High-frequency high-voltage transformer for electrostatic dust collection optimum design method
Technical field:
The present invention designs a kind of Optimum Design of Transformers method more particularly to a kind of high-frequency high-voltage transformer for electrostatic dust collection
Optimum design method belongs to technical field of environment protection equipment.
Technical background:
With the continuous development of industry, atmosphere pollution is on the rise, and the environmental consciousness of the mankind is gradually reinforced, to industrial dedusting
Technology proposes requirements at the higher level.Wherein, high-frequency and high-voltage electrostatic precipitator is more and more extensive in industrial dedusting use.
High frequency transformer is the very important link of electrostatic dust collection equipment.With the continuous development of power technology, become
The overall performance of depressor has obtained steady promotion, but from the point of view of current transformer service condition, there is also many problems,
It must be optimized, to realize the maximization of transformer efficiency performance.Especially because its work is under high frequency condition,
Operating status is more complicated, and high-frequency loss increases, therefore how to reduce loss and improve efficiency that there is still a need for more researchs.
Particle swarm optimization algorithm (Particle Swarm Optimization-PSO), genetic algorithm (Genetic
It Algorithm-GA) is all the evolution algorithm (Evolutionary Algorithm-EA) having many uses.PSO algorithm and GA are calculated
Method is similar, from RANDOM SOLUTION, finds optimal solution by iteration.103310055 B of Patent No. CN 104317979 A and CN
Document describe genetic algorithm and particle swarm algorithm respectively to optimize design of transformer, but particle swarm algorithm is due between individual
Information sharing is not carried out directly, easily falls into local optimum, and genetic algorithm does not account for the situation of every generation individual, easily falls into office
Portion is optimal, therefore two kinds of algorithm limitations.
Summary of the invention:
The object of the present invention is to provide a kind of high-frequency high-voltage transformer for electrostatic dust collection optimum design methods.In order to mention
The efficiency of transmission of high high frequency high voltage transformer, makes it meet the requirement of high frequency condition, and the present invention is excellent by using Genetic Particle Swarm
Change algorithm, meets the performance requirement of its various aspects, delivery efficiency requires to reach 85% or more, and keeps the operation shape of transformer
State is stablized.
The purpose of the present invention is mainly achieved through the following technical solutions:
A kind of high-frequency high-voltage transformer for electrostatic dust collection optimum design method, specifically includes:
(1) transformer technology parameter is determined according to design objective;
(2) resonant parameter is analyzed, and the selection of resonant parameter should meet fs<1/2fr, f in formulasFor switching frequency, frFor resonance
Intrinsic frequency,Ce=(CpCs)/(Cs+Cp), LsFor leakage inductance, CpFor distribution capacity, CsFor series resonance electricity
Hold, power work is in discontinuous conduct mode;
(3) according to relevant parameters such as power, frequencies by following formula reference area product ApValue:
K is form factor in formula, and D is duty ratio complete cycle, in order to make ApValue meets maximum requirement, and taking D is maximum value, D
=1;KuFor window utilization factor, value range 0.2-0.4, J are current density, empirical value 150-400A/cm2;F is work
Working frequency;BmFor peakflux density, takeAwFor entire magnetic core window area;AeFor effective sectional area;
(4) core material is selected, is the ultramicro-crystal alloy that Germany's VAC company model is Vitroperm 500;
(5) turn ratio theoretical value is calculatedU in formulap-minFor the specified minimum value of primary side input voltage;
(6) Genetic Particle Swarm Algorithm is used, is that optimization becomes with the turn ratio and duty ratio using transformer loss as optimization aim
Amount continues in next step, otherwise to re-start optimization, majorized function can be by following formula letter if optimum results are consistent with target efficiency
