CN109165793A - A kind of blending ore sintering basic characteristic forecasting procedure based on PSO-ELM algorithm - Google Patents
A kind of blending ore sintering basic characteristic forecasting procedure based on PSO-ELM algorithm Download PDFInfo
- Publication number
- CN109165793A CN109165793A CN201811072418.8A CN201811072418A CN109165793A CN 109165793 A CN109165793 A CN 109165793A CN 201811072418 A CN201811072418 A CN 201811072418A CN 109165793 A CN109165793 A CN 109165793A
- Authority
- CN
- China
- Prior art keywords
- basic characteristic
- blending ore
- ore
- blending
- data
- 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
- 238000002156 mixing Methods 0.000 title claims abstract description 107
- 238000005245 sintering Methods 0.000 title claims abstract description 61
- 238000000034 method Methods 0.000 title claims abstract description 38
- 239000002245 particle Substances 0.000 claims abstract description 67
- 238000013528 artificial neural network Methods 0.000 claims abstract description 51
- 239000000126 substance Substances 0.000 claims abstract description 37
- 239000011159 matrix material Substances 0.000 claims abstract description 28
- 238000012360 testing method Methods 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 19
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 claims description 20
- 229910052742 iron Inorganic materials 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 9
- 239000000203 mixture Substances 0.000 claims description 8
- 239000007791 liquid phase Substances 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 239000011230 binding agent Substances 0.000 claims description 4
- 210000005036 nerve Anatomy 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000005457 optimization Methods 0.000 description 25
- 230000008569 process Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 239000012071 phase Substances 0.000 description 2
- 239000000843 powder Substances 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- WETINTNJFLGREW-UHFFFAOYSA-N calcium;iron;tetrahydrate Chemical compound O.O.O.O.[Ca].[Fe].[Fe] WETINTNJFLGREW-UHFFFAOYSA-N 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 244000144992 flock Species 0.000 description 1
- 239000003292 glue Substances 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000002028 premature Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012795 verification 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
- 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
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- General Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Manufacturing & Machinery (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Primary Health Care (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Manufacture And Refinement Of Metals (AREA)
Abstract
The present invention provides a kind of blending ore sintering basic characteristic forecasting procedure based on PSO-ELM algorithm, is related to field of computer technology.The present invention is the following steps are included: step 1: obtaining blending ore sample data to be detected, sample data is divided into training set and test set, and the blending ore sample data includes chemical component, scaling loss and its sintering basic characteristic；Step 2: establishing blending ore sintering basic characteristic forecasting model；Step 3: basic characteristic forecasting model being sintered to blending ore by the data of test set and is verified.The present invention uses the extreme learning machine neural network optimized with optimal weights matrix, optimal bias vector particle swarm algorithm to handle ore blender chemical compositionx and scaling loss, and the sintering basic characteristic precision of obtained blending ore sample is higher and this method is high-efficient, at low cost.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of blending ore based on PSO-ELM algorithm to be sintered basis
Prediction on Characteristics method.
Background technique
Iron ore sintering basic characteristic refers to the high-temperature physics chemical property that iron ore is shown during the sintering process, it
The main assimilation performance including iron ore, Liquid phase flowability, binder strength performance and calcium ferrite generate performance etc..Iron ore is burnt
The it is proposed of knot basic property is conducive to improve sintering theory, improves sinter quality and optimization sintering process process.Traditional pair
The experimental method of iron ore sintering basic characteristic forecast, required time is long, consumptive material is more, costly.In order to correctly hold iron ore
Behavior and its interaction under hot conditions during the sintering process, realizes the rational ore matching of sintering production, is technological parameter
It reasonably selects, optimization offer theoretical foundation, the sintering basic characteristic of iron ore is fast and accurately forecast significant.
Summary of the invention
It is a kind of based on PSO-ELM calculation the technical problem to be solved by the present invention is in view of the above shortcomings of the prior art, provide
The blending ore of method is sintered basic characteristic forecasting procedure, has the advantages that high-efficient, at low cost and with high accuracy.
