CN108182490A - A kind of short-term load forecasting method under big data environment - Google Patents
A kind of short-term load forecasting method under big data environment Download PDFInfo
- Publication number
- CN108182490A CN108182490A CN201711442212.5A CN201711442212A CN108182490A CN 108182490 A CN108182490 A CN 108182490A CN 201711442212 A CN201711442212 A CN 201711442212A CN 108182490 A CN108182490 A CN 108182490A
- Authority
- CN
- China
- Prior art keywords
- neural network
- weights
- threshold value
- particle
- value
- 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 13
- 238000013528 artificial neural network Methods 0.000 claims abstract description 48
- 239000002245 particle Substances 0.000 claims abstract description 43
- 238000012545 processing Methods 0.000 claims abstract description 5
- 210000002569 neuron Anatomy 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 14
- 238000000034 method Methods 0.000 claims description 14
- 238000003062 neural network model Methods 0.000 claims description 9
- 238000013480 data collection Methods 0.000 claims description 7
- 238000012937 correction Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims 1
- 210000005036 nerve Anatomy 0.000 claims 1
- 230000005611 electricity Effects 0.000 description 3
- 241000208340 Araliaceae Species 0.000 description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 235000008434 ginseng Nutrition 0.000 description 2
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005194 fractionation Methods 0.000 description 1
- 210000002364 input neuron Anatomy 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 238000002407 reforming Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000002834 transmittance Methods 0.000 description 1
- 238000010977 unit operation Methods 0.000 description 1
- 238000005303 weighing 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
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Tourism & Hospitality (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Water Supply & Treatment (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Operations Research (AREA)
- Public Health (AREA)
- Quality & Reliability (AREA)
- Primary Health Care (AREA)
- Entrepreneurship & Innovation (AREA)
Abstract
The invention discloses the short-term load forecasting methods under a kind of big data environment, carry out distributed storage and processing to mass data using Hadoop framework, improve load prediction speed.The improved particle cluster algorithm of the present invention optimizes traditional BP neural network, improves load prediction precision.
Description
Technical field
The present invention relates to Short Term load Forecasting Technique, more particularly to the short-term load forecasting side under a kind of big data environment
Method.
Background technology
Load Prediction In Power Systems level becomes one of mark for weighing power system management modernization.Short-term load forecasting
It is the important component of load forecast.As electricity market is kept reforming, the precision of power-system short-term load forecasting
Directly affect the economic benefit in power grid and power plant.
Existing Forecasting Methodology still has certain limitation.Traditional BP neural network learning process convergence rate is slow, holds
Local minimum point easily is absorbed in, robustness is bad.Meanwhile greatly developing with intelligent grid, the links such as generate electricity, transmit electricity, dispatching
Mass data is emerged, proposes higher requirement, traditional BP nerve nets in the environment of big data to predetermined speed and precision
Network Forecasting Methodology is in the case of mass data, and due to needing to find neighbour for each test point, operand is very big, unit operation
Time it is very long.
Invention content
Goal of the invention:The object of the present invention is to provide the short-term load forecasting method under a kind of big data environment, Neng Gouyou
Effect solves the problems, such as the precision and arithmetic speed of load prediction under big data environment.
Technical solution:Short-term load forecasting method under big data environment of the present invention, includes the following steps:
S1:Obtain historical load data collection;
S2:Using the MapReduce data processing systems based on Hadoop framework, load data collection is split as small data
According to collection, it is stored in each back end of distributed file system;
S3:BP neural network is built, initializes BP neural network parameter;
S4:The initial parameter of BP neural network is optimized using particle cluster algorithm, weights is obtained and threshold value uploads to
In distributed file system;
S5:In the Map stages, the parameter in distributed file system is read, including weights, threshold value, is opened in each subtask
During the beginning, BP neural network is restored;According to subtask distribute the input signal that data carry out BP neural network it is positive transmit with
The backpropagation of error signal obtains the correction amount of the weights, threshold value of BP neural network under current data set, and according to key assignments
To input parameter of the form as the Reduce stages;
S6:In the Reduce stages, after BP neural network trains all data sets, according to input layer, hidden layer and defeated
Go out the corresponding key-value pair of layer neuron<key,value>In key values, count after all load data sample trainings to each
The influence amount of neuron weights, threshold value, result is exported into distributed file system;
S7:Judge under current iteration task, if reach convergence precision or reach preset iterations;If so,
According to weights, threshold value ginseng in the number of plies and its distributed file system of the input layer of BP neural network, hidden layer and output layer
Number establishes distributed BP neural network model;If it is not, carry out the amendment of BP neural network weights, threshold parameter;
S8:According to distributed BP neural network model, input prediction day data predicted, obtain the load work(of prediction day
Rate data.
