CN105426959B - Aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms - Google Patents
Aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms Download PDFInfo
- Publication number
- CN105426959B CN105426959B CN201510750725.7A CN201510750725A CN105426959B CN 105426959 B CN105426959 B CN 105426959B CN 201510750725 A CN201510750725 A CN 201510750725A CN 105426959 B CN105426959 B CN 105426959B
- Authority
- CN
- China
- Prior art keywords
- bacterium
- neural network
- individual
- emission
- algorithms
- 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.)
- Active
Links
- 229910052782 aluminium Inorganic materials 0.000 title claims abstract description 47
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 35
- 239000004411 aluminium Substances 0.000 title claims abstract description 31
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 25
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 16
- 241000894006 Bacteria Species 0.000 claims abstract description 64
- 239000005431 greenhouse gas Substances 0.000 claims abstract description 28
- 238000004519 manufacturing process Methods 0.000 claims abstract description 19
- 235000013305 food Nutrition 0.000 claims abstract description 12
- 238000005457 optimization Methods 0.000 claims abstract description 11
- 230000006870 function Effects 0.000 claims description 39
- 210000002569 neuron Anatomy 0.000 claims description 22
- 230000036541 health Effects 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 10
- 210000004027 cell Anatomy 0.000 claims description 7
- AZDRQVAHHNSJOQ-UHFFFAOYSA-N alumane Chemical compound [AlH3] AZDRQVAHHNSJOQ-UHFFFAOYSA-N 0.000 claims description 6
- 230000001174 ascending effect Effects 0.000 claims description 6
- 230000008901 benefit Effects 0.000 claims description 6
- 230000005540 biological transmission Effects 0.000 claims description 5
- 238000013459 approach Methods 0.000 claims description 4
- 230000006399 behavior Effects 0.000 claims description 4
- 230000001580 bacterial effect Effects 0.000 claims description 3
- 230000004087 circulation Effects 0.000 claims description 3
- 239000003792 electrolyte Substances 0.000 claims description 3
- 239000003016 pheromone Substances 0.000 claims description 3
- 108090000623 proteins and genes Proteins 0.000 claims description 3
- 238000009790 rate-determining step (RDS) Methods 0.000 claims description 3
- 230000009897 systematic effect Effects 0.000 claims description 3
- 230000005611 electricity Effects 0.000 claims description 2
- 241000207187 Candidatus Legionella jeonii Species 0.000 claims 1
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 238000005868 electrolysis reaction Methods 0.000 abstract description 14
- 230000008569 process Effects 0.000 abstract description 9
- 238000004364 calculation method Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000007935 neutral effect Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000004131 Bayer process Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000010429 evolutionary process Effects 0.000 description 1
- 239000007789 gas Substances 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 239000012774 insulation material Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000004886 process control Methods 0.000 description 1
- 230000001568 sexual effect Effects 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- TXEYQDLBPFQVAA-UHFFFAOYSA-N tetrafluoromethane Chemical compound FC(F)(F)F TXEYQDLBPFQVAA-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- C—CHEMISTRY; METALLURGY
- C25—ELECTROLYTIC OR ELECTROPHORETIC PROCESSES; APPARATUS THEREFOR
- C25C—PROCESSES FOR THE ELECTROLYTIC PRODUCTION, RECOVERY OR REFINING OF METALS; APPARATUS THEREFOR
- C25C3/00—Electrolytic production, recovery or refining of metals by electrolysis of melts
- C25C3/06—Electrolytic production, recovery or refining of metals by electrolysis of melts of aluminium
- C25C3/20—Automatic control or regulation of cells
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Physiology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Electrochemistry (AREA)
- Materials Engineering (AREA)
- Metallurgy (AREA)
- Organic Chemistry (AREA)
- Feedback Control In General (AREA)
- Water Treatment By Electricity Or Magnetism (AREA)
Abstract
The present invention provides a kind of aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms, first, aluminum electrolysis process is modeled using BP neural network, then, using the multiple target bacterium based on adaptive step look for food optimization algorithm production process model is optimized, obtain the one group of optimal solution and the corresponding current efficiency of the optimal solution and greenhouse gas emissions of each decision variable, wherein, when being optimized to production process model, the step-length for tending to operation is adjusted into Mobile state according to flora Evolving State, to ensure the diversity of population and convergence.This method determines the optimal value of technological parameter during aluminum electrolysis, effectively increases current efficiency, reduces greenhouse gas emissions, is really achieved the purpose of energy-saving and emission-reduction.
Description
Technical field
The present invention relates to the automatic control technology during aluminum electrolysis, and in particular to one kind based on BP neural network with
The aluminium electroloysis energy-saving and emission-reduction method of adaptive M BFO algorithms.