In order to solve the above technical problems, the technical solution used in the present invention is: a kind of mixing based on PSO-ELM algorithm
Mine is sintered basic characteristic forecasting procedure, the specific steps are as follows:
Step 1: obtaining blending ore sample data to be detected, sample data is divided into training set and test set, the mixing
Sample ore notebook data includes the iron of blending ore sintering basic characteristic, the corresponding chemical component of blending ore sintering basic characteristic, scaling loss
Miberal powder data, wherein including n feature in the chemical composition data of every group of blending ore sample；
Step 2: establishing blending ore sintering basic characteristic forecasting model；Pass through particle swarm algorithm using the data in training set
Optimal weights matrix, the optimal bias vector of extreme learning machine neural network are obtained, to obtain blending ore sintering basic characteristic
Forecasting model；
Step 3: basic characteristic forecasting model being sintered to blending ore by the data of test set and is verified.
Basic characteristic in the step 1 includes assimilation temperature, Liquid phase flowability index and binder strength；
The step 2 includes following sub-step:
Step 2.1: the parameter for the extreme learning machine neural network that setting particle swarm algorithm optimizes, wherein setting input layer
Node number be n+1, be arranged output layer node be 1；
Step 2.2: the limit that the chemical component of the blending ore in training set and scaling loss are optimized as particle swarm algorithm
Practise the input data of machine neural network；The extreme learning machine that the sintering basic characteristic data of blending ore optimize as particle swarm algorithm
The output data of neural network；
Step 2.3: using particle swarm algorithm obtain the extreme learning machine neural network optimal weights matrix, it is optimal partially
Set vector；And obtained optimal weights arranged in matrix is sintered to the weight matrix of basic characteristic forecasting model for blending ore；Will
To optimal bias vector be set as blending ore sintering basic characteristic forecasting model bias vector；To obtain blending ore sintering
Basic characteristic forecasting model.
Preferably, the extreme learning machine neural network of the particle swarm algorithm optimization is being obtained most using particle swarm algorithm
When excellent weight matrix and optimal bias vector, in each round iteration, inertia weight ω (k) is obtained according to the first formula, it is described
First formula are as follows:
ω (k)=ωstart-(ωstart-ωend)·k/Tmax, wherein ωstart=0.9, ωend=0.4, k are current change
Generation number, TmaxFor maximum number of iterations.
Preferably, in each round iteration, according to the individual extreme value P of the second formula more new particleibWith group extreme value Pg, institute
State the second formula are as follows:
Wherein, PiFor the position of i-th of particle, f () is the fitness function of particle.
The extreme learning machine neural network is using Sigmoid function as activation primitive.
The beneficial effects of adopting the technical scheme are that provided by the invention a kind of based on PSO-ELM algorithm
Blending ore be sintered basic characteristic forecasting procedure, using with optimal weights matrix, optimal bias vector particle swarm algorithm optimization
Extreme learning machine neural network ore blender chemical compositionx and scaling loss are handled, the sintering base of obtained blending ore sample
Plinth feature accuracy is higher and this method is high-efficient, at low cost.
Detailed description of the invention
Fig. 1 is that the blending ore provided in an embodiment of the present invention based on PSO-ELM algorithm is sintered basic characteristic forecasting procedure
Flow chart；
Fig. 2 is the schematic diagram provided in an embodiment of the present invention established blending ore and be sintered basic characteristic forecasting model；
Fig. 3 is that the blending ore in test data provided in an embodiment of the present invention is sintered basic characteristic detection data and blending ore
It is sintered the comparison diagram of basic characteristic real data；Wherein, (a) assimilates the output of temperature test collection；(b) Liquid phase flowability test set is defeated
Out；(c) binder strength test set exports.
Specific embodiment
With reference to the accompanying drawings and examples, specific embodiments of the present invention will be described in further detail.Implement below
Example is not intended to limit the scope of the invention for illustrating the present invention.
As shown in Figure 1, the forecast side for the blending ore sintering basic characteristic that present embodiment discloses a kind of based on PSO-ELM
Method, comprising the following steps:
Step 1: obtaining blending ore sample data to be detected, sample data is divided into training set and test set, the mixing
Sample ore notebook data includes the iron of blending ore sintering basic characteristic, the corresponding chemical component of blending ore sintering basic characteristic, scaling loss
Miberal powder data, wherein including n feature in the chemical composition data of every group of blending ore sample；
The step of chemical component and scaling loss of blending ore sample are obtained in the step 1 are as follows:
Multiple chemical composition test is carried out to blending ore sample, and the chemical composition data repeatedly tested progress arithmetic is put down
The chemical composition data as the blending ore sample afterwards, wherein include n spy in the chemical composition data of the test
Sign.
