CN101504689B - Radiator optimizing parameter confirming method - Google Patents

Abstract

The invention relates to a method for determining optimization design parameters of a heat radiator, which is characterized by comprising: determining a group of parameter combinations which minimize the value of a comprehensive object function (M(X)) in a solution space of design parameter values of the heat radiator according to thermodynamic characteristic parameters of the heat radiator, wherein the comprehensive object function is positively correlated with the dynamic response characteristic and the pressure drop loss characteristic of the heat radiator, and the processing of the group of parameter combinations which minimize the value of the comprehensive object function comprises parameter optimization of the comprehensive object function so as to obtain the optimization design parameters meeting comprehensive assessment indexes of the heat radiator.

Description

The heating radiator parameters optimization is determined method

Technical field

The present invention relates to a kind of heating radiator Optimization Design, belong to thermal technology's science and technology field based on genetic algorithm.

Background technology

Along with the development of microelectric technique, integrated level height, the silicon that power is big, volume is little more and more in depth use the every field of industry, and the something in common of these chips is exactly that the heat dissipation capacity of single chip is very big.The mid-80 just has 10 on each chip ⁶Individual element, though the power of each element is very little, high like this integrated level makes heat flow density up to 5 * 10 ⁵W/m ²When the heat flow density increase of chip, the temperature of electronic equipment just rises rapidly, and this is having a strong impact on the reliability and security of electronic equipment.

For the electronic component that thermal value is higher maintains under the suitable temperature, will adopt corresponding heat dissipation technology.Wind-cooling heat dissipating is the main mode of high power electronic chip cooling at present, the principle of work of electronic chip wind-cooling heat dissipating is by heat radiator the heat that chip produces to be conducted out, rotate by fan again, strengthen the convection heat transfer of air and heat radiator, the mode by forced convertion is dispersed into surrounding environment with the heat on the heat radiator.The main effect of heating radiator is in order to increase area of dissipation, to improve heat-sinking capability; And, improve air current composition by using fan to improve wind speed.

In industrial applicability, proposed much to have the difform fin of high heat dispersion, as the waveform fin, have the bar shaped fin of bifurcated, hollow rib adds the fin of window shutter, and the hexagon rib, taper rib etc.Because have heating radiator of a great variety of different fin shapes, its working condition is divided into natural convection and forced convertion again, is very important so find out the thermodynamic behaviour of difformity heating radiator and how to make it reach optimum Working in design.Thereby, should in design process, introduce optimized Measures, according to designing requirement each parameter relevant with optimization aim is all assessed, draw the optimum solution of having taken all factors into consideration all these parameters.But in engineering reality, find out such optimum solution not a duck soup, because different design criteria correspondences different Optimization Model, and each of a parameter may value all corresponding one by target function value that Optimization Model determined, therefore, press for the effective ways that design object carried out the parameter optimization design.

The researchist of association area is exploring different Optimization Design always.The existing method of optimization problem optimum solution or approximate optimal solution of asking mainly contains: enumerative technique, heuritic approach and searching algorithm.Wherein enumerative technique can enumerate all feasible solutions in the feasible solution set, obtain accurate optimum solution, but its efficient is lower, and is very consuming time; Heuritic approach is sought optimum solution by seeking a kind of heuristic rule that produces feasible solution, though efficient is higher, heuristic rule does not have versatility; Searching algorithm is by carrying out optimum solution or the approximate optimal solution that search operation is sought problem in the subclass of feasible solution set, it can reach balance preferably with finding the solution in the quality of approximate solution on the efficient.

At the optimal design problem of heating radiator, researchist's emphasis is also had nothing in common with each other, and mainly determines optimization aim from following several respects: a. heat radiator thermal resistance minimum; B. the manufacturing expense of heating radiator is minimum; C. economical with materials; D. pressure drop is less.No matter single goal optimization or multiple-objection optimization though existing Optimization Design has been considered the characteristics such as heat transfer property of heating radiator, all do not considered the dynamic perfromance of heating radiator.A kind of Optimization Design of taking all factors into consideration thermal resistance, time constant and the pressure drop of heating radiator not only can reflect the heat transfer characteristic of heating radiator but also reflect its dynamic response characteristic; Realize that by genetic algorithm optimizing process not only can guarantee the quality of finding the solution but also can improve and find the solution efficient.