Further, in the step S4, the tool that is optimized using particle cluster algorithm to the initial parameter of BP neural network
Body process is:The initial weight of BP neural network and threshold value set are mapped as population, that is, setting the position element of population is
The optimal weights of population and threshold value is obtained in connection weight and threshold value between all nodes of BP neural network, each iteration,
The weights and threshold value of global optimum are finally obtained, assign BP neural network;The speed and location updating equation of population is:
In formula (1), ω(t)The inertia weight factor for the t times iteration;c1And c2Be all Studying factors or be all accelerate it is normal
Number;r1And r2It is all the uniform random number in the range of [0,1];vidFor the speed of i-th of particle d dimension, xidFor i-th of particle
The position of d dimensions, pidFor the optimal location that i-th of particle lives through, pgdThe optimal location lived through for entire population.
Further, in the step S5, the correction amount of the weights, threshold value of BP neural network under current data set is obtained
Process is:
If error criterion function is:
In formula (2), YiFor the desired network output vector of i-th of sample;Yi' for i-th of sample reality network export to
Amount;P is number of samples;The vector that w is made of network weight and threshold value;ei(w) error for i-th of sample;
If wkRepresent the weights of kth time iteration and the vector that threshold value is formed, the vectorial w that new weights and threshold value are formedk +1=wk+Δw;Value increase Δ w calculation formula are as follows:
Δ w=[JT(w)J(w)+μI]-1JT(w)e(w) (3)
In formula (3), I is unit matrix;λ is user-defined learning rate;E (w) is error;J (w) is Jacobian squares
Battle array, i.e.,:
In formula (4), wnThe vector being made of nth iteration weights and threshold value.
Further, the ω(t)It is obtained by following formula:
ω(t)=μ ω(t-1)(1-ω(t-1)) (5)
In formula (5), μ is chaology parameter;
Further, the step S4 specifically includes following steps:
S4.1:Particle swarm parameter is set, including population, maximum iteration, the fitness limits of error, inertia weight and
Practise the factor;
S4.2:Initialize speed and the position of all particles;
S4.3:The fitness function value of each particle is calculated using mean square error as fitness function, and carry out step S4.4 and
The judgement of step S4.5;
S4.4:If the current fitness function value of particle is better than its history optimal value, history is substituted with current location
Optimal location;
S4.5:If the history optimal value of particle is better than global optimum, substituted with the history optimal value of the particle complete
Office's optimal value;
S4.6:The update of speed and position is carried out to each particle;
S4.7:Check whether the speed of particle and position exceed the range of setting:If gone beyond the scope, made with boundary value
Speed and position for particle;
S4.8:Iterations add 1, check whether and reach termination condition:If reached, stop iteration, output weights and
Threshold value;Otherwise, step S4.3 is gone to;
S4.9:BP neural network model is built with obtained weights and threshold value.
Further, in the step S4.8, termination condition is:Reaching maximum iteration or reach minimal error will
It asks.
Advantageous effect:The invention discloses the short-term load forecasting method under a kind of big data environment, using Hadoop framves
Structure carries out distributed storage and processing to mass data, improves load prediction speed;Optimized with improved particle cluster algorithm
Traditional BP neural network improves load prediction precision.
Description of the drawings
Fig. 1 is the structure chart of BP neural network in the specific embodiment of the invention;
Fig. 2 is the work flow diagram of MapReduce in the specific embodiment of the invention;
Fig. 3 is the flow chart of S4 in the specific embodiment of the invention.
