CN109034275A - Prediction technique, system, medium and the equipment of polycrystalline reduction process energy consumption value - Google Patents
Prediction technique, system, medium and the equipment of polycrystalline reduction process energy consumption value Download PDFInfo
- Publication number
- CN109034275A CN109034275A CN201811011310.8A CN201811011310A CN109034275A CN 109034275 A CN109034275 A CN 109034275A CN 201811011310 A CN201811011310 A CN 201811011310A CN 109034275 A CN109034275 A CN 109034275A
- Authority
- CN
- China
- Prior art keywords
- frog
- reduction process
- energy consumption
- algorithm
- group
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- C—CHEMISTRY; METALLURGY
- C01—INORGANIC CHEMISTRY
- C01B—NON-METALLIC ELEMENTS; COMPOUNDS THEREOF; METALLOIDS OR COMPOUNDS THEREOF NOT COVERED BY SUBCLASS C01C
- C01B33/00—Silicon; Compounds thereof
- C01B33/02—Silicon
- C01B33/021—Preparation
- C01B33/027—Preparation by decomposition or reduction of gaseous or vaporised silicon compounds other than silica or silica-containing material
- C01B33/03—Preparation by decomposition or reduction of gaseous or vaporised silicon compounds other than silica or silica-containing material by decomposition of silicon halides or halosilanes or reduction thereof with hydrogen as the only reducing agent
-
- 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
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
Abstract
This application discloses a kind of prediction techniques of polycrystalline reduction process energy consumption value, comprising: obtains the influence factor of polycrystalline reduction process energy consumption value;Influence factor is modeled using wavelet neural network algorithm, obtains initialization model;Initialization model is optimized using SFLA algorithm, obtains object module;It is predicted using power consumption values of the object module to target polycrystalline reduction process.As it can be seen that the precision of prediction of polycrystalline reduction process energy consumption value can be significantly improved by the method in the application.Correspondingly, forecasting system, medium and the equipment of a kind of polycrystalline reduction process energy consumption value disclosed in the present application, equally have above-mentioned beneficial effect.
Description
Technical field
The present invention relates to energy consumption in production process to predict field, in particular to a kind of prediction of polycrystalline reduction process energy consumption value
Method, system, medium and equipment.
Background technique
The polysilicon making material one of best as photovoltaic converter, 95% solar battery is all with polysilicon
As raw material, main production method has Siemens Method, metallurgy method and carbothermic method etc..Wherein, Siemens Method is
Polysilicon technology the most mature is produced, which mainly includes trichlorosilane synthesis process, trichlorosilane rectifying
Purifying technique, trichlorosilane reduction technique and tail gas recycle recycle technique.And trichlorosilane reduction technique is production of polysilicon
Core technology, monopolized for a long time by overseas enterprise, China's technology level fall behind, it is every production 1kg polysilicon will consume
The trichlorosilane of 10~15kg, energy consumption is huge, and utilize prediction model can power consumption values to polycrystalline reduction process into
Row prediction, thus staff can according to the power consumption values of polycrystalline reduction process to the production procedure of polycrystalline reduction process into
Row control, and then achieve the purpose that improve production of polysilicon efficiency, this method is currently on the market using relatively broad.But
In the prior art, when most of prediction model predicts polycrystalline reduction process energy consumption value, all there is prediction essence
Spend lower problem, it can be seen that, the prediction of polycrystalline reduction process energy consumption value how is improved using a kind of better method
Precision is those skilled in the art's urgent problem to be solved.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of prediction technique of polycrystalline reduction process energy consumption value, system,
Medium and equipment, to improve the precision of prediction of polycrystalline reduction process energy consumption value.Its concrete scheme is as follows:
A kind of prediction technique of polycrystalline reduction process energy consumption value, comprising:
Obtain the influence factor of polycrystalline reduction process energy consumption value;
The influence factor is modeled using wavelet neural network algorithm, obtains initialization model;
The initialization model is optimized using SFLA algorithm, obtains object module;
It is predicted using power consumption values of the object module to target polycrystalline reduction process.
Preferably, described that the influence factor is modeled using wavelet neural network algorithm, obtain initialization model
Process before, further includes:
Dimension-reduction treatment is carried out to the influence factor.