Step 2: establishing blending ore sintering basic characteristic forecasting model；Pass through particle swarm algorithm using the data in training set
Optimal weights matrix, the optimal bias vector of extreme learning machine neural network are obtained, by optimal weights matrix, optimal is biased towards
It measures and determines that blending ore is sintered basic characteristic forecasting model；
Step 3: basic characteristic forecasting model being sintered to blending ore by the data of test set and is verified.
Wherein, the blending ore sintering basic characteristic forecasting model is in advance using multiple mixed known to sintering basic characteristic
What the chemical component and scaling loss of even sample ore product were established, the blending ore sintering basic characteristic forecasting model is particle swarm algorithm optimization
Extreme learning machine neural network, blending ore sintering basic characteristic forecasting model has after particle swarm algorithm optimizes
Optimal weights matrix and optimal bias vector, the blending ore sintering basic characteristic forecasting model include n feature for handling
Ore blender chemical compositionx and scaling loss blending ore data to obtain the sintering basic characteristic of corresponding blending ore.
Further, the method, it is described chemical to the blending ore using blending ore sintering basic characteristic forecasting model
Before ingredient and scaling loss are handled, further includes:
It include the sample of ore blender chemical compositionx and scaling loss with the blending ore of corresponding sintering basic characteristic using multiple groups
Data establish the basic characteristic forecasting model of the blending ore of the extreme learning machine neural network optimized based on particle swarm algorithm.
It should be appreciated that the basis for establishing the blending ore of the extreme learning machine neural network optimized based on particle swarm algorithm is special
Property forecasting model the step of the chemical component and scaling loss should be carried out in the basic characteristic forecasting model using blending ore
Before processing.What the chemical component and scaling loss for obtaining blending ore sample to be detected can be optimized in foundation based on particle swarm algorithm
Before the step of basic characteristic forecasting model of the blending ore of extreme learning machine neural network, or later.
Specifically, the extreme learning machine that blending ore sample data to be detected is optimized in foundation based on particle swarm algorithm is obtained
Before the blending ore sintering basic characteristic forecasting model step of neural network, it is adapted in the modelling phase.
Specifically, chemical component, scaling loss and its corresponding sintering basic characteristic number of the blending ore sample obtained
According to.And it regard the random a part of selecting of these blending ore samples as the training sample for establishing model, another part sample work
For the test sample for verifying model.
Specifically, the data for obtaining blending ore sample to be detected are establishing the limit study optimized based on particle swarm algorithm
The blending ore of machine neural network was sintered after the step of basic characteristic forecasting model, was adapted to after model foundation use model
Carry out the stage that basic characteristic forecast is sintered to blending ore.
Specifically, special on the blending ore sintering basis for establishing the extreme learning machine neural network optimized based on particle swarm algorithm
Property forecasting model after, to the unknown blending ore sample of sintering basic characteristic of acquisition, using obtained corresponding mixing
The chemical component and scaling loss data of mine.These pre- samples are handled using the blending ore sintering basic characteristic forecasting model of foundation
Chemical component and scaling loss data obtain the corresponding sintering basic characteristic of pre- sample.
By training step above-mentioned and verification step, the predicted value error of pre- sample sintering basic characteristic is smaller, can
To meet production actual needs.
Specifically, extreme learning machine (Extreme Learning Machine, hereinafter referred to as ELM) neural network is
A kind of single hidden layer feedforward neural network that Huang was proposed in 2004.Traditional BP neural network is mainly based upon under gradient
Algorithm drops, and this method would generally become very slowly since operation is improper or falls into local minimum.And ELM neural network
Feature is randomly generated hidden layer node parameter, and training pattern speed is fast, and has better Generalization Capability.
Usually in extreme learning machine neural network, weight matrix and bias vector be it is given at random, this will lead to pole
The part weight and deviation for limiting learning machine cannot be optimal, so the result difference exported each time is bigger.
Particle group optimizing (Particle Swarm Optimization, hereinafter referred to as PSO) algorithm is a kind of global optimization
Algorithm.Therefore, extreme learning machine neural network and particle swarm optimization algorithm combine, can weight square to extreme learning machine
Battle array and bias vector carry out global optimization, but traditional PSO is used in the renewal process of particle rapidity using invariable
Property weight, it will cause the PSO later period to fall into local optimum, therefore in method of the invention in particle swarm optimization algorithm lead-in
Inertia weight that property is successively decreased overcomes the disadvantage.Using the above particle swarm optimization algorithm come the power to extreme learning machine neural network
Weight matrix, bias vector carry out global optimization, and acquisition has optimal weights matrix, the particle swarm algorithm of optimal bias vector excellent
The extreme learning machine neural network of change can be used for handling ore blender chemical compositionx and scaling loss data to obtain the corresponding burning of blending ore
Tie basic characteristic.
