-08-10 | 2013-08-09 | 아주대학교산학협력단 | Heating supply determination method for district heating network |
| WO2015138535A1 (en) * | 2014-03-11 | 2015-09-17 | General Electric Company | Systems and methods for utility crew forecasting |
| CN103955768A (en) * | 2014-04-30 | 2014-07-30 | 河北省电力勘测设计研究院 | CSP radiation and heat energy predication method based on BP neural network model |
| CN105528466A (en) * | 2014-09-28 | 2016-04-27 | 国家电网公司 | Wind power optimal planning modeling method considering adaptability and economy of power system |
| CN104504451A (en) * | 2014-11-04 | 2015-04-08 | 国家电网公司 | Optimization method of day-ahead unit commitment of thermal power system of thermal power plant capable of operating under multiple states |
| CN105389980A (en) * | 2015-11-09 | 2016-03-09 | 上海交通大学 | Short-time traffic flow prediction method based on long-time and short-time memory recurrent neural network |
| CN105805822A (en) * | 2016-03-24 | 2016-07-27 | 常州英集动力科技有限公司 | Heat supply energy saving control method and system based on neural network prediction |
| CN105913141A (en) * | 2016-04-06 | 2016-08-31 | 上海电器科学研究院 | Load prediction method supporting big data |
| CN106126492A (en) * | 2016-06-07 | 2016-11-16 | 北京高地信息技术有限公司 | Statement recognition methods based on two-way LSTM neutral net and device |
| CN106127303A (en) * | 2016-06-15 | 2016-11-16 | 国网山东省电力公司菏泽供电公司 | A kind of short-term load forecasting method towards multi-source data |
| CN106447152A (en) * | 2016-06-21 | 2017-02-22 | 中国南方电网有限责任公司电网技术研究中心 | Energy center-based power-gas-heat system co-scheduling method and system |
| CN106339772A (en) * | 2016-08-11 | 2017-01-18 | 清华大学 | Heat supply pipe network thermal storage efficiency-based thermal-electric joint optimal dispatching method |
Non-Patent Citations (2)
| Title |
|---|
| 刘桐彤: "《基于长短期记忆神经网络的短期负荷预测方法》", 《科技论坛》 * |
| 陈亮等: "《深度学习框架下LSTM网络在短期电力负荷预测中的应用》", 《电力信息与通信技术》 * |
Cited By (69)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108051035A (en) * | 2017-10-24 | 2018-05-18 | 清华大学 | The pipe network model recognition methods of neural network model based on gating cycle unit |
| CN107909206B (en) * | 2017-11-15 | 2021-06-04 | 电子科技大学 | PM2.5 prediction method based on deep structure recurrent neural network |
| CN107909206A (en) * | 2017-11-15 | 2018-04-13 | 电子科技大学 | A kind of PM2.5 Forecasting Methodologies based on deep structure Recognition with Recurrent Neural Network |
| CN109936464B (en) * | 2017-12-15 | 2022-04-01 | 中国移动通信集团浙江有限公司 | Communication capacity open load prediction method and device based on multilayer LSTMRNN |
| CN109936464A (en) * | 2017-12-15 | 2019-06-25 | 中国移动通信集团浙江有限公司 | The open load forecasting method of communication capacity based on multilayer LSTMRNN and device |
| CN108319132A (en) * | 2018-01-11 | 2018-07-24 | 合肥工业大学 | Decision system and method for unmanned plane aerial opposition |
| CN108319132B (en) * | 2018-01-11 | 2021-01-26 | 合肥工业大学 | Decision-making system and method for unmanned aerial vehicle air countermeasure |
| CN108416690A (en) * | 2018-01-19 | 2018-08-17 | 中国矿业大学 | Load Forecasting based on depth LSTM neural networks |
| CN110084357A (en) * | 2018-01-25 | 2019-08-02 | 三星电子株式会社 | Accelerate shot and long term memory network via selectivity trimming |
| CN110084357B (en) * | 2018-01-25 | 2024-02-27 | 三星电子株式会社 | Acceleration of long and short term memory networks via selective pruning |
| CN108648829A (en) * | 2018-04-11 | 2018-10-12 | 平安科技（深圳）有限公司 | Disease forecasting method and device, computer installation and readable storage medium storing program for executing |
| CN108920888A (en) * | 2018-04-26 | 2018-11-30 | 浙江工业大学 | It is a kind of that autoclave reaction process discrimination method is continuously stirred based on deep neural network |
| CN108920888B (en) * | 2018-04-26 | 2021-08-03 | 浙江工业大学 | Continuous stirred tank type reaction process identification method based on deep neural network |
| CN108711847A (en) * | 2018-05-07 | 2018-10-26 | 国网山东省电力公司电力科学研究院 | A kind of short-term wind power forecast method based on coding and decoding shot and long term memory network |
| CN108711847B (en) * | 2018-05-07 | 2019-06-04 | 国网山东省电力公司电力科学研究院 | A kind of short-term wind power forecast method based on coding and decoding shot and long term memory network |
