CN114611239A - Design method and application of LED chip radiator adopting hierarchical genetic evolution - Google Patents

Design method and application of LED chip radiator adopting hierarchical genetic evolution Download PDF

Info

Publication number
CN114611239A
CN114611239A CN202210241788.XA CN202210241788A CN114611239A CN 114611239 A CN114611239 A CN 114611239A CN 202210241788 A CN202210241788 A CN 202210241788A CN 114611239 A CN114611239 A CN 114611239A
Authority
CN
China
Prior art keywords
population
individuals
radiator
heat
led chip
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210241788.XA
Other languages
Chinese (zh)
Inventor
潘中良
陈翎
李炜
陈倩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
South China Normal University
Original Assignee
South China Normal University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by South China Normal University filed Critical South China Normal University
Priority to CN202210241788.XA priority Critical patent/CN114611239A/en
Publication of CN114611239A publication Critical patent/CN114611239A/en
Pending legal-status Critical Current

Links

Images

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
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L33/00—Semiconductor devices having potential barriers specially adapted for light emission; Processes or apparatus specially adapted for the manufacture or treatment thereof or of parts thereof; Details thereof
    • H01L33/48—Semiconductor devices having potential barriers specially adapted for light emission; Processes or apparatus specially adapted for the manufacture or treatment thereof or of parts thereof; Details thereof characterised by the semiconductor body packages
    • H01L33/64—Heat extraction or cooling elements
    • H01L33/648—Heat extraction or cooling elements the elements comprising fluids, e.g. heat-pipes
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Geometry (AREA)
  • Biophysics (AREA)
  • Computer Hardware Design (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Microelectronics & Electronic Packaging (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Molecular Biology (AREA)
  • Computational Mathematics (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Genetics & Genomics (AREA)
  • Physiology (AREA)
  • Manufacturing & Machinery (AREA)
  • Power Engineering (AREA)
  • Cooling Or The Like Of Semiconductors Or Solid State Devices (AREA)

Abstract

The invention discloses a design method and application of an LED chip radiator adopting hierarchical genetic evolution. The invention carries out the geometric modeling of the radiator for a given power type LED chip; establishing an objective function for optimally designing the radiator according to the requirement of thermal design; solving the optimal solution of the objective function by using a self-adaptive hierarchical genetic evolution algorithm to obtain a radiator structure with better heat transfer characteristic; the algorithm is composed of two populations, a parallel competition and isolation mechanism is adopted in the evolution process, each population adopts a respective evolution strategy, the evolution algorithm can jump from a neighborhood of a local extreme value to a neighborhood of a global optimal solution on the whole, the search can be performed in the neighborhood of the global optimal solution with high precision, and the trapping of the evolution algorithm into the local optimal solution can be avoided to a certain extent. For the power type LED chip, the radiator designed by the invention has better performance and can be widely applied in practice.