It specifically, include ore blender chemical compositionx and scaling loss and corresponding sintering base using multiple groups in the step 2
The sample data of the blending ore of plinth characteristic establishes the blending ore of the extreme learning machine neural network optimized based on particle swarm algorithm
Basic characteristic forecasting model, specific steps include:
Obtain the mixing that multiple groups include ore blender chemical compositionx and sintering basic characteristic corresponding with the chemical component
The integrated data of mine, wherein include n feature and scaling loss data in the sinter chemical composition of every group of integrated data.
Step 2.1: the parameter for the extreme learning machine neural network that setting particle swarm algorithm optimizes, wherein setting input layer
Node number be n+1, be arranged output layer node be 1.
Step 2.2: using in the blending ore integrated data ore blender chemical compositionx and scaling loss calculate as the population
The input data of the extreme learning machine neural network of method optimization.
The limit that sintering basic characteristic data in the blending ore integrated data are optimized as the particle swarm algorithm
Learn the output data of machine neural network.
Step 2.3: using particle swarm algorithm obtain the extreme learning machine neural network optimal weights matrix, it is optimal partially
Set vector.
By the weight for the extreme learning machine neural network that the optimal weights arranged in matrix is particle swarm algorithm optimization
Matrix.
Set the optimal bias vector to the biasing of the extreme learning machine neural network of the particle swarm algorithm optimization
Vector.
The extreme learning machine neural network of the particle swarm algorithm optimization of acquisition is blending ore sintering basic characteristic forecast mould
Type.
The extreme learning machine neural network is described as follows:
Assuming that extreme learning machine neural network has L hidden layer node, there is N number of training sample data (xi,ti)。
Wherein, xi=[xi1 xi2 L xin]T∈Rn, ti=[ti1 ti2 L tim]T∈Rm.Wherein i=1,2 ... L；Rn、Rm
Refer to the dimension of vector；The then output of the neural network are as follows:
Wherein, j=1,2 ... N；ωi=[ωi1 ωi2 … ωin]TFor i-th of hiding node layer of connection and input node
Input weight.βiFor the output weight for connecting i-th of hiding node layer and output node.biFor the inclined of i-th hiding node layer
It sets.G (x) represents the output of hidden layer neuron, for addition type concealed nodes, G (x) are as follows: G (ωi,bi,xj)=g (ωi·xj
+bi).Allow neural network reality output and desired output it is equal, also may indicate that are as follows:
H β=T (2)
Wherein, H is hidden layer output matrix, and β is that hidden layer exports weight matrix, and T is desired output.
The algorithm is random selection input weight and hidden layer deviation, and it is linear that this network structure of training is equivalent to solution
The least square solution of system H β=T
Minimize:||Hβ-T|| (5)
The minimum value of the last professor Huang verified least-square solution of linear system is:
Here: H+It is the Moore-Penrose generalized inverse matrix of H, and the minimum value of the least square solution of H β=T is only
One.
The particle swarm algorithm is described as follows:
Particle swarm algorithm is a kind of global optimization approach proposed by Kennedy and Eberhart, is preyed on derived to flock of birds
The research of behavior.The basic thought of particle swarm algorithm is optimal to find by cooperation between individual in population and information sharing
Solution.Since particle swarm algorithm structure is simple, easy to accomplish, and without the adjusting of many parameters, it has been widely used at present
The fields such as function optimization, neural metwork training.
Particle swarm optimization algorithm realizes that process is as follows: in a population, each bird is conceptualized as a particle, and prolongs
Reach N-dimensional space, position X of the particle i in N-dimensional spacei=(Xi1,Xi2,L,XiN), the speed V of flighti=(Vi1,Vi2,L,ViN),
Each particle has the fitness value determined by objective function, wherein i=1,2 ..., m, wherein m represents particle number.
In each iteration, the best position pbest that particle is undergone by tracking particle itselfi=(pbesti1,
pbesti2,L,pbestiN) and the best position gbest that is passed through of entire groupi=(gbesti1,gbesti2,L,
gbestiN), and speed and position are constantly updated according to formula (7) and (8).