Background technology
Aluminium electroloysis is a complicated industrial processes, and generally use Bayer process is smelted, however, this method consumes energy
Huge and efficiency is low.At the same time, a large amount of greenhouse gases can be produced during aluminum electrolysis, environmental pollution is serious.Therefore, exist
Ensure how aluminium cell steadily on the premise of production, improves current efficiency, reduce energy consumption, reduce polluted gas discharge capacity, with
Realize that efficient, energy saving, emission reduction has become the productive target of aluminium electroloysis enterprise.But material chemistry complicated inside aluminium cell
Change causes groove intrinsic parameter more with exterior a variety of uncertain operation factors, and the spy such as non-linear, strong coupling is showed between parameter
Point, and the parameter such as pole span, thermal insulation material thickness is difficult to measurement in real time, adjustment, gives aluminum electrolysis process control optimization band
Carry out certain difficulty.
The content of the invention
The application by providing a kind of aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms,
With huge energy consumption, efficiency are low caused by it can not obtain optimal procedure parameters during solution in the prior art aluminum electrolysis
And the technical problem of serious pollution environment.
In order to solve the above technical problems, the application is achieved using following technical scheme：
A kind of aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms, includes the following steps：
S1：Selection forms decision variable X=[x to current efficiency and the influential control parameter of greenhouse gas emissions1,
x2,…,xM], M is the number of selected parameter；
S2：Selected aluminium electrolytic industry scene, collection N group decision variables X1, X2..., XNAnd its corresponding current efficiency y1,
y2..., yNWith greenhouse gas emissions z1, z2..., zNAs data sample, with each decision variable XiAs input, difference
With corresponding current efficiency yiWith greenhouse gas emissions ziAs output, sample is trained with BP neural network, is examined
Test, establish aluminium cell production process model；
S3：Looked for food using multiple target bacterium and optimize algorithm, i.e. MBFO algorithms, to two production process moulds obtained by step S2
Type optimizes, and obtains one group of optimizing decision variable XbestAnd its corresponding current efficiency ybestWith greenhouse gas emissions zbest,
During optimization, the step-length for tending to operation is adjusted into Mobile state according to flora Evolving State, to ensure the diversity of population and convergence
Property；
The step-length dynamic adjustment for tending to operation is as follows：
The step-length of i-th bacterium, the t+1 times iteration is
Step-length Tuning function is
In formula, Ct(i) it is the step-length of i-th bacterium, the t times iteration, diFor the crowding distance of i-th bacterium, μ, λ ∈ (0,
1) it is step-length Dynamic gene, For the crowding distance of i-th bacterium, the t times iteration, L is bacterial community
Size；
S4：According to the optimizing decision variable X obtained by step S3bestIn control parameter come it is selected in rate-determining steps S2
Aluminium electrolytic industry scene, reaches energy-saving and emission-reduction.
With reference to practical conditions, 8 parameters are have selected in step S1 and form decision variables, be respectively potline current, under
Expect number, molecular proportion, aluminum yield, aluminium level, electrolyte level, bath temperature and tank voltage.
In order to meet modeling requirement, the BP neural network in step S2 is made of input layer, hidden layer and output layer；
For the production process model constructed by current efficiency, its input layer uses 8 neuron nodes, hidden layer
Using 13 neuron nodes, output layer uses 1 neuron node, and input layer to transmission function between hidden layer is Tansig
Function, hidden layer to the function between output layer are Purelin functions, and iterations during sample training is 500；
For the production process model constructed by greenhouse gas emissions, its input layer uses 8 neuron nodes,
Hidden layer uses 12 neuron nodes, and output layer uses 1 neuron node, and input layer is to transmission function between hidden layer
Logsig functions, hidden layer to the function between output layer are Purelin functions, and iterations during sample training is 500.
Further, the MBFO algorithms in step S3 comprise the following steps：
S31：The value of decision variable X is considered as bacterium position, is generated at random according to the scope of parameters in decision variable X
L bacterium forms flora initial position；
S32：Systematic parameter is initialized, including tends to times Nc, times N of advancing in approach behaviors, breed times Nre, drive
Dissipate times Ned, disperse probability Ped, external archive scale K；
External archive is used to preserve the Pareto solutions that individual produces in search procedure and meter is exported at the end of algorithm
Calculate result；
S33：Perform and tend to operation, including overturn and advance；
Assuming that i-th bacterium is in the position that jth time tends to operation kth time duplication operation and disperses for the l times after operation
θi(j, k, l), i=1,2 ..., L
Then θi(j+1, k, l)=θi(j,k,l)+C(i)*dcti,
In formula, dctiSelected random vector direction when being the last upset of i-th bacterium, C (i) are them along dcti
The paces length that direction is advanced, andΔiIt is the vector of [- 1,1] interior random number for each component, it is vectorial
Dimension is identical with the dimension of decision variable X；
S34：According to the pheromone concentration J between individualccExecution is bunched operation：
S35：The health function of flora is calculated, and is carried out descending arrangement, eliminates the small half bacterium of health function value,
The other half big bacterium of health function value is bred, and careful bacterium ability of looking for food keeps consistent with parent；
To given k, l, the health function of every bacterium isIn formula,Represent the
The energy of i bacterium, J (i, j, k, l) represent that bacterium i replicates operation in jth time trend operation kth time and disperses operation with the l times
Fitness function value afterwards, NcRepresent to tend to number,It is bigger, represent that the ability of looking for food of bacterium i is stronger；
S36：The S35 floras produced are merged with the preceding flora for once iterating to calculate generation, calculate this stylish flora individual
The crowding distance of Pareto solutions is simultaneously ranked up according to the Pareto crowding distances solved, and L advantage individual is formed next before selection
For flora；
S37：Disperse：After bacterium experience several generations replicates, to disperse probability PedDispersed the optional position into search space；
S38：Judge to optimize whether algorithm meets termination condition, such as meet, then export Pareto forward positions, that is, optimizing decision and become
Measure XbestAnd its corresponding current efficiency ybestWith greenhouse gas emissions zbest, such as it is unsatisfactory for, then jumps to step S33 circulations
Perform.