Specific embodiment
With reference to embodiment and attached drawing, technical scheme of the present invention is described further.
Present embodiment discloses the short-term load forecasting method under a kind of big data environment, includes the following steps:
S1:Obtain historical load data collection.
S2:Using the MapReduce data processing systems based on Hadoop framework, load data collection is split as small data
According to collection, it is stored in each back end of distributed file system.
S3:BP neural network as shown in Figure 1 is built, initializes BP neural network parameter.
S4:The initial parameter of BP neural network is optimized using particle cluster algorithm, obtains weights and threshold value.
S5:The workflow of Hadoop by the weights in S4 and threshold value as shown in Fig. 2, be stored in distributed field system first
In system, data fractionation is carried out according to the quantity of working node.
S6:In the Map stages, the parameter in distributed file system is read, including weights, threshold value, is opened in each subtask
During the beginning, BP neural network is restored;According to subtask distribute the input signal that data carry out BP neural network it is positive transmit with
The backpropagation of error signal obtains the correction amount of the weights, threshold value of BP neural network under current data set, and according to key assignments
To input parameter of the form as the Reduce stages.
S7:In the Reduce stages, after BP neural network trains all data sets, according to input layer, hidden layer and defeated
Go out the corresponding key-value pair of layer neuron<key,value>In key values, count after all load data sample trainings to each
The influence amount of neuron weights, threshold value, result is exported into distributed file system.
S8:Judge under current iteration task, if reach convergence precision or reach preset iterations;If so,
According to weights, threshold value ginseng in the number of plies and its distributed file system of the input layer of BP neural network, hidden layer and output layer
Number establishes distributed BP neural network model;If it is not, carry out the amendment of BP neural network weights, threshold parameter.
S9:According to distributed BP neural network model, input prediction day data predicted, obtain the load work(of prediction day
Rate data.
Fig. 1 is the structure chart of typical three layers of BP neural network, it is assumed that input neuron number is M, and hidden layer is refreshing
It is I through first number, output layer neuron number is J.M-th of neuron of input layer is denoted as am, i-th of neuron of hidden layer be denoted as
kiJ-th of neuron of output layer is denoted as yj.From amTo kiConnection weight be wmi, from kiTo yjConnection weight be wij。
(1) the positive transmittance process of input signal
According to the structure chart of BP neural network in Fig. 1, the output of input layer is equal to the input signal of network:
vm M(n)=a (n)
The input of i-th of neuron of hidden layer is equal to vm M(n) weighted sum:
Assuming that f () is implicit layer functions, then the output of i-th of neuron of hidden layer is equal to:
vi I(n)=f (ui I(n))
The input of j-th of neuron of output layer is equal to vi I(n) weighted sum:
Assuming that g () is output layer functions, the output of j-th of neuron of output layer is equal to:
vj J(n)=g ((uj J(n))
The error of j-th of neuron of output layer:
ej(n)=dj(n)-vj J(n)
The overall error of network:
(2) back-propagation process of error signal
The output error of each layer neuron is successively calculated first by output layer, is then used according to error level
Levenberg-Marquardt (LM) algorithm adjusts the weights and threshold value of each layer, enables the final output of network mapping after adjusting
Close to desired value.
In step S5,The process that LM methods obtain the correction amount under current data set of weights, threshold value of BP neural network is:
If error criterion function is:
In formula (2), YiFor the desired network output vector of i-th of sample;Yi' for i-th of sample reality network export to
Amount;P is number of samples;The vector that w is made of network weight and threshold value;ei(w) error for i-th of sample;
If wkRepresent the weights of kth time iteration and the vector that threshold value is formed, the vectorial w that new weights and threshold value are formedk +1=wk+Δw;Value increase Δ w calculation formula are as follows:
Δ w=[JT(w)J(w)+λI]-1JT(w)e(w) (3)
In formula (3), I is unit matrix;λ is user-defined learning rate;E (w) is error;J (w) is Jacobian squares
Battle array, i.e.,:
In formula (4), wnThe vector being made of nth iteration weights and threshold value.