Vi k+1=ω Vi k+c1rand()(pbesti-Xi k)+c2rand()(gbesti-Xi k) (7)
Wherein, k is current iteration number, c1,c2For Studying factors, ω is inertia weight.
In the method for the embodiment of the present invention, using the inertia weight of linear decrease, so that Premature Convergence is more effectively avoided,
Stability is increased to network.
Specifically, in the method, the pole that the particle swarm algorithm goes optimization is being obtained using particle swarm optimization algorithm
When the optimal weights matrix and optimal bias vector of limit study machine neural network, in each round iteration, obtained according to the first formula
Take inertia weight ω, first formula are as follows:
ω (k)=ωstart-(ωstart-ωend)·k/Tmax, wherein ωstart=0.9, ωend=0.4, k are current change
Generation number, TmaxFor maximum number of iterations.
In each round iteration, according to the individual extreme value P of the second formula more new particleibWith group extreme value Pg, described second
Formula are as follows:
Wherein, PiFor the position of i-th of particle, f () is the fitness function of particle.
Specifically, in the method, in the extreme learning machine neural network of the particle swarm algorithm optimization, limit study
Machine neural network is using Sigmoid function as activation primitive.
Below in conjunction with concrete application, the chemical component and scaling loss Data Detection to the embodiment of the present invention based on blending ore are mixed
The method of even mine sintering basic characteristic is illustrated.
Step 1: collecting blending ore sample, share 100 samples, and chemical component detection is carried out to it.
Specifically, the ore blender chemical compositionx of acquisition and sintering basic characteristic data share 100 groups, wherein choosing 70 groups of works
For training data, 30 groups are used as test data.
Step 2: the Particle Swarm Optimization with optimal weights matrix, optimal bias vector is established using 70 groups of training datas
The extreme learning machine neural network of method optimization, which is used as, is sintered basic characteristic forecasting model based on blending ore.As shown in Fig. 2, of the invention
The method for establishing blending ore sintering basic characteristic forecasting model based on PSO-ELM algorithm of example, steps are as follows:
Step 2.1: ore blender chemical compositionx and scaling loss data being inputed to blending ore sintering basic characteristic forecasting model, often
A ore blender chemical compositionx data have 5 features, therefore the input data of model is 6 features, and output data is 1 feature.
Step 2.2: primary group algorithm and ELM neural network, the maximum number of iterations of particle are 50, population quantity
It is 30；Select Sigmoid function as the activation primitive of ELM neural network.
Step 2.3: using ELM algorithm to the training of training set blending ore data to obtain the fitness value of population, to suitable
It answers angle value to be judged, then saves speed, position and the particle length of optimal fitness value and population.
Optimizing is just exited when network reaches maximum number of iterations, the extreme learning machine mind of particle swarm algorithm optimization at this moment
There is optimal weights matrix, optimal bias vector through network.
Step 3: handling 30 groups of test datas, the sintering of obtained blending ore using the blending ore basic characteristic forecasting model
The testing result of basic characteristic is as shown in Figure 3.Wherein, figure a, b, c is respectively blending ore assimilation temperature, Liquid phase flowability and glues
Tie the forecast result of phase intensity.
Table is first is that the test set for listing 30 groups of blending ore samples carries out basis using method proposed by the invention
Precision, root-mean-square error and the related coefficient of Prediction on Characteristics.It can be seen that the pole proposed by the present invention based on particle group optimizing
The method for limiting learning machine neural network algorithm forecast blending ore sintering basic characteristic, time-consuming is shorter, cost is relatively low and precision of prediction
It is higher, it can satisfy industrial demand.
Table 1 is sintered basic characteristic forecast result
In conclusion the blending ore of the embodiment of the present invention is sintered basic characteristic forecasting model method, using particle swarm algorithm
The blending ore sintering basic characteristic forecasting model for optimizing extreme learning machine neural network is pre- in blending ore sintering basic characteristic
It gives the correct time accurate, efficient.The method of the blending ore sintering basic characteristic forecast based on PSO-ELM algorithm of the embodiment of the present invention is passing through
There is very big advantage and important practical application value in terms of Ji, speed, accuracy.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify to technical solution documented by previous embodiment, or some or all of the technical features are equal
Replacement；And these are modified or replaceed, model defined by the claims in the present invention that it does not separate the essence of the corresponding technical solution
It encloses.