Description

Design method and application of LED chip radiator adopting hierarchical genetic evolution
Technical Field
The invention belongs to the field of thermal design of LED chips, and relates to a design method and application of an LED chip radiator adopting hierarchical genetic evolution.
Background
An led (light Emitting diode) is a semiconductor light Emitting device, and when a forward current is applied to a PN junction of a semiconductor thereof, recombination of electrons and holes releases a part of energy in the form of light, thereby Emitting various types of light such as red, blue, green, and visible light. In recent years, LEDs have been widely used in many fields such as daily lighting, screen display, advertisement decoration, automobile brake lights and turn lights, because of their advantages of low power consumption, high brightness, long service life, environmental protection, etc.
In the LED, electrons cross the PN junction and diffuse from the N-type semiconductor to the P-region, and holes diffuse from the P-type semiconductor to the N-region, so that a potential barrier is formed at the PN junction to prevent the diffusion of electrons and holes due to the mutual diffusion of carriers, and an equilibrium state is achieved. If a forward bias voltage is applied to the PN junction, the P-type material is connected to the positive electrode, the N-type material is connected to the negative electrode, the potential barrier of the PN junction is lowered, electrons in the N region are injected into the P region, holes in the P region are injected into the N region, an unbalanced state is formed, the injected electrons and holes meet and are recombined at the PN junction, and redundant energy is released in the form of light, so that light is emitted. Due to different materials used by different LEDs, the energy levels occupied by electrons and holes of the LEDs are different; this difference in energy levels affects the photons generated when electrons recombine with holes, thus producing different wavelengths and different colors of light for different LEDs.
The LED belongs to a cold light source with high heating, the photoelectric conversion efficiency of the LED in the prior art only reaches 10% -20%, the rest is converted into heat, and the heat cannot be released completely by means of heat radiation. If the heat generated by an LED chip is concentrated inside the chip with a small size and cannot be effectively dissipated, the temperature of the whole chip is increased, which causes non-uniform distribution of thermal stress, and finally leads to reduction of the light emitting efficiency of the LED chip.
The variation of LED performance due to the increase of junction temperature is mainly reflected in the following aspects: the light emitting quantity of the LED is reduced, the wavelength of the light emitted by the LED is shifted, and the color of the light is shifted; shortening the life of the LED. Therefore, for the high-power LED, good heat dissipation treatment is carried out on the high-power LED, which is a key technical problem for applying the high-power LED to daily illumination and realizing industrialization.
For the heat dissipation of a high-power LED (i.e., a power LED), the heat dissipation can be handled by two ideas, i.e., increasing the electro-optic conversion efficiency of the LED, thereby reducing the electro-thermal conversion efficiency, and guiding the heat to the external environment through the LED package and the heat sink. For technical reasons, the latter is mainly used in current research and application, that is, a more reasonable heat dissipation scheme is selected and used by analyzing main factors and dissipation paths affecting the heat dissipation performance of the LED. The level of LED heat dissipation can be divided into package-level heat dissipation and device-level heat dissipation.
The current power type LED heat dissipation modes mainly include three modes, namely natural cooling, air cooling, heat pipe cooling and the like: (1) natural cooling is the simplest, economical and reliable method of heat dissipation, but the amount of heat dissipation is limited by the environmental space; if the radiator is designed and used, the radiating area can be increased on the whole, and the radiating efficiency is improved; (2) the air cooling is to add a fan in the passive heat dissipation structure to enhance the convection intensity of the LED and the ambient air; (3) the heat pipe cooling is to use a heat pipe to dissipate heat of the LED, and for example, a pulsating heat pipe, a loop heat pipe, a laminated heat pipe, or the like may be used to dissipate heat of the LED.
When a heat dissipation structure of a power type LED is designed, how to design a heat sink having good heat dissipation performance and small volume and mass of the heat sink is a key problem to be solved.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a design method of an LED chip radiator adopting hierarchical genetic evolution.
The invention further aims to provide application of the design method of the LED chip radiator adopting hierarchical genetic evolution.
The purpose of the invention is realized by the following technical scheme: a design method of an LED chip radiator adopting hierarchical genetic evolution comprises the following steps:
performing geometric modeling of the LED chip radiator on a given power type LED chip, namely determining the shape of the geometric structure of the radiator;
establishing an objective function for optimally designing the radiator according to the requirement of thermal design;
and (3) solving the optimal solution of the objective function by using a self-adaptive hierarchical genetic evolution algorithm to obtain the radiator structure with better heat transfer characteristic.
The LED chip radiator mainly comprises a base and at least one fin; at least one flat micro heat pipe is embedded in the fin.
The flat micro heat pipe comprises a heat pipe wall, a capillary core and an internal working medium; the capillary core and the internal working medium are arranged in the heat pipe wall.
The geometric modeling is to determine at least one of the following parameters of the LED chip heat radiator: the heat pipe comprises a heat radiator, a heat radiator base, a heat radiator, a heat pipe and a heat pipe, wherein the heat radiator comprises a heat radiator material density, a length of the heat radiator base, a width of the heat radiator base, a thickness of the heat radiator base, a height of fins, a thickness of the fins, the number of the fins, a mass of the heat pipe and the number of the heat pipes.
The thermal design requirements are preferably that the heat sink has good heat dissipation performance and has small volume and mass.
The optimization design is that FsIs minimized in order to minimize the junction temperature T of the LED chipJAnd the mass of the heat sink are minimized.
The objective function is
Figure BDA0003542494440000031
Wherein:
Figure BDA0003542494440000032
is a constant greater than 0; wherein T isJJunction temperature of the LED chip after the LED chip works to reach thermal equilibrium; gsIs the mass of the heat sink.
The self-adaptive hierarchical genetic evolution algorithm consists of two populations U (t) and L (t), a parallel competition and isolation mechanism is adopted in the evolution process, and each population U (t) and L (t) adopts a respective evolution strategy.
Preferably, the adaptive hierarchical genetic evolution algorithm performs migration of some individuals between populations u (t) and l (t), and selects μ individuals with higher fitness from population l (t) to replace μ individuals randomly selected from population u (t).