Yet further, the crowding distance that Pareto solutions are calculated in step S36 includes the following steps：
A1：Ascending sort is carried out to all optimum individuals in external archive；
A2：Calculating each solves distance of the two adjacent individuals in each optimization aim spatially；
A3：These distances are added up to the crowding distance of required Pareto solutions, and set the crowding distance of Boundary Solutions as nothing
It is poor big；
I-th individual crowding distance be：
In formula,For i-th of individual distance in target j, R is the individual sequence number of crowding distance maximum after ascending order arrangement,
M is object space dimension；
When being updated in step S3 to external archive, it is assumed that external archive A maximum capacities are q, and ith iteration calculates shape
Into non-domination solution be Q, specifically comprise the following steps：
B1：All individual crowding distances of external archive are calculated, and make descending arrangement；
B2：Update external archive：In ith iteration, if Q ＞ A, A is replaced with Q；
B3：Judge whether external archive A capacity meets the requirement of maximum capacity, if number of individuals n ＜ q in A, replicate Q to A
In, form new external archive A1；
B4：Judge in new external archive A1 with the presence or absence of the identical individual of desired value, if only retaining one in the presence of if；
B5：If number of individuals n1≤q in new external archive A1, carries out next iteration, if n1 ＞ q, calculate crowded
Distance di, and descending arranges, and deletes n1-q minimum individual of crowding distance.
Compared with prior art, the technical solution that the application provides, the technique effect or advantage having are：This method determines
The optimal value of technological parameter, effectively increases current efficiency, reduces greenhouse gas emissions, very during aluminum electrolysis
Just achieve energy-saving and emission reduction purposes.
Brief description of the drawings
Fig. 1 is flow chart of the method for the present invention；
Fig. 2 is current efficiency prediction result figure；
Fig. 3 is CF4Forecasting of discharged quantity result figure；
Fig. 4 is current efficiency prediction-error image；
Fig. 5 is CF4Forecasting of discharged quantity Error Graph.
Embodiment
By providing, a kind of aluminium electroloysis based on BP neural network and adaptive M BFO algorithms is energy saving to be subtracted the embodiment of the present application
Discharge method, it is huge to consume energy during solution in the prior art aluminum electrolysis caused by it can not obtain optimal procedure parameters
Greatly, efficiency it is low and it is serious pollution environment technical problem.
In order to better understand the above technical scheme, in conjunction with appended figures and specific embodiments, it is right
Above-mentioned technical proposal is described in detail.
Embodiment
As shown in Figure 1, a kind of aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms, it is closed
Key is, includes the following steps：
S1：Selection forms decision variable X=[x to current efficiency and the influential control parameter of greenhouse gas emissions1,
x2,…,xM], M is the number of selected parameter；
Implementation is by counting influential on current efficiency and greenhouse gas emissions original during aluminum electrolysis
Variable, and therefrom determine to influence big parameter as decision variable X to current efficiency and greenhouse gas emissions；
By being counted to obtain to current efficiency and greenhouse gas emission to measurement parameter during actual industrial production
Amount influences maximum variable：Potline current x1, blanking number x2, molecular proportion x3, aluminum yield x4, the horizontal x of aluminium5, electrolyte level
x6, bath temperature x7, tank voltage x8Totally 8 variables.
S2：Selected aluminium electrolytic industry scene, collection N group decision variables X1, X2..., XNAnd its corresponding current efficiency y1,
y2..., yNWith greenhouse gas emissions z1, z2..., zNAs data sample, with each decision variable XiAs input, difference
With corresponding current efficiency yiWith greenhouse gas emissions ziAs output, sample is trained with BP neural network, is examined
Test, establish aluminium cell production process model；
In the present embodiment, the 223# groove electrolytic cells in Chongqing Tiantai Aluminium Industry Co., Ltd. 170KA series electrolytic cells are gathered
400 groups of data in actual production process are as experimental data, wherein first 380 groups are used as training sample, latter 20 groups as test
Sample, data sample are as shown in table 1 below.
1 data sample of table
In neutral net design, the number of hidden nodes number be the key for determining neural network model quality, it is and refreshing
Through the difficult point in network design, the number of nodes of hidden layer is determined using trial and error procedure here.