Claims (5)
1. a kind of blending ore based on PSO-ELM algorithm is sintered basic characteristic forecasting procedure, it is characterised in that: including following step
It is rapid:
Step 1: obtaining blending ore sample data to be detected, sample data is divided into training set and test set, the mixing sample ore
Notebook data includes blending ore sintering basic characteristic, the corresponding chemical component of blending ore sintering basic characteristic, the Iron Ore Powder of scaling loss
Data, wherein including n feature in the chemical composition data of every group of blending ore sample；
Step 2: establishing blending ore sintering basic characteristic forecasting model；It is obtained using the data in training set by particle swarm algorithm
Optimal weights matrix, the optimal bias vector of extreme learning machine neural network, it is true by optimal weights matrix, optimal bias vector
Determine blending ore sintering basic characteristic forecasting model；
Step 3: basic characteristic forecasting model being sintered to blending ore by the data of test set and is verified.
2. a kind of blending ore based on PSO-ELM algorithm according to claim 1 is sintered basic characteristic forecasting procedure, special
Sign is: the basic characteristic in the step 1 includes assimilation temperature, Liquid phase flowability index and binder strength.
3. a kind of blending ore based on PSO-ELM algorithm according to claim 1 is sintered basic characteristic forecasting procedure, special
Sign is: step 2 includes following sub-step:
Step 2.1: the parameter for the extreme learning machine neural network that setting particle swarm algorithm optimizes, wherein the section of input layer is set
Point number is n+1, and the node that output layer is arranged is 1；
Step 2.2: the extreme learning machine that the chemical component of the blending ore in training set and scaling loss are optimized as particle swarm algorithm
The input data of neural network；The extreme learning machine nerve that the sintering basic characteristic data of blending ore optimize as particle swarm algorithm
The output data of network；
Step 2.3: obtaining the optimal weights matrix of the extreme learning machine neural network using particle swarm algorithm, optimal be biased towards
Amount；And obtained optimal weights arranged in matrix is sintered to the weight matrix of basic characteristic forecasting model for blending ore；By what is obtained
Optimal bias vector is set as the bias vector of blending ore sintering basic characteristic forecasting model；To obtain blending ore sintering basis
Prediction on Characteristics model.
4. a kind of blending ore based on PSO-ELM algorithm according to claim 1 or 3 is sintered basic characteristic forecasting procedure,
It is characterized by: the extreme learning machine neural network is using Sigmoid function as activation primitive.
5. a kind of blending ore based on PSO-ELM algorithm according to claim 3 is sintered basic characteristic forecasting procedure, special
Sign is: in the step 2.3 in the optimal weights matrix that obtains extreme learning machine neural network using particle swarm algorithm and
When optimal bias vector, in each round iteration, inertia weight ω (k) is obtained according to the first formula, first formula are as follows:
ω (k)=ωstart-(ωstart-ωend)·k/Tmax, wherein ωstart=0.9, ωend=0.4, k are current iteration time
Number, TmaxFor maximum number of iterations；
In each round iteration, according to the individual extreme value P of the second formula more new particleibWith group extreme value Pg, second formula
Are as follows:
Wherein, PiFor the position of i-th of particle, f () is the fitness function of particle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811072418.8A CN109165793A (en) | 2018-09-14 | 2018-09-14 | A kind of blending ore sintering basic characteristic forecasting procedure based on PSO-ELM algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811072418.8A CN109165793A (en) | 2018-09-14 | 2018-09-14 | A kind of blending ore sintering basic characteristic forecasting procedure based on PSO-ELM algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109165793A true CN109165793A (en) | 2019-01-08 |
Family
ID=64879345
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811072418.8A Pending CN109165793A (en) | 2018-09-14 | 2018-09-14 | A kind of blending ore sintering basic characteristic forecasting procedure based on PSO-ELM algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109165793A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112651164A (en) * | 2021-01-20 | 2021-04-13 | 华东理工大学 | Creep fatigue life prediction method based on machine learning |
| CN112992285A (en) * | 2020-12-31 | 2021-06-18 | 无锡东研信科科技研发有限公司 | IPSO-HKELM-based blast furnace molten iron silicon content prediction method |