Summary of the invention

According to an aspect of the present invention, a kind of method of optimal design parameter of definite heating radiator is provided, it is characterized in that comprising: according to the thermodynamic behaviour parameter of described heating radiator, in the solution space of described Design of for heat sinks parameter value, determine to make one group of parameter combinations of the value minimum of an integrated objective function (M (X)), the dynamic response characteristic of described integrated objective function and described heating radiator and the positive correlation of droop loss characteristic

Wherein, the processing of one group of parameter combinations of the described value minimum of determining to make described integrated objective function comprises: described integrated objective function is carried out parameter optimization, thereby be met the optimal design parameter of heating radiator comprehensive evaluation index.

According to another aspect of the present invention, provide a kind of heating radiator, having comprised with parameters optimization:

A base,

Be arranged on a plurality of fins on the described base, described a plurality of fins are lined up array of fins,

The design parameter value that it is characterized in that described array of fins determines that according to a kind of definite method this determines that method comprises:

Thermodynamic behaviour parameter according to described heating radiator, in the solution space of described design parameter value, determine to make one group of parameter combinations of the value minimum of an integrated objective function, the dynamic response characteristic of described integrated objective function and described heating radiator and the positive correlation of droop loss characteristic, wherein, the processing of one group of parameter combinations of the described value minimum of determining to make described integrated objective function comprises: described integrated objective function is carried out parameter optimization, thereby be met the optimal design parameter of heating radiator comprehensive evaluation index.

Description of drawings

Fig. 1 is the heating radiator optimal design process flow diagram that comprises embodiments of the invention.

Fig. 2 is cylindrical needle rib heat spreader structures figure according to an embodiment of the invention

Fig. 3 is the vertical view of cylindrical needle rib heating radiator according to an embodiment of the invention.

Fig. 4 is the multiparameter concatenated coding and decoding synoptic diagram of related according to one embodiment of present invention decision variable.

Fig. 5 is related according to one embodiment of present invention single-point intersection synoptic diagram.

Fig. 6 is related according to one embodiment of present invention even variation synoptic diagram.

Fig. 7 carries out the algorithm evolution figure that integration objective is optimized for utilization genetic algorithm according to an embodiment of the invention to cylindrical needle rib heating radiator.

Embodiment

Genetic algorithm (Genetic Algorithm, note by abridging be GA) is a kind of random search algorithm with global optimization ability that is proposed by professor J.Holland, and it is biological evolution development in natural imitation circle.According to " survival of the fittest " principle in the Darwinian theory of biological evolution, in the evolutionary process of nature biotechnology, the probability of adaptable bion existence is bigger.And the parent individuality of Zu Chenging can generate more good offspring individual thus.What genetic algorithm was used is exactly this reason, and the simulation of occurring in nature biological evolution rule is applied in the application of searching problem optimum solution.Studies show that genetic algorithm has very strong ability of searching optimum, especially solving aspect the complicated optimization problem, as multi-objective problem, multi-peak, nonlinear problem aspect.GA algorithm strong robustness is a kind of adaptable global optimum searching algorithm.In the research of taking all factors into consideration multinomial evaluation index problem, can find the Pareto of optimization problem to separate.It is one group of equilibrium solution of optimization problem that Pareto separates, be called noninferior solution again, be meant and do not have more excellent separates all more excellent than this each target function value of separating, can be expressed as: establish x* ∈ R, if there is not x ∈ R, satisfy F (x)≤F (x*), claim that then x* is effectively separating (or claiming that Pareto separates) of multi-objective problem.

The present invention is directed to the problem that faces in the heating radiator optimization adopts genetic algorithm that heating radiator is optimized design, provide a kind of effective utilization genetic algorithm to carry out the chip radiator parameters optimization method, this method provides a benchmark for the optimal design of heat transfer characteristic, dynamic response characteristic and the droop loss characteristic of taking all factors into consideration heating radiator.

In solution of the present invention, a kind of chip radiator method for optimally designing parameters based on genetic algorithm is consuming time and can not realize the problem of heating radiator parametric synthesis optimal design to solve traditional heat-dissipating device Parameters design; Dynamic response characteristic and droop loss characteristic with the heating radiator studied when the thermodynamic behaviour of research heating radiator are the objective function that the complex optimum target makes up algorithm optimization, in the solution space of multidimensional parameter value, seek the one group of parameter combinations that makes integrated objective function value minimum, to obtain making designed heating radiator satisfy the model parameter system that not only dynamic response is fast but also pressure loss is little.