In formula, p is hidden neuron number of nodes, and n is input layer number, and m is output layer neuron number, k 1-10
Between constant, the arrange parameter of BP neural network is as shown in table 2 below in this example.
Table 2BP nerve arrange parameters
Specially：For the production process model constructed by current efficiency, its input layer uses 8 neuron sections
Point, hidden layer use 13 neuron nodes, and output layer uses 1 neuron node, and input layer is to transmitting letter between hidden layer
Number is Tansig functions, and hidden layer to the function between output layer is Purelin functions, and iterations during sample training is
500；
For the production process model constructed by greenhouse gas emissions, its input layer uses 8 neuron nodes,
Hidden layer uses 12 neuron nodes, and output layer uses 1 neuron node, and input layer is to transmission function between hidden layer
Logsig functions, hidden layer to the function between output layer are Purelin functions, and iterations during sample training is 500.
Carried out in the training process of neutral net essentially according to following steps：
X is setk=[xk1,xk2,…,xkM] (k=1,2 ..., N) be input vector, N is training sample number,
For the g times iteration when input layer M and hidden layer I it
Between weighted vector, WJP(g) weighted vector when being the g times iteration between hidden layer J and output layer P, Yk(g)=[yk1(g),yk2
(g),…,ykP(g)] reality output of network, d when (k=1,2 ..., N) is the g times iterationk=[dk1,dk2,…,dkP] (k=
1,2 ..., N) it is desired output；
Aluminum electrolysis process model is established in step S2 to specifically comprise the following steps：
S21：Initialization, if iterations g initial values are 0, is assigned to W respectivelyMI(0)、WJP(0) (0,1) section it is random
Value；
S22：Stochastic inputs sample Xk；
S23：To input sample Xk, the input signal and output signal of every layer of neuron of forward calculation BP neural network；
S24：According to desired output dkWith reality output Yk(g), calculation error E (g)；
S25：Whether error in judgement E (g) meets the requirements, and is such as unsatisfactory for, then enters step S26, such as meets, then enters step
S29；
S26：Judge whether iterations g+1 is more than maximum iteration, such as larger than, then enter step S29, otherwise, into
Enter step S27；
S27：To input sample XkThe partial gradient δ of every layer of neuron of backwards calculation；
S28：Modified weight amount Δ W is calculated, and corrects weights, calculation formula is： In formula, η is learning efficiency；G=g+1 is made, jumps to step S23；
S29：Judge whether to complete all training samples, if it is, completing modeling, otherwise, continue to jump to step
S22。
By the above process, BP neural network prediction effect is can obtain as shown in Fig. 2,3,4,5.Aluminum electrolysis process is excellent
The basis of change is the foundation of Optimized model, and model accuracy directly affects optimum results.By analyzing Fig. 2,3,4,5, warp
BP neural network is trained, and the largest prediction error of current efficiency is -0.407%, greenhouse gases carbon tetrafluoride CF4Forecasting of discharged quantity
Error 0.0165%, model prediction accuracy is high, meets modeling demand.
S3：Looked for food using multiple target bacterium and optimize algorithm, i.e. MBFO algorithms, to two production process moulds obtained by step S2
Type optimizes, and obtains one group of optimizing decision variable XbestAnd its corresponding current efficiency ybestWith greenhouse gas emissions zbest,
During optimization, the step-length for tending to operation is adjusted into Mobile state according to flora Evolving State, to ensure the diversity of population and convergence
Property；
The step-length dynamic adjustment for tending to operation is as follows：
The step-length of i-th bacterium, the t+1 times iteration is
Step-length Tuning function is
In formula, Ct(i) it is the step-length of i-th bacterium, the t times iteration, diFor the crowding distance of i-th bacterium, μ, λ ∈ (0,
1) it is step-length Dynamic gene, For the crowding distance of i-th bacterium, the t times iteration, L is bacterial community
Size；
Further, the MBFO algorithms in step S3 comprise the following steps：
S31：The value of decision variable X is considered as bacterium position, is generated at random according to the scope of parameters in decision variable X
L bacterium forms flora initial position；
S32：Systematic parameter is initialized, including tends to times Nc, times N of advancing in approach behaviors, breed times Nre, drive
Dissipate times Ned, disperse probability Ped, external archive scale K；
S33：Perform and tend to operation, including overturn and advance；
Assuming that i-th bacterium is in the position that jth time tends to operation kth time duplication operation and disperses for the l times after operation
θi(j, k, l), i=1,2 ..., L, then θi(j+1, k, l)=θi(j,k,l)+C(i)*dcti；
In formula, dctiSelected random vector direction when being the last upset of i-th bacterium, C (i) are them along dcti
The paces length that direction is advanced, andΔiIt is the vector of [- 1,1] interior random number for each component, it is vectorial
Dimension is identical with the dimension of decision variable X；
S34：According to the pheromone concentration J between individualccExecution is bunched operation：
S35：The health function of flora is calculated, and is carried out descending arrangement, eliminates the small half bacterium of health function value,
The other half big bacterium of health function value is bred, and careful bacterium ability of looking for food keeps consistent with parent；
To given k, l, the health function of every bacterium isIn formula,Represent the
The energy of i bacterium, J (i, j, k, l) represent that bacterium i replicates operation in jth time trend operation kth time and disperses operation with the l times
Fitness function value afterwards, NcRepresent to tend to number,It is bigger, represent that the ability of looking for food of bacterium i is stronger；
After trend operation by a cycle, according to the rule of " survival of the fittest ", bacterium colony will be obtained according to every bacterium
Energy carry out duplication operation.In operating process is replicated, every bacterium is ranked up according to the gross energy found, and obtains energy
Low half bacterium will die, and obtain the high half bacterium of energy into line splitting, maintain bacterium colony total number of bacteria constant with this.