| CN108921341A (en) * | 2018-06-26 | 2018-11-30 | 国网山东省电力公司电力科学研究院 | A kind of steam power plant's short term thermal load forecasting method encoded certainly based on gate |
| CN109242236A (en) * | 2018-07-05 | 2019-01-18 | 西安图迹信息科技有限公司 | Based on the PMU primary frequency modulation load forecasting method for being associated with full Connection Neural Network and LSTM |
| CN109242236B (en) * | 2018-07-05 | 2021-04-20 | 西安图迹信息科技有限公司 | PMU primary frequency modulation load prediction method based on correlation full-connection neural network and LSTM |
| CN109146162A (en) * | 2018-08-07 | 2019-01-04 | 河海大学 | A kind of probability wind speed forecasting method based on integrated Recognition with Recurrent Neural Network |
| CN109146162B (en) * | 2018-08-07 | 2019-07-23 | 河海大学 | A kind of probability wind speed forecasting method based on integrated Recognition with Recurrent Neural Network |
| CN109165727A (en) * | 2018-09-04 | 2019-01-08 | 成都品果科技有限公司 | A kind of data predication method based on Recognition with Recurrent Neural Network |
| CN109325624A (en) * | 2018-09-28 | 2019-02-12 | 国网福建省电力有限公司 | A kind of monthly electric power demand forecasting method based on deep learning |
| CN109002942A (en) * | 2018-09-28 | 2018-12-14 | 河南理工大学 | A kind of short-term load forecasting method based on stochastic neural net |
| CN109063939A (en) * | 2018-11-01 | 2018-12-21 | 华中科技大学 | A kind of wind speed forecasting method and system based on neighborhood door shot and long term memory network |
| CN109583565A (en) * | 2018-11-07 | 2019-04-05 | 河海大学 | Forecasting Flood method based on the long memory network in short-term of attention model |
| CN109583565B (en) * | 2018-11-07 | 2022-08-26 | 河海大学 | Flood prediction method based on attention model long-time and short-time memory network |
| CN109670629B (en) * | 2018-11-16 | 2021-09-07 | 浙江蓝卓工业互联网信息技术有限公司 | Coal-fired boiler thermal efficiency prediction method based on long-short term memory neural network |
| CN109670629A (en) * | 2018-11-16 | 2019-04-23 | 浙江蓝卓工业互联网信息技术有限公司 | Coal-burning boiler thermal efficiency forecast method based on shot and long term Memory Neural Networks |
| CN109253985B (en) * | 2018-11-28 | 2022-01-11 | 东北林业大学 | Method for identifying wood grade for koto panel by near infrared spectrum based on neural network |
| CN109253985A (en) * | 2018-11-28 | 2019-01-22 | 东北林业大学 | The method of near infrared light spectrum discrimination Chinese zither panel grading of timber neural network based |
| CN109492839A (en) * | 2019-01-17 | 2019-03-19 | 东华大学 | A kind of mineral hot furnace operating condition prediction technique based on RNN-LSTM network |
| CN110007613A (en) * | 2019-04-10 | 2019-07-12 | 北方工业大学 | Warming prediction method and system for heat storage type electric heater and storage medium |
| CN110007613B (en) * | 2019-04-10 | 2022-04-29 | 北方工业大学 | Warming prediction method and system for heat storage type electric heater and storage medium |
| CN110135634A (en) * | 2019-04-29 | 2019-08-16 | 广东电网有限责任公司电网规划研究中心 | Long-medium term power load forecasting device |
| CN110381524A (en) * | 2019-07-15 | 2019-10-25 | 安徽理工大学 | The mobile flow on-line prediction method of large scene based on Bi-LSTM, system and storage medium |
| CN110381524B (en) * | 2019-07-15 | 2022-12-20 | 安徽理工大学 | Bi-LSTM-based large scene mobile flow online prediction method, system and storage medium |
| CN110363289A (en) * | 2019-07-17 | 2019-10-22 | 山东浪潮人工智能研究院有限公司 | A kind of industrial steam amount prediction technique and device based on machine learning |
| CN110428053A (en) * | 2019-08-08 | 2019-11-08 | 东北大学 | A kind of steam produces the dynamic prediction method of consumption |
| CN110705743B (en) * | 2019-08-23 | 2023-08-18 | 国网浙江省电力有限公司 | New energy consumption electric quantity prediction method based on long-term and short-term memory neural network |
| CN110705743A (en) * | 2019-08-23 | 2020-01-17 | 国网浙江省电力有限公司 | New energy consumption electric quantity prediction method based on long-term and short-term memory neural network |
| CN110619389A (en) * | 2019-09-23 | 2019-12-27 | 山东大学 | Load prediction method and system of combined cooling heating and power system based on LSTM-RNN |