The technical scheme of embodiments of the invention as shown in Figure 1 mainly comprises the steps (the frame of broken lines middle part is divided into the genetic algorithm flow process among Fig. 1):

(1) determines with time constant and droop loss minimum to be the heating radiator optimization mathematical model of comprehensive performance evaluation index

The main thermodynamic behaviour parameter of heating radiator has thermal capacity C _t, thermal resistance R and timeconstant, droop loss Δ P, heat dissipation capacity Q, they are index amounts of reflection heating radiator static characteristics and dynamic perfromance.There is the parameter of direct or indirect influence that the geometric parameter of heating radiator, the material of heating radiator, the type and the cooling condition (laminar flow, turbulent flow) of heat eliminating medium are arranged to the radiator heat mechanical characteristic;

According to thermal capacitance, thermal resistance, time constant and the pressure loss of heating radiator, determine to take all factors into consideration timeconstant and pressure loss Δ P as performance evaluation index to designed heating radiator.In the used pin rib heating radiator of present embodiment was optimized, decision variable was the rib spacing of pin rib, and constraint condition is the radiating requirements that heating radiator satisfies institute's cooling electronic device, determined the mathematical model of optimization problem according to this.

(2) at heating radiator Optimization Model in (1), the utilization genetic algorithm is optimized heating radiator.Adjust program running parameter, seek the Pareto optimum solution of genetic algorithm optimization, comprising:

A. generate initial population.At first (adopt the binary cascade coding method) the heating radiator feasible solution is encoded, generating P group chromosome length is the individual chromosome combination of 2L; Determine corresponding coding/decoding method then, obtain the phenotype of individuality (separating);

B. determine ideal adaptation degree evaluation method.Calculate each individual pairing target function value M (X), determine its fitness function value F (X) by transformation rule;

C. carry out selection operation (selection operator).According to selecting Probability p _s, according to the principle of " survival of the fittest " from t for selecting the higher individuality of fitness the P of colony (t) as the parent individuality, carry out the genetic manipulation of P (t+1) for colony; Use optimum conversation strategy, the individuality that fitness is the highest in the population is preserved be genetic directly to the next generation;

D. carry out interlace operation (crossover operator).At first to the pairing in twos at random of parent individuality, according to crossover probability p _c, adopt the chromosome dyad between a pair of individuality of single-point cross exchange, generate new individuality;

E. carry out mutation operation (mutation operator).According to the variation Probability p _m, the mode that employing evenly makes a variation changes the genic value on individual certain some locus, thereby generates new individuality;

F. repeat b, c, d, e goes on foot until satisfying the loop ends condition.

(3) satisfy the loop ends condition, the output optimum solution.

(4) the heating radiator optimal design finishes

The concrete of principle of the present invention is suitable for:

(1) carries out the necessity of heating radiator optimal design

The purpose of fin surface ribbing is to strengthen heat radiation by rib.The heat dissipation capacity on surface is Q when supposing no fin, and its heat dissipation capacity is Q after the ribbing _f, then according to the magnitude relationship of the two, can be divided into three kinds of situations: (1) .Q＜Q _f, this moment, fin worked to strengthen heat radiation; (2) .Q=Q _f, illustrate that fin is inoperative this moment; (3) .Q＞Q _f, this moment, fin will weaken the effect of heat radiation.Under the situation of the longitudinal size of rib, the heat transfer problem of rib can be by the approximate one dimensional heat transfer problem that is considered as not having endogenous pyrogen at the lateral dimension of rib:

$\frac{\∂T}{\∂t}=a^2\frac{{\∂}^{2}T}{{\∂x}^2}---\left(2\right)$

Do not consider rib tip end-face heat sinking in the calculating of common actual fin, promptly the end boundaries condition is:

$\left\{\begin{array}{c}x=0,T=T_0\\ x=l,\mathrm{dT}/\mathrm{dx}=0\end{array}\right.---\left(3\right)$

In the formula, 1 is the rib height, and T is the temperature of rib, T ₀Be the rib base temperature.Then the size that can get the heating radiator heat transfer capacity of ribbing by (2) (3) is:

Q＝αA _fΔtη _f (4)

α is the surface coefficient of heat transfer of radiator material in the formula; A _fEfficiently radiates heat area (m for heating radiator ²); T is the temperature difference of spreader surface medial temperature and ambient temperature; η _fFor the efficient of rib then the efficient of rib can be expressed as:

${\mathrm{\&eta;}}_f=\frac{\mathrm{<figure-callout\; id="0"\; label="Q\; Q"\; filenames="H2009100808671E00031.png,H2009100808671E00032.png"\; state="\{\{state\}\}">Q</figure-callout>}}{Q_{}}=\frac{\mathrm{th}\left(\mathrm{ml}\right)}{\left(\mathrm{ml}\right)}---\left(5\right)$

In the formula, U is the section girth of rib, A _cBe the cross-sectional area of rib, Q ₀Be the heat transfer capacity of rib ideally, m serves as reasons by the normal value of the geometric parameter decision of rib, the rib of different geometries, its expression way difference.As seen, the heat dissipation capacity of rib and its geometric parameter have confidential relation, and be therefore very crucial in Design of for heat sinks middle rib Parameter Optimization.

(2) optimization aim chooses

Thermal capacity C to heating radiator _t, thermal resistance R, timeconstant analyze as can be known that timeconstant is approximately thermal capacity C _tProduct with thermal resistance R.Be the heating radiator that thermal resistance is little or thermal capacity is little, its time constant is less, and dynamic response is rapid; The heating radiator that thermal resistance is big or thermal capacity is big, its time constant is big, and dynamic responding speed is slower.The pressure loss characteristic also has material impact to the performance of heating radiator, and it is determining whether designed heating radiator is energy-conservation.According to analysis to the radiator heat mechanical characteristic, the present invention takes all factors into consideration the time constant and the objective function of pressure loss for optimizing of heating radiator, the rib spacing of heating radiator is a decision variable, and the minimum heat amount that heating radiator will satisfy is that constraint condition is optimized heating radiator.Be intended to be optimized by the Design of for heat sinks parameter make it not only thermal resistance be less, dynamic response rapidly, less energy consumption, and can satisfy certain radiating requirements.Set up the optimization mathematical model of heating radiator, the utilization genetic algorithm is searched in the feasible solution space, adjust genetic algorithm parameter, optimizing obtains making the pareto of heating radiator comprehensive evaluation index optimum to separate, thereby is preferably satisfied the optimal design parameter of heating radiator comprehensive evaluation index.If genetic algorithm is described as a homogeneous Markov chain Pt={P (t), t 〉=0} carries out the convergence analysis as can be known to it, and the probability that the genetic algorithm of use optimized individual conversation strategy can converge on optimum solution is 1.

The present invention compares with existing fansink designs technology and has the following advantages:

(1) genetic algorithm is based on the overall adaptive probability search optimized Algorithm that the coding of feasible solution is operated, and very strong ability of searching optimum is arranged.The utilization genetic algorithm is carried out parameter optimization, as long as population quantity, to select probability, crossover probability, variation probability to select proper, algorithm can be searched in whole solution space, thereby has reduced to be absorbed in the possibility of local optimum;

(2) the parameters optimization model of being set up has been considered the dynamic response characteristic of heating radiator, is intended to study heating radiator changes response to ambient temperature speed; Also considered the droop loss characteristic of heating radiator, purpose is from energy-conservation angle heating radiator to be designed.

(3) the utilization genetic algorithm is optimized design to heating radiator, can obtain good design result by less calculation cost.Not only can be optimized design, and can optimize the optimal design result under do not coexisted design criteria and the strategy, thereby the design criteria and the layout strategy of more realistic demand are determined in contrast at specific design criteria and layout strategy.

Introduce the concrete utilization of genetic algorithm in heating radiator is optimized below in conjunction with an instantiation and accompanying drawing, in this example heating radiator be aluminium matter cylindrical needle rib heating radiator (Fig. 2, Fig. 3).Note pin rib base plate of radiator thickness is δ (mm), and base areas is of a size of 70 * 70 (mm ²), the rib height is 1, flowing to the direction rib spacing along refrigerating gas is d ₁, the edge is d with the rib spacing that refrigerating gas flows to the vertical direction of direction ₂Refrigerating gas is an air, and the fan wind speed is u (m/s), guarantees in the laminar flow situation.