Preferably, the process that dimension-reduction treatment is carried out to the influence factor, comprising:
Dimension-reduction treatment is carried out to the influence factor using PCA algorithm.
Preferably, the model expression of the wavelet neural network algorithm are as follows:
In formula, n is the number of hidden node, wijFor the weight for connecting hidden layer node i and input layer node j, φx,yFor
Continuous wavelet, x be wavelet neural network algorithm in currently imply node coefficient of dilatation, y be wavelet neural network algorithm in when
The translation coefficient of preceding implicit node, p are the number of input layer node, wjkFor the power for connecting hidden layer node j and input layer node k
Value.
Preferably, described the initialization model to be optimized using SFLA algorithm, obtain object module process it
Afterwards, further includes:
The precision of prediction of the object module is detected using test data.
Preferably, described that the initialization model is optimized using SFLA algorithm, the process of object module is obtained, is wrapped
It includes:
Initiation parameter is established using SFLA algorithm, to determine the quantity F of the frog and quantity m of group in frog group, and is counted
Calculate the adaptive value of each frog in the frog group;
All frogs are subjected to descending arrangement according to the size of adaptive value, and by F frog after sequence according to default point
It distributes with condition to m group;
The frog in each group is made to optimize using cultural gene algorithm, obtains optimization frog;
Optimization frog in the m group is mixed, to determine the optimal solution in the frog group;
Judge whether the optimal solution meets default screening conditions;
If so, the optimal solution is input to the initialization model, to obtain the object module.
Preferably, it is described judge the process whether optimal solution meets default screening conditions after, further includes:
All frogs are subjected to descending arrangement according to the size of adaptive value if it is not, then executing again, and only by the F after sequence
The step of frog distributes the F frog to m group according to default distributive condition.
Correspondingly, the invention also discloses a kind of forecasting systems of polycrystalline reduction process energy consumption value, comprising:
Factor obtains module, for obtaining the influence factor of polycrystalline reduction process energy consumption value;
Model initialization module is obtained just for being modeled using wavelet neural network algorithm to the influence factor
Beginningization model;
Model optimization module obtains object module for optimizing using SFLA algorithm to the initialization model;
Model prediction module, it is pre- for being carried out using power consumption values of the object module to target polycrystalline reduction process
It surveys.
Correspondingly, the invention also discloses a kind of computer readable storage medium, on the computer readable storage medium
It is stored with computer program, the computer program realizes polycrystalline reduction process energy as previously disclosed when being executed by processor
The step of prediction technique of consumption value.
Correspondingly, the invention also discloses a kind of pre- measurement equipments of polycrystalline reduction process energy consumption value, comprising:
Memory, for storing computer program;
Processor realizes the pre- of aforementioned disclosed polycrystalline reduction process energy consumption value when for executing the computer program
The step of survey method.
As it can be seen that being to obtain the influence factor for influencing polycrystalline reduction process energy consumption value, then, benefit first in the present invention
Influence factor is modeled with wavelet neural network algorithm, obtains initialization model, it is clear that due to wavelet neural network algorithm
With self-learning function, it is possible to more be influenced using wavelet neural network algorithm to polycrystalline reduction process is influenced
Factor is analyzed, therefore, using the initialization model of wavelet neural network algorithm creation to target polycrystalline reduction process energy
When consumption value is predicted, it will be able to significantly improve the precision of prediction of target polycrystalline reduction process energy consumption value.Also, it is obtaining just
After beginningization model, initialization model is optimized using SFLA algorithm once more, obtains object module, to avoid initialization
Parameter in model falls into locally optimal solution, so, it can be further improved the precision of prediction of object module using this kind of method,
Obviously, it when being predicted using the object module in the present invention target polycrystalline reduction process energy consumption value, will further mention
The precision of prediction of high target polycrystalline reduction process energy consumption value.Correspondingly, a kind of polycrystalline reduction process energy disclosed by the invention
Forecasting system, medium and the equipment of consumption value equally have above-mentioned beneficial effect.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The embodiment of invention for those of ordinary skill in the art without creative efforts, can also basis
The attached drawing of offer obtains other attached drawings.