| CN111222677A (en) * | 2019-10-22 | 2020-06-02 | 浙江运达风电股份有限公司 | Wind speed prediction method and system based on long-short term memory time neural network |
| CN112712189A (en) * | 2019-10-25 | 2021-04-27 | 北京市热力集团有限责任公司 | Heat supply demand load prediction method |
| CN111353631A (en) * | 2019-11-26 | 2020-06-30 | 国网山东省电力公司电力科学研究院 | Thermal power plant condenser vacuum degree prediction method based on multilayer LSTM |
| CN110889494A (en) * | 2019-11-29 | 2020-03-17 | 北京龙田华远科技有限公司 | Coal mining machine positioning method and device |
| CN110991761B (en) * | 2019-12-13 | 2022-06-28 | 新奥数能科技有限公司 | Heat supply load prediction method and device |
| CN110991761A (en) * | 2019-12-13 | 2020-04-10 | 新奥数能科技有限公司 | Heat supply load prediction method and device |
| CN111160659A (en) * | 2019-12-31 | 2020-05-15 | 国家电网公司西南分部 | Power load prediction method considering temperature fuzzification |
| CN111275169A (en) * | 2020-01-17 | 2020-06-12 | 北京石油化工学院 | Method for predicting building thermal load in short time |
| CN111613338A (en) * | 2020-05-19 | 2020-09-01 | 中国电子科技集团公司第三十八研究所 | Method and system for constructing spike-slow complex wave detection model |
| CN111738520A (en) * | 2020-06-24 | 2020-10-02 | 中国电子科技集团公司第二十八研究所 | System load prediction method fusing isolated forest and long-short term memory network |
| CN111895986A (en) * | 2020-06-30 | 2020-11-06 | 西安建筑科技大学 | MEMS gyroscope original output signal noise reduction method based on LSTM neural network |
| CN111861013A (en) * | 2020-07-23 | 2020-10-30 | 长沙理工大学 | Power load prediction method and device |
| CN111861013B (en) * | 2020-07-23 | 2023-09-08 | 长沙理工大学 | Power load prediction method and device |
| CN111932015A (en) * | 2020-08-12 | 2020-11-13 | 武汉中电节能有限公司 | Method and system for predicting cold and heat load of regional cooling and heating |
| CN112598222A (en) * | 2020-11-30 | 2021-04-02 | 河南汇祥通信设备有限公司 | Comprehensive pipe rack operation and maintenance prediction method based on long-time and short-time memory cyclic neural network |
| CN112832744A (en) * | 2021-01-07 | 2021-05-25 | 中国石油大学(华东) | Pumping unit well pump detection period prediction method based on LSTM neural network |
| CN112733452B (en) * | 2021-01-13 | 2024-03-29 | 北京嘀嘀无限科技发展有限公司 | Track prediction method, track prediction device, electronic equipment and readable storage medium |
| CN112733452A (en) * | 2021-01-13 | 2021-04-30 | 北京嘀嘀无限科技发展有限公司 | Trajectory prediction method and apparatus, electronic device and readable storage medium |
| CN112733457A (en) * | 2021-01-18 | 2021-04-30 | 武汉大学 | Load electricity utilization characteristic modeling method for improving double-layer long-short term memory network |
| CN112733457B (en) * | 2021-01-18 | 2022-03-15 | 武汉大学 | Load electricity utilization characteristic modeling method for improving double-layer long-short term memory network |
| CN113623719B (en) * | 2021-06-23 | 2022-08-19 | 国家电投集团东北电力有限公司大连开热分公司 | Heat exchange station prediction control method based on effective room temperature detection |
| CN113623719A (en) * | 2021-06-23 | 2021-11-09 | 国家电投集团东北电力有限公司大连开热分公司 | Heat exchange station prediction control method based on effective room temperature detection |
| CN113420508B (en) * | 2021-07-07 | 2024-02-27 | 华北电力大学 | Unit combination calculation method based on LSTM |
| CN113420508A (en) * | 2021-07-07 | 2021-09-21 | 华北电力大学 | Unit combination calculation method based on LSTM |
| CN116361709A (en) * | 2023-03-31 | 2023-06-30 | 山东省计算中心（国家超级计算济南中心） | Self-adaptive power load identification method and device |
| CN116361709B (en) * | 2023-03-31 | 2023-10-31 | 山东省计算中心（国家超级计算济南中心） | Self-adaptive power load identification method and device |
| CN116362139A (en) * | 2023-04-14 | 2023-06-30 | 应急管理部沈阳消防研究所 | Multi-parameter fire detection method based on hierarchical long-short-time memory network |
| CN116362139B (en) * | 2023-04-14 | 2024-01-30 | 应急管理部沈阳消防研究所 | Multi-parameter fire detection method based on hierarchical long-short-time memory network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107239859B (en) | 2018-05-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107239859A (en) | The heating load forecasting method of Recognition with Recurrent Neural Network is remembered based on series connection shot and long term | |