In step S4, the detailed process optimized using particle cluster algorithm to the initial parameter of BP neural network is:It will
The initial weight and threshold value set of BP neural network are mapped as population, that is, the position element for setting population is BP neural network
Connection weight and threshold value between all nodes, each iteration are obtained the optimal weights of population and threshold value, finally obtain the overall situation
Optimal weights and threshold value assign BP neural network;The speed and location updating equation of population is:
In formula (1), ω(t)The inertia weight factor for the t times iteration;c1And c2Be all Studying factors or be all accelerate it is normal
Number;r1And r2It is all the uniform random number in the range of [0,1];vidFor the speed of i-th of particle d dimension, xidFor i-th of particle
The position of d dimensions, pidFor the optimal location that i-th of particle lives through, pgdThe optimal location lived through for entire population.
ω(t)It is obtained by following formula:
ω(t)=μ ω(t-1)(1-ω(t-1)) (5)
In formula (5), μ is chaology parameter.
As shown in figure 3, step S4 specifically includes following steps:
S4.1:Particle swarm parameter is set, including population, maximum iteration, the fitness limits of error, inertia weight and
Practise the factor;
S4.2:Initialize speed and the position of all particles;
S4.3:The fitness function value of each particle is calculated using mean square error as fitness function, and carry out step S4.4 and
The judgement of step S4.5;
S4.4:If the current fitness function value of particle is better than its history optimal value, history is substituted with current location
Optimal location;
S4.5:If the history optimal value of particle is better than global optimum, substituted with the history optimal value of the particle complete
Office's optimal value;
S4.6:The update of speed and position is carried out to each particle;
S4.7:Check whether the speed of particle and position exceed the range of setting:If gone beyond the scope, made with boundary value
Speed and position for particle;
S4.8:Iterations add 1, check whether and reach termination condition:If reached, stop iteration, output weights and
Threshold value;Otherwise, step S4.3 is gone to;
S4.9:BP neural network model is built with obtained weights and threshold value.
In step S4.8, termination condition is:Reach maximum iteration or reach minimal error requirement.
Claims (6)
1. a kind of short-term load forecasting method under big data environment, it is characterised in that:Include the following steps:
S1:Obtain historical load data collection;
S2:Using the MapReduce data processing systems based on Hadoop framework, load data collection is split as small-sized data
Collection, is stored in each back end of distributed file system;
S3:BP neural network is built, initializes BP neural network parameter;
S4:The initial parameter of BP neural network is optimized using particle cluster algorithm, weights is obtained and threshold value uploads to distribution
In formula file system;
S5:In the Map stages, the parameter in distributed file system is read, including weights, threshold value, when starting in each subtask,
Restore BP neural network;The positive transmission for the input signal that data carry out BP neural network and error letter are distributed according to subtask
Number backpropagation, obtain the correction amount of the weights, threshold value of BP neural network under current data set, and according to key-value pair form
Input parameter as the Reduce stages;
S6:In the Reduce stages, after BP neural network trains all data sets, according to input layer, hidden layer and output layer
The corresponding key-value pair of neuron<key,value>In key values, count after all load data sample trainings to each nerve
The influence amount of first weights, threshold value, result is exported into distributed file system;
S7:Judge under current iteration task, if reach convergence precision or reach preset iterations;If so, foundation
Weights, threshold parameter in the number of plies and its distributed file system of the input layer of BP neural network, hidden layer and output layer, build
Vertical distribution BP neural network model;If it is not, carry out the amendment of BP neural network weights, threshold parameter;
S8:According to distributed BP neural network model, input prediction day data predicted, obtain the load power number of prediction day
According to.