The associated hot mechanics parameters of heating radiator

The heat-conduction coefficient λ of the thermodynamic property of heating radiator and radiator material _f, the geometric configuration of heat radiator fin and wind speed that heating radiator is cooled off size relevant, its main thermal property has heat dissipation capacity Q, thermal capacity C _t, thermal resistance R, timeconstant and droop loss characteristic Δ P.Prerequisite to its thermodynamic behaviour analysis is to think that the coefficient of heat conductivity of timber material is a constant, and the surface coefficient of heat transfer between rib and the environment is a constant, and ambient temperature is a constant.

A. radiator heat-dissipation amount Q, the heat dissipation capacity of uniform cross section cylindrical needle rib heating radiator is:

Q=α A _fΔ t η _f, the efficient of cylindrical needle rib heating radiator is: η _f=th (ml)/(ml), wherein

B. thermal capacity C _t, be the tolerance of heating radiator heat absorption capacity.For given heating radiator and the temperature difference t between the environment, the heat storage capacity of the system that thermal capacity is bigger is stronger, and its dynamic perfromance is also poor more accordingly, is shown below:

C _t＝ρcV＝ρc(A ₂δ+n ₁n ₂l·πd ²/4) (6)

In the formula, ρ is a radiator material density, and c is the specific heat capacity of radiator material (as aluminium), n ₁, n ₂Be respectively the pin rib number that heating radiator flows to and flows to perpendicular to cooling fluid along cooling fluid: n ₁=(0.07+d ₁)/(d+d ₁), n ₂=(0.07+d ₂)/(d+d ₂).

C. the thermal resistance R of heating radiator is made up of two parts, and a part is the thermal-convection resistance between pin rib and the surrounding air, and another part is the thermal conduction resistance between base plate of radiator and the pin rib, and is as follows:

R＝1/αA ₁η _f+δ/λ _fA ₂ (7)

In the formula,

$\mathrm{\&alpha;}=\mathrm{\&lambda;}/L\&CenterDot;1.4{(\mathrm{uL}/v)}^{0.28}{\mathrm{<figure-callout\; id="1"\; label="p\; r"\; filenames="H2009100808671E00021.png,H2009100808671E00031.png"\; state="\{\{state\}\}">p</figure-callout>}}_r^{},$

η _f＝th(ml)/ml，

$m=\sqrt{4\mathrm{\&alpha;}/\mathrm{\&lambda;d}},$

A ₂＝0.0049，A _f＝n ₁n ₂πdl，

A ₁＝A ₂+A _f。

In the formula, A ₁The total area (m for rib ²), A ₂Be base plate of radiator area (m ²), A _fBe efficiently radiates heat area (m ²), L is effective length (m), Nu is a nusselt number, P _rBe the Prandtl constant

P _r=v/ α, R _eBe Reynolds number, v is average fluid kinetic viscosity (m ²/ s), η _fBe fin efficiency, λ _fBe the heat transfer coefficient (W/m K) of timber material, λ is air thermal conductivity (W/m K).

D. timeconstant is represented the heating radiator speed of variation of ambient temperature response to external world, and has reflected the dynamic perfromance of heating radiator.According to thermal conduction study knowledge, the time constant that volume and surface area are respectively the solid matter system with arbitrary shape of V and A is

$\mathrm{\&tau;}=\frac{\mathrm{\&rho;cV}}{\mathrm{\&alpha;A}}=\frac{C_t}{\mathrm{\&alpha;A}}---\left(8\right)$

In the formula, α is the material surface heat transfer coefficient.

For pin rib heating radiator,

A parameter meaning is the same in the formula.

E. the pressure loss Δ P of heating radiator, the resistance that streams generation when the atmospheric pressure loss of the pin rib heating radiator of flowing through is the fluid cross-flow tube bank causes.The pressure loss Δ P of fluid can be expressed as when fluid flow through pin rib heating radiator:

$\mathrm{\&Delta;p}={\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="H2009100808671E00021.png,H2009100808671E00031.png"\; state="\{\{state\}\}">c</figure-callout>}}_1{\mathrm{Re}}^{-1/5}\&CenterDot;2{\mathrm{\&rho;}}_f\&CenterDot;u_{\max }^2{(\mathrm{\&mu;}/{\mathrm{\&mu;}}_w)}^{0.14}\&CenterDot;m---\left(5\right)$

In the formula, c ₁Be constant; ρ _fDensity for cooling fluid; u _MaxBe the Peak Flow Rate in cross section, fluid passage, u _Max=u (d+d ₁)/d ₁μ, μ _wBe respectively the coefficient of dynamic viscosity and the wall coefficient of kinetic viscosity of fluid; M is the pin rib row number that flows to along cooling fluid, m=(W ₁+ d ₂)/(d+d ₂).Wherein, W ₁For flow to the base plate of radiator width of direction along cooling fluid.