Fig. 1 is a kind of flow chart of the prediction technique of polycrystalline reduction process energy consumption value provided in an embodiment of the present invention;
Fig. 2 is the flow chart of the prediction technique of another polycrystalline reduction process energy consumption value provided in an embodiment of the present invention;
Fig. 3 is a kind of structure chart of the forecasting system of polycrystalline reduction process energy consumption value provided in an embodiment of the present invention;
Fig. 4 is a kind of structure chart of the pre- measurement equipment of polycrystalline reduction process energy consumption value provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The embodiment of the invention discloses a kind of prediction techniques of polycrystalline reduction process energy consumption value, as shown in Figure 1, this method
Include:
Step S11: the influence factor of polycrystalline reduction process energy consumption value is obtained;
In the present embodiment, in order to which the power consumption values to polycrystalline reduction process are predicted, firstly, it is necessary to which it is more to obtain influence
Crystal silicon restores the influence factor of process, such as: the arrangement of silicon rod in the structure of reduction furnace, reduction furnace deposits in reduction furnace
The temperature of object, the pressure in reduction furnace, the proportion of trichlorosilane and hydrogen, the electric current of silicon rod and voltage etc. in reduction furnace.When
So, in practical application, there are also many other influence factors that can influence polycrystalline reduction process, no longer carry out one herein
One enumerates.
Step S12: influence factor is modeled using wavelet neural network algorithm, obtains initialization model;
Step S13: initialization model is optimized using SFLA algorithm, obtains object module;
It is understood that the simple, fast convergence rate because of wavelet neural network algorithm structure, has powerful self study
Ability, further, it is possible to be carried out by scale contraction-expansion factor in wavelet neural network algorithm and translation factor pair input data more
The effective information in more input datas is extracted in dimensional analysis, so, it is to utilize wavelet neural network in the present embodiment
Algorithm models influence factor, and obtain can be to the initialization mould that target polycrystalline reduction process energy consumption value is predicted
Type.
Further, since having some parameters in the middle using the initialization model of wavelet neural network algorithm creation is that comparison is fixed
, it is easy so that the initialization model of creation falls into locally optimal solution, in this case, initialization model can not be to target
The power consumption values of polycrystalline reduction process carry out optimal prediction, so, in the present embodiment, also using SFLA algorithm, (leapfrog calculation
Method) initialization model created based on wavelet neural network algorithm is optimized, it is fallen into avoid the parameter in initialization model
Enter locally optimal solution, then obtains object module.
Step S14: it is predicted using power consumption values of the object module to target polycrystalline reduction process.
It is also contemplated that being optimized when using SFLA algorithm to initialization model, search out in initialization model
Optimized parameter, after obtaining object module, so that it may using object module come the power consumption values to target polycrystalline reduction process into
Row prediction, obtains the power consumption values of target polycrystalline reduction process.Obviously, when the energy consumption for obtaining target polycrystalline reduction process is predicted
After value, it will be able to which the energy consumption predicted value of the target polysilicon obtained according to prediction adjusts the reduction process of target polysilicon
Whole and improvement, to improve the production efficiency of target polysilicon.
It should be noted that target polysilicon herein refers to the polysilicon of reduction process energy consumption value to be predicted, above-mentioned step
The polysilicon being mentioned to during rapid refers to the data sample of influence polycrystalline reduction process used when creation initialization model
This.Also, a kind of prediction technique of polycrystalline reduction process energy consumption value provided in this embodiment, can also be applied to it is other with it is more
Crystal silicon restores the similar process flow of process, such as: iron restores the power consumption values prediction etc. of process and copper reduction process.
As it can be seen that be to obtain the influence factor for influencing polycrystalline reduction process energy consumption value first in the present embodiment, then,
Influence factor is modeled using wavelet neural network algorithm, obtains initialization model, it is clear that since wavelet neural network is calculated
Method has self-learning function, it is possible to using wavelet neural network algorithm to the influence more shadows of polycrystalline reduction process
The factor of sound is analyzed, therefore, using the initialization model of wavelet neural network algorithm creation to target polycrystalline reduction process
When power consumption values are predicted, it will be able to significantly improve the precision of prediction of target polycrystalline reduction process energy consumption value.Also, it is obtaining
After initialization model, initialization model is optimized using SFLA algorithm once more, obtains object module, to avoid initial
The parameter changed in model falls into locally optimal solution, so, it can be further improved the prediction essence of object module using this kind of method
Degree, it is clear that, will be into one when being predicted using the object module in the present embodiment target polycrystalline reduction process energy consumption value
Step improves the precision of prediction of target polycrystalline reduction process energy consumption value.