It is single to indicate:
P in formulaloss(n, D) is the loss expression formula of transformer;N is the transformer primary pair side turn ratio;D is the duty of waveform
Than;
(7) the transformer primary pair side turn ratio that optimization obtains in (6) is utilized, determines former secondary side the number of turns,
n2=n × n1, n1、n2Respectively former and deputy side the number of turns;
(8) Main Insulation Structure of Electric designs, using oil-immersed transformer, major insulation using transformer oil as insulating materials,
Using oily shielding insulation structure;By formulaFind out the window of transformer core
Mouth width degree LwAnd the volume V of transformerc;Wherein L1Insulation distance between primary side winding and magnetic core, L2For primary side winding
Every thickness degree, L3For transformer primary side pair side insulation distance, L4For the every thickness degree of vice-side winding, n2cFor the winding number of plies on secondary side, L5
For transformer secondary to the insulation distance on secondary side, AcFor the geometric cross section product of transformer, h1Indicate magnetic core window height;
(9) winding line gauge, the bare conductor sectional area S of primary side winding are calculatedp=Ip/ J, the height of the copper foil of primary sideThe bare conductor sectional area of vice-side windingSecondary wire line footpathWherein IsFor secondary current;It is former
Side electric current Ip=IS×n;dpFor copper thickness;
(10) can verifying magnetic core window area meet formula Spn1+Ssn2≤0.2Di 2It is required that if satisfied, continue in next step,
Otherwise (9) are returned to recalculate;Wherein DiIndicate magnetic core window width;
(11) winding copper loss is calculated, by formula
Pcu=2 (Pp-cu+Ps-cu) calculate winding copper loss;
Wherein kp-s、kp-xRespectively indicate the skin effect coefficient and kindred effect coefficient of primary side winding, ks-s、ks-xRespectively indicate secondary side around
The skin effect coefficient and kindred effect coefficient of group, ρ are resistivity, and δ is kelvin effect penetration depth, r0For wire radius, r0=
dp/ 1.772, r1=ds/ 2, Δ0=dp/ δ, Δ1=ds/ δ, la-p、la-sIndicate that the average turn of former vice-side winding is long, Rac-p、Rac-s
Respectively indicate the AC resistance of former vice-side winding, Pp-cu、Ps-cuRespectively indicate the copper loss of former vice-side winding, PcuFor winding copper loss;
(12) core loss is calculatedMagnetic core density p in formulacore=7.2g/cm3, VcFor
Volume of transformer, k, α, β are empirical coefficient, D1/2For the duty ratio of half period;
(13) according to effectiveness formulaWhether verification efficiency η is more than or equal to 80%, if satisfied, continue in next step,
Otherwise it returns to (1) to recalculate, wherein PoFor output power, PΣFor total losses of transformer, P∑=Pcu+PFe;
(14) distribution parameter and series resonant capacitance of transformer are determined, the frequency for judging whether it meets in (2) is wanted
It asks, if satisfied, then continuing in next step, otherwise to return to (1) and recalculate;
(15) transformer temperature rise and heat dissipation calculate, and temperature rise is approximatelyIf satisfied, then continuing in next step, otherwise
(1) is returned to recalculate;Wherein PΣFor total losses of transformer;KkFor heat transfer coefficient, StFor the total surface area of transformer;
(16) short-circuit impedance calculates, and short-circuit impedance isIn 9%-
10.5%, then it meets the requirements, if satisfied, then continuing in next step, otherwise to return to (1) and recalculate;The wherein total circle of W armature winding
Number, I are armature winding rated current, and K is additional reactance coefficient, can be tabled look-up, and ρ is Rockwell coefficient, etIt is every for high-low pressure winding
Circle voltage, Σ D refer to the equivalent area of leakage field, HkIt must be averaged reactance height for two windings.
The purpose of the present invention can be further realized by following technical measures:
Aforementioned electrostatic dedusting high frequency high voltage transformer optimum design method, wherein in step (3), when input waveform is positive
When string wave, K=4.44;When input waveform is square wave, K=4, KuTake 0.4;J takes 150A/cm2;BmValue 0.4T.
