CN109033489A - Based on horse shoe flame glass furnace efficiency optimization method, the system for improving particle swarm algorithm - Google Patents
Based on horse shoe flame glass furnace efficiency optimization method, the system for improving particle swarm algorithm Download PDFInfo
- Publication number
- CN109033489A CN109033489A CN201810529748.9A CN201810529748A CN109033489A CN 109033489 A CN109033489 A CN 109033489A CN 201810529748 A CN201810529748 A CN 201810529748A CN 109033489 A CN109033489 A CN 109033489A
- Authority
- CN
- China
- Prior art keywords
- particle
- energy consumption
- fitness
- factor
- avg
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P40/00—Technologies relating to the processing of minerals
- Y02P40/50—Glass production, e.g. reusing waste heat during processing or shaping
Abstract
The invention discloses based on horse shoe flame glass furnace efficiency optimization method, the system for improving particle swarm algorithm, step A, acquire the creation data of horse shoe flame glass furnace, the energy consumption model constrained based on clear quality is established, and credit rating mapping is become to the clarification factor of quantization using the energy consumption model constrained based on clear quality；Step B, by convergence factor Δ and Colony fitness variance σ2Particle swarm algorithm is improved to judge the Evolving State of particle group as evaluation index；Improved particle swarm algorithm combination Multiplier Penalty Function is constrained processing method by step C, applied to the energy consumption optimal solution in the energy consumption model under clear quality constraint, being guaranteed under glass smelting quality condition.Quantitative analysis is carried out in the complexity that clarifying area escapes to bubble, the key influence factor of energy consumption and credit rating is analyzed, obtains optimal parameter and combine and then instruct actual production, to reduce energy consumption while achieving the purpose that ensure clear quality.
Description
Technical field
The present invention relates to glass furnace fields, more particularly to based on the horse shoe flame glass furnace efficiency optimization for improving particle swarm algorithm
Method, system.
Background technique
The combustion space of heat accumulating type horse shoe flame glass melter has a flame flow stock to be rotarily formed U-shaped flame and at revolution
The torrid zone is formed, due to the limitation to the length of flame and to the requirement of rotary motive power, the short width of kiln shape is compact-sized,
Cross-sectional view of the structure and flame is gentle flows away to as shown in Figure 1.Glass melting process is generated comprising silicate, and glass is formed, glass
Liquid clarification, five stages such as homogenizing and cooling.Traditional melted technology is based primarily upon artificial experience to adjust related process parameters,
This mode not only low efficiency, and it is difficult to adapt to the diversified demand of glassware.Air blister defect of the glass in melting process
Product quality is seriously affected, brings serious loss to enterprise.
Summary of the invention
It is an object of the invention to propose based on improve particle swarm algorithm horse shoe flame glass furnace efficiency optimization method, be
System obtains optimal parameter and combines and then instruct actual production, to reduce energy consumption while achieving the purpose that ensure clear quality.
To achieve this purpose, the present invention adopts the following technical scheme:
A kind of horse shoe flame glass furnace efficiency optimization method based on improvement particle swarm algorithm, comprising the following steps:
Step A acquires the creation data of horse shoe flame glass furnace, establishes the energy consumption model constrained based on clear quality, and benefit
Credit rating mapping is become to the clarification factor of quantization with the energy consumption model constrained based on clear quality；Step B, by convergence factor
Δ and Colony fitness variance σ2Particle swarm algorithm is changed to judge the Evolving State of particle group as evaluation index
Into；Improved particle swarm algorithm combination Multiplier Penalty Function is constrained processing method by step C, is applied under clear quality constraint
Energy consumption model in, the energy consumption optimal solution that is guaranteed under glass smelting quality condition.
It is preferably based on the energy consumption model establishment process of clear quality constraint are as follows:
Step A1 acquires the creation data of horse shoe flame glass furnace, obtains the tank furnace thermal efficiency using counterbalancing analysis method
Model:
Wherein,TarchFor arch top temperature
Degree, α are coefficient of excess air, and x is fuel flow rate, and f is radiation position orifice area, and Φ is radiation position aperture of door coefficient, CtFor
Position specific heat is radiated,For the low heat value of fuel, tfireFor the temperature of fuel preheating, CfireFor by fuel preheating to tfire
When avergae specific heat, taFor the preheating temperature of combustion air, CaT is preheating to for combustion airaWhen specific heat capacity, tsmokeFor compression
Preheating of air temperature, VsmokeFor the volume of compressed air, CsmokeT is preheating to for compressed airsmokeWhen specific heat capacity, heject
For enthalpy of discharging fume, λ is thermal coefficient, and δ is brick material thickness, and F is heat dissipation area, and L is combustion-supporting air quantity, hkFor cooling air enthalpy
Value；
Step A2, by the residence time t of production streamglassTime t required for being escaped with bubblebubbleRatio as clarification
Factor R F, then
Wherein,ρlFor the density of glass melt, ρbFor gas
The density of gas in bubble, μ are the coefficient of viscosity of bubble, and r is bubble radius, labFor the height of settling section, lbcFor clarifying area
Length, b are the width of tank furnace, and n is the amount of the substance of gas,It is t for temperature0When combustion product specific heat capacity；
Step A3, by tank furnace thermal efficiency model η (Tarch, α, x) and factor R F is clarified as objective function, to obtain base
It is in the energy consumption model of clear quality constraint
Preferably, credit rating mapping is become to the clarification factor of quantization are as follows:
Credit rating is divided into A, B, C, D and E a total of five grade by step A4, and each grade has maximum to bubble straight
The technique requirement of diameter: grade A is to be not allow for bubble, and grade B is that largest air bubbles diameter is not more than 0.1mm, and grade C is most atmosphere
Bulb diameter is not more than 0.2mm, and grade D is that largest air bubbles diameter is not more than 0.5mm, and grade E is that largest air bubbles diameter is not more than
1.0mm；
Step A5, if temperature gradientFor definite value k, then pass through the pass between bubble radius r and clarification factor R F
Credit rating mapping is become the clarification factor of quantization, obtained by system:
The credit rating of step A5 is increased to the energy consumption model based on clear quality constraint by step A6, it may be assumed that
Preferably, particle swarm algorithm improves specifically:
Step B1, if the number of particles of population is z, in the y times iteration, fiFor the fitness of i-th of particle, fmTable
Show the fitness of optimal particle, favgFor average fitness, and fitness in particle is enabled to be greater than favgFitness average value be
favg', then using the difference of individual adaptive optimal control degree and population average fitness as convergence factor Δ: Δ=| fm-favg' |,Convergence factor Δ is used to evaluate the degree of convergence of population: convergence factor Δ is smaller, then the particle
More tend to Premature Convergence；
Step B2 passes through Colony fitness variance σ2Degree of scatter to particle in population is evaluated:
Colony fitness variance σ2Smaller, particle more disperses, on the contrary then particle
More concentrate；
Step B3, by convergence factor Δ and Colony fitness variance σ2As evaluation index, by the Evolving State of population into
Row is sorted out, and inertia weight value ω is dynamically modified according to different Evolving States:
As the f of particle groupi> favg' or favg< fi< f 'avgWhen, set inertia weight value ω=ω-(ω-
ωmin)·e-γ, wherein ωminBeing set as 0.5, γ is evolutionary generation；
As the f of particle groupi< favgAnd σ2When greater than preset range value, inertia weight value is set:
As the f of particle groupi< favgAnd σ2When less than preset range value, inertia weight value is set:
Preferably, the acquisition of the energy consumption optimal solution specifically:
Energy consumption model under clear quality constraint is set as d dimension target search space, ties up target search in d by step C1
There is a population in spaceRepresent m particle, in which:
I=1,2 ..., m are the vector point that i-th of particle ties up solution space in d,I=1,2 ..., m are the search speed of i-th of particle, and i-th of particle search arrives optimal
Position is pi=[pi1, pi2..., pim], i=1,2 ..., m, the optimal location that collective search arrives are pg=[pi1, pi2...,
pim],
vi(s+1)=ω vi(s)+c1ε1(pi-xi(s))+c2ε2(pi-xi(s)),
xi(s+1)=xi(s)+vi(s), ε1, ε2Random number between 0 to 1, S are the number of iterations, Studying factorsStudying factors
Step C2, in solution space in random initializtion group particle position and speed, setting current location be piIf
Setting optimum position in initial population is pg；
Step C3, according to the speed formula v in step C1i(s+1) and location formula xi(s+1), change particle speed and
Location status；
Penalty function and Lagrangian are combined construction Multiplier Penalty Function by step C4:
Wherein θ is decision variable, f
(θ) is objective function, hj(θ)=0, j=q+1 ..., m is equality constraint；
The fitness of particle is calculated according to Multiplier Penalty Function；
Step C5 carries out the adaptive adjustment of inertia weight value ω according to current Evolving State, and by current adaptive value
Compared with itself optimal solution: if current solution is better than itself optimal value, it is optimal that current adaptive value being just set as itself
Solution；
Step C6, by the fitness of the particle and history optimal location pi=[pi1, pi2..., pim] fitness compare
Relatively decide whether to update current globally optimal solution；
Step C7, whether detection is evolved reaches maximum set value or reaches preset convergence precision, if meeting the requirements,
The evolutionary process terminates, the energy consumption optimal solution being guaranteed under glass smelting quality condition；Otherwise C4 is returned to step.