S36：The S35 floras produced are merged with the preceding flora for once iterating to calculate generation, calculate this stylish flora individual
The crowding distance of Pareto solutions is simultaneously ranked up according to the Pareto crowding distances solved, and L advantage individual is formed next before selection
For flora；
S37：Disperse：After bacterium experience several generations replicates, to disperse probability PedDispersed the optional position into search space；
After experience several generations replicates operation, bacterium colony will gather, and make its various sexual involution.In order to ensure the various of bacterium colony
Property, some individuals in bacterium colony will be with small probability PedDispersed, and reappear in position new in region of search.Although disperse
Operation destroys the approach behavior of bacterium, but bacterium is also possible to therefore occur in the abundanter region of food.
S37：Judge to optimize whether algorithm meets termination condition, such as meet, then export Pareto forward positions, that is, optimizing decision and become
Measure XbestAnd its corresponding current efficiency ybestWith greenhouse gas emissions zbest, such as it is unsatisfactory for, then jumps to step S33 circulations
Perform.
Further, the crowding distance that advantage individual is calculated in step S36 includes the following steps：
A1：Ascending sort is carried out to all optimum individuals in external archive；
A2：Calculating each solves distance of the two adjacent individuals in each optimization aim spatially；
A3：These distances are added the crowding distance up to required advantage individual, and set the crowding distance of Boundary Solutions as nothing
It is poor big；
I-th individual crowding distance be：
In formula,For i-th of individual distance in target j, R is the individual sequence number of crowding distance maximum after ascending order arrangement,
M is object space dimension；
The big individual of crowding distance has more chances for participating in next iteration and calculating, so as to maintain the more of algorithm population
Sample so that algorithm can converge to an equally distributed Pareto curved surface.Therefore, it is exterior by the use of crowding distance as renewal
The foundation of archives can make algorithm preferably converge to Pareto curved surfaces.
External archive is used to preserve the Pareto solutions that individual produces in search procedure and meter is exported at the end of algorithm
Calculate as a result, at the same time also can bootstrap algorithm evolutionary process to ensure algorithm global convergence.It is outer in order to control to safeguard external archive
The maximum-norm of portion's archives, while ensure that the diversity of Pareto solutions and optimum individual are not lost, have to algorithm important
Meaning.The present invention using based on crowding distance sequence external archive more new strategy so that external archive dynamic it is adaptive
Renewal, so that flora more rapidly moves in searching process towards target, so as to fast on the premise of population diversity is ensured
Speed convergence.
When being updated in step S3 to external archive, it is assumed that external archive A maximum capacities are q, and ith iteration calculates shape
Into non-domination solution be Q, specifically comprise the following steps：
B1：All individual crowding distances of external archive are calculated, and make descending arrangement；
B2：Update external archive：In ith iteration, if Q ＞ A, A is replaced with Q；
B3：Judge whether external archive A capacity meets the requirement of maximum capacity, if number of individuals n ＜ q in A, replicate Q to A
In, form new external archive A1；
B4：Judge in new external archive A1 with the presence or absence of the identical individual of desired value, if only retaining one in the presence of if；
B5：If number of individuals n1≤q in new external archive A1, carries out next iteration, if n1 ＞ q, calculate crowded
Distance di, and descending arranges, and deletes n1-q minimum individual of crowding distance.
Aluminum electrolysis process is optimized by above-mentioned steps can obtain 100 groups of optimal decision variables with it is corresponding defeated
Go out value, choose wherein most rational 3 groups and be listed in the table below in 3.
3 optimized producing parameter of table
The average value of contrast wherein optimal operating parameter and annual record in 2013 understands that current efficiency improves
4.25%th, CF4Discharge capacity reduces 0.64kg.
S4：According to the optimizing decision variable X obtained by step S3bestIn control parameter come it is selected in rate-determining steps S2
Aluminium electrolytic industry scene, reaches energy-saving and emission-reduction.