2. the short-term load forecasting method under big data environment according to claim 1, it is characterised in that:The step S4
In, the detailed process optimized using particle cluster algorithm to the initial parameter of BP neural network is:By the first of BP neural network
Beginning weights and threshold value set are mapped as population, that is, the position element for setting population is between all nodes of BP neural network
The optimal weights of population and threshold value is obtained in connection weight and threshold value, each iteration, finally obtains the weights and threshold of global optimum
Value assigns BP neural network;The speed and location updating equation of population is:
In formula (1), ω(t)The inertia weight factor for the t times iteration;c1And c2It is all Studying factors or is all aceleration pulse;r1
And r2It is all the uniform random number in the range of [0,1];vidFor the speed of i-th of particle d dimension, xidIt is tieed up for i-th of particle d
Position, pidFor the optimal location that i-th of particle lives through, pgdThe optimal location lived through for entire population.
3. the short-term load forecasting method under big data environment according to claim 1, it is characterised in that:The step S5
In, the process for obtaining the correction amount of the weights, threshold value of BP neural network under current data set is:
If error criterion function is:
In formula (2), YiFor the desired network output vector of i-th of sample;Yi' network the output vector for i-th of sample reality;p
For number of samples;The vector that w is made of network weight and threshold value;ei(w) error for i-th of sample;
If wkRepresent the weights of kth time iteration and the vector that threshold value is formed, the vectorial w that new weights and threshold value are formedk+1=
wk+Δw;Value increase Δ w calculation formula are as follows:
Δ w=[JT(w)J(w)+μI]-1JT(w)e(w) (3)
In formula (3), I is unit matrix;μ is user-defined learning rate;E (w) is error;J (w) is Jacobian matrixes, i.e.,:
In formula (4), wuThe vector being made of nth iteration weights and threshold value, 1≤u≤n.
4. the short-term load forecasting method under big data environment according to claim 2, it is characterised in that:The ω(t)It is logical
Following formula is crossed to obtain:
ω(t)=μ ω(t-1)(1-ω(t-1)) (5)
In formula (5), μ is chaology parameter.
5. the short-term load forecasting method under big data environment according to claim 1, it is characterised in that:The step S4
Specifically include following steps:
S4.1:Particle swarm parameter is set, including population, maximum iteration, the fitness limits of error, inertia weight and study because
Son;
S4.2:Initialize speed and the position of all particles;
S4.3:The fitness function value of each particle is calculated using mean square error as fitness function, and carries out step S4.4 and step
The judgement of S4.5;
S4.4:If the current fitness function value of particle is better than its history optimal value, it is optimal to substitute history with current location
Position;
S4.5:If the history optimal value of particle is better than global optimum, the overall situation is substituted most with the history optimal value of the particle
The figure of merit;
S4.6:The update of speed and position is carried out to each particle;
S4.7:Check whether the speed of particle and position exceed the range of setting:If gone beyond the scope, by the use of boundary value as grain
The speed of son and position;
S4.8:Iterations add 1, check whether and reach termination condition:If reached, stop iteration, export weights and threshold value;
Otherwise, step S4.3 is gone to;
S4.9:BP neural network model is built with obtained weights and threshold value.
6. the short-term load forecasting method under big data environment according to claim 5, it is characterised in that:The step
In S4.8, termination condition is:Reach maximum iteration or reach minimal error requirement.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711442212.5A CN108182490A (en) | 2017-12-27 | 2017-12-27 | A kind of short-term load forecasting method under big data environment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711442212.5A CN108182490A (en) | 2017-12-27 | 2017-12-27 | A kind of short-term load forecasting method under big data environment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108182490A true CN108182490A (en) | 2018-06-19 |