It is preferably, a kind of based on the horse shoe flame glass furnace efficiency optimization system for improving particle swarm algorithm, comprising:
Energy consumption modeling module establishes the energy constrained based on clear quality for acquiring the creation data of horse shoe flame glass furnace
Model is consumed, and credit rating mapping is become to the clarification factor of quantization using the energy consumption model constrained based on clear quality；Algorithm
Optimization module is used for convergence factor Δ and Colony fitness variance σ2As evaluation index, to judge the evolution of particle group
State improves particle swarm algorithm；With energy optimization module, for improved particle swarm algorithm combination multiplier to be penalized letter
Number constraint processing method is guaranteed under glass smelting quality condition applied in the energy consumption model under clear quality constraint
Energy consumption optimal solution.
The horse shoe flame glass furnace efficiency optimization method based on improvement particle swarm algorithm is guaranteeing clarification process quality
On the basis of study melting technology energy consumption, analyzed by the thermal balance to melting furnace structure, and by analysis clarifying area in
The glass liquid stream characteristics of motion carries out quantitative analysis in the complexity that clarifying area escapes to bubble, establishes clear quality
Energy optimization model under constraint, analyzes the key influence factor of energy consumption and credit rating, and with improved adaptive grain
Swarm optimization obtains optimal parameter and combines and then instruct actual production to the object module optimizing, so that reaching ensures to clarify matter
The purpose of energy consumption is reduced while amount.
Detailed description of the invention
The present invention will be further described for attached drawing, but the content in attached drawing does not constitute any limitation of the invention.
Fig. 1 is the heat accumulating type horse shoe flame glass melter structure chart of the one of embodiment of the present invention
Fig. 2 is the glass melter effluogram of the one of embodiment of the present invention.
Specific embodiment
To further illustrate the technical scheme of the present invention below with reference to the accompanying drawings and specific embodiments.
Embodiment one
The horse shoe flame glass furnace efficiency optimization method based on improvement particle swarm algorithm of the present embodiment, comprising the following steps:
Step A acquires the creation data of horse shoe flame glass furnace, establishes the energy consumption model constrained based on clear quality, and using based on clear
Credit rating mapping is become the clarification factor of quantization by the energy consumption model of clear quality constraint；
Step B, by convergence factor Δ and Colony fitness variance σ2As evaluation index, to judge the evolution of particle group
State improves particle swarm algorithm；Improved particle swarm algorithm combination Multiplier Penalty Function is constrained processing side by step C
Method, applied to the energy consumption optimal solution in the energy consumption model under clear quality constraint, being guaranteed under glass smelting quality condition.
The horse shoe flame glass furnace efficiency optimization method based on improvement particle swarm algorithm is guaranteeing clarification process quality
On the basis of study melting technology energy consumption, analyzed by the thermal balance to melting furnace structure, and by analysis clarifying area in
The glass liquid stream characteristics of motion carries out quantitative analysis in the complexity that clarifying area escapes to bubble, establishes clear quality
Energy optimization model under constraint, analyzes the key influence factor of energy consumption and credit rating, and with improved adaptive grain
Swarm optimization obtains optimal parameter and combines and then instruct actual production to the object module optimizing, so that reaching ensures to clarify matter
The purpose of energy consumption is reduced while amount.
Air blister defect is that there are visible air inclusions mass defects in vitreum, in batch melting and glass metal shape
A large amount of CO are had during2And H2The gases such as O, until glass shaping process, still some is not completely from glass metal
In escape completely, therefore formed bubble quality defect, reduce the credit rating of glass product, and affect its mechanical strength.
Bubble quality defect is the major defect of tank furnace working region, and in the ideal situation, clarification stage is not because complete
Discharge and caused by air blister defect, just at the key factor of influence current glass liquid quality problems.As shown in Fig. 2, glass metal
Circulation is formed in melting end, the powerful reflux of the flow direction cooperation bed of material can effectively stop the dross of molten surface in circulation, and kiln
The setting of bank forces production stream to be flipped up, and enables liquid stream to reach glass metal surface, and then obtain high temperature in hot spot region
Clarification, is layered on rear side of weir after clarification and turns back, and enters work department by dog-hole in bottom, high-quality so as to obtain
The glass metal of amount.The main function of settling section is exactly to eliminate air blister defect, and the superiority and inferiority of clarifying process is directly related to glass liquid stream
Quality problems.Minute bubbles in glass liquid stream, which readily diffuse into, to be allowed to diameter in the biggish bubble of volume and becomes larger, and diameter is bigger
Its buoyance lift speed is faster, and time of liberation is fewer.When bubble diameter is less than 10 μm, the effect of surface tension can be such that microbubble dissolves
In glass liquid stream.
Assuming that the gas inside bubble is considered as perfect condition, and internal component and temperature are uniformly distributed；Bubble shape is approximate
Think spherical；It is not chemically reacted between the gas of each component in bubble；And the bubble in glass melt does not influence glass
The heat transfer of liquid stream；Then bubble radius is the key factor for restricting speed in bubble uphill process, and bubble radius can be with temperature
Change and certain variation occurs, therefore it is extremely important to the relationship between bubble radius and temperature.