| CN103390116B (en) | Use the photovoltaic power station power generation power forecasting method of stepping mode | |
| CN107909220A (en) | Electric heating load prediction method | |
| CN104636985A (en) | Method for predicting radio disturbance of electric transmission line by using improved BP (back propagation) neural network | |
| CN113112077B (en) | HVAC control system based on multi-step prediction deep reinforcement learning algorithm | |
| CN108446794A (en) | One kind being based on multiple convolutional neural networks combination framework deep learning prediction techniques | |
| CN106529719A (en) | Method of predicting wind power of wind speed fusion based on particle swarm optimization algorithm | |
| CN113489015B (en) | Multi-time-scale reactive voltage control method for power distribution network based on reinforcement learning | |
| CN104392394B (en) | A kind of detection method of micro-capacitance sensor energy storage nargin | |
| CN107609667A (en) | Heating load forecasting method and system based on Box_cox conversion and UFCNN | |
| CN107453396A (en) | A kind of Multiobjective Optimal Operation method that distributed photovoltaic power is contributed | |
| CN106980910A (en) | Long Electric Power Load calculating system and method | |
| Ahmad et al. | A bio-inspired heuristic algorithm for solving optimal power flow problem in hybrid power system | |
| CN109214565A (en) | A kind of subregion system loading prediction technique suitable for the scheduling of bulk power grid subregion | |
| CN106295857A (en) | A kind of ultrashort-term wind power prediction method | |
| Naeem et al. | Bees algorithm based solution of non-convex dynamic power dispatch issues in thermal units | |
| CN105633950B (en) | A kind of probabilistic multiple target Random-fuzzy Dynamic Optimal Power Flow Problem method for solving of consideration wind-powered electricity generation injection | |
| Duong et al. | Optimal placement of wind power plants in transmission power networks by applying an effectively proposed metaheuristic algorithm | |
| CN108303898B (en) | Intelligent scheduling method of novel solar-air energy coupling cold and heat cogeneration system | |
| Yang et al. | A novel exhaust gas temperature prediction method of hot blast stove | |
| CN103489037B (en) | A kind of Forecasting Methodology that can power generating wind resource | |
| CN107895215A (en) | The prediction of community network influence power and maximization System and method for based on neutral net | |
| CN107423811A (en) | The streamflow change attribution recognition methods combined based on BP artificial neural networks and Scene Simulation | |
| Nainwal et al. | Comparison of multi linear regression and artificial neural network to predict the energy consumption of residential buildings | |
| CN107622319A (en) | A kind of microgrid power Forecasting Methodology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CB03 | Change of inventor or designer information |
Inventor after: Lu Kuan Inventor after: Han Yingkun Inventor after: Li Kelei Inventor after: Gao Song Inventor after: Yu Qingbin Inventor after: Lin Bo Inventor after: Yan Qing Inventor after: Cheng Yan Inventor after: Zhang Yong Inventor after: Ma Changhui Inventor after: Li Guanglei Inventor after: Su Jianjun Inventor after: Li Honghai Inventor after: Luan Yinglin Inventor after: Zhao Cuan Inventor after: Xu Nan Inventor after: Zhu Zijian Inventor after: Zhao Yan Inventor after: Bi Zhenfu Inventor after: Lang Chengyu Inventor after: Meng Xiangrong Inventor after: Sun Wenxue Inventor after: Pang Xiangkun Inventor after: Wang Wenkuan Inventor before: Lu Kuan Inventor before: Han Yingkun Inventor before: Pang Xiangkun Inventor before: Li Guanglei Inventor before: Zhang Yong Inventor before: Wang Shibai Inventor before: Su Jianjun Inventor before: Zhao Yan Inventor before: Bi Zhenfu Inventor before: Lang Chengyu Inventor before: Meng Xiangrong Inventor before: Ma Changhui Inventor before: Wang Wenkuan Inventor before: Sun Wenxue |
|
| CB03 | Change of inventor or designer information |