Aforementioned electrostatic dedusting high frequency high voltage transformer optimum design method, wherein the Genetic Particle Swarm optimization of step (6)
Algorithm, the specific steps of which are as follows:
(1) population initializes: initializing the number of iterations G of population, particle dimension n, population size m;
(2) initialize particle position and speed: creation m row n column location matrix pop, and by its assignment 0 to 1 it
Between, every a line represents a particle, and first row represents the turn ratio, and secondary series represents duty ratio, and third column represent transformer loss, and
Make MinX (j)≤pop (i, j)≤MaxX (j), wherein 1≤i≤m, 1≤j≤n, MinX (j) are the corresponding minimum of jth column variable
Value, MaxX (j) are the corresponding maximum value of jth column variable;The rate matrices v for creating m row n column simultaneously, is worth for [0,0.02];
And current location pop (i, j) of particle is initialized as to the optimal location ibest of particle itself, according to the initial position of particle
The fitness value that all particles are found out with fitness function finds out the optimal particle position of fitness, complete as population
The initial value gbest of office's optimal location;
(3) particle fitness calculates: as population updates the progress of iteration, grain after finding out update with fitness function
The fitness value of son, the foundation that the superiority and inferiority of fitness value is updated as particle;
(4) roulette selection: generating line number r, 1≤r≤m at random, then generates the r row selection of pop matrix new
M row n column matrix newpop in carry out following cross and variation etc. operation;
(5) crossover operation: one crossover probability pc of selection, value range is [0.4,0.9], when the probability generated at random is small
Crossover operation is then carried out when pc, the m row of matrix newpop is assigned a value of global optimum gbest;
(6) mutation operation: suitable mutation probability pm is selected, value range is [0.01,0.1], general when what is be randomly generated
Then particle makes a variation when rate is less than this probability, i.e., its value is become random value, and the m row of matrix newpop is assigned a value of the overall situation most
Figure of merit gbest;
(7) it the optimal location ibest of more new particle itself and population global optimum position gbest: calculates and judges more
The fitness value of particle after new will if its value is better than the fitness value that particle is obtained at itself optimal location ibest
Ibest is updated to the position of current particle, does not otherwise update;If its value is better than in population global optimum, the place gbest, position
Gbest is then updated to current particle position, otherwise not updated by the fitness value of acquirement.
(8) particle position and speed are updated: according to the speed and location update formula of following population, more new particle
Speed and position, and its value is constrained within limits;
V (i :) and=w*v (i :)+c1*r1*(ibest(i,:)-pop(i,:))+c2*r2*(gbest-pop(i,:))
Pop (i :)=pop (i :)+0.3*v (i :)
Wherein w is the inertia weight factor, and for controlling the influence that current particle speed updates next-generation particle, w is bigger,
Influence is bigger, and the speed of the update of particle is faster, and group is more active, is conducive to global optimizing;Conversely, speed updates slowly, favorably
In local optimal searching;C1, c2 are the constant on [0,2], referred to as positive weighting constant, for controlling particle to itself optimal location
Ibest and global optimum position gbest close speed, generally takes c1=c2=2;R1, r2 are the constant on [0,1], often
Secondary iteration Shi Douhui is randomly generated;Pop (i :), v (i :) in formula are respectively all members of the i-th row of pop matrix and v matrix
Element, 1≤i≤m;
(9) judge whether to meet termination condition, satisfaction then stops iteration, otherwise continues.
Aforementioned electrostatic dedusting high frequency high voltage transformer optimum design method, wherein crossover probability pc value is 0.8.
Aforementioned electrostatic dedusting high frequency high voltage transformer optimum design method, wherein mutation probability pm, value 0.04.
Compared with prior art, the beneficial effects of the present invention are: the characteristic of high frequency transformer is directed to, in high-frequency and high-voltage transformation
Population is combined with genetic algorithm on the basis of device design method, is optimization with the duty ratio of transformer turns ratio and waveform
Variable optimizes transformer using output loss as optimization aim.The present invention uses Genetic Particle Swarm Algorithm, i.e., in grain
The selection algorithm of genetic algorithm, crossover operation and mutation operation are introduced into particle swarm algorithm on the basis of swarm optimization,
Inertia weight is added on the basis of basic particle group algorithm in Genetic Particle Swarm Algorithm, to the global and local optimizing ability of algorithm
It is improved, and is combined with genetic algorithm, joined defect and deficiency that the cross and variation factor makes up particle swarm algorithm.This
Algorithm iteration process is simple, has the advantages that be readily appreciated that, be easily achieved, convergence rate is very fast, efficiency is more excellent etc..