Preferably, the adaptive hierarchical genetic evolution algorithm is adaptively changeable and updatable in the evolution process, such as the population scale of the population u (t), the number of intersection points when two individuals perform intersection operation, and the mutation probability when individuals in the population perform mutation operation.
Preferably, the adaptive hierarchical genetic evolution algorithm firstly performs mutation operation on some individuals in the population L (t) and performs cross operation on other individuals to generate a new population Q (t); then, a selection operation is performed to select N individuals from the population L (t) and the population q (t), and the new generation population L (t +1) is formed by the N individuals.
The method for solving the optimal solution of the objective function by using the adaptive hierarchical genetic evolution algorithm comprises the following steps:
(1) representing a radiator structure by using a group of parameters of a radiator, and coding the parameters of the group of radiators to obtain an individual;
(2) carrying out genetic evolution on a population consisting of a plurality of individuals by using a self-adaptive hierarchical genetic evolution algorithm to obtain individuals with better performance; the radiator structure corresponding to the individual is the radiator structure with better heat dissipation performance;
fitness (fitness) of an individual X is defined as:
Figure BDA0003542494440000033
Fsof heat sink structures corresponding to individual X
Figure BDA0003542494440000034
The value of (c).
The adaptive hierarchical genetic evolution algorithm is specifically as follows:
s1, setting the value of parameter t equal to 0, i.e. t equal to 0; the number of individuals in both the initial populations U (0) and L (0) is N; randomly generating each individual of the initial populations U (0) and L (0);
s2, performing genetic evolution operation on individuals in the populations U (t) and L (t) respectively; this step is repeated ω times, where ω is a positive integer given in advance;
s3, transferring some individuals between the populations U (t) and L (t),
S4, randomly removing one individual from the current population U (t), and adding an optimal individual in the previous generation population U (t-1) into the current population U (t); similarly, randomly removing one individual from the current population L (t), and adding a best individual in the previous generation population L (t-1) to the current population L (t); thus, by this step, the best individuals in the population u (t) and the population l (t) always survive into the next generation population;
s5, if the termination condition is met, ending the execution of the algorithm; otherwise, set t: (t +1), go to step S2.
The numbers of individuals in the initial populations U (0) and L (0) described in step S1 are both N, where N is a positive integer given in advance.
The population u (t), t ═ 1,2, ·, in step S2, changes in its population size (i.e., the number of individuals in the population) as evolution progresses.
The population u (t) described in step S2 is obtained by conventional genetic operations such as selection, crossover and mutation, and the evolution process is similar to that of conventional genetic algorithms.
The population u (t) described in step S2 also uses three adaptive evolution strategies:
firstly, setting the number of individuals in an initial population U (0) as N; for t ═ 0,1,2, ·, the population size of each population U (t +1) is adaptively updated as follows:
Figure BDA0003542494440000041
wherein beta istAnd betat+1The population sizes of the populations U (t) and U (t +1), respectively; f. ofmaxIs the fitness of the optimal individual in the population U (t); f. ofavgIs the average fitness of all individuals in population U (t);
next, the number of intersections of the two individuals when performing the intersection operation is adaptively updated as follows:
Figure BDA0003542494440000042
wherein, KtAnd Kt+1The number of crossing points of the individuals in the populations U (t) and U (t +1), betatIs the population size of U (t), navgIs the average value of the individual numbers of the first three generations of populations U (t), U (t-1) and U (t-2);
thirdly, the mutation probability when mutation operation is carried out on the individuals in the population is adaptively updated as follows:
Figure BDA0003542494440000051
wherein λtAnd λt+1The mutation probabilities used by populations U (t) and U (t +1), respectively; the parameter ξ is a constant with a small value, for example, ξ is 0.001.
The population l (t), t ═ 1,2, ·, which does not change in size, is N in step S2.
(t) evolving the population l described in step S2 with algorithm 2; the algorithm 2 is specifically as follows:
p1, calculating the fitness of each individual in the current population L (t);
p2, carrying out mutation operation on some individuals in the population L (t), and carrying out cross operation on other individuals to generate a new population Q (t);
p3, selecting N individuals from the population L (t) and the population Q (t), and forming a new generation population L (t +1) by the N individuals;
p4, t: (t + 1). If the value of t is less than ω, then go to P1, otherwise terminate execution of the algorithm.
The mutation operation described in step P2 takes the following form: let X be mutated to generate a new individual X', assuming that they all have n components, i.e. X ═ X1,x2,···,xn),X′＝(x1′,x2′,···,xn'). The individual X is given the parameter variance vector θ for its variation.
θi′＝θi·exp(τ′·N(0,1)+τ·Ni(0,1))
xi′＝xi+N(0,θi′) i＝1,2,···,n.
Wherein, thetaiAnd xiRespectively mutated to new values thetai' and xi'; n (0,1) and Ni(0,1) are mutually independent standard normal distribution random variables, N (0, theta)i') is a normally distributed random variable;
Figure BDA0003542494440000052
the selecting operation described in step P3 is: firstly, directly selecting the individual with the maximum fitness in the populations L (t) and Q (t) and entering the next generation population; secondly, randomly selecting (N-2) multiplied by 0.2 individuals from the population L (t), and randomly selecting (N-2) multiplied by 0.8 individuals from the population Q (t); thirdly, the selected N individuals are formed into a new generation population L (t + 1).
In step S3, some individuals are migrated between populations u (t) and l (t) by the following specific method:
and (3) selecting the mu individuals with higher fitness from the population L (t) to replace the mu individuals randomly selected from the population U (t). Where mu is a positive integer having a value of N.times.0.3 or less, and N is the size of the population U (t).
The termination condition described in step S5 is that the optimal individual has been found or that the maximum number of evolutionary iterations has been reached.
The design method of the LED chip radiator adopting hierarchical genetic evolution is applied to the preparation of the LED chip radiator.
The principle of the invention is as follows:
generally, the overall structure of a power type LED system is generally composed of a plurality of parts, such as an LED chip, an aluminum-based PCB board, and a heat sink including aluminum fins. The aluminum-based PCB is composed of a copper-clad layer, a dielectric layer, an aluminum-based layer and the like, and the thickness, the heat conductivity and the like of the materials of all layers are different. If the small amount of heat dissipated on the surface of the LED chip is neglected, it is obvious that the most dominant transfer path of heat in the power type LED system is: the LED chip comprises an LED chip, an aluminum-based circuit board (an aluminum-based PCB board), a heat-conducting silica gel layer, an aluminum finned radiator and outside air.