Based on the above embodiment, the present embodiment is further described and optimizes to above-described embodiment, specifically, above-mentioned step
Rapid S12: modeling influence factor using wavelet neural network algorithm, before obtaining the process of initialization model, further includes:
Dimension-reduction treatment is carried out to influence factor.
It is understood that influence polycrystalline reduction process energy consumption value influence factor it is varied, these influence because
In element, necessarily has some influence factors and show linearly related characteristic, so, in the present embodiment, in order to reduce
Data processing quantity when handling in follow-up process step data, also to the influence polycrystalline reduction process got
Influence factor has carried out dimension-reduction treatment, to further increase the speed of creation initialization model.
Specifically, above-mentioned steps: carrying out the process of dimension-reduction treatment to influence factor, comprising:
Dimension-reduction treatment is carried out to influence factor using PCA algorithm.
Specifically, being using PCA algorithm (Principal Component Analysis) come to shadow in the present embodiment
The influence factor for ringing polycrystalline reduction process energy consumption value carries out dimension-reduction treatment, that is, getting rid of influence polysilicon using PCA algorithm
Characteristic quantity irrelevant in the influence factor of process energy consumption value is restored, is influenced to reach reduction and influence polycrystalline reduction process
The purpose of factor, and then data processing quantity when handling in follow-up process step data can be significantly reduced.
Certainly, in practical operation, LDA algorithm (Linear Discriminant Analysis, line can also be utilized
Property diagnostic method), KPCA (Kernel Principal Component Analysis, the Principal Component Analysis Algorithm based on core) come
Dimension-reduction treatment is carried out to the influence factor for influencing polycrystalline reduction process, is not limited specifically herein.
Based on the above embodiment, the present embodiment is further described and optimizes to above-described embodiment, specifically, small echo is refreshing
Model expression through network algorithm are as follows:
In formula, n is the number of hidden node, wijFor the weight for connecting hidden layer node i and input layer node j, φx,yFor
Continuous wavelet, x be wavelet neural network algorithm in currently imply node coefficient of dilatation, y be wavelet neural network algorithm in when
The translation coefficient of preceding implicit node, p are the number of input layer node, wjkFor the power for connecting hidden layer node j and input layer node k
Value.
In the present embodiment, there is provided the mathematical model expression formula of wavelet neural network algorithm, pass through the mathematical model
Expression formula can model the influence factor for influencing polycrystalline reduction process energy consumption value, obtain initialization model.
Furthermore, it is desirable to which explanation, is using gradient descent method come in network model in wavelet neural network algorithm
Parameter to be determined, such as: output layer weight wij, hidden layer weight wjk, contraction-expansion factor xj, shift factor yjIt is solved, is counted
Calculation obtains the gradient value in wavelet neural network algorithm.It is also contemplated that more shadows can be extracted using this kind of method
Ring the influence factor of polycrystalline reduction process.
Wherein, in wavelet neural network algorithm objective function mathematic(al) representation are as follows:
In formula,For the real output value of i-th of output node of k input sample,For withCorresponding small echo mind
Prediction output valve through network, k are the sum of input sample, and n is the number of dimensions for exporting space.
The weight of input layer in wavelet neural network algorithm are as follows:
In formula,For the real output value of the output node of k input sample,For withCorresponding wavelet neural
The prediction output valve of network, k are the sum of input sample, φx,yFor continuous wavelet, x is current hidden in wavelet neural network algorithm
Coefficient of dilatation containing node, y are the translation coefficient that node is currently implied in wavelet neural network algorithm;
The weight of output layer in wavelet neural network algorithm are as follows:
In formula,For the real output value of i-th of output node of k input sample,For withCorresponding small echo mind
Prediction output valve through network, k are the sum of input sample, φx,yFor continuous wavelet, x is current in wavelet neural network algorithm
The coefficient of dilatation of implicit node, y are the translation coefficient that node is currently implied in wavelet neural network algorithm, xjFor contraction-expansion factor.