Detailed description of the invention:
Fig. 1 is electrostatic precipitation high-frequency and high-voltage power supply system construction drawing;
Fig. 2 is the high-frequency high-voltage transformer for electrostatic dust collection design flow diagram based on Genetic Particle Swarm Algorithm;
Fig. 3 is Genetic Particle Swarm Optimization Algorithm flow chart;
Fig. 4 transformer simplified model.
Specific embodiment:
The present invention will be further explained below with reference to the attached drawings and specific examples.
As shown in Figure 1, being electrostatic precipitation high-frequency and high-voltage power supply system construction drawing, system main circuit specifically includes that 380V
Three-phase alternating current 1, electric-power wire filter 2, rectified three-phase circuit 3, filter capacitor 4, the first full bridge inverter 5, first buffering
Circuit 6, the second full bridge inverter 7, the second buffer circuit 8, resonance circuit 9, high frequency transformer 10, high-voltage rectification silicon stack 11,
Damping resistance 12, equivalent load 13.System control loop specifically includes that analog signal conditioning circuit 14, governor circuit 15, number
Signal conditioning circuit 16, driving plate 17, iFIX ipc monitor unit 18.
380V three-phase alternating current 1 first passes through concatenated electric-power wire filter 2 except making an uproar, then passes sequentially through concatenated three phase rectifier
Circuit 3 and filter capacitor 4 are changed into 530V direct current, then pass through concatenated first full bridge inverter 5, the first buffer circuit
6,530V direct current is changed into the high-frequency ac of 20kHz by the second full bridge inverter 7, the second buffer circuit 8, resonance circuit 9
Electricity, and give to high frequency transformer 10 and boost, finally after the rectification of high-voltage rectification silicon stack 11, export 72kV high voltage direct current.?
During this, control loop acquisition major loop and other coherent signals, the signal of acquisition successively pass through analog signal conditioning circuit
14, driving plate 17 and iFIX ipc monitor unit 18, driving plate are re-fed into after governor circuit 15, digital signal conditioning circuit 16
17 export the PWM waves of corresponding frequencies to the first full bridge inverter 5 and the second full bridge inverter 7 according to the signal of acquisition,
The operation conditions of each signal is then carried out display and troubleshooting etc. by iFIX ipc monitor unit 18, common realization power supply
Output control.
It is illustrated in figure 4 transformer simplified model.It is as shown in Figure 2 and Figure 3: the electrostatic precipitation based on Genetic Particle Swarm Algorithm
With high frequency high voltage transformer design method, the specific steps are as follows:
(1) each primitive technology requirement is determined according to design objective, such as working frequency f=20kHZ, primary side input voltage Up
In 510V~530V range, secondary side output voltage Us=80V, secondary side export electric current Is=1.07A, output power Po=
85KVA, efficiency eta >=80%, 45 DEG C of < of temperature rise Δ t etc..
(2) resonant parameter is analyzed, and the selection of resonant parameter should meet fs<1/2fr, f in formulasFor switching frequency, frFor resonance
Intrinsic frequency,Ce=(CpCs)/(Cs+Cp), LsFor leakage inductance, CpFor distribution capacity, CsFor series resonance electricity
Hold, choosing suitable parameter makes power work in discontinuous conduct mode.
(3) with according to power, the relevant parameters such as frequency calculate A by following formulapValue.
(4) core material is selected, with according to ApValue choosing and actual demand select its size.Using selection VAC company type, Germany
Number be Vitroperm 500 ultramicro-crystal alloy.Using two-way magnetized state, using the nanometer of small to high-frequency loss, high magnetic conduction
Ultracrystallite circular cross-section square shape magnetic core takes special high-voltage line wrapping technique, reduces window size, reduce iron core
Loss.Using stainless steel cover board, the generation of high-frequency vortex is avoided.