In general, after the LED chip is in thermal equilibrium, its junction temperature T can be adjustedJExpressed as the following expression:
TJ＝TH+RE·P
here THDenotes the ambient temperature, REP is the thermal power, which is the thermal resistance of the LED chip itself plus the total thermal resistance after the thermal resistance of the package. Since the photoelectric conversion efficiency of an LED is low, it is generally considered that 80% to 85% of its input electric power is converted into heat energy.
Denote the thermal resistance of the package as RDR may beDFurther decomposition into two parts: rD＝RC+RZI.e. chip-level thermal resistance R in packageCAnd package-level thermal resistance R in packageZ. In order to reduce the operating temperature of the LED chip while ensuring the light-emitting quality and reliability, it is obvious that REShould be as small as possible. In the design and packagingWhen the LED chip is high in power, the common measures are as follows: firstly, the packaging structure in the LED chip is optimized, the structure layer is simplified, and the thermal resistance R between the LED chip and the radiator is reducedC(ii) a Secondly, the external radiator is improved, the heat dissipation capacity of the external radiator is improved, the interface thermal resistance between the LED module and the external radiator is reduced, and the thermal resistance R of the packaging part to the external environment is ensuredZAs low as possible, thereby improving the heat exchange capability of the radiator.
In order to improve the heat transfer capacity of the radiator, a plurality of flat micro heat pipes are embedded in the fins of the radiator. The structural block diagram of the heat pipe is shown in fig. 2, and the heat pipe mainly comprises a heat pipe wall, a capillary core, an internal working medium and the like, wherein the capillary core mainly functions to provide certain power for the backflow of the working medium. The heat pipe can be divided into an evaporation section, a heat insulation section, a condensation section and the like according to the state of the working medium in the heat pipe. Generally, an evaporation section is placed at a heat source with higher temperature, and at the moment, a liquid working medium is vaporized into a gas working medium (a vapor working medium) after being heated; the gas working medium is conveyed to the condensing section through the heat insulation section; the condensing section releases heat brought by the gas working medium to the environment where the condensing section is located, and the heat is liquefied and released, so that the gas working medium is converted into a liquid working medium. Then, the liquid working medium flows back to the evaporation section, and the process is continuously and circularly carried out.
The structure of the heat radiator is shown in figure 3, and mainly comprises a base and a plurality of fins, wherein the LED chip is placed below the base, is in contact with the lower plane of the base and is tightly attached to the lower plane of the base. Mass G for radiatorsIf the fin parts of the radiator are all considered as common rectangular section ribs, the specific expression is as follows:
Gs＝γ·(Hs·Ws+Hn·Wn·Ns)·Ls+Mr·Nr；
wherein gamma is the material density of the heat sink, LsAnd HsLength and width, W, of the base of the radiatorsIs the thickness of the base of the heat sink, HnIs the fin height, WnIs fin thickness, NsIs the number of fins, MrBeing a heat pipeMass, NrThe number of the heat pipes.
Order to
Figure BDA0003542494440000071
Wherein
Figure BDA0003542494440000072
Is a constant greater than 0. The heat radiator is optimally designed to FsIn order to minimize the junction temperature T of the LED chipJAnd the mass of the heat sink are minimized. In this regard, the present invention uses an adaptive hierarchical genetic evolution algorithm, Algorithm 1, to pair FsAnd (6) processing.
First, a heat sink structure is represented by an individual, and then the individual is encoded. An individual is mainly composed of a plurality of parts such as the length and the height of a radiator, the thickness, the height and the thickness of fins, the number of the fins, the number of heat pipes and the like, each part is coded by binary, and then the binary codes are combined to obtain the individual.
Secondly, carrying out genetic evolution on a population consisting of a plurality of individuals by using an algorithm 1 to obtain individuals with better performance, wherein the radiator structure corresponding to the individual is the radiator structure with better heat dissipation performance.
Third, fitness (fitness) of an individual X is defined as:
Figure BDA0003542494440000073
wherein, F in the above formulasOf heat sink structure corresponding to individual X
Figure BDA0003542494440000074
The value of (c).
Algorithm 1 is based on an adaptive hierarchical genetic evolution strategy, and consists of two populations u (t) and l (t), where u (t) can be regarded as a population in high-level evolution, l (t) can be regarded as a population in low-level evolution, and co-evolution is achieved by exchanging information between u (t) and l (t).
The number of individuals in both the initial populations U (0) and L (0) is N, where N is a positive integer given in advance. For the population u (t), t is 1,2, the population size (i.e. the number of individuals in the population) changes as the evolution progresses. For the population L (t), t is 1,2, the population scale of which is not changed and is N.
Algorithm 1
Step 1: setting the value of the parameter t equal to 0, namely t equal to 0; the number of individuals in both the initial populations U (0) and L (0) is N; each individual in the initial populations U (0) and L (0) is randomly generated.
Step 2: and (3) performing genetic evolution operation on individuals in the populations U (t) and L (t) respectively. This step is repeated ω times, where ω is a positive integer given in advance.
And step 3: some individuals are migrated between populations U (t) and L (t).
And 4, step 4: randomly removing one individual from the current population U (t), and adding an optimal individual in the previous generation population U (t-1) to the current population U (t). Similarly, one individual is randomly removed from the current population L (t), and an optimal individual in the previous generation population L (t-1) is added to the current population L (t). Thus, by this step, the best individuals in the population u (t) and the population l (t) always survive into the next generation population.
And 5: if the termination condition is met, ending the execution of the algorithm; otherwise, setting t: (t +1) and turning to the step 2.
In its entirety, algorithm 1 employs a mechanism of parallel competition and isolation for the evolutionary processes of populations u (t) and l (t). The populations U (t) and L (t) adopt their own respective evolutionary modes, which are different. The detailed implementation of algorithm 1 is as follows.
(1) For the population U (t) in the algorithm 1, conventional genetic operations such as selection, crossover and mutation are used, and the evolution process is similar to that of a conventional genetic algorithm. In step 2 of algorithm 1, the following three adaptive evolution strategies are also used for the population u (t).
First, the number of individuals in the initial population U (0) is set to N. For t ═ 0,1,2, ·, the population size of each population U (t +1) is adaptively updated as follows:
Figure BDA0003542494440000081