Family
ID=62547567
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711442212.5A Pending CN108182490A (en) | 2017-12-27 | 2017-12-27 | A kind of short-term load forecasting method under big data environment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108182490A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110068759A (en) * | 2019-05-22 | 2019-07-30 | 四川华雁信息产业股份有限公司 | A kind of fault type preparation method and device |
CN110262897A (en) * | 2019-06-13 | 2019-09-20 | 东北大学 | A kind of Hadoop calculating task primary distribution method based on load estimation |
CN111353582A (en) * | 2020-02-19 | 2020-06-30 | 四川大学 | Particle swarm algorithm-based distributed deep learning parameter updating method |
CN111695667A (en) * | 2020-05-27 | 2020-09-22 | 江苏信息职业技术学院 | MapReduce-based distributed particle swarm clustering algorithm |
CN112231489A (en) * | 2020-10-19 | 2021-01-15 | 中国科学技术大学 | Knowledge learning and transferring method and system for epidemic prevention robot |
CN113065693A (en) * | 2021-03-22 | 2021-07-02 | 哈尔滨工程大学 | Traffic flow prediction method based on radial basis function neural network |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103164742A (en) * | 2013-04-02 | 2013-06-19 | 南京邮电大学 | Server performance prediction method based on particle swarm optimization nerve network |
CN103729695A (en) * | 2014-01-06 | 2014-04-16 | 国家电网公司 | Short-term power load forecasting method based on particle swarm and BP neural network |
CN104361393A (en) * | 2014-09-06 | 2015-02-18 | 华北电力大学 | Method for using improved neural network model based on particle swarm optimization for data prediction |
CN104715282A (en) * | 2015-02-13 | 2015-06-17 | 浙江工业大学 | Data prediction method based on improved PSO-BP neural network |
CN106022521A (en) * | 2016-05-19 | 2016-10-12 | 四川大学 | Hadoop framework-based short-term load prediction method for distributed BP neural network |
CN106229964A (en) * | 2016-07-22 | 2016-12-14 | 南京工程学院 | A kind of based on the electrical power distribution network fault location method improving binary particle swarm algorithm |
CN106372756A (en) * | 2016-09-07 | 2017-02-01 | 南京工程学院 | Thermal power plant load optimization distribution method based on breeding particle swarm optimization |
CN106779177A (en) * | 2016-11-28 | 2017-05-31 | 国网冀北电力有限公司唐山供电公司 | Multiresolution wavelet neutral net electricity demand forecasting method based on particle group optimizing |
CN106777449A (en) * | 2016-10-26 | 2017-05-31 | 南京工程学院 | Distribution Network Reconfiguration based on binary particle swarm algorithm |
CN107301475A (en) * | 2017-06-21 | 2017-10-27 | 南京信息工程大学 | Load forecast optimization method based on continuous power analysis of spectrum |
CN107316099A (en) * | 2017-05-22 | 2017-11-03 | 沈阳理工大学 | Ammunition Storage Reliability Forecasting Methodology based on particle group optimizing BP neural network |
-
2017
- 2017-12-27 CN CN201711442212.5A patent/CN108182490A/en active Pending
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103164742A (en) * | 2013-04-02 | 2013-06-19 | 南京邮电大学 | Server performance prediction method based on particle swarm optimization nerve network |
CN103729695A (en) * | 2014-01-06 | 2014-04-16 | 国家电网公司 | Short-term power load forecasting method based on particle swarm and BP neural network |
CN104361393A (en) * | 2014-09-06 | 2015-02-18 | 华北电力大学 | Method for using improved neural network model based on particle swarm optimization for data prediction |
CN104715282A (en) * | 2015-02-13 | 2015-06-17 | 浙江工业大学 | Data prediction method based on improved PSO-BP neural network |
CN106022521A (en) * | 2016-05-19 | 2016-10-12 | 四川大学 | Hadoop framework-based short-term load prediction method for distributed BP neural network |
CN106229964A (en) * | 2016-07-22 | 2016-12-14 | 南京工程学院 | A kind of based on the electrical power distribution network fault location method improving binary particle swarm algorithm |
CN106372756A (en) * | 2016-09-07 | 2017-02-01 | 南京工程学院 | Thermal power plant load optimization distribution method based on breeding particle swarm optimization |
CN106777449A (en) * | 2016-10-26 | 2017-05-31 | 南京工程学院 | Distribution Network Reconfiguration based on binary particle swarm algorithm |
CN106779177A (en) * | 2016-11-28 | 2017-05-31 | 国网冀北电力有限公司唐山供电公司 | Multiresolution wavelet neutral net electricity demand forecasting method based on particle group optimizing |