The above are the mathematical model expression formulas of key parameter models some in wavelet neural network algorithm and function, it is clear that
It, can be to the influence of polycrystalline reduction process energy consumption value by above-mentioned model expression based on wavelet neural network theory of algorithm
Factor is modeled, and initialization model is obtained.
Based on the above embodiment, the present embodiment is further described and optimizes to above-described embodiment, specifically, above-mentioned step
Rapid S13: optimizing initialization model using SFLA algorithm, after obtaining the process of object module, further includes:
Utilize the precision of prediction of test data detection object module.
In order to further detect object module to the precision of prediction of polycrystalline reduction process energy consumption value, in the present embodiment,
Also detected using precision of prediction of the test data to object module.Specifically, can be and obtaining in practical operation
When taking the influence factor of polycrystalline reduction process energy consumption value, these influence factors are divided into training data and test data, so
Afterwards, initialization model is trained using training data, and then using test data to the precision of prediction of object module into
Performing check judges the levels of precision of object module with this.Obviously, by such method, staff can know in time
The precision of prediction of object module, and production procedure of target polycrystalline reduction process regulated and controled with this.
Based on the above embodiment, the present embodiment is further described and optimizes to above-described embodiment, as shown in Fig. 2, tool
Body, above-mentioned steps S13: optimizing initialization model using SFLA algorithm, obtains the process of object module, comprising:
Step S131: establishing initiation parameter using SFLA algorithm, to determine the quantity F of frog and the number of group in frog group
M is measured, and calculates the adaptive value of each frog in frog group;
Step S132: all frogs are subjected to descending arrangement according to the size of adaptive value, and F frog after sequence is pressed
It distributes according to default distributive condition to m group;
In the present embodiment, it will be specifically described and how initialization model to be optimized using SFLA algorithm, obtained
Object module.The cardinal principle of SFLA algorithm is: assuming that a group frog that lives in a piece of wetland, frog passes through different
Stone carries out the more place of jump search of food.Every frog realizes the exchange of information, and every frog by cultural exchanges
It can be defined as a solution of problem, then, entire frog group be divided into different groups, each group also has certainly
Oneself culture, that is, frog executes local searching strategy, find locally optimal solution by jumping in group.In addition, group
In every frog can all have oneself culture, also, the culture of these frogs can evolve with the evolution of sub-group, group
After Swarm Evolution to certain phase, the exchange of information is carried out between each sub-group again, to realize the mixing between sub-group
Operation has looked for globally optimal solution until meeting preset condition.
That is, be to establish initiation parameter using SFLA algorithm first, to determine the quantity F of frog and group in frog group
Quantity m, and calculate the adaptive value of each frog in frog group.When be calculated the frog group in each frog adaptive value it
Afterwards, so that it may all frogs are subjected to descending arrangement according to the size of adaptive value, and by F frog after sequence according to default point
It distributes with condition to m group.
Default distributive condition herein is illustrated by a specific example, it is assumed that the quantity of frog is in frog group
9, the quantity of group is 3, then after this 9 frogs carry out descending arrangement according to the size of adaptive value, so that it may will arrange
It is placed into first group in primary frog, deputy frog will be come and be placed in second group, it will
The frog for coming third position is placed in third group, later, the frog for coming the 4th is placed on first group
In the middle, the frog for coming the 5th is placed in second group, the frog for coming the 6th is placed on third race
In group, and so on, until this 9 frogs are placed in 3 groups.
Step S133: the frog in each group is made to optimize using cultural gene algorithm, obtains optimization frog;
Step S134: the optimization frog in m group is mixed, to determine the optimal solution in frog group;
Step S135: judge whether optimal solution meets default screening conditions;
Step S136: if so, optimal solution is input to initialization model, to obtain object module.