wherein beta istAnd betat+1The population sizes of the populations U (t) and U (t +1), respectively; f. ofmaxIs the fitness of the optimal individual in the population U (t); f. ofavgIs the average fitness of all individuals in the population u (t). By adopting the updating mode for the population scale, the population scale can be dynamically changed and is related to the fitness of the optimal individual in the population and the average fitness of all the individuals in the population.
Next, the number of intersections of the two individuals when performing the intersection operation is adaptively updated as follows:
Figure BDA0003542494440000082
wherein KtAnd Kt+1The number of crossing points of the individuals in the populations U (t) and U (t +1), betatIs the population size of U (t), navgIs the average of the individual numbers of the first three generations of populations U (t), U (t-1) and U (t-2). By using this update manner of the number of intersections, when the number of individuals in the current population U (t +1) is smaller than the average value of the numbers of individuals in the population of the previous three generations, the number of intersections at the time of the intersection operation will increase.
Thirdly, the mutation probability when mutation operation is carried out on the individuals in the population is adaptively updated as follows:
Figure BDA0003542494440000083
wherein λtAnd λt+1Are respectively a populationThe mutation probabilities used for U (t) and U (t + 1); the parameter ξ is a constant with a small value, for example ξ ═ 0.001. By using this update of the mutation probability, the number of individuals having mutations (variations) can be increased with a relatively small population size.
(2) For the population L (t) in step 2 of the algorithm 1, the specific steps of the evolution mode thereof are shown in the algorithm 2.
Algorithm 2
The first step is as follows: calculating the fitness of each individual in the current population L (t).
The second step is that: performing mutation operation on some individuals in the population L (t), and performing cross operation on other individuals to generate a new population Q (t).
The third step: and (4) selecting N individuals from the population L (t) and the population Q (t), and forming a new generation population L (t +1) by the N individuals.
The fourth step: set t ═ t + 1. If the value of t is less than omega, the first step is carried out, otherwise, the execution of the algorithm is terminated.
The specific implementation of the mutation operation and the selection operation in the algorithm 2 will be described below.
In the second step of algorithm 2, a large fraction (e.g., 60%) of the individuals is selected for mutation by a deterministic method. When the selected individuals are subjected to mutation operation, each selected individual is subjected to mutation, and a new individual is generated. For the rest of individuals in the population, a crossover operation is used, namely, two-point crossover is adopted through pairwise pairing to generate new individuals.
The mutation operation is performed in the following way: let X be mutated to generate a new individual X', assuming that they all have n components, i.e. X ═ X1,x2,···,xn),X′＝(x1′,x2′,···,xn'). The individual X is given the parameter variance vector θ for its variation.
θi′＝θi·exp(τ′·N(0,1)+τ·Ni(0,1))
xi′＝xi+N(0,θi′) i＝1,2,···,n.
Here, θiAnd xiRespectively mutated into new values thetai' and xi'; n (0,1) and Ni(0,1) are standard normal distribution random variables, N (0, theta) independent of each otheri') is a normally distributed random variable;
Figure BDA0003542494440000091
when the selection operation is performed in the third step of the algorithm 2, firstly, the individual with the maximum fitness in the populations L (t) and Q (t) is directly selected and enters the next generation population; next, (N-2). times.0.2 individuals were randomly selected from the population L (t), and (N-2). times.0.8 individuals were randomly selected from the population Q (t). Thirdly, the selected N individuals constitute a new generation population L (t + 1).
(3) In step 3 of algorithm 1, the method of migrating some individuals between populations u (t) and l (t) is as follows.
And (3) selecting the mu individuals with higher fitness from the population L (t) to replace the mu individuals randomly selected from the population U (t). Where mu is a positive integer having a value of N.times.0.3 or less, and N is the size of the population U (t).
(4) For the termination condition in step 5 of algorithm 1, it is defined as: the optimal individual has been found or the maximum number of evolutionary iterations has been reached. Here, the maximum value of the number of evolutionary iterations of algorithm 1 is a predetermined constant, for example 600, and algorithm 1 ends after performing up to 600 iterative computations.
The algorithm 1 adopts an adaptive hierarchical genetic evolution strategy and consists of two populations U (t) and L (t), a parallel competition and isolation mechanism is adopted in the evolution process, and each population U (t) and L (t) adopts respective evolution modes, so that the algorithm can jump from a neighborhood of a local extreme value to a neighborhood of a global optimal solution on the whole, can search in the neighborhood of the global optimal solution with high precision, and can avoid falling into local optimal to a certain extent. A better individual obtained by using algorithm 1 corresponds to a better heat sink structure that achieves a lower junction temperature of the LED chip and a lower quality heat sink.
Compared with the prior art, the invention has the following advantages and effects:
when a heat dissipation structure of a power type LED is designed, how to design a heat sink having good heat dissipation performance and small volume and mass of the heat sink is a key problem to be solved. The invention provides a design method of an LED chip radiator adopting hierarchical genetic evolution, and the design method comprises the following steps that firstly, a plurality of flat micro heat pipes are embedded into fins of the radiator so as to improve the heat conduction efficiency; secondly, the volume, the quality and the like of the radiator are optimized by using a self-adaptive hierarchical genetic evolution algorithm, so that the designed radiator has better performance and can be widely applied in practice. The self-adaptive hierarchical genetic evolution algorithm is composed of two populations U (t) and L (t), a parallel competition and isolation mechanism is adopted in the evolution process, and each population U (t) and L (t) adopts a respective evolution mode, so that the evolution algorithm can jump from a neighborhood of a local extreme value to a neighborhood of a global optimal solution on the whole, can search in the neighborhood of the global optimal solution with high precision, and can avoid falling into local optimal to a certain extent.
Drawings
Fig. 1 is a flow chart of the calculation of the heat sink design for power LED chips according to the present invention.
Fig. 2 is a block diagram of a flat micro heat pipe used in the present invention.
Fig. 3 is a schematic view of the heat sink structure of the present invention.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but the present invention is not limited thereto.
The invention provides a design method of an LED chip radiator adopting hierarchical genetic evolution for a power type LED chip, and a calculation flow chart of the design method is shown in figure 1. For a given power type LED chip, firstly, geometric modeling of a radiator is carried out, namely the shape of the geometric structure of the radiator is determined; secondly, establishing an objective function for optimally designing the radiator according to the requirement of thermal design; thirdly, calculating the objective function by using a self-adaptive hierarchical genetic evolution algorithm, namely algorithm 1, and obtaining the radiator structure with better heat transfer characteristic. The specific implementation steps are as follows:
(1) the radiator mainly comprises a base and a plurality of fins, and the LED chip is placed below the base, is in contact with the lower plane of the base and is tightly attached to the lower plane of the base. A plurality of flat micro heat pipes are embedded in fins of the radiator to improve the heat conduction efficiency. For a given power type LED chip, establishing an objective function
Figure BDA0003542494440000111
Wherein
Figure BDA0003542494440000112
Is a constant greater than 0, TJIs the junction temperature, G, of the LED chip when it is in thermal equilibriumsIs the mass of the heat sink.
(2) Solving an objective function F using Algorithm 1sThe minimum value point of (2) is that an individual with better performance is obtained by carrying out genetic evolution on a population consisting of a plurality of individuals. The radiator structure corresponding to the individual is the radiator structure with better heat dissipation performance.
The following describes the specific implementation steps of the present invention by taking an LED chip as an example. The LED chip comprises 120 LED lamp beads, and the input electric power of each LED lamp bead is 1.5W. During packaging, the 120 lamp beads are divided into eight groups, each group of 15 lamp beads is connected in series, and then the eight groups are connected in parallel. When designing the heat sink of the LED chip, the requirements given by the design specifications are: the heat sink is designed such that the junction temperature of the entire LED chip is not higher than 80 ℃.
The invention relates to a radiator design for an LED chip, wherein a plurality of flat micro heat pipes are embedded in fins of the radiator. Considering that the phase transition process inside the micro heat pipe is complicated, the micro heat pipe is equivalent to a solid material with anisotropic thermal conductivity, wherein the thermal conductivity along the Y axis is 2500W/(m · K), and the thermal conductivity along the Z axis and the X axis is 500W/(m · K).
When implemented, the function F is processedsParameter (2) of
Figure BDA0003542494440000113
The value of (d) was taken to be 0.6. The length and the width of the radiator base are respectively set as: l iss＝0.6m，Hs0.45 m. Thickness W of base of heat sinksIs set as Ws0.01 m. The fins are made of aluminum materials, and the corresponding heat conductivity coefficient is 205W/(m.K).
For function FsHere, it is mainly operated on its following variables: fin height HnThickness W of finnNumber of fins NsAnd the number N of heat pipesr. This is done by using the heat conduction equation to build a thermal model of the heat sink and to calculate the junction temperature T of the LED deviceJAnd heat transfer from the heat sink. The specific form of the heat transfer equation is as follows: let T (x, y, z, T) denote the temperature value of medium D at time T and position (x, y, z) for the heat conduction on a given medium. From the fourier law in heat transfer science, one can derive:
Figure BDA0003542494440000114
here, equation (1) is a heat conduction equation of a non-uniform isotropic medium, where k, ρ and c are a heat conduction coefficient, density and specific heat of the medium, respectively. If the medium is a homogeneous medium, i.e., k, ρ and c are all constants, equation (1) can be simplified to the following form:
Figure BDA0003542494440000121
wherein, alpha is the thermal diffusion coefficient,
Figure BDA0003542494440000122
the above equations (1) and (2) represent the heat conduction equation in the absence of a heat source. Some modifications to the heat source distribution of the media are required if they are considered. Assuming that the amount of heat generated per unit volume per unit time is g (x, y, z, t), the heat transfer equations (1) and (2) at this time should be rewritten as follows:
Figure BDA0003542494440000123
Figure BDA0003542494440000124
carrying out finite element mesh subdivision on the three-dimensional structure areas of the LED chip and the radiator, and then solving the formula (3) and the formula (4) by adopting a finite element method to calculate the junction temperature T of the LED chipJAnd heat transfer from the heat sink.
In algorithm 1, the maximum value of the number of iterations is set to 600, i.e., the operation is stopped when the number of iterations of the algorithm is greater than or equal to 600; the number N of individuals in the initial population U (0) is 20; the number N of individuals in the population L (t) is 20; setting the value of the parameter ω to 10; for the population U (t), when the cross operation is carried out, the cross probability is 0.95; when mutation operation is performed, the initial mutation probability is 0.001, and then the mutation probability lambda is calculatedtAnd (3) carrying out self-adaptive updating:
Figure BDA0003542494440000125
the value of parameter xi here is 0.001.
In step 3 of algorithm 1, the method of migrating some individuals between populations u (t) and l (t) is: for the parameter mu, its value is taken as 5, i.e. 5 individuals with higher fitness are selected from the population L (t) to replace the 5 individuals randomly selected from the population U (t).
Using these parameters as above, algorithm 1 is set, and through the operation of algorithm 1, the result of optimizing the radiator is: height of the fins: 42 mm; thickness of the fin: 2.2 mm; fin spacing: 7.8 mm; the number of the heat pipes is as follows: 21, the number of the cells is 21; junction temperature of the LED chip: 68.7 ℃. This result shows that by using the method provided by the present invention to design the heat sink for the LED chip, a heat sink structure can be obtained in which the junction temperature of the LED chip reaches a relatively small value, and the junction temperature of the LED chip of the obtained heat sink is only 68.7 ℃ which is lower than 80 ℃ required by the design specification. Therefore, the radiator structure obtained by using the method provided by the invention can better realize the heat dissipation of the LED chip and can maintain the temperature of the LED chip within a normal range.
The design of the heat sink was performed for the above example of the LED chip, and the result of the optimization calculation by using the conventional genetic algorithm (basic genetic algorithm) was as follows: the parameters of the conventional genetic algorithm are set as: the population scale (i.e. the number of individuals in the population) is 40, the crossover probability is 0.95, the variation probability is 0.001, and the maximum value of the iteration times is 6000; by performing calculations using conventional genetic operations such as selection, crossover, and mutation, the obtained result of optimizing the heat sink is: height of the fins: 48 mm; thickness of the fin: 2.1 mm; fin spacing: 7.7 mm; the number of the heat pipes is as follows: 15, the number of the cells is 15; junction temperature of the LED chip: 76.2 ℃. Comparing the result with the optimization result of the algorithm 1 on the radiator, it can be known that the junction temperature of the LED chip obtained by the algorithm 1 is 68.7 ℃ lower than the junction temperature of the LED chip obtained by the conventional genetic algorithm is 76.2 ℃, therefore, the performance of the algorithm 1 of the invention is superior to that of the conventional genetic algorithm when the radiator is designed on the LED chip, and the radiator structure obtained by the algorithm 1 can better realize the heat radiation on the LED chip.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.