In above-described embodiment of the application, by providing a kind of aluminium electricity based on BP neural network and adaptive M BFO algorithms
Energy-saving and emission-reduction method is solved, first, aluminum electrolysis process is modeled using BP neural network, then, using based on crowded
The multiple target bacterium of distance look for food optimization algorithm production process model is optimized, obtain one group of optimal solution of each decision variable
And the corresponding current efficiency of the optimal solution and greenhouse gas emissions, wherein, when being optimized to production process model, root
The step-length for tending to operation is adjusted into Mobile state according to flora Evolving State, to ensure the diversity of population and convergence.This method
The optimal value of technological parameter during aluminum electrolysis is determined, effectively increases current efficiency, reduces greenhouse gas emission
Amount, is really achieved the purpose of energy-saving and emission-reduction.
It should be pointed out that it is limitation of the present invention that described above, which is not, the present invention is also not limited to the example above,
What those skilled in the art were made in the essential scope of the present invention changes, is modified, adds or replaces, and also should
Belong to protection scope of the present invention.
Claims (5)
1. a kind of aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms, includes the following steps：
S1：Selection forms decision variable X=[x to current efficiency and the influential control parameter of greenhouse gas emissions1,x2,…,
xM], M is the number of selected parameter；
S2：Selected aluminium electrolytic industry scene, collection N group decision variables X1, X2..., XNAnd its corresponding current efficiency y1, y2...,
yNWith greenhouse gas emissions z1, z2..., zNAs data sample, with each decision variable XiAs input, respectively with correspondence
Current efficiency yiWith greenhouse gas emissions ziAs output, sample is trained with BP neural network, is examined, established
Aluminium cell production process model；
S3：Looked for food optimization algorithm using multiple target bacterium, i.e. MBFO algorithms, to two production process models obtained by step S2 into
Row optimization, obtains one group of optimizing decision variable XbestAnd its corresponding current efficiency ybestWith greenhouse gas emissions zbest, optimization
When, the step-length for tending to operation is adjusted into Mobile state according to flora Evolving State, to ensure the diversity of population and convergence；
The step-length dynamic adjustment for tending to operation is as follows：
The step-length of i-th bacterium, the t+1 times iteration is
Step-length Tuning function is
In formula, Ct(i) it is the step-length of i-th bacterium, the t times iteration, diFor the crowding distance of i-th bacterium, μ, λ ∈ (0,1) are
Step-length Dynamic gene, For the crowding distance of i-th bacterium, the t times iteration, L is the big of bacterial community
It is small；
S4：According to the optimizing decision variable X obtained by step S3bestIn control parameter carry out selected aluminium electricity in rate-determining steps S2
Industry spot is solved, reaches energy-saving and emission-reduction.
2. the aluminium electroloysis energy-saving and emission-reduction method according to claim 1 based on BP neural network Yu adaptive M BFO algorithms,
It is characterized in that, have selected 8 parameters in step S1 forms decision variables, be respectively potline current, blanking number, molecular proportion,
Aluminum yield, aluminium level, electrolyte level, bath temperature and tank voltage.
3. the aluminium electroloysis energy-saving and emission-reduction side according to claim 1 or 2 based on BP neural network Yu adaptive M BFO algorithms
Method, it is characterised in that the BP neural network in step S2 is made of input layer, hidden layer and output layer；
For the production process model constructed by current efficiency, its input layer uses 8 neuron nodes, and hidden layer uses
13 neuron nodes, output layer use 1 neuron node, and input layer to transmission function between hidden layer is Tansig letters
Number, hidden layer to the function between output layer be Purelin functions, and iterations during sample training is 500；
For the production process model constructed by greenhouse gas emissions, its input layer uses 8 neuron nodes, hides
Layer uses 12 neuron nodes, and output layer uses 1 neuron node, and input layer is to transmission function between hidden layer
Logsig functions, hidden layer to the function between output layer are Purelin functions, and iterations during sample training is 500.
4. the aluminium electroloysis energy-saving and emission-reduction method according to claim 1 based on BP neural network Yu adaptive M BFO algorithms,
It is characterized in that, the MBFO algorithms in step S3 comprise the following steps：
S31：The value of decision variable X is considered as bacterium position, L are generated at random according to the scope of parameters in decision variable X
Bacterium forms flora initial position；
S32：Systematic parameter is initialized, including tends to times Nc, times N of advancing in approach behaviors, breed times Nre, disperse number
Ned, disperse probability Ped, external archive scale K；
External archive is used to preserve the Pareto solutions that individual produces in search procedure and the output calculating knot at the end of algorithm
Fruit；
S33：Perform and tend to operation, including overturn and advance；
Assuming that i-th bacterium is θ in the position that jth time tends to operation kth time duplication operation and disperses for the l times after operationi(j,
K, l), i=1,2 ..., L,
Then θi(j+1, k, l)=θi(j,k,l)+C(i)*dcti,
In formula, dctiSelected random vector direction when being the last upset of i-th bacterium, C (i) are them along dctiDirection
The paces length of advance, andΔiIt is the vector of [- 1,1] interior random number for each component, vectorial dimension
It is identical with the dimension of decision variable X；
S34：According to the pheromone concentration J between individualccExecution is bunched operation：
S35：The health function of flora is calculated, and is carried out descending arrangement, eliminates the small half bacterium of health function value, health
The other half big bacterium of functional value is bred, and careful bacterium ability of looking for food keeps consistent with parent；
To given k, l, the health function of every bacterium isIn formula,Represent i-th thin
The energy of bacterium, J (i, j, k, l) represent that bacterium i tends to after operation kth time replicates operation and disperse for the l time and operate in jth time
Fitness function value, NcRepresent to tend to number,It is bigger, represent that the ability of looking for food of bacterium i is stronger；
S36：The S35 floras produced are merged with the preceding flora for once iterating to calculate generation, calculate this stylish flora individual
The crowding distance of Pareto solutions is simultaneously ranked up according to the Pareto crowding distances solved, and L advantage individual is formed next before selection
For flora；
S37：Disperse：After bacterium experience several generations replicates, to disperse probability PedDispersed the optional position into search space；
S38：Judge to optimize whether algorithm meets termination condition, such as meet, then export Pareto forward positions, that is, optimizing decision variable Xbest
And its corresponding current efficiency ybestWith greenhouse gas emissions zbest, such as it is unsatisfactory for, then jumps to step S33 circulations and perform.