It is preferably based on the energy consumption model establishment process of clear quality constraint are as follows:
Step A1 acquires the creation data of horse shoe flame glass furnace, obtains the tank furnace thermal efficiency using counterbalancing analysis method
Model:
Wherein,TarchFor arch top temperature
Degree, α is coefficient of excess air, and x is fuel flow rate, units/kg/h；F is radiation position orifice area, unit m2；Φ is irradiation unit
Position aperture of door coefficient, CtTo radiate position specific heat, unit kj/ (m3·℃)；For the low heat value of fuel, unit kj/kg；
tfireFor the temperature of fuel preheating, unit DEG C, generally 115 DEG C；CfireFor by fuel preheating to tfireWhen avergae specific heat, it is single
Position kj/ (m3·℃)；taFor the preheating temperature of combustion air, unit DEG C；CaT is preheating to for combustion airaWhen specific heat capacity, it is single
Position kj/ (kg DEG C)；tsmokeFor the preheating temperature of compressed air, unit DEG C；VsmokeFor the volume of compressed air, unit m3；
CsmokeT is preheating to for compressed airsmokeWhen specific heat capacity, unit kj/ (m3·℃)；hejectFor enthalpy of discharging fume, unit kJ/kg；
λ is thermal coefficient, and δ is brick material thickness, unit m；F is heat dissipation area, unit m2；R is gas constant；L is combustion-supporting air quantity, single
Position m3；hkFor cooling air enthalpy, unit kJ/kg；
Step A2, by the residence time t of production streamglassTime t required for being escaped with bubblebubbleRatio as clarification
Factor R F, then
RF value is bigger, and time of the production stream in clarifying area is longer, and the time of bubble evolution metal level is shorter, clarification
Effect is better；
Wherein,ρlFor the density of glass melt, unit
kg/m3；ρbFor the density of gas in bubble, units/kg/m3；μ is the coefficient of viscosity of bubble, and r is bubble radius, unit m；labFor
The height of settling section, unit m；lbcFor the length of clarifying area, unit m；B is the width of tank furnace, unit m；N is the substance of gas
Amount, unit mol；It is t for temperature0When combustion product specific heat capacity, unit kj/ (kg DEG C)；
Step A3, by tank furnace thermal efficiency model η (Tarch, α, x) and factor R F is clarified as objective function, to obtain base
It is in the energy consumption model of clear quality constraint
Origin of heat is in the heat that by ingredient and cullet and atomizing medium, combustion air is brought into tank furnace.Glass
Heat required for chemically reacting in melting process is the heat effectively paid, and is specifically heated to fusing required for glass metal
Degassing product is heated to heat consumption required for fusion temperature by heat consumption when temperature, generates the heat consumption of silicate, is formed
The heat consumption of glass melts, the heat consumption of evaporation water, others expenditure heat have smoke exhaust heat, and kiln body radiates, heat loss through radiation,
Overflow heat dissipation etc., these are the heat effectively paid, as shown in table 1.
Table 1
Radiating position includes the positions such as feed opening, nozzle brick hole, thermometer hole, drum, work department, but because charging
Mouthful, nozzle brick hole, thermometer hole etc. is lower by the space temperature at radiation, and it is lower to radiate position aperture of door coefficient Φ, therefore only
Only consider the thermal loss of radiation direction glass metal work department.Counter balancing method can be relatively easy to determine the size of various heat loss,
And then understand the operating condition of tank furnace, therefore solve the thermal efficiency with counterbalancing analysis method:
Wherein:q2=xCfiretfire, q3=xLCata, q4=xVsmokeCsmoketsmoke,
According to tank furnace thermal efficiency model it is found that influencing more than the factor of the tank furnace thermal efficiency and complicated, principal element are as follows:
Coping temperature Tarch, characterization parameter of the coping temperature usually as glass metal actual temperature in glass melting process
It is controlled.Melting rate can be improved to a certain extent by improving coping temperature, but as the rising of coping temperature can aggravate kiln
The radiation loss of body, it is necessary to the better fire resisting materials for wall of thermal insulation property and thermal insulation material, to increase cost.
Coefficient of excess air α is the ratio of air capacity required for the air capacity and theory of actual consumption, passes through flue gas point
Analyzer analyzes CO, O in flue2、CO2Content calculate:ω(O2) be flue gas in oxygen percentage
Than.Will increase if coefficient of excess air α is excessive flue gas discharge thermal losses, otherwise it is too small just will increase it is unburnt
Thermal loss.Inappropriate coefficient of excess air α also affects the pressure and temperature inside kiln.Therefore worked in melting furnaces
It has to adjust coefficient of excess air α in optimum range in journey.
Fuel flow rate x needs to consume a large amount of energy during furnace equipment work, and fuel is direct as energy supply source
Affect the thermal efficiency of kiln.And in order to which the operational efficiency for effectively promoting kiln also usually needs root in kiln actual motion
The ratio between fuel and combustion air is controlled according to actual production situation.
Preferably, credit rating mapping is become to the clarification factor of quantization are as follows:
Credit rating is divided into A, B, C, D and E a total of five grade by step A4, and each grade has maximum to bubble straight
The technique requirement of diameter: grade A is to be not allow for bubble, and grade B is that largest air bubbles diameter is not more than 0.1mm, and grade C is most atmosphere
Bulb diameter is not more than 0.2mm, and grade D is that largest air bubbles diameter is not more than 0.5mm, and grade E is that largest air bubbles diameter is not more than
1.0mm；
Step A5, if temperature gradientFor definite value k, then pass through the pass between bubble radius r and clarification factor R F
Credit rating mapping is become the clarification factor of quantization, obtained by system:
The credit rating of step A5 is increased to the energy consumption model based on clear quality constraint by step A6, it may be assumed that
The diameter of largest air bubbles is the important indicator of glass quality grade classification, and the present embodiment is using " mountain " shape temperature
Curve, the advantage of the temperature schedule are that hot spot is prominent, and foam line is steady and audible.Partial region before hotspot range is because of cooperation
Material cover glass liquid surface cause heat transfer is obstructed, so temperature rise it is relatively slow, with batch area of coverage distance
Increase, heat absorption is consequently increased, thus temperature rise it is relatively fast.But the temperature schedule is difficult to give full play to the latent of melting furnaces
Power needs to take steps to improve prefusing temperature, with this to reinforce the melting capacity of batch in the region of batch covering
Simultaneously also it has to be ensured that the temperature gradient in the clarifying and homogenizing area after batch area meets the manufacturing technique requirent of clarifying and homogenizing,
Therefore it is required that the temperature gradient of clarifying area meets technique requirement, investigation is it is found that by temperature gradient ideal when formulating technological standards
Turn to definite value.
Preferably, particle swarm algorithm improves specifically:
Step B1, if the number of particles of population is z, in the y times iteration, fiFor the fitness of i-th of particle, fmTable
Show the fitness of optimal particle, favgFor average fitness, and fitness in particle is enabled to be greater than favgFitness average value be
favg', then using the difference of individual adaptive optimal control degree and population average fitness as convergence factor Δ: Δ=| fm-favg' |,Convergence factor Δ is used to evaluate the degree of convergence of population: convergence factor Δ is smaller, then the particle
More tend to Premature Convergence；
Step B2 passes through Colony fitness variance σ2Degree of scatter to particle in population is evaluated:
Colony fitness variance σ2Smaller, particle more disperses, on the contrary then particle
More concentrate；
Step B3, by convergence factor Δ and Colony fitness variance σ2As evaluation index, by the Evolving State of population into
Row is sorted out, and inertia weight value ω is dynamically modified according to different Evolving States:
As the f of particle groupi> favg' or favg< fi< f 'avgWhen, set inertia weight value ω=ω-(ω-
ωmin)·e-γ, wherein ωminBeing set as 0.5, γ is evolutionary generation；
As the f of particle groupi< favgAnd σ2When greater than preset range value, inertia weight value is set:
As the f of particle groupi< favgAnd σ2When less than preset range value, inertia weight value is set:
Particle swarm algorithm uses " speed-displacement " evolution Model, and easy to operate, parameter is few, and memory function can dynamic
Track-while-scan situation is a kind of very efficient parallel search optimization algorithm.But the algorithm has some limitations: grain
Subgroup evolve it is several instead of after, may all particles be all gathered in a local optimum position, the phenomenon that population assembles is exactly
Premature Convergence, therefore how to allow particle swarm algorithm that can escape in time local optimum during scanning for solution space to be
There is currently the problem of.