When F frog is placed into m group, so that it may using cultural gene algorithm to each group in
Frog optimize, obtain optimization frog, it is clear that this operation purpose be in order to enable each frog is more nearly group
Optimal value in the middle later mixes the optimization frog in m group, that is, the frog in each group
It has been carried out after the evolution of cultural gene algorithm, has allowed each frog to jump in each group, to each race
Frog in group is reconfigured, and is reconsolidated to obtain a new frog group, then, it is green to be recalculated each in new frog group
The adaptive value of the frog, and descending arrangement is carried out to the frog in new frog group again according to the size of adaptive value, to determine new frog group
In the frog with adaptive optimal control value, that is, determine the optimal solution in new frog group, when determining the optimal solution in new frog group
When, judge whether the optimal solution meets default screening conditions, that is, judge whether the optimal solution reaches stopping criterion for iteration, if
Reach stopping criterion for iteration, then stops the optimizing iteration to optimal solution, and optimal solution is input in initialization model, with
To object module.
Correspondingly, above-mentioned steps S135: after judging the process whether optimal solution meets default screening conditions, further includes:
All frogs are subjected to descending arrangement according to the size of adaptive value if it is not, then executing again, and only by the F after sequence
The step of frog distributes F frog to m group according to default distributive condition.
It is understood that illustrating the optimal solution if searching for obtained optimal solution and being unsatisfactory for default screening conditions
It is not the globally optimal solution in frog group, needs to again return to execution step S132 at this time, redefines out in frog group most
The figure of merit optimizes purpose to the parameter in initialization model to reach, to obtain object module.
Correspondingly, the invention also discloses a kind of forecasting systems of polycrystalline reduction process energy consumption value, as shown in figure 3, should
System includes:
Factor obtains module 21, for obtaining the influence factor of polycrystalline reduction process energy consumption value;
Model initialization module 22 is obtained initial for being modeled using wavelet neural network algorithm to influence factor
Change model;
Model optimization module 23 obtains object module for optimizing using SFLA algorithm to initialization model;
Model prediction module 24, for being predicted using power consumption values of the object module to target polycrystalline reduction process.
Correspondingly, being stored on computer readable storage medium the invention also discloses a kind of computer readable storage medium
There is computer program, the computer program realizes polycrystalline reduction process energy consumption value as previously disclosed when being executed by processor
Prediction technique the step of.
Correspondingly, the invention also discloses a kind of pre- measurement equipments of energy consumption of polycrystalline reduction process, as shown in figure 4, this sets
It is standby to include:
Memory 31, for storing computer program;
Processor 32 realizes the pre- of polycrystalline reduction process energy consumption value as previously disclosed when for executing computer program
The step of survey method.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with it is other
The difference of embodiment, same or similar part may refer to each other between each embodiment.For being filled disclosed in embodiment
For setting, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is referring to method part
Explanation.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that the process, method, article or equipment for including a series of elements not only includes that
A little elements, but also including other elements that are not explicitly listed, or further include for this process, method, article or
The intrinsic element of equipment.In the absence of more restrictions, the element limited by sentence "including a ...", is not arranged
Except there is also other identical elements in the process, method, article or apparatus that includes the element.
It to a kind of prediction technique of polycrystalline reduction process energy consumption value provided by the present invention, system, medium and sets above
Standby to be described in detail, used herein a specific example illustrates the principle and implementation of the invention, above
The explanation of embodiment is merely used to help understand method and its core concept of the invention;Meanwhile for the general skill of this field
Art personnel, according to the thought of the present invention, there will be changes in the specific implementation manner and application range, in conclusion this
Description should not be construed as limiting the invention.
Claims (10)
1. a kind of prediction technique of polycrystalline reduction process energy consumption value characterized by comprising
Obtain the influence factor of polycrystalline reduction process energy consumption value;
The influence factor is modeled using wavelet neural network algorithm, obtains initialization model;
The initialization model is optimized using SFLA algorithm, obtains object module;
It is predicted using power consumption values of the object module to target polycrystalline reduction process.
2. the method according to claim 1, wherein it is described using wavelet neural network algorithm on the influence because
Element is modeled, before obtaining the process of initialization model, further includes:
Dimension-reduction treatment is carried out to the influence factor.