| CN112992284A (en) * | 2020-12-31 | 2021-06-18 | 无锡东研信科科技研发有限公司 | Sintered mineral quality forecasting method based on PSO-ELM algorithm |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0742520B2 (en) * | 1986-02-14 | 1995-05-10 | 日新製鋼株式会社 | Sintered ore manufacturing method |
| CN102540889A (en) * | 2011-12-30 | 2012-07-04 | 中冶长天国际工程有限责任公司 | Burning through point forecasting method and system |
| CN104200268A (en) * | 2014-09-03 | 2014-12-10 | 辽宁大学 | PSO (Particle Swarm Optimization) extremity learning machine based strip steel exit thickness predicting method |
-
2018
- 2018-09-14 CN CN201811072418.8A patent/CN109165793A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0742520B2 (en) * | 1986-02-14 | 1995-05-10 | 日新製鋼株式会社 | Sintered ore manufacturing method |
| CN102540889A (en) * | 2011-12-30 | 2012-07-04 | 中冶长天国际工程有限责任公司 | Burning through point forecasting method and system |
| CN104200268A (en) * | 2014-09-03 | 2014-12-10 | 辽宁大学 | PSO (Particle Swarm Optimization) extremity learning machine based strip steel exit thickness predicting method |
Non-Patent Citations (1)
| Title |
|---|
| LIU SONG,ETC.: "Research on prediction model of basic sintering characteristics of mixed Iron ore and sinter quality", 《INTERNATIONAL CONFERENCE ON MANAGEMENT，COMPUTER AND EDUCATION INFORMATIZATION》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112992285A (en) * | 2020-12-31 | 2021-06-18 | 无锡东研信科科技研发有限公司 | IPSO-HKELM-based blast furnace molten iron silicon content prediction method |
| CN112992284A (en) * | 2020-12-31 | 2021-06-18 | 无锡东研信科科技研发有限公司 | Sintered mineral quality forecasting method based on PSO-ELM algorithm |
| CN112651164A (en) * | 2021-01-20 | 2021-04-13 | 华东理工大学 | Creep fatigue life prediction method based on machine learning |
| WO2022156110A1 (en) * | 2021-01-20 | 2022-07-28 | 华东理工大学 | Creep fatigue life prediction method based on machine learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109165793A (en) | A kind of blending ore sintering basic characteristic forecasting procedure based on PSO-ELM algorithm | |
| Jiang et al. | Parameter estimation of solar cells and modules using an improved adaptive differential evolution algorithm | |
| CN109030388B (en) | A kind of iron ore all iron content detection method based on spectroscopic data | |
| Lombardo et al. | Accelerated optimization methods for force‐field parametrization in battery electrode manufacturing modeling | |
| CN106786524B (en) | Load model parameters discrimination method based on noise-like signal and improved differential evolution | |
| CN106778059A (en) | A kind of colony's Advances in protein structure prediction based on Rosetta local enhancements | |
| CN104657596B (en) | A kind of large-scale new compressor performance prediction fast modeling method migrated based on model | |
| CN107609298A (en) | A kind of Jiles Atherton model parameter identification methods and device | |
| CN108983127B (en) | A kind of bloodstone magnetism rate detection method based on spectroscopic data | |
| CN109154588A (en) | Automatic biological process development | |
| CN113066527B (en) | Target prediction method and system for siRNA knockdown mRNA | |
| Long et al. | Hysteresis compensation of the Prandtl-Ishlinskii model for piezoelectric actuators using modified particle swarm optimization with chaotic map | |
| Ma et al. | Multicriteria optimal Latin hypercube design-based surrogate-assisted design optimization for a permanent-magnet vernier machine | |
| CN106096077B (en) | Medium optimization method | |
| CN110246205A (en) | A kind of flat work pieces automatic composing method | |
| CN105259795B (en) | Internal driving parameter spread method for electrokinetic cell simulator | |
| CN111539508B (en) | Generator excitation system parameter identification algorithm based on improved gray wolf algorithm | |
| CN113935556B (en) | Temperature sensor optimal arrangement method based on DNA genetic algorithm | |
| CN115221787A (en) | Cylindrical permanent magnet linear motor multi-objective optimization method and system based on NSGA-II | |
| Cui et al. | Swarm intelligence in bioinformatics: Methods and implementations for discovering patterns of multiple sequences | |
| CN112992284A (en) | Sintered mineral quality forecasting method based on PSO-ELM algorithm | |
| CN103778327A (en) | Curve fitting method | |
| CN109147867A (en) | A kind of group's Advances in protein structure prediction based on dynamic fragment length | |
| Leon et al. | Adaptive differential evolution supports automatic model calibration in furnace optimized control system | |
| Yu et al. | Research on quality prediction of iron ore green pellets based on optimized neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190108 |