5. the aluminium electroloysis energy-saving and emission-reduction method according to claim 4 based on BP neural network Yu adaptive M BFO algorithms,
It is characterized in that, the crowding distance that Pareto solutions are calculated in step S36 includes the following steps：
A1：Ascending sort is carried out to all optimum individuals in external archive；
A2：Calculating each solves distance of the two adjacent individuals in each optimization aim spatially；
A3：These distances are added up to the crowding distance of required Pareto solutions, and set the crowding distance of Boundary Solutions as infinity；
I-th individual crowding distance be：
In formula,For i-th of individual distance in target j, R is the individual sequence number of crowding distance maximum after ascending order arrangement, and m is
Object space dimension；
When being updated in step S3 to external archive, it is assumed that external archive A maximum capacities are q, and ith iteration calculates what is formed
Non-domination solution is Q, is specifically comprised the following steps：
B1：All individual crowding distances of external archive are calculated, and make descending arrangement；
B2：Update external archive：In ith iteration, if Q ＞ A, A is replaced with Q；
B3：Judge whether external archive A capacity meets the requirement of maximum capacity, if number of individuals n ＜ q in A, replicate in Q to A,
Form new external archive A1；
B4：Judge in new external archive A1 with the presence or absence of the identical individual of desired value, if only retaining one in the presence of if；
B5：If number of individuals n1≤q in new external archive A1, carries out next iteration, if n1 ＞ q, calculate crowding distance
di, and descending arranges, and deletes n1-q minimum individual of crowding distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510750725.7A CN105426959B (en) | 2015-11-06 | 2015-11-06 | Aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510750725.7A CN105426959B (en) | 2015-11-06 | 2015-11-06 | Aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105426959A CN105426959A (en) | 2016-03-23 |
| CN105426959B true CN105426959B (en) | 2018-05-01 |
Family
ID=55505154
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510750725.7A Active CN105426959B (en) | 2015-11-06 | 2015-11-06 | Aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105426959B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107309213B (en) * | 2017-08-16 | 2020-01-17 | 吉林省农业科学院 | Red date ultrasonic cleaner control method based on BP neural network |
| CN107511823B (en) * | 2017-08-29 | 2019-09-27 | 重庆科技学院 | The method of robot manipulating task track optimizing analysis |
| CN110165297A (en) * | 2019-05-31 | 2019-08-23 | 重庆市科学技术研究院 | A kind of procedure parameter selection method of film lithium cell solid electrolyte |
| CN117452828B (en) * | 2023-12-22 | 2024-03-01 | 中电行唐生物质能热电有限公司 | Method and system for controlling emission of harmful gas in boiler based on neural network |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6745169B1 (en) * | 1995-07-27 | 2004-06-01 | Siemens Aktiengesellschaft | Learning process for a neural network |
| CN102184454A (en) * | 2011-05-26 | 2011-09-14 | 浙江迦南科技股份有限公司 | Granulator formula generation method based on neural network system |
| CN104657777A (en) * | 2015-03-16 | 2015-05-27 | 长春工业大学 | Electric wire extrusion molding process parameter optimization method for vehicles based on chaotic wavelet neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8538901B2 (en) * | 2010-02-05 | 2013-09-17 | Toyota Motor Engineering & Manufacturing North America, Inc. | Method for approximation of optimal control for nonlinear discrete time systems |
-
2015
- 2015-11-06 CN CN201510750725.7A patent/CN105426959B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6745169B1 (en) * | 1995-07-27 | 2004-06-01 | Siemens Aktiengesellschaft | Learning process for a neural network |
| CN102184454A (en) * | 2011-05-26 | 2011-09-14 | 浙江迦南科技股份有限公司 | Granulator formula generation method based on neural network system |
| CN104657777A (en) * | 2015-03-16 | 2015-05-27 | 长春工业大学 | Electric wire extrusion molding process parameter optimization method for vehicles based on chaotic wavelet neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105426959A (en) | 2016-03-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105404926B (en) | Aluminum electrolysis production technique optimization method based on BP neural network Yu MBFO algorithms | |
| CN105321000B (en) | Aluminum electrolysis process parameter optimization method based on BP neural network Yu MOBFOA algorithms | |
| CN105447567B (en) | Aluminium electroloysis energy-saving and emission-reduction control method based on BP neural network Yu MPSO algorithms | |
| CN105426959B (en) | Aluminium electroloysis energy-saving and emission-reduction method based on BP neural network Yu adaptive M BFO algorithms | |
| CN109932903A (en) | The air-blower control Multipurpose Optimal Method of more parent optimization networks and genetic algorithm | |
| CN106529818B (en) | Water quality assessment Forecasting Methodology based on Fuzzy Wavelet Network | |
| CN109634121A (en) | More parent genetic algorithm air source heat pump multiobjective optimization control methods based on radial basis function neural network | |
| CN105404142B (en) | Aluminium electroloysis multi parameters control method based on BP neural network Yu MBFO algorithms | |