Adaptive impovement is carried out to particle swarm algorithm, needs population according to different evolution in the different phase of evolution
State independently changes inertia weight value ω, improves Evolution of Population efficiency by this adaptive inertia weight mode, and then obtain
The globally optimal solution of the group.The solution space of particle swarm algorithm is the value range of the solution of actual optimization problem, and particle is more
Trivial solution may be become beyond the bounds set when itself new speed and position.And the adaptation of these trivial solutions
Value calculation times increase the workload of algorithm, it is therefore desirable to carry out BORDER PROCESSING to the particle beyond boundary, guarantee that it is returned to
Effective search space, continues searching.
Wherein [xMin, j, xMax, j] it is the range of definition that jth ties up particle.On
The BORDER PROCESSING stated indicates that the particle after updating reaches a new position, if detecting that the position has exceeded boundary,
It moves to particle on boundary, then updates its speed again.
The prematurity convergence degree for reasonably judging out the algorithm is most important for the rational modification of algorithm.Convergence factor Δ
The similarity degree between the maximum individual of fitness in population is reflected, the possible some bad shadows of poor particle have been evaded
It rings, extremely accurate describes population at individual Premature Convergence degree.But convergence factor Δ is not distinguished population and is currently at point
Dead state or local optimum state are dissipated, so there are certain drawbacks when judging Evolution of Population state,
Therefore Colony fitness variance σ is introduced2, the degree of scatter of particle in population is evaluated, and comprehensive group's fitness variances sigma2
Rational evaluation is carried out to Evolution of Population state in the case where convergence factor Δ.Inertia weight value ω for adjust it is global explore and
Local optimal searching ability.Biggish inertia weight value ω accelerates the global developing ability of each particle in population, accelerates the receipts of algorithm
Speed is held back, lesser inertia weight value ω is more conducive to the local exploring ability of each particle, can increase the uniformity coefficient of solution.
Therefore it needs dynamically to modify inertia weight value ω according to different Evolving States, guarantees to remain in any period particle of evolution
Enough there is certain search capability, so that it is guaranteed that particle effectively jumps out local optimum.
As the f of particle groupi> favg' when, it indicates that the particle is particle more outstanding in group, is approaching the overall situation
Optimal or favg< fi< favg' when, indicate that these particles have good local optimal searching and global exploring ability in group,
State should continue to keep；Therefore lesser inertia weight value ω is set, local optimal searching ability is reinforced.
As the f of particle groupi< favgAnd σ2When greater than preset range value, indicate that algorithm stays cool, particle distribution
It is more loose, it needs suitably to reduce inertia weight value ω at this time, reinforces local optimal searching ability；When population is close to local optimum position
When setting, the update of speed is mainly ω vi(s) it determines, but the inertia weight value ω of usually population is both less than 1, so
The speed of particle is smaller and smaller, this just considerably increases the probability of Premature Convergence.Therefore inertia weight value ω is selected from 1.2 herein
It begins to decline.
As the f of particle groupi< favgAnd σ2When less than preset range value, indicate that algorithm is in local convergence state, grain
Son distribution is more concentrated, and should suitably be increased inertia weight value ω at this time, be increased the global investigation ability of particle, to jump as early as possible
Local optimum state out.
Preferably, the acquisition of the energy consumption optimal solution specifically:
Energy consumption model under clear quality constraint is set as d dimension target search space, ties up target search in d by step C1
There is a population in spaceRepresent m particle, in which:
I=1,2 ..., m are the vector point that i-th of particle ties up solution space in d,I=1,2 ..., m are the search speed of i-th of particle, and i-th of particle search arrives optimal
Position is pi=[pi1, pi2..., pim], i=1,2 ..., m, the optimal location that collective search arrives are pg=[pi1, pi2...,
pim],
vi(s+1)=ω vi(s)+c1ε1(pi-xi(s))+c2ε2(pi-xi(s)),
xi(s+1)=xi(s)+vi(s), ε1, ε2Random number between 0 to 1, S are the number of iterations, Studying factorsStudying factors
Step C2, in solution space in random initializtion group particle position and speed, solution space asks for actual optimization
The value range of the solution of topic, setting current location are pi, it is p that optimum position in initial population, which is arranged,g；
Step C3, according to the speed formula v in step C1i(s+1) and location formula xi(s+1), change particle speed and
Location status；
Penalty function and Lagrangian are combined construction Multiplier Penalty Function by step C4:
Wherein θ is decision variable, f
(θ) is objective function, hj(θ)=0, j=q+1 ..., m is equality constraint, vjLagrange to use in iteration j multiplies
Son；
The fitness of particle is calculated according to Multiplier Penalty Function；
Step C5 carries out the adaptive adjustment of inertia weight value ω according to current Evolving State, and by current adaptive value
Compared with itself optimal solution: if current solution is better than itself optimal value, it is optimal that current adaptive value being just set as itself
Solution；
Step C6, by the fitness of the particle and history optimal location pi=[pi1, pi2..., pim] fitness compare
Relatively decide whether to update current globally optimal solution；
Step C7, whether detection is evolved reaches maximum set value or reaches preset convergence precision, if meeting the requirements,
The evolutionary process terminates, the energy consumption optimal solution being guaranteed under glass smelting quality condition；Otherwise C4 is returned to step.It will multiply
Sub- penalty function method combined with improved adaptive particle swarm optimization algorithm to clear quality constraint under energy consumption model into
Row solves, and the thermal efficiency optimal value for meeting clarification factor value range is obtained, to provide theoretical direction to actual production.
Constrained optimization problem will guarantee the balance between objective function and constraint in searching process, if optimization process mistake
Divide the condition limitation that suffers restraints, it would be possible that missing objective function fitness best region；If excessively pursuing the quality of solution, just
Feasible zone may be surmounted.Therefore the key of processing constrained optimization problem is processing constraint condition, and currently processed constraint condition is most
Extensive method is penalty function.Penalty function method is based on sequence without constrained minimization principle, according to the concrete property of constraint condition,
Suitable penalty is constructed, and increases this penalty item in objective function, so that restricted problem is converted into
Unconstrained optimization problem.
The description of penalty are as follows:H (y) is penalty function power
Degree, y is algorithm current iteration number, and H (θ) is penalty factor.Consider optimal solution in constrained optimization obtained in boundary it is general
Rate is bigger, therefore using boundary value as reference items, and fully ensures that and construct Multiplier Penalty Function on the basis of constraint condition.
The possibility solution of particle expression objective function in particle swarm algorithm, and these particles iterative search in solution space, should
Search speed can dynamically be adjusted according to itself experience and social groups' experience.Studying factors c1For experience, study because
Sub- c1The search in itself field can be guided when larger, it is therefore desirable to set Studying factors c1Larger, the later stage of evolution at the initial stage of evolution
It is smaller；Studying factors c2Group's experience of expression, Studying factors c2Particle can be guided to carry out global search when larger, because
This needs to set Studying factors c2Evolution initial stage is smaller, and later stage of evolution is larger.
Embodiment two
The present embodiment is in order to verify the validity of particle swarm algorithm after a kind of improvement of embodiment, using single constraint and multiple constraint
Function carries out test and comparison respectively.
The mono- constraint test function of Goldstein-Price is
F (x)=[1+ (x1+x2+1)2(19-14x1+3x1 2-14x2+6x1x2+3x2 2)]
×[30+(2x1-3x2)2(18-32x1+12x1 2+48x2-36x1x2+27x2 2)]
-2≤x1, x2≤2
The known test function optimal solution is 3.0, and is not at boundary position.Test selects 50 particles to do 15 examinations
It tests, setting maximum number of iterations is 300.Experimental result is as shown in table 2, as can be known from Table 2 using conventional particle group algorithm 77
Optimal solution is obtained after generation, and particle swarm algorithm in 63 generations obtains optimal solution after improving, particle swarm algorithm is restraining after improving
It is promoted in performance.