The mathematical model of heating radiator parameter optimization

For heating radiator, big thermal capacity is indicating that heating radiator will could reach thermal equilibrium with surrounding environment with the long period; Low thermal resistance represents that the radiator heat-dissipation ability is good, and time constant is little, can within a short period of time and external environment reach thermal equilibrium.Time constant then can the reflect heat capacity and the synthesis result of two kinds of influences of thermal resistance.The big explanation radiator heat of time constant capacity resistance big or heat flow path is big, so radiator temperature changes fast; Time constant novel bare radiator thermal capacity resistance little or heat flow path is little, so temperature variation is slow.Therefore, in an embodiment, taking all factors into consideration the heating radiator response time and the pressure loss minimum is optimization aim, adopts the weight coefficient method of changing that the integration objective optimization problem is converted into the single goal optimization problem; Time constant and pressure drop are carried out nondimensionalization, set up Optimization Model shown in the formula (6):

$\left\{\begin{array}{cc}\min M\left(X\right)=& a\&CenterDot;\mathrm{\&tau;}({\mathrm{<figure-callout\; id="1"\; label="x"\; filenames="H2009100808671E00021.png,H2009100808671E00031.png"\; state="\{\{state\}\}">x</figure-callout>}}_1,x_2)/{\mathrm{\&tau;}}_{\max }+(1-a)\&CenterDot;\mathrm{\&Delta;p}({\mathrm{<figure-callout\; id="1"\; label="x"\; filenames="H2009100808671E00021.png,H2009100808671E00031.png"\; state="\{\{state\}\}">x</figure-callout>}}_1,x_2)/{\mathrm{\&Delta;p}}_{\max }\\ \mathrm{subject}& \mathrm{to}:x_{1\min }<x_1<x_{1\max },x_{2\min },<x_2<x_{2\max }\\ Q_{\min }\&le;Q& \end{array}\right.---\left(6\right)$

0＜a in the formula＜1 is a weight coefficient, and it is a constant; Δ P is the pressure drop of pin rib heating radiator; τ _MaxWith Δ P _MaxBe respectively the maximal value of time constant and pressure drop, τ _MaxThe time constant of base plate of radiator when getting apleuria, Δ P _MaxBe taken as the maximum pressure drop of fan, determine by the characteristic of concrete used fan.x ₁Represent spacing d ₁, x ₂Expression rib spacing d ₂, Q _MinIt is the thermal source thermal value.

The utilization genetic algorithm is carried out parameter optimization

A. generate initial population

Setting the population size is P, selects the multiparameter cascade coding method for use, and the note chromosome length is 2L.Generate the string of binary characters that the P group length is 2L at random, constraint condition x is satisfied in representative _i∈ (x _Imin, x _Imax) (i=1,2, L, P n) is individual, and each individuality has all been represented a kind of feasible solution of the problem of finding the solution.Get L=10 among the present invention, the actual value (as Fig. 4) of individuality being decoded and obtaining decision variable according to formula (7):

$x_i=(x_{i\max }-x_{i\min })\&CenterDot;\frac{y_i}{2^{(L-1)}}+x_{i\min }(i=\mathrm{1,2},L,n)---\left(7\right)$

Y in the formula _iBe the pairing decimal system numerical value of string of binary characters.

B. determine ideal adaptation degree evaluation method

Calculate individual goal functional value f (X), objective function f (the X)=M=a τ/τ of heating radiator optimization problem _Max+ (1-a) Δ p/ Δ p _Max, note ideal adaptation degree value is Fit, individual Fit value is determined according to formula (8), and according to this individuality is assessed;

$\mathrm{Fit}=\left\{\begin{array}{cc}{C}_{\max }-f\left(X\right),& \mathrm{iff}\left(X\right)<{\mathrm{<figure-callout\; id="0"\; label="C\; max"\; filenames="H2009100808671E00031.png,H2009100808671E00032.png"\; state="\{\{state\}\}">C</figure-callout>}}_{\max }\\ 0,& \mathrm{iff}\left(X\right)\&GreaterEqual;C_{\max }\end{array}\right.---\left(8\right)$

When independent variable does not satisfy constraint condition, need to introduce penalty function in the fitness function to guarantee that fitness is for just.Penalty function is expressed as: P (X)=c _i* (Q _Min-Q), (i=1,2 ..., n), c wherein _i(i=1,2 ..., n) being penalty factor, it is the constant between 0 to 1.

C. individuality is carried out selection operation

Selection operation is based upon on the basis that the fitness of individuality is estimated, and the individual inheritance that the present invention adopts roulette system of selection and optimum conversation strategy to choose from parent colony to have higher fitness is to colony of future generation.The fitness value of remembering individual i is Fit _i, its selected probability is:

$p_{\mathrm{is}}={\mathrm{Fit}}_i/\underset{j=1}{\overset{P}{\mathrm{\&Sigma;}}}{\mathrm{Fit}}_j---\left(9\right)$

At last, operate to determine each individual selected number of times with the simulation of the random number between 0-1 gambling dish.Because the roulette Select Error is bigger, so adopt optimum conversation strategy to guarantee the algorithm global convergence.Its way is: begin to find out individuality and minimum individuality and the best up to now individuality of fitness that fitness is the highest the current population from the first generation, replace the poorest individuality with best up to now individuality, thereby guarantee that best individuality can not destroy because of relatively poor, mutation operation.

D. individuality is carried out interlace operation

Produce new individual main method in the crossing operation formula genetic algorithm, the present invention adopts single-point intersection (as Fig. 5).At first, the P in the colony individual mode with at random partnered in twos, form

Individual combination; Then, the position of setting at random behind a certain locus is the point of crossing, and then every pair of individuality has L-1 possible point of crossing; At last according to crossover probability p _cExchange each to the chromosome of individuality behind the point of crossing, generate new individual; New individuality is assessed, abandoned the new individuality that does not satisfy constraint condition, this is carried out interlace operation again to individuality, until generating qualified new individuality.

E. individuality is carried out mutation operation

Mutation operation changes some genic value on the individual chromosome coded strings with less probability, can improve the local search ability of genetic algorithm by mutation operation, keeps the diversity of population.In to the design of the parameter optimization of heating radiator, adopt even mutation operation, specify successively that each locus is a change point in the coded strings of decision variable.For binary-coded individuality, according to the variation Probability p _mWhether the gene of judging each change point successively morphs, and generates new individuality (as shown in Figure 6) if morph then " 0 " to be replaced with " 1 " or " 1 " is replaced with " 0 "; Also to assess at last, abandon the new individuality that does not satisfy constraint condition, this is carried out mutation operation again to individuality, until generating qualified new individuality new individuality.

F. utilize among the b ideal adaptation degree evaluation index that new individuality is estimated, and judge whether to reach maximum evolutionary generation G

Utilize among the b ideal adaptation degree evaluation index that new individuality is estimated, and judge whether to reach maximum evolutionary generation G.If evolutionary generation T＜G repeats b, c, d, the e step is proceeded to evolve; If T=G then satisfies the evolution termination condition, finish to evolve output heating radiator parameter optimization result.Obtaining the chip cooling amount as example is Q=10W, and the utilization genetic algorithm is carried out evolution curve such as Fig. 7 that integration objective is optimized to the embodiment of the invention during weight coefficient a=0.5.

Below invention has been described in conjunction with specific embodiments, but should be understood that, the invention is not restricted to above-mentioned specifically described embodiment, on the contrary, under the prerequisite that does not depart from the scope of the present invention with spirit, can carry out various distortion, replacement and/or correction to the present invention, these distortion, replace and/or revise the scope of the present invention that all belongs to appended claims and limited.

Claims (4)

1. the method for the optimal design parameter of a definite heating radiator is characterized in that comprising:

Thermodynamic behaviour parameter according to described heating radiator, in the solution space of described Design of for heat sinks parameter value, determine to make one group of parameter combinations of the value minimum of an integrated objective function (M (X)), the dynamic response characteristic of described integrated objective function and described heating radiator and the positive correlation of droop loss characteristic

Wherein, the processing of one group of parameter combinations of the described value minimum of determining to make described integrated objective function comprises:

Described integrated objective function is carried out parameter optimization, thereby is met the optimal design parameter of heating radiator comprehensive evaluation index,

Described dynamic response characteristic comprises the time constant (τ) of the temperature variation speed that reflects described heating radiator, and described droop loss characteristic comprises that the heat eliminating medium that is adopted flows through the droop loss of described heating radiator (Δ p),