CN107316099A (en) * | 2017-05-22 | 2017-11-03 | 沈阳理工大学 | Ammunition Storage Reliability Forecasting Methodology based on particle group optimizing BP neural network |
CN107301475A (en) * | 2017-06-21 | 2017-10-27 | 南京信息工程大学 | Load forecast optimization method based on continuous power analysis of spectrum |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110068759A (en) * | 2019-05-22 | 2019-07-30 | 四川华雁信息产业股份有限公司 | A kind of fault type preparation method and device |
CN110068759B (en) * | 2019-05-22 | 2021-11-09 | 华雁智能科技(集团)股份有限公司 | Fault type obtaining method and device |
CN110262897A (en) * | 2019-06-13 | 2019-09-20 | 东北大学 | A kind of Hadoop calculating task primary distribution method based on load estimation |
WO2020248226A1 (en) * | 2019-06-13 | 2020-12-17 | 东北大学 | Initial hadoop computation task allocation method based on load prediction |
CN110262897B (en) * | 2019-06-13 | 2023-01-31 | 东北大学 | Hadoop calculation task initial allocation method based on load prediction |
CN111353582A (en) * | 2020-02-19 | 2020-06-30 | 四川大学 | Particle swarm algorithm-based distributed deep learning parameter updating method |
CN111695667A (en) * | 2020-05-27 | 2020-09-22 | 江苏信息职业技术学院 | MapReduce-based distributed particle swarm clustering algorithm |
CN112231489A (en) * | 2020-10-19 | 2021-01-15 | 中国科学技术大学 | Knowledge learning and transferring method and system for epidemic prevention robot |
CN112231489B (en) * | 2020-10-19 | 2021-11-02 | 中国科学技术大学 | Knowledge learning and transferring method and system for epidemic prevention robot |
CN113065693A (en) * | 2021-03-22 | 2021-07-02 | 哈尔滨工程大学 | Traffic flow prediction method based on radial basis function neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108182490A (en) | A kind of short-term load forecasting method under big data environment | |
CN109102126B (en) | Theoretical line loss rate prediction model based on deep migration learning | |
Li et al. | Prediction for tourism flow based on LSTM neural network | |
CN108694467B (en) | Method and system for predicting line loss rate of power distribution network | |
CN106022521B (en) | Short-term load prediction method of distributed BP neural network based on Hadoop architecture | |
Gill et al. | Training back propagation neural networks with genetic algorithm for weather forecasting | |
CN104217258B (en) | A kind of electric load sigma-t Forecasting Methodology | |
CN107704875A (en) | Based on the building load Forecasting Methodology and device for improving IHCMAC neutral nets | |
CN103105246A (en) | Greenhouse environment forecasting feedback method of back propagation (BP) neural network based on improvement of genetic algorithm | |
CN104636985A (en) | Method for predicting radio disturbance of electric transmission line by using improved BP (back propagation) neural network | |
CN109445935A (en) | A kind of high-performance big data analysis system self-adaption configuration method under cloud computing environment | |
CN104376389A (en) | Master-slave type micro-grid power load prediction system and master-slave type micro-grid power load prediction method based on load balancing | |
CN111723839B (en) | Method for predicting line loss rate of transformer area based on edge calculation | |
CN106296044B (en) | Power system risk scheduling method and system | |
CN109614580A (en) | Antidetonation bulk testing model update method based on online Xgboost algorithm | |
CN108171319A (en) | The construction method of the adaptive depth convolution model of network connection | |
CN112633577A (en) | Short-term household electrical load prediction method, system, storage medium and equipment | |
CN108805346A (en) | A kind of hot continuous rolling force forecasting method based on more hidden layer extreme learning machines | |
CN109214565A (en) | A kind of subregion system loading prediction technique suitable for the scheduling of bulk power grid subregion | |
CN116340006A (en) | Computing power resource idle prediction method based on deep learning and storage medium | |
CN114493052A (en) | Multi-model fusion self-adaptive new energy power prediction method and system | |
CN110163419A (en) | A kind of method of middle and small river river basin flood forecast | |
CN109190749A (en) | A kind of prediction technique and device for the intelligent electric meter service life | |
Showkati et al. | Short term load forecasting using echo state networks | |
CN103763123A (en) | Method and device for evaluating health condition of network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180619 |