Table 2
High-dimensional multiple constraint function is
F (x)=(x1-10)2+5(x2-12)2+x3 4+3(x4-11)2+10x5 2
+7x6 2+x7 2-4x6x7-10x6-8x7
s.t.-127+2x1 2+3x2 4+x3+4x4 2+5x5≤0
-282+7x1+3x2+10x3 2+x4-x5≤0
-196+23x1+x2 2+6x6 2-8x7≤ 0, -10≤xi≤ 10, i=1 ..., 7
4x1 2+x2 2-3x1x2+2x3 2+5x6-11x7≤0
Known high-dimensional multiple constraint Function Optimization solution is 680.630057, and is not at boundary position.Test selects 50
Particle does 15 tests, and setting maximum number of iterations is 300.Experimental result is as shown in table 3, as can be known from Table 3 using traditional grain
Swarm optimization obtains optimal solution after 82 generations, and particle swarm algorithm obtains optimal solution, particle swarm algorithm warp after 66 generations after improving
It is promoted on constringency performance after crossing improvement.
Table 3
Embodiment three
The present embodiment solves the energy consumption model based on clear quality constraint in embodiment one using Matlab platform,
Obtain optimal solution.It needs to do homework to simulated environment and parameter before solving, steps are as follows:
(1) experimental situation: Intel (R) Core (TM) i7-5500U CPU, 2.40GHZ, 12GB, MATLAB2017a is set
Environment in carry out l-G simulation test.
(2) coping temperature Tarch, coefficient of excess air α and fuel flow rate x will clarify factor R F as variable to be optimized
Credit rating be chosen to be grade B, and willM file is write as objective function.
(3) it is configured by SMOPSO1 function to based on the energy consumption model that clear quality constrains, improved grain is set
The relevant parameter and return value of swarm optimization are saved in argument structure body.The configuration of its relevant parameter is as follows: population invariable number 150,
Maximum number of iterations is 500, ωmin=0.3, ωmax=0.9.
(4) it according to the number and value range of setting variable, for objective function and argument structure body, and calls
SMOPSO2 function is to its optimizing, to obtain optimal solution, as shown in table 4.
Table 4
When selected credit rating is B, the value range of RF all meets 3.12 < RF < 3.55, is provided by table 4
Optimum results it is found that the result meets credit rating standard, and by with collection in worksite to data compare and analyze, it is known that
Improved particle swarm algorithm thermal efficiency in the case where guaranteeing product quality is significantly improved.After demonstrating improvement simultaneously
Algorithm it is highly effective.With should credit rating select other grades when it is equally applicable.
Example IV
The horse shoe flame glass furnace efficiency optimization system based on improvement particle swarm algorithm of the present embodiment, comprising:
Energy consumption modeling module establishes the energy constrained based on clear quality for acquiring the creation data of horse shoe flame glass furnace
Model is consumed, and credit rating mapping is become to the clarification factor of quantization using the energy consumption model constrained based on clear quality；
Algorithm optimization module is used for convergence factor Δ and Colony fitness variance σ2As evaluation index, to judge grain
The Evolving State of sub-group, improves particle swarm algorithm；
It is answered with energy optimization module for improved particle swarm algorithm combination Multiplier Penalty Function to be constrained processing method
Energy consumption optimal solution for being guaranteed under glass smelting quality condition in the energy consumption model under clear quality constraint.
Preferably, the energy consumption modeling module includes:
Submodule A1 obtains tank furnace using counterbalancing analysis method for acquiring the creation data of horse shoe flame glass furnace
Thermal efficiency model:
Wherein,TarchFor arch top temperature
Degree, α is coefficient of excess air, and x is fuel flow rate, units/kg/h；F is radiation position orifice area, unit m2；Φ is irradiation unit
Position aperture of door coefficient, CtTo radiate position specific heat, unit kj/ (m3·℃)；For the low heat value of fuel, unit kj/kg；
tfireFor the temperature of fuel preheating, unit DEG C, generally 115 DEG C；CfireFor by fuel preheating to tfireWhen avergae specific heat, it is single
Position kj/ (m3·℃)；taFor the preheating temperature of combustion air, unit DEG C；CaT is preheating to for combustion airaWhen specific heat capacity, it is single
Position kj/ (kg DEG C)；tsmokeFor the preheating temperature of compressed air, unit DEG C；VsmokeFor the volume of compressed air, unit m3；
CsmokeT is preheating to for compressed airsmokeWhen specific heat capacity, unit kj/ (m3·℃)；hejectFor enthalpy of discharging fume, unit kJ/kg；
λ is thermal coefficient, and δ is brick material thickness, unit m；F is heat dissipation area, unit m2；L is combustion-supporting air quantity, unit m3；hkFor cooling
Air enthalpy, unit kJ/kg；
Submodule A2, for the residence time t of stream will to be producedglassTime t required for being escaped with bubblebubbleRatio make
To clarify factor R F, then
Wherein,ρlFor the density of glass melt, unit
kg/m3；ρbFor the density of gas in bubble, units/kg/m3；μ is the coefficient of viscosity of bubble, and r is bubble radius, unit m；labFor
The height of settling section, unit m；lbcFor the length of clarifying area, unit m；B is the width of tank furnace, unit m；N is the substance of gas
Amount, unit mol；It is t for temperature0When combustion product specific heat capacity, unit kj/ (kg DEG C)；
With submodule A3, it is used for tank furnace thermal efficiency model η (Tarch, α, x) and clarification factor R F as objective function, from
And obtain based on clear quality constrain energy consumption model be
Preferably, the energy consumption modeling module further include:
Submodule A4, for credit rating to be divided into A, B, C, D and E a total of five grade, each grade has to bubble
The technique requirement of maximum gauge: grade A is to be not allow for bubble, and grade B is that largest air bubbles diameter is not more than 0.1mm, and grade C is
Largest air bubbles diameter be not more than 0.2mm, grade D be largest air bubbles diameter be not more than 0.5mm, grade E be largest air bubbles diameter not
Greater than 1.0mm；
With submodule A5, for setting temperature gradientFor definite value k, then pass through bubble radius r and clarification factor R F
Between relationship, by credit rating mapping become quantization the clarification factor, obtain:
Submodule A6, for the credit rating of submodule A5 to be increased to the energy consumption model based on clear quality constraint, it may be assumed that
Preferably, the algorithm optimization module includes:
Submodule B1, for setting the number of particles of population as z, in the y times iteration, fiFor the adaptation of i-th of particle
Degree, fmIndicate the fitness of optimal particle, favgFor average fitness, and fitness in particle is enabled to be greater than favgFitness it is average
Value is favg', then using the difference of individual adaptive optimal control degree and population average fitness as convergence factor Δ: Δ=| fm-favg' |,Convergence factor Δ is used to evaluate the degree of convergence of population: convergence factor Δ is smaller, then the particle is got over
Tend to Premature Convergence；
Submodule B2, for passing through Colony fitness variance σ2Degree of scatter to particle in population is evaluated:
Colony fitness variance σ2Smaller, particle more disperses, on the contrary then particle
More concentrate；
With submodule B3, it is used for convergence factor Δ and Colony fitness variance σ2As evaluation index, by population into
Change state is sorted out, and dynamically modifies inertia weight value according to different Evolving States:
As the f of particle groupi> favg' or favg< fi< f 'avgWhen, set inertia weight value ω=ω-(ω-
ωmin)·e-γ, wherein ωminBeing set as 0.5, γ is evolutionary generation；
As the f of particle groupi< favgAnd σ2When greater than preset range value, inertia weight value is set:
As the f of particle groupi< favgAnd σ2When less than preset range value, inertia weight value is set:
Preferably, the energy optimization module includes:
Submodule C1 ties up mesh in d for the energy consumption model under clear quality constraint to be set as d dimension target search space
Mark has a population in search spaceRepresent m particle, in which:
I=1,2 ..., m are the vector point that i-th of particle ties up solution space in d,I=1,2 ..., m are the search speed of i-th of particle, and i-th of particle search arrives optimal
Position is pi=[pi1, pi2..., pim], i=1,2 ..., m, the optimal location that collective search arrives are pg=[pi1, pi2...,
pim],
vi(s+1)=ω vi(s)+c1ε1(pi-xi(s))+c2ε2(pi-xi(s)),
xi(s+1)=xi(s)+vi(s), ε1, ε2Random number between 0 to 1, S are the number of iterations, Studying factorsStudying factors
Submodule C2, in solution space in random initializtion group particle position and speed, be arranged current location
For pi, it is p that optimum position in initial population, which is arranged,g；
Submodule C3, for according to the speed formula v in step C1i(s+1) and location formula xi(s+1), change particle
Speed and location status；
Submodule C4, for penalty function and Lagrangian to be combined construction Multiplier Penalty Function:
Wherein θ is decision variable, f