Described integrated objective function has following form:

M(X)＝a·τ/τ _max+(1-a)·Δp/Δp _max

Describedly thereby described integrated objective function is carried out the processing that parameter optimization is met the optimal design parameter of heating radiator comprehensive evaluation index can be expressed as:

$\left\{\begin{array}{cc}\min M\left(X\right)=& a\&CenterDot;\mathrm{\&tau;}/{\mathrm{\&tau;}}_{\max }+(1-a)\&CenterDot;\mathrm{\&Delta;p}/{\mathrm{\&Delta;p}}_{\max }\\ \mathrm{subject}& \mathrm{to}:x_{1\min }<x_1<x_{1\max },x_{2\min }<x_2<x_{2\max }\\ Q_{\min }\&le;Q& \end{array}\right.$

Wherein,

τ represents to reflect the time constant of the temperature variation speed of described heating radiator,

Δ p represents that the heat eliminating medium that is adopted flows through the droop loss of described heating radiator,

x ₁Representative flows to the rib spacing (d of direction along cooling fluid ₁),

x _1maxExpression x ₁Maximal value ₁,

x ₂Expression is along the rib spacing (d that flows to direction perpendicular to cooling fluid ₂),

x _2maxExpression x ₂Maximal value,

Q _MinThe heat dissipation capacity of representing the thermal source that described heating radiator is used for,

0＜a＜1st, weight coefficient, it is a constant,

τ _MaxThe time constant of base plate of radiator during expression heating radiator apleuria,

Δ p _MaxThe maximal value of expression pressure drop Δ p.

The described processing that described integrated objective function is carried out parameter optimization comprises:

Obtain making the pareto of heating radiator comprehensive evaluation index optimum to separate, thereby be met the optimal design parameter of heating radiator comprehensive evaluation index.

2. according to the method for the optimal design parameter of definite heating radiator of claim 1, it is characterized in that:

Described heating radiator has along the longitudinal forms array with transversely arranged a plurality of fins,

The described thermodynamic behaviour parameter of described heating radiator comprises the type and the cooling condition of the height of the size of the base plate of described heating radiator, described fin, described heating radiator and the material of fin, described heat eliminating medium,

Described optimal design parameter comprises that described fin flows to the rib spacing (x of direction along cooling fluid ₁) and/or flow to the rib spacing (x of direction perpendicular to cooling fluid ₂).

3. according to the method for the optimal design parameter of definite heating radiator of claim 2, thereby it is characterized in that the processing that the described pareto that obtains making heating radiator comprehensive evaluation index optimum separates the optimal design parameter that is met the heating radiator comprehensive evaluation index further comprises:

A. generate initial population, its population size is P, comprises adopting the binary cascade coding method that the feasible solution that makes heating radiator comprehensive evaluation index optimum is encoded, and generating P group chromosome length is the individual chromosome combination of 2L; Determine corresponding coding/decoding method then, obtain individual phenotype;

B. determine ideal adaptation degree evaluation method, comprise the value of calculating each individual pairing described integrated objective function (M (X)) and determine its fitness value (Fit);

C. carry out selection operation (selection operator), comprise according to selecting probability (p _s), according to the principle of " survival of the fittest " from t for selecting the higher individuality of fitness the P of colony (t) as the parent individuality, carry out the genetic manipulation of P (t+1) for colony; Use optimum conversation strategy, the individuality that fitness is the highest in the population is preserved be genetic directly to the next generation;

D. carrying out interlace operation (crossover operator) comprises to the pairing in twos at random of parent individuality, according to crossover probability (p _c), adopt the chromosome dyad between a pair of individuality of single-point cross exchange, generate new individuality;

E. carry out mutation operation (mutation operator), comprise according to variation probability (p _m), adopt the genic value on individual certain some locus of the change that evenly makes a variation, thereby generate new individuality;

F. utilize ideal adaptation degree evaluation index that new individuality is estimated, and judge whether to reach maximum evolutionary generation (G); If "No" is repeating said steps b, c, d, e then, proceed to evolve; If "Yes" then finishes to evolve.

6. according to the method for the optimal design parameter of definite heating radiator of claim 4, it is characterized in that:

Described population size P value is 80;

Described L value is 10;

Described crossover probability is between 0.66-0.8;

Described variation probability is between 0.15-0.24.

Images