3. according to the method described in claim 2, it is characterized in that, the mistake for carrying out dimension-reduction treatment to the influence factor
Journey, comprising:
Dimension-reduction treatment is carried out to the influence factor using PCA algorithm.
4. the method according to claim 1, wherein the model expression of the wavelet neural network algorithm are as follows:
In formula, n is the number of hidden node, wijFor the weight for connecting hidden layer node i and input layer node j, φx,yIt is continuous
Small echo, x are the coefficient of dilatation that node is currently implied in wavelet neural network algorithm, and y is current hidden in wavelet neural network algorithm
Translation coefficient containing node, p are the number of input layer node, wjkFor the weight for connecting hidden layer node j and input layer node k.
5. the method according to claim 1, wherein described carry out the initialization model using SFLA algorithm
Optimization, after obtaining the process of object module, further includes:
The precision of prediction of the object module is detected using test data.
6. method according to any one of claims 1 to 5, which is characterized in that described to utilize SFLA algorithm to described initial
Change model to optimize, obtain the process of object module, comprising:
Initiation parameter is established using SFLA algorithm, to determine the quantity F of the frog and quantity m of group in frog group, and calculates institute
State the adaptive value of each frog in frog group;
All frogs are subjected to descending arrangement according to the size of adaptive value, and by F frog after sequence according to default distribution item
Part is distributed to m group;
The frog in each group is made to optimize using cultural gene algorithm, obtains optimization frog;
Optimization frog in the m group is mixed, to determine the optimal solution in the frog group;
Judge whether the optimal solution meets default screening conditions;
If so, the optimal solution is input to the initialization model, to obtain the object module.
7. according to the method described in claim 6, it is characterized in that, described judge whether the optimal solution meets default screening item
After the process of part, further includes:
All frogs are subjected to descending arrangement according to the size of adaptive value if it is not, then executing again, and by F frog after sequence
The step of F frog is distributed to m group according to default distributive condition.
8. a kind of forecasting system of polycrystalline reduction process energy consumption value characterized by comprising
Factor obtains module, for obtaining the influence factor of polycrystalline reduction process energy consumption value;
Model initialization module is initialized for being modeled using wavelet neural network algorithm to the influence factor
Model;
Model optimization module obtains object module for optimizing using SFLA algorithm to the initialization model;
Model prediction module, for being predicted using power consumption values of the object module to target polycrystalline reduction process.
9. a kind of computer readable storage medium, which is characterized in that be stored with computer on the computer readable storage medium
Program, the computer program realize polycrystalline reduction process as described in any one of claim 1 to 7 when being executed by processor
The step of prediction technique of power consumption values.
10. a kind of pre- measurement equipment of energy consumption of polycrystalline reduction process characterized by comprising
Memory, for storing computer program;
Processor realizes polycrystalline reduction work as described in any one of claim 1 to 7 when for executing the computer program
The step of prediction technique of sequence power consumption values.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811011310.8A CN109034275A (en) | 2018-08-31 | 2018-08-31 | Prediction technique, system, medium and the equipment of polycrystalline reduction process energy consumption value |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811011310.8A CN109034275A (en) | 2018-08-31 | 2018-08-31 | Prediction technique, system, medium and the equipment of polycrystalline reduction process energy consumption value |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109034275A true CN109034275A (en) | 2018-12-18 |
Family
ID=64622625
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811011310.8A Pending CN109034275A (en) | 2018-08-31 | 2018-08-31 | Prediction technique, system, medium and the equipment of polycrystalline reduction process energy consumption value |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109034275A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110362558A (en) * | 2019-06-12 | 2019-10-22 | 广东工业大学 | A kind of energy consumption data cleaning method based on neighborhood propagation clustering |
CN115032891A (en) * | 2022-08-11 | 2022-09-09 | 科大智能物联技术股份有限公司 | Polycrystalline silicon reduction furnace control method based on time series prediction |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102063094A (en) * | 2009-11-18 | 2011-05-18 | 上海宝信软件股份有限公司 | Optimization method for power distribution among steel rolling process sets |
CN103837859A (en) * | 2014-03-05 | 2014-06-04 | 湖州师范学院 | Method for calculating position of sound emission source through wavelet neural network of mixed leapfrog |
-
2018
- 2018-08-31 CN CN201811011310.8A patent/CN109034275A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102063094A (en) * | 2009-11-18 | 2011-05-18 | 上海宝信软件股份有限公司 | Optimization method for power distribution among steel rolling process sets |
CN103837859A (en) * | 2014-03-05 | 2014-06-04 | 湖州师范学院 | Method for calculating position of sound emission source through wavelet neural network of mixed leapfrog |
Non-Patent Citations (4)
Title |
---|
乔非 等著: "《钢铁企业能源管理模型与系统节能技术》", 31 December 2016 * |
牛凡超: "基于改进蛙跳算法的小波神经网络短期电力负荷预测研究", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
韩志艳 等: "《面向语音与面部表情信号的多模式情感识别技术研究》", 31 January 2017 * |
高盛祥 等: "基于蚁群优化的小波神经网络钢铁生产能耗预测模型", 《昆明理工大学学报(自然科学版)》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110362558A (en) * | 2019-06-12 | 2019-10-22 | 广东工业大学 | A kind of energy consumption data cleaning method based on neighborhood propagation clustering |
CN110362558B (en) * | 2019-06-12 | 2022-12-16 | 广东工业大学 | Energy consumption data cleaning method based on neighborhood propagation clustering |
CN115032891A (en) * | 2022-08-11 | 2022-09-09 | 科大智能物联技术股份有限公司 | Polycrystalline silicon reduction furnace control method based on time series prediction |
CN115032891B (en) * | 2022-08-11 | 2022-11-08 | 科大智能物联技术股份有限公司 | Polycrystalline silicon reduction furnace control method based on time series prediction |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Lin et al. | An improved moth-flame optimization algorithm for support vector machine prediction of photovoltaic power generation | |
CN106373025B (en) | Power information acquisition system based on outlier detection anti-electricity-theft monitoring method in real time | |
Zhang et al. | An adaptive particle swarm optimization algorithm for reservoir operation optimization | |
Li et al. | An empirical comparison of multiple linear regression and artificial neural network for concrete dam deformation modelling | |
CN106446571B (en) | The error assessment method of wind farm power prediction | |
CN106796577A (en) | Use the energy foundation facility sensor data correction of regression model | |
Chen et al. | Improved particle swarm optimization-based form-finding method for suspension bridge installation analysis | |
CN104376371B (en) | A kind of distribution based on topology is layered load forecasting method | |
CN109919421A (en) | Short-term power load prediction model establishment method based on VMD-PSO-BPNN | |
CN105320834B (en) | Method for calculating number of people based on using state of electric appliance and monitoring system thereof | |
CN110334879A (en) | Power grid bus reactive load forecasting method and device | |
CN104573857A (en) | Power grid load rate prediction method based on intelligent algorithm optimization and combination | |
CN112990500A (en) | Transformer area line loss analysis method and system based on improved weighted gray correlation analysis | |
CN109034275A (en) | Prediction technique, system, medium and the equipment of polycrystalline reduction process energy consumption value | |
CN103617563B (en) | A kind of water supply network theoretical based on geo-statistic spatial analysis is without monitoring node pressure determination statement | |
CN110969306A (en) | Power distribution low-voltage distribution area load prediction method and device based on deep learning | |
CN113591368A (en) | Comprehensive energy system multi-energy load prediction method and system | |
Fu et al. | CPS information security risk evaluation system based on Petri net | |
Yan et al. | Short-term load forecasting of smart grid based on load spatial-temporal distribution | |
CN109034497A (en) | Prediction technique, system, medium and the equipment of polycrystalline reduction process energy consumption value | |
El-Ghandour et al. | Groundwater management using a new coupled model of flow analytical solution and particle swarm optimization | |
CN116316617B (en) | Multi-station intelligent fusion new energy generation power region prediction method and system | |
CN109961160A (en) | A kind of power grid future operation trend predictor method and system based on trend parameter | |
Sha et al. | Prediction of commercial building lighting energy consumption based on EPSO-BP | |
Lu et al. | Forecasting China’s per capita living energy consumption by employing a novel DGM (1, 1, tα) model with fractional order accumulation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181218 |
|
RJ01 | Rejection of invention patent application after publication |