(θ) is objective function, hj(θ)=0, j=q+1 ..., m is equality constraint；
The fitness of particle is calculated according to Multiplier Penalty Function；
Submodule C5, for carrying out the adaptive adjustment of inertia weight value ω according to current Evolving State, and will be current
Adaptive value is compared with itself optimal solution: if current solution is better than itself optimal value, being just set as current adaptive value certainly
Body optimal solution；
Submodule C6, for by the fitness of the particle and history optimal location pi=[pi1, pi2..., pim] adaptation
Degree decides whether to update current globally optimal solution compared to relatively；
Submodule C7 evolves for detection and whether reaches maximum set value or reach preset convergence precision, if meeting
It is required that then the evolutionary process terminates, the energy consumption optimal solution being guaranteed under glass smelting quality condition；Otherwise implementation sub-module
C4。
The technical principle of the invention is described above in combination with a specific embodiment.These descriptions are intended merely to explain of the invention
Principle, and shall not be construed in any way as a limitation of the scope of protection of the invention.Based on the explanation herein, the technology of this field
Personnel can associate with other specific embodiments of the invention without creative labor, these modes are fallen within
Within protection scope of the present invention.
Claims (10)
1. a kind of based on the horse shoe flame glass furnace efficiency optimization method for improving particle swarm algorithm, which is characterized in that including following step
It is rapid:
Step A acquires the creation data of horse shoe flame glass furnace, establishes the energy consumption model constrained based on clear quality, and utilize base
Credit rating mapping is become into the clarification factor quantified in the energy consumption model of clear quality constraint；
Step B, by convergence factor Δ and Colony fitness variance σ2As evaluation index, to judge the Evolving State of particle group,
Particle swarm algorithm is improved；
Improved particle swarm algorithm combination Multiplier Penalty Function is constrained processing method by step C, is applied under clear quality constraint
Energy consumption model in, the energy consumption optimal solution that is guaranteed under glass smelting quality condition.
2. according to claim 1 based on the horse shoe flame glass furnace efficiency optimization method for improving particle swarm algorithm, feature
It is, the energy consumption model establishment process based on clear quality constraint are as follows:
Step A1 acquires the creation data of horse shoe flame glass furnace, obtains tank furnace thermal efficiency model using counterbalancing analysis method:
Wherein,TarchFor coping temperature, α
For coefficient of excess air, x is fuel flow rate, and f is radiation position orifice area, and Φ is radiation position aperture of door coefficient, CtFor radiation
Position specific heat,For the low heat value of fuel, tfireFor the temperature of fuel preheating, CfireFor by fuel preheating to tfireWhen
Avergae specific heat, taFor the preheating temperature of combustion air, CaT is preheating to for combustion airaWhen specific heat capacity, tsmokeFor compressed air
Preheating temperature, VsmokeFor the volume of compressed air, CsmokeT is preheating to for compressed airsmokeWhen specific heat capacity, hejectFor row
Cigarette enthalpy, λ are thermal coefficient, and δ is brick material thickness, and F is heat dissipation area, and L is combustion-supporting air quantity, hkFor cooling air enthalpy；
Step A2, by the residence time t of production streamglassTime t required for being escaped with bubblebubbleRatio as clarification the factor
RF, then
Wherein,ρlFor the density of glass melt, ρbFor in bubble
The density of gas, μ are the coefficient of viscosity of bubble, and r is bubble radius, labFor the height of settling section, lbcFor the length of clarifying area
Degree, b are the width of tank furnace, and n is the amount of the substance of gas,It is t for temperature0When combustion product specific heat capacity；
Step A3, by tank furnace thermal efficiency model η (Tarch, α, x) and factor R F is clarified as objective function, to obtain based on clear
The energy consumption model of clear quality constraint is
3. according to claim 2 based on the horse shoe flame glass furnace efficiency optimization method for improving particle swarm algorithm, feature
It is, credit rating mapping is become to the clarification factor of quantization are as follows:
Credit rating is divided into A, B, C, D and E a total of five grade by step A4, and each grade has to bubble maximum gauge
Technique requirement: grade A is to be not allow for bubble, and grade B is that largest air bubbles diameter is not more than 0.1mm, and grade C is that largest air bubbles are straight
Diameter is not more than 0.2mm, and grade D is that largest air bubbles diameter is not more than 0.5mm, and grade E is that largest air bubbles diameter is not more than 1.0mm；
Step A5, if temperature gradientIt will then by the relationship between bubble radius r and clarification factor R F for definite value k
Credit rating mapping becomes the clarification factor of quantization, obtains:
The credit rating of step A5 is increased to the energy consumption model based on clear quality constraint by step A6, it may be assumed that
4. according to claim 3 based on the horse shoe flame glass furnace efficiency optimization method for improving particle swarm algorithm, feature
It is, particle swarm algorithm improves specifically:
Step B1, if the number of particles of population is z, in the y times iteration, fiFor the fitness of i-th of particle, fmIt indicates most
The fitness of excellent particle, favgFor average fitness, and fitness in particle is enabled to be greater than favgFitness average value be favg', then
Using the difference of individual adaptive optimal control degree and population average fitness as convergence factor Δ: Δ=| fm-favg' |,Convergence factor Δ is used to evaluate the degree of convergence of population: convergence factor Δ is smaller, then the particle is got over
Tend to Premature Convergence；
Step B2 passes through Colony fitness variance σ2Degree of scatter to particle in population is evaluated:
Colony fitness variance σ2Smaller, particle more disperses, and on the contrary then particle more collects
In；
Step B3, by convergence factor Δ and Colony fitness variance σ2As evaluation index, the Evolving State of population is returned
Class dynamically modifies inertia weight value ω according to different Evolving States:
As the f of particle groupi> favg' or favg< fi< favg' when, set inertia weight value ω=ω-(ω-ωmin).e-γ,
Wherein ωminBeing set as 0.5, γ is evolutionary generation；
As the f of particle groupi< favgAnd σ2When greater than preset range value, inertia weight value is set:
As the f of particle groupi< favgAnd σ2When less than preset range value, inertia weight value is set:
5. according to claim 4 based on the horse shoe flame glass furnace efficiency optimization method for improving particle swarm algorithm, feature
It is, the acquisition of the energy consumption optimal solution specifically:
Energy consumption model under clear quality constraint is set as d dimension target search space, ties up target search space in d by step C1
In have a populationRepresent m particle, in which:
A vector point of solution space is tieed up in d for i-th of particle,For the search speed of i-th of particle, i-th of particle search is arrived most
Excellent position is pi=[pi1, pi2..., pim], i=1,2 ..., m, the optimal location that collective search arrives are pg=[pi1,
pi2..., pim],
vi(s+1)=ω vi(s)+c1ε1(pi-xi(s))+c2ε2(pi-xi(s)),
xi(s+1)=xi(s)+vi(s), ε1, ε2Random number between 0 to 1, S are the number of iterations, Studying factorsStudying factors
Step C2, in solution space in random initializtion group particle position and speed, setting current location be pi, setting is just
Optimum position is p in beginning groupg；
Step C3, according to the speed formula v in step C1i(s+1) and location formula xi(s+1), change speed and the position of particle
State；
Penalty function and Lagrangian are combined construction Multiplier Penalty Function by step C4:
Wherein θ is decision variable, and f (θ) is mesh
Scalar functions, hj(θ)=0, j=q+1 ..., m is equality constraint；
The fitness of particle is calculated according to Multiplier Penalty Function；
Step C5 carries out the adaptive adjustment of inertia weight value ω according to current Evolving State, and by current adaptive value and certainly
Body optimal solution compares: if current solution is better than itself optimal value, current adaptive value being just set as itself optimal solution；
Step C6, by the fitness of the particle and history optimal location pi=[pi1, pi2..., pim] fitness compared to relatively determining
It is fixed whether to need to update current globally optimal solution；
Step C7, whether detection is evolved reaches maximum set value or reaches preset convergence precision, should be into if meeting the requirements
Change process terminates, the energy consumption optimal solution being guaranteed under glass smelting quality condition；Otherwise C4 is returned to step.