Claims (10)

1. A design method of an LED chip radiator adopting hierarchical genetic evolution is characterized by comprising the following steps: performing geometric modeling of the LED chip radiator on a given power type LED chip, namely determining the shape of the geometric structure of the radiator;
establishing an objective function for optimally designing the radiator according to the requirement of thermal design;
solving the optimal solution of the objective function by using a self-adaptive hierarchical genetic evolution algorithm to obtain a radiator structure with better heat transfer characteristic;
the self-adaptive hierarchical genetic evolution algorithm consists of two populations U (t) and L (t), a parallel competition and isolation mechanism is adopted in the evolution process, and each population U (t) and L (t) adopts a respective evolution strategy.
2. The method of claim 1, wherein the method comprises the steps of:
the LED chip radiator mainly comprises a base and at least one fin; at least one flat micro heat pipe is embedded in the fin;
the flat micro heat pipe comprises a heat pipe wall, a capillary core and an internal working medium; the capillary core and the internal working medium are arranged in the heat pipe wall;
the geometric modeling is to determine at least one of the following parameters of the LED chip heat radiator: the heat pipe comprises a heat radiator, a heat radiator base, a heat radiator, a heat pipe and a heat pipe, wherein the heat radiator comprises a heat radiator material density, a length of the heat radiator base, a width of the heat radiator base, a thickness of the heat radiator base, a height of fins, a thickness of the fins, the number of the fins, a mass of the heat pipe and the number of the heat pipes.
3. The method of claim 1, wherein the method comprises the steps of:
the optimization design is that FsThe value of (d) is minimized;
the objective function is
Figure FDA0003542494430000011
Wherein:
Figure FDA0003542494430000012
is a constant greater than 0; wherein T isJJunction temperature of the LED chip after the LED chip works to reach thermal equilibrium; gsIs the mass of the heat sink.
4. The method of claim 1, wherein the method comprises the steps of:
the self-adaptive hierarchical genetic evolution algorithm is characterized in that a plurality of individuals are migrated between populations U (t) and L (t), and mu individuals with higher fitness are selected from the population L (t) to replace mu individuals randomly selected from the population U (t).
5. The method of claim 1, wherein the method comprises the steps of:
the self-adaptive hierarchical genetic evolution algorithm can be self-adaptively changed and updated in the evolution process, such as the population scale of the population U (t), the number of cross points when two individuals are subjected to cross operation, the variation probability when the individuals in the population are subjected to variation operation, and the like.
6. The method of claim 1, wherein the method comprises the steps of:
the self-adaptive hierarchical genetic evolution algorithm firstly performs mutation operation on some individuals in the population L (t) and performs cross operation on other individuals to generate a new population Q (t); then, a selection operation is performed to select N individuals from the population L (t) and the population q (t), and the new generation population L (t +1) is formed by the N individuals.
7. The method of claim 1, wherein the method comprises the steps of:
the method for solving the optimal solution of the objective function by using the adaptive hierarchical genetic evolution algorithm comprises the following steps:
(1) representing a radiator structure by using a group of parameters of a radiator, and coding the parameters of the group of radiators to obtain an individual;
(2) carrying out genetic evolution on a population consisting of a plurality of individuals by using a self-adaptive hierarchical genetic evolution algorithm to obtain individuals with better performance; the radiator structure corresponding to the individual is the radiator structure with better heat dissipation performance;
fitness (fitness) of an individual X is defined as:
Figure FDA0003542494430000021
Fsof heat sink structures corresponding to individual X
Figure FDA0003542494430000022
A value of (d);
the adaptive hierarchical genetic evolution algorithm is specifically as follows:
s1, setting the value of parameter t equal to 0, i.e. t equal to 0; the number of individuals in both the initial populations U (0) and L (0) is N; randomly generating each individual in the initial populations U (0) and L (0);
s2, performing genetic evolution operation on individuals in the populations U (t) and L (t) respectively; this step is repeated ω times, where ω is a positive integer given in advance;
s3, migrating some individuals between the populations U (t) and L (t);
s4, randomly removing one individual from the current population U (t), and adding an optimal individual in the previous generation population U (t-1) into the current population U (t); similarly, randomly removing one individual from the current population L (t), and adding a best individual in the previous generation population L (t-1) to the current population L (t); thus, by this step, the best individuals in the population u (t) and the population l (t) always survive into the next generation population;
s5, if the termination condition is met, ending the execution of the algorithm; otherwise, set t: (t +1), go to step S2.
8. The method of claim 7, wherein the method comprises the steps of:
the population u (t) described in step S2 also uses three adaptive evolution strategies:
firstly, setting the number of individuals in an initial population U (0) as N; for t-0, 1,2, …, the population size of each population U (t +1) is adaptively updated as follows:
Figure FDA0003542494430000031
wherein beta istAnd betat+1The population sizes of the populations U (t) and U (t +1), respectively; f. ofmaxIs the fitness of the optimal individual in the population U (t); f. ofavgIs the average fitness of all individuals in population U (t);
next, the number of intersections of the two individuals when performing the intersection operation is adaptively updated as follows:
Figure FDA0003542494430000032
wherein, KtAnd Kt+1The number of crossing points of the individuals in the populations U (t) and U (t +1), betatIs the population size of U (t), navgIs the average value of the individual numbers of the first three generations of populations U (t), U (t-1) and U (t-2);
thirdly, the mutation probability when mutation operation is carried out on the individuals in the population is adaptively updated as follows:
Figure FDA0003542494430000033
wherein λtAnd λt+1Respectively, the populations U (t) and U(t +1) mutation probability used; the parameter xi is a constant with a small value;
(t) evolving the population l described in step S2 with algorithm 2; the algorithm 2 is specifically as follows:
p1, calculating the fitness of each individual in the current population L (t);
p2, carrying out mutation operation on some individuals in the population L (t), and carrying out cross operation on other individuals to generate a new population Q (t);
p3, selecting N individuals from the population L (t) and the population Q (t), and forming a new generation population L (t +1) by the N individuals;
p4, position t: (t + 1); if the value of t is less than ω, then go to P1, otherwise terminate execution of the algorithm.
9. The method of claim 8, wherein the method comprises the steps of:
the mutation operation described in step P2 takes the following form: let X be mutated to generate a new X', assuming that they all have n components, i.e. X ═ X1,x2,…,xn),X′＝(x1′,x2′,…,xn') to a host; assigning to the individual X a parameter of the variance vector θ for its variation;
θi′＝θi·exp(τ′·N(0,1)+τ·Ni(0,1))；
xi′＝xi+N(0,θi′)i＝1,2,…,n；
wherein, thetaiAnd xiRespectively mutated into new values thetai' and xi'; n (0,1) and Ni(0,1) are standard normal distribution random variables, N (0, theta) independent of each otheri') is a normally distributed random variable;
Figure FDA0003542494430000041
the selecting operation described in step P3 is: firstly, directly selecting the individual with the maximum fitness in the populations L (t) and Q (t) and entering the next generation population; secondly, randomly selecting (N-2) multiplied by 0.2 individuals from the population L (t), and randomly selecting (N-2) multiplied by 0.8 individuals from the population Q (t); thirdly, the selected N individuals form a new generation of population L (t + 1);
in step S3, some individuals are migrated between populations u (t) and l (t) by the following specific method: selecting mu individuals with higher fitness from the population L (t) to replace the mu individuals randomly selected from the population U (t); where mu is a positive integer having a value of N.times.0.3 or less, N being the size of the population U (t).
10. Use of the method of any one of claims 1 to 9 for designing an LED chip heat sink using hierarchical genetic evolution for the manufacture of an LED chip heat sink.
CN202210241788.XA 2022-03-11 2022-03-11 Design method and application of LED chip radiator adopting hierarchical genetic evolution Pending CN114611239A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210241788.XA CN114611239A (en) 2022-03-11 2022-03-11 Design method and application of LED chip radiator adopting hierarchical genetic evolution