(5) turn ratio theoretical value is calculated
(6) Genetic Particle Swarm Algorithm is used, is that optimization becomes with the turn ratio and duty ratio using transformer loss as optimization aim
Amount continues in next step, otherwise to re-start optimization, majorized function can be by following formula letter if optimum results are consistent with target loss
It is single to indicate:
P in formulaloss(n, D) is the loss expression formula of transformer;
N is the transformer primary pair side turn ratio;
D is the duty ratio of waveform;
The specific method of optimization is as follows:
1) population initializes: initializing the number of iterations G of population, particle dimension n, population size m etc..
2) initialize particle position and speed: creation m row n column location matrix pop, and by its assignment 0 to 1 it
Between, every a line represents a particle, and first row represents the turn ratio, and secondary series represents duty ratio, and third column represent transformer loss, wrap
Copper loss and iron loss are included, and makes MinX (j)≤pop (i, j)≤MaxX (j), wherein 1≤i≤m, 1≤j≤n, MinX (j) are jth
The corresponding minimum value of column variable, MaxX (j) are the corresponding maximum value of jth column variable.The velocity moment of m row n column is created simultaneously
Battle array v, is worth for [0,0.02].And current location pop (i, j) of particle is initialized as to the optimal location ibest of particle itself, root
The fitness value for finding out all particles with fitness function according to the initial position of particle, finds out the optimal particle position of fitness,
As the initial value gbest of population global optimum position.
3) particle fitness calculates: as population updates the progress of iteration, grain after finding out update with fitness function
The fitness value of son, the foundation that the superiority and inferiority of fitness value is updated as particle.
4) roulette selection: generating line number r, 1≤r≤m at random, then generates the r row selection of pop matrix newly
The operation such as following cross and variation is carried out in the matrix newpop of m row n column.
5) crossover operation: one crossover probability pc of selection generally takes [0.4,0.9], and value is 0.8 best, when random raw
At probability be less than pc when then carry out crossover operation, cross method takes method are as follows: the first row is exchanged with m row, the second row and
The exchange of m-1 row, and so on.The m row of matrix newpop is assigned a value of global optimum gbest.
6) mutation operation: selecting suitable mutation probability pm, generally take [0.01,0.1], and value is 0.04 best, when with
Then particle makes a variation when the probability that machine generates is less than this probability, i.e., its value is become to the random value in parameter area.Matrix
The m row of newpop is assigned a value of global optimum gbest.
7) the optimal location ibest of more new particle itself and population global optimum position gbest.
8) particle position and speed are updated: according to the speed and location update formula of following population, more new particle
Speed and position, and its value is constrained within limits.Speed more new formula and displacement more new formula are as follows:
V (i :) and=w*v (i :)+c1*r1*(ibest(i,:)-pop(i,:))+c2*r2*(gbest-pop(i,:))
Pop (i :)=pop (i :)+0.3*v (i :)
The weighting constant that c1, c2 are positive in formula leans on for adjusting particle to particle itself and entire population optimal location
Close speed, generally takes c1=c2=2.
W is the inertia weight factor.
R1, r2 are the random number between 0 to 1.
I indicates line number, 1≤i≤m.
9) judge whether to meet termination condition, termination condition herein is the total degree for requiring iteration, meets and then stops changing
Otherwise in generation, continues.
10) output result obtains turn ratio n=156, duty ratio D=0.69, it can be seen that the theoretical value of the turn ratio and optimal value are non-
Very close to loss is also met the requirements.Then once calculated using optimum results.
(7) the transformer primary pair side turn ratio that optimization obtains in (6) is utilized, determines former secondary side the number of turns,
(8) Main Insulation Structure of Electric designs, and the present invention uses oil-immersed transformer, and major insulation is using transformer oil as insulation
Material, using oily shielding insulation structure, by formulaCalculate to obtain Lw=
1.7cm, Vc=3.196 × 103cm3。
(9) winding line gauge, the bare conductor sectional area S of primary side winding are calculatedp=Ip/ J=167 × 102/ 150=111mm, it is former
The height h of the copper foil on sidep=Sp/dp=111/0.5=222mm, the bare conductor sectional area S of vice-side windings=Is/ J=1.07 ×
102/ 150=0.713mm2, secondary wire line footpathWherein IsFor secondary current;
Primary current Ip=IS× n=1.07 × 156=167A;dpFor copper thickness.
(10) can verifying magnetic core window area meet formula Spn1+Ssn2≤0.2Di 2It is required that if satisfied, continue in next step,
Otherwise (9) are returned to recalculate.The data band substituted into the present invention obtains (111 × 7+0.713 × 1092=1555.6) < (0.2
×1702=5780), it was demonstrated that meet the requirements.
(11) winding copper loss is calculated, by following formula
Obtain former secondary side around
Group copper loss, finally calculate winding copper loss be Pcu=2 (Pp-cu+Ps-cu)=2 × (68.64+3468.6)=7074.48W,
Middle kp-s、kp-xRespectively indicate the skin effect coefficient and kindred effect coefficient of primary side winding, ks-s、ks-xRespectively indicate vice-side winding
Skin effect coefficient and kindred effect coefficient, ρ is resistivity, and δ is kelvin effect penetration depth, r0For wire radius, r0=
dp/ 1.772, r1=ds/ 2, Δ0=dp/ δ, Δ1=ds/ δ, la-p、la-sIndicate that the average turn of former vice-side winding is long, Rac-p、Rac-s
Respectively indicate the AC resistance of former vice-side winding, Pp-cu、Ps-cuRespectively indicate the copper loss of former vice-side winding.
(12) core loss is calculated by following formula
In formula
Magnetic core density pcore=7.2g/cm3, VcFor volume of transformer, k, α, β are empirical coefficient, D1/2For the duty ratio of half period, by
(6) step optimum results take 0.345.
(13) according to effectiveness formulaWhether verification efficiency η is more than or equal to 80%, if satisfied, continue in next step,
Otherwise (1) is returned to recalculate;Each parameter is brought into the present invention
Meet efficiency requirements.
(14) distribution parameter and series resonant capacitance of transformer are determined, the frequency for judging whether it meets in (2) is wanted
It asks, if satisfied, then continuing in next step, otherwise to return to (1) and recalculate.The leakage inductance L that transformer winding generates in the present inventions=13
μ H, distribution capacity Cp=6 μ H, series resonant capacitance Cs=1 μ H, is computed fr=47HZ, meets the requirements.
(15) transformer temperature rise and heat dissipation calculate, it is assumed that the even heat that magnetic core and winding generate dissipates, then temperature rise is approximate
Are as follows:
Wherein, Δ t is to allow temperature rise;PΣFor total losses of transformer;KkFor heat transfer coefficient, to oil-immersed transformer, Kk=5
×10-3W/(℃·cm2);For dry-type transformer, Kk=1.25 × 10-3W/(℃·cm2);StFor the total surface area of transformer.
(16) short-circuit impedance calculates, the calculation formula of short-circuit impedance can abbreviation be following formula:
Wherein Uk(%) is short-circuit impedance UkReactive component percentage value, f be transformer working frequency, the total circle of W armature winding
Number, I are armature winding rated current, and K is additional reactance coefficient, can be tabled look-up, and ρ is Rockwell coefficient, etIt is every for high-low pressure winding
Circle voltage, Σ D refer to the equivalent area of leakage field, HkIt must be averaged reactance height for two windings.
The relationship of inventive algorithm and transformer design parameter are as follows: objective function, i.e. fitness letter are gone out according to the derivation of equation
Number, formula include optimized variable and predetermined optimizing target parameter, can calculate optimized parameter by algorithm iteration.
In addition to the implementation, the present invention can also have other embodiments, all to use equivalent substitution or equivalent transformation shape
At technical solution, be all fallen within the protection domain of application claims.