6. a kind of based on the horse shoe flame glass furnace efficiency optimization system for improving particle swarm algorithm characterized by comprising
Energy consumption modeling module establishes the energy consumption mould constrained based on clear quality for acquiring the creation data of horse shoe flame glass furnace
Type, and credit rating mapping is become into the clarification factor quantified using the energy consumption model constrained based on clear quality；
Algorithm optimization module is used for convergence factor Δ and Colony fitness variance σ2As evaluation index, to judge particle group
Evolving State, particle swarm algorithm is improved；
It is applied to energy optimization module for improved particle swarm algorithm combination Multiplier Penalty Function to be constrained processing method
In energy consumption model under clear quality constraint, the energy consumption optimal solution that is guaranteed under glass smelting quality condition.
7. according to claim 6 based on the horse shoe flame glass furnace efficiency optimization system for improving particle swarm algorithm, feature
It is, the energy consumption modeling module includes:
Submodule A1 obtains tank furnace thermal effect using counterbalancing analysis method for acquiring the creation data of horse shoe flame glass furnace
Rate model:
Wherein,TarchFor coping temperature, α
For coefficient of excess air, x is fuel flow rate, and f is radiation position orifice area, and Φ is radiation position aperture of door coefficient, CtFor radiation
Position specific heat,For the low heat value of fuel, tfireFor the temperature of fuel preheating, CfireFor by fuel preheating to tfireWhen
Avergae specific heat, taFor the preheating temperature of combustion air, CaT is preheating to for combustion airaWhen specific heat capacity, tsmokeFor compressed air
Preheating temperature, VsmokeFor the volume of compressed air, CsmokeT is preheating to for compressed airsmokeWhen specific heat capacity, hejectFor row
Cigarette enthalpy, λ are thermal coefficient, and δ is brick material thickness, and F is heat dissipation area, and L is combustion-supporting air quantity, hkFor cooling air enthalpy；
Submodule A2, for the residence time t of stream will to be producedglassTime t required for being escaped with bubblebubbleRatio as clear
Clear factor R F, then
Wherein,ρlFor the density of glass melt, ρbFor in bubble
The density of gas, μ are the coefficient of viscosity of bubble, and r is bubble radius, labFor the height of settling section, lbcFor the length of clarifying area
Degree, b are the width of tank furnace, and n is the amount of the substance of gas,It is t for temperature0When combustion product specific heat capacity；
With submodule A3, it is used for tank furnace thermal efficiency model η (Tarch, α, x) and factor R F is clarified as objective function, thus
It is to based on the energy consumption model that clear quality constrains
8. according to claim 7 based on the horse shoe flame glass furnace efficiency optimization system for improving particle swarm algorithm, feature
It is, the energy consumption modeling module further include:
Submodule A4, for credit rating to be divided into A, B, C, D and E a total of five grade, each grade has to bubble maximum
The technique requirement of diameter: grade A is to be not allow for bubble, and grade B is that largest air bubbles diameter is not more than 0.1mm, and grade C is maximum
Bubble diameter is not more than 0.2mm, and grade D is that largest air bubbles diameter is not more than 0.5mm, and grade E is that largest air bubbles diameter is not more than
1.0mm；
With submodule A5, for setting temperature gradientFor definite value k, then pass through between bubble radius r and clarification factor R F
Relationship, by credit rating mapping become quantization the clarification factor, obtain:
Submodule A6, for the credit rating of submodule A5 to be increased to the energy consumption model based on clear quality constraint, it may be assumed that
9. according to claim 8 based on the horse shoe flame glass furnace efficiency optimization system for improving particle swarm algorithm, feature
It is, the algorithm optimization module includes:
Submodule B1, for setting the number of particles of population as z, in the y times iteration, fiFor the fitness of i-th of particle, fm
Indicate the fitness of optimal particle, favgFor average fitness, and fitness in particle is enabled to be greater than favgFitness average value be
favg', then using the difference of individual adaptive optimal control degree and population average fitness as convergence factor Δ: Δ=| fm-favg' |,Convergence factor Δ is used to evaluate the degree of convergence of population: convergence factor Δ is smaller, then the particle is got over
Tend to Premature Convergence；
Submodule B2, for passing through Colony fitness variance σ2Degree of scatter to particle in population is evaluated:
Colony fitness variance σ2Smaller, particle more disperses, and on the contrary then particle more collects
In；
With submodule B3, it is used for convergence factor Δ and Colony fitness variance σ2As evaluation index, by the Evolving State of population
Sorted out, inertia weight value dynamically modified according to different Evolving States:
As the f of particle groupi> favg' or favg< fi< favg' when, set inertia weight value ω=ω-(ω-ωmin).e-γ,
Wherein ωminBeing set as 0.5, γ is evolutionary generation；
As the f of particle groupi< favgAnd σ2When greater than preset range value, inertia weight value is set:
As the f of particle groupi< favgAnd σ2When less than preset range value, inertia weight value is set:
10. according to claim 9 based on the horse shoe flame glass furnace efficiency optimization system for improving particle swarm algorithm, feature
It is, the energy optimization module includes:
Submodule C1 is searched for the energy consumption model under clear quality constraint to be set as d dimension target search space in d dimension target
There is a population in rope spaceRepresent m particle, in which:
A vector point of solution space is tieed up in d for i-th of particle,For the search speed of i-th of particle, i-th of particle search is arrived most
Excellent position is pi=[pi1, pi2..., pim], i=1,2 ..., m, the optimal location that collective search arrives are pg=[pi1,
pi2..., pim],
vi(s+1)=ω vi(s)+c1ε1(pi-xi(s))+c2ε2(pi-xi(s)),
xi(s+1)=xi(s)+vi(s), ε1, ε2Random number between 0 to 1, S are the number of iterations, Studying factorsStudying factors
Submodule C2, in solution space in random initializtion group particle position and speed, setting current location be pi,
It is p that optimum position in initial population, which is arranged,g；
Submodule C3, for according to the speed formula v in step C1i(s+1) and location formula xi(s+1 changes the speed of particle
And location status；
Submodule C4, for penalty function and Lagrangian to be combined construction Multiplier Penalty Function:
Wherein θ is decision variable, and f (θ) is mesh
Scalar functions, hj(θ)=0, j=q+1 ..., m is equality constraint；
The fitness of particle is calculated according to Multiplier Penalty Function；
Submodule C5 for carrying out the adaptive adjustment of inertia weight value ω according to current Evolving State, and will be adapted to currently