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210241788.XA CN114611239A (en) 2022-03-11 2022-03-11 Design method and application of LED chip radiator adopting hierarchical genetic evolution

Publications (1)

Publication Number Publication Date
CN114611239A true CN114611239A (en) 2022-06-10

Family

ID=81863536

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210241788.XA Pending CN114611239A (en) 2022-03-11 2022-03-11 Design method and application of LED chip radiator adopting hierarchical genetic evolution

Country Status (1)

Country Link
CN (1) CN114611239A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117688803A (en) * 2023-11-27 2024-03-12 广东乐城科技有限公司 Design method and system of excimer discharge light source

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117688803A (en) * 2023-11-27 2024-03-12 广东乐城科技有限公司 Design method and system of excimer discharge light source

Similar Documents

Publication Publication Date Title
Bender et al. Solid-state lighting: A concise review of the state of the art on LED and OLED modeling
Ramos-Alvarado et al. Comparison and optimization of single-phase liquid cooling devices for the heat dissipation of high-power LED arrays
Bélanger et al. Thermoelectric generator sandwiched in a crossflow heat exchanger with optimal connectivity between modules
CN114611239A (en) Design method and application of LED chip radiator adopting hierarchical genetic evolution
Bing On thermal structure optimization of a power LED lighting
CN105320822A (en) Double-objective comprehensive optimization design method for LED radiator structure parameters
CN107944190A (en) High-power LED lamp life-span prediction method based on artificial neural network
Sahbaz et al. Thermal analysis and optimization of high power led armature
CN103827579A (en) Heat sink and lighting apparatus having same
Feng et al. Design of a novel LED bulb with entire surface thermally activated for passive cooling
CN111353231B (en) Genetic algorithm-based LED radiator design method and system
KR20130075742A (en) Led-light heatsink and led lamp
CN117195817A (en) Optimal heat dissipation layout method and heat dissipation system for chip
Zhu et al. Mathematical model and computation of heat distribution for LED heat sink
CN102620268B (en) Non-uniformly distributed semiconductor illumination radiator
Wenzl et al. White light quality of phosphor converted LEDs from a phosphor materials perspective of view: an evaluation based on combined thermal and optical simulations
Zhu et al. Modelling and numerical simulations of heat distribution for LED heat sink
CN101995008A (en) Ceramic radiator with conductive circuit
CN111159867B (en) Construction method of heat dissipation device, heat dissipation device and electronic equipment
Kıyak A Study on the Working Performance of Dimming Methods for Single‐and Multichip Power LEDs
CN209909625U (en) High-power LED lamp and radiator thereof
Paul et al. Modelling and thermal analysis of honey comb heat sink for LEDs in street lighting applications
CN201851917U (en) High-power LED (Light Emitting Diode) lamp conducting heat by using average-temperature plate
Gielen et al. Development of an intelligent integrated LED system-in-package
Huang et al. Study on packaging structure of high power multi-chip LED

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