| CN105302973A (en) | MOEA/D algorithm based aluminum electrolysis production optimization method | |
| CN107203590B (en) | Personalized movie recommendation method based on improved NSGA-II | |
| CN109085752B (en) | Aluminum electrolysis preference multi-objective optimization algorithm based on angle domination relationship | |
| CN108805268A (en) | Deeply learning strategy network training method based on evolution algorithm | |
| CN105608295B (en) | The multi-objective genetic algorithm of coking furnace pressure and RBF neural Optimization Modeling method | |
| Cheng et al. | An elitism strategy based genetic algorithm for streaming pattern discovery in wireless sensor networks | |
| CN108445756B (en) | Aluminum electrolysis energy-saving emission-reduction intelligent control method based on AR domination relationship | |
| CN116307149A (en) | Blast furnace performance optimization method based on attention LSTM and KBNSGA | |
| CN105426960B (en) | Aluminium electroloysis energy-saving and emission-reduction control method based on BP neural network Yu MBFO algorithms | |
| Hu et al. | Hybrid prediction model for the interindustry carbon emissions transfer network based on the grey model and general vector machine | |
| CN105334824A (en) | Aluminum electrolysis production optimization method based on NSGA-II algorithm | |
| CN105302976A (en) | Aluminum electrolysis production optimization method based on SPEA2 algorithm | |
| CN105420760B (en) | Aluminium electroloysis multi-parameters optimization method based on adaptive step bacterial foraging algorithm | |
| CN108734349A (en) | Distributed generation resource addressing constant volume optimization method based on improved adaptive GA-IAGA and system | |
| CN109086469B (en) | Aluminum electrolysis modeling and optimizing method based on recurrent neural network and preference information | |
| Wang et al. | Intelligent access selection in cognitive networks: A fuzzy neural network approach | |
| Hu et al. | A classification surrogate model based evolutionary algorithm for neural network structure learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20160323 Assignee: Guangzhou nuobi Electronic Technology Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2023980052372 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20231220 Application publication date: 20160323 Assignee: Lingteng (Guangzhou) Electronic Technology Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2023980052367 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20231220 Application publication date: 20160323 Assignee: Guangzhou Taipu Intelligent Technology Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2023980052361 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20231220 Application publication date: 20160323 Assignee: GUANGZHOU GUOCHUANG TECHNOLOGY Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2023980052357 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20231220 Application publication date: 20160323 Assignee: GUANGZHOU YIJUN TECHNOLOGY Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2023980052341 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20231220 Application publication date: 20160323 Assignee: GUANGZHOU XINGYIN TECHNOLOGY Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2023980052337 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20231220 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20160323 Assignee: WUZHOU JINZHENGYUAN ELECTRONIC TECH. Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2023980053985 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20231227 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20160323 Assignee: Liaoning Higher Education Technology Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2024980000653 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20240119 Application publication date: 20160323 Assignee: Silk Road Inn (Chongqing) Trading Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2024980000638 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20240119 Application publication date: 20160323 Assignee: Hengdian Wuxia Film and Television (Chongqing) Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2024980000634 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20240119 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20160323 Assignee: Foshan WanChen Technology Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2024980004249 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20240415 Application publication date: 20160323 Assignee: FOSHAN ZHENGRONG TECHNOLOGY Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2024980004248 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20240415 Application publication date: 20160323 Assignee: FOSHAN DOUQI TECHNOLOGY Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2024980004247 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20240415 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20160323 Assignee: Foshan helixing Technology Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2024980004524 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20240419 Application publication date: 20160323 Assignee: Foshan qianshun Technology Co.,Ltd. Assignor: Chongqing University of Science & Technology Contract record no.: X2024980004523 Denomination of invention: Energy saving and emission reduction method for aluminum electrolysis based on BP neural network and adaptive MBFO algorithm Granted publication date: 20180501 License type: Common License Record date: 20240419 |