Value is compared with itself optimal solution: if current solution is better than itself optimal value, current adaptive value being just set as itself most
Excellent solution；
Submodule C6, for by the fitness of the particle and history optimal location pi=[pi1, pi2..., pim] fitness compare
Relatively decide whether to update current globally optimal solution；
Submodule C7 evolves for detection and whether reaches maximum set value or reach preset convergence precision, if meeting the requirements,
Then the evolutionary process terminates, the energy consumption optimal solution being guaranteed under glass smelting quality condition；Otherwise implementation sub-module C4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810529748.9A CN109033489B (en) | 2018-05-29 | 2018-05-29 | Improved particle swarm algorithm-based horseshoe flame glass kiln energy efficiency optimization method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810529748.9A CN109033489B (en) | 2018-05-29 | 2018-05-29 | Improved particle swarm algorithm-based horseshoe flame glass kiln energy efficiency optimization method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109033489A true CN109033489A (en) | 2018-12-18 |
| CN109033489B CN109033489B (en) | 2022-06-07 |
Family
ID=64611462
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810529748.9A Active CN109033489B (en) | 2018-05-29 | 2018-05-29 | Improved particle swarm algorithm-based horseshoe flame glass kiln energy efficiency optimization method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109033489B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111259540A (en) * | 2020-01-14 | 2020-06-09 | 广东工业大学 | Energy efficiency analysis and optimization method for photovoltaic glass calendering and molding process |
| CN114638467A (en) * | 2022-01-31 | 2022-06-17 | 南通鑫鑫医药药材有限公司 | Medical glass container production quality detection system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| SU1201236A1 (en) * | 1982-10-25 | 1985-12-30 | Предприятие П/Я М-5314 | Method of heating bath glassmaking furnace |
| US5764544A (en) * | 1995-11-16 | 1998-06-09 | Gas Research Institute | Recuperator model for glass furnace reburn analysis |
| US20090103989A1 (en) * | 2007-10-17 | 2009-04-23 | Remco International , Inc. | Method of dynamic energy-saving superconductive transporting of medium flow |
| WO2016001310A1 (en) * | 2014-07-01 | 2016-01-07 | Magma Giessereitechnologie Gmbh | A method and algorithm for simulating the influence of thermally coupled surface radiation in casting processes |
| CN106863136A (en) * | 2017-01-15 | 2017-06-20 | 复旦大学 | CCOS glossing full frequency band converged paths planing methods |
| CN107301303A (en) * | 2017-07-14 | 2017-10-27 | 吴康 | 3D glass heat bender mold heating system colony intelligence Optimization Designs |
-
2018
- 2018-05-29 CN CN201810529748.9A patent/CN109033489B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| SU1201236A1 (en) * | 1982-10-25 | 1985-12-30 | Предприятие П/Я М-5314 | Method of heating bath glassmaking furnace |
| US5764544A (en) * | 1995-11-16 | 1998-06-09 | Gas Research Institute | Recuperator model for glass furnace reburn analysis |
| US20090103989A1 (en) * | 2007-10-17 | 2009-04-23 | Remco International , Inc. | Method of dynamic energy-saving superconductive transporting of medium flow |
| WO2016001310A1 (en) * | 2014-07-01 | 2016-01-07 | Magma Giessereitechnologie Gmbh | A method and algorithm for simulating the influence of thermally coupled surface radiation in casting processes |
| CN106863136A (en) * | 2017-01-15 | 2017-06-20 | 复旦大学 | CCOS glossing full frequency band converged paths planing methods |
| CN107301303A (en) * | 2017-07-14 | 2017-10-27 | 吴康 | 3D glass heat bender mold heating system colony intelligence Optimization Designs |
Non-Patent Citations (2)
| Title |
|---|
| PROCESSESKANG-KANG XU: "Dual least squares support vector machines based spatiotemporalmodeling for nonlinear distributed thermal processes", 《JOURNAL OF PROCESS CONTROL》 * |
| 邵景楚: "玻璃窑炉节能优化设计初探", 《玻璃》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111259540A (en) * | 2020-01-14 | 2020-06-09 | 广东工业大学 | Energy efficiency analysis and optimization method for photovoltaic glass calendering and molding process |
| CN111259540B (en) * | 2020-01-14 | 2023-04-11 | 广东工业大学 | Energy efficiency analysis and optimization method for photovoltaic glass calendering and molding process |
| CN114638467A (en) * | 2022-01-31 | 2022-06-17 | 南通鑫鑫医药药材有限公司 | Medical glass container production quality detection system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109033489B (en) | 2022-06-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109033489A (en) | Based on horse shoe flame glass furnace efficiency optimization method, the system for improving particle swarm algorithm | |
| CN201737965U (en) | Automatic optimizing combustion intelligent control system of hot blast stove | |
| CN108489108A (en) | A kind of duty control method based on electric heater load group model | |
| CN105157057A (en) | Combustion control method and system for hot blast heater | |
| CN105864797A (en) | Real-time prediction system and method for boiler entering heat value of circulating fluidized bed household garbage incineration boiler | |
| CN108764450A (en) | A kind of parameter optimization and method of estimation based on broad sense particle cluster algorithm | |
| SA516370507B1 (en) | Submerged combustion melters and methods | |
| CN105631759B (en) | The Multiobjective Scheduling planning device of steel mill consideration hot metal supply condition | |
| CN107330597A (en) | Blast funnace hot blast stove power consumption state various dimensions appraisal procedure based on Fuzzy Pattern Recognition | |
| CN103645694A (en) | PS copper converter converting process intelligent decision and terminal forecast method and device | |
| CN107994574B (en) | New energy consumption-oriented centralized temperature control load side demand response decision method | |
| CN103544273A (en) | Method for assessing integral states of furnace conditions by aid of pattern recognition technology | |
| CN105045104A (en) | Kaldo furnace parameter optimization control method based on fuzzy clustering genetic algorithm | |
| CN201748797U (en) | Novel immersion-tube aluminum alloy melting holding furnace | |
| CN103135443A (en) | AOD (argon oxygen decarburization) furnace energy consumption optimization method based on energy carrier | |
| CN106232533B (en) | Glass melting and method of refining and device | |
| CN108151242A (en) | A kind of central air-conditioner control method towards cluster demand response | |
| CN110059940A (en) | A kind of steel smelting-continuous casting interface linking power-economizing method and system | |
| Tian et al. | Application of genetic algorithm combined with BP neural network in soft sensor of molten steel temperature | |
| CN208468689U (en) | Utilize the system of hot molten slag production insulation board | |
| CN106629628B (en) | A kind of combustion furnace | |
| CN109598643B (en) | Method and equipment for distributing zonal gas flow of annealing furnace and storage equipment | |
| TW201111727A (en) | Furnace for melting batch raw materials with optimized preheating zone | |
| Wang et al. | Hybrid model for predicting oxygen consumption in BOF steelmaking process based on cluster analysis | |
| CN205917166U (en) | Glass bead's apparatus for producing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |