VCSEL array oxidation aperture intelligent optimization design method

Abstract

The invention discloses a VCSEL (Vertical Cavity Surface Emitting Laser) array oxidation aperture intelligent optimization design method, which is used for improving the temperature distribution uniformity of a VCSEL array so as to improve the overall performance of a laser. According to the method, firstly, a BP neural network model is trained to replace a thermoelectric feedback model, the mapping relation between the oxidation aperture and the steady-state junction temperature is established, then the oxidation aperture of each unit of the VCSEL array is directly selected to carry out chromosome coding, a fitness function representing the temperature uniformity of the VCSEL array is constructed to provide an optimization standard, and the optimization accuracy of the temperature uniformity of the VCSEL array is improved. And finally, adopting a genetic algorithm to realize the optimal optimization design of the oxidized aperture of the VCSEL array with any shape and any number of units. According to the method, the labor cost and the time cost in seeking the optimal oxidation aperture design of the VCSEL array are effectively reduced. And finally, oxidation aperture optimization design is carried out by taking a hexagonal VCSEL array with 61 units as an example, and the effectiveness and practicability of the method are proved.

Description

Technical Field

[0001] The invention relates to the field of semiconductor lasers and intelligent computing technology, and in particular to a VCSEL array oxidation aperture intelligent optimization design method. Background Art

[0002] Vertical cavity surface emitting lasers (VCSELs) are widely used in communications, lidar, medical treatment, sensing and other fields due to their advantages such as low threshold current, high beam quality, stable single longitudinal mode operation and easy two-dimensional array integration.

[0003] Since the output optical power provided by a single-hole VCSEL is very limited, in order to increase the output optical power, a parallel structure of multiple small-sized VCSEL units with independent cavities is usually adopted, that is, a two-dimensional VCSEL array. However, in the VCSEL array, due to the uneven heat dissipation of each unit and the thermal coupling effect between each other, the junction temperature distribution of the VCSEL array will be uneven. The bias current with a positive temperature coefficient will also cause uneven current distribution, forming thermal photoelectric feedback, causing thermal reversal of the output optical power, thereby seriously limiting the thermal photoelectric performance of the VCSEL array.

[0004] Considering that the unit oxidation aperture size in the VCSEL array is one of the important factors that determine the size of its injection current, the oxidation aperture design will directly affect the temperature distribution of the VCSEL array. Therefore, the prior art usually adopts the method of reducing the unit oxidation aperture in the central area of ​​the VCSEL array to reduce thermal coupling, thereby improving the uniformity of temperature distribution. However, the above method currently only optimizes the array design by manually adjusting the VCSEL unit oxidation aperture. With the increase in the integration of VCSEL arrays, this method will consume a lot of manpower and time costs, has low accuracy and is difficult to design the oxidation aperture of large-scale arrays. In view of the above problems, there is an urgent need for an effective method for intelligently optimizing the oxidation aperture design of the VCSEL array to improve the uniformity of temperature distribution and improve the overall performance of the device.

[0005] As we all know, with the rapid development of artificial intelligence and deep learning, neural networks, as a powerful learning model, have been widely used in various complex prediction, classification, regression and other tasks. BP (Back Propagation) neural network is the most commonly used type of neural network. It is trained through multi-layer nonlinear transformation and back propagation algorithm, so that the network can gradually adjust parameters to approach the expected output.

[0006] At the same time, Genetic Algorithm (GA), as a global optimization algorithm that simulates the natural evolution process, is based on Darwin's theory of evolution and genetic principles. It searches for the global optimal solution by simulating operations such as selection, crossover and mutation of biological chromosomes, and is widely used in various optimization problems.

[0007] The present invention combines genetic algorithm with BP neural network and applies them together to the optimization design of VCSEL array oxidation aperture. Its main features are that a fast and accurate BP neural network model is used to replace the complex and time-consuming thermoelectric feedback model to obtain the temperature distribution to calculate the fitness function; through the genetic algorithm, the oxidation aperture of the VCSEL unit can be directly chromosomally encoded and operated, avoiding the limitation of the number of VCSEL array units; this method does not require the fitness function to be continuous or differentiable, and is not constrained by the VCSEL array modeling rules or special requirements, showing excellent global optimization capabilities. Therefore, BP neural network and genetic algorithm are very suitable for solving the optimization problem of VCSEL array oxidation aperture design. Summary of the invention

[0008] The purpose of the present invention is to solve the problems in the prior art of using traditional manual adjustment and design of the oxidation aperture of each unit of the VCSEL array, which brings huge manpower and time costs and cannot achieve the optimal design, and to provide a VCSEL array oxidation aperture intelligent optimization design method.

[0009] To achieve the above object, the present invention adopts the following technical solutions:

[0010] The present invention proposes a VCSEL array oxidation aperture intelligent optimization design method, which applies BP neural network and genetic algorithm to the oxidation aperture optimization design of VCSEL array. This method replaces the thermoelectric feedback model with the BP neural network model to obtain the mapping relationship between the oxidation aperture and the steady-state junction temperature, and establishes a fitness function that reflects the array temperature uniformity to provide an optimization standard. Then, a genetic algorithm is used to perform chromosome encoding on the oxidation aperture of a VCSEL array of any shape and any number of units. The gene of each chromosome corresponds to the oxidation aperture of a VCSEL unit, and each chromosome corresponds to an oxidation aperture design method. Finally, the optimal oxidation aperture optimization design is obtained through iteration. This method improves the optimization efficiency and greatly reduces the labor cost and time cost. The specific steps of a VCSEL array oxidation aperture intelligent optimization design method are as follows:

[0011] Step 1: Select the oxidation aperture R corresponding to each unit in the VCSEL array with n units to be optimized i(i=1,2,3,...,n) as the input variables of the BP neural network; the steady-state junction temperature of each unit in the VCSEL array As the output variable of BP neural network;

[0012] Step 2: The allowed variation range of the oxide aperture of each unit in the VCSEL array (R min ~R max ) randomly generates N groups of VCSEL array unit oxidation aperture sample values, each group of samples contains n oxidation aperture values ​​R i =R min +rand()·(R max -R min ), where Rand() represents a random number between 0 and 1;

[0013] Step 3: Set the n oxidation pore size values ​​R in each group of samples i The temperature distribution is simulated in the VCSEL array thermoelectric feedback model to obtain the corresponding steady-state junction temperature T i (i=1,2,3,...,n), thereby generating a data set for the BP neural network;

[0014] Step 4: Use the data set obtained in step 3 to train and test the BP neural network, wherein 80% of the samples in the data set are randomly selected as the training data of the neural network, and the remaining 20% ​​are used as the test data, until the error between the predicted value of the neural network and the sample value is limited to the allowable error range of 1K, and the construction of the BP neural network is completed;

[0015] Step 5: Let the oxide aperture of the jth VCSEL unit be R j , the allowed variation range of the oxide aperture of each unit in the VCSEL array (R min ~R max ) generates M chromosomes, the number of which is between 100 and 200, and each chromosome contains n oxidation pore genes that satisfy: R j =R min +rand()·(R max -R min ), where rand() represents a random number between 0 and 1, and each chromosome represents an oxidation aperture optimization method, constituting the initial population;

[0016] Step 6: Import each chromosome in the population into the BP neural network constructed in step 4 to obtain the steady-state junction temperature of each unit. Constructing the fitness function L that reflects the temperature uniformity of the VCSEL array p , that is, using temperature variance as the evaluation criterion to calculate the fitness value of the chromosome;

[0017] Step 7: Perform a selection operation on the population based on the reverse ranking according to the fitness value, and select the top M chromosomes for inheritance;

[0018] Step 8: Determine whether the genetic algorithm has reached the maximum number of iterations. If so, execute step 12; otherwise, execute step 9. The number of iterations is between 500 and 1000.

[0019] Step 9: For all chromosomes selected in step 7, perform a one-point crossover operation, select two chromosomes for crossover in the order of crossover probability, the crossover point is a random integer from 2 to n-1, and add the new chromosome generated after crossover to the population, the crossover probability is between 0.4 and 0.6;

[0020] Step 10: For all chromosomes in step 9, a mutation operation based on the oxidation aperture gene point is performed, one chromosome is selected according to the mutation probability for random oxidation aperture gene mutation, and the mutated new chromosome is added to the population, and the mutation probability is between 0.01 and 0.1;

[0021] Step 11: Generate a population of offspring chromosomes, return to step 6, import the BP neural network to calculate the fitness value of the chromosome, and start the next round of genetic algorithm iterative optimization;

[0022] Step 12: The genetic algorithm iteration is completed and the chromosome ranked first is extracted;

[0023] Step 13: Decode the chromosome to obtain the optimal VCSEL array oxidation aperture size information;

[0024] Step 14: Import the optimal VCSEL array oxidation aperture size information into the thermoelectric feedback model to simulate the temperature distribution;

[0025] Step 15: The optimized steady-state temperature distribution of the VCSEL array is obtained, and the optimization is completed.

[0026] The technical problem to be solved by the present invention can also be further achieved by the following technical solution. In step 3, the thermoelectric feedback model simulating the temperature distribution includes the following steps:

[0027] (1) Using ANSYS to establish a VCSEL array structure model based on the oxidation aperture sample value or the oxidation aperture information corresponding to the optimal chromosome and structural parameters including VCSEL array shape, center-to-center distance between adjacent units, and substrate size;

[0028] (2) Setting the operating conditions of the VCSEL array including injection current and ambient temperature;

[0029] (3) Simulate the temperature distribution of the VCSEL array and extract the peak junction temperature T M0;

[0030] (4) Use MATLAB to calculate the heat flux density of each unit in the VCSEL array;

[0031] (5) Extract the heat flux density in (4) and load it into the VCSEL array. Use ANSYS to simulate the temperature distribution of the VCSEL array and extract the peak junction temperature T M1 ;

[0032] (6) Judgement |T M1 -T M0 |Is it less than 1K? If it is less than 1K, execute step (7); otherwise, make T M0 =T M1 , return to step (4);

[0033] (7) Output the steady-state junction temperature T1, T2, T3, ..., T of each unit in the VCSEL array n .

[0034] The step 4 of constructing the BP neural network model includes the following steps:

[0035] (1) Establish a three-layer BP neural network structure including input layer, hidden layer and output layer, where the number of neurons in the input layer is n, and the input variables are R1, R2, R3, ..., R n ; The number of neurons in the output layer is n, and the output variable is The number of neurons in the hidden layer is q is a constant (q∈[1,10]);

[0036] (2) The tansig function is used as the activation function of the hidden layer, and the purelin linear function is used as the activation function of the output layer:

[0037]

[0038] purelin(x)=x

[0039] (3) Use the mean square error function to calculate the error of the BP neural network:

[0040]

[0041] in, is the steady-state junction temperature predicted by the BP neural network, T i is the sample steady-state junction temperature corresponding to the test data in the data set;

[0042] (4) Use the gradient descent method trainlm function to train the BP neural network.

[0043] The fitness function L in step 6P , select the VCSEL array temperature uniformity as the evaluation criterion, use the variance of the VCSEL array temperature to calculate, the fitness function of the pth chromosome is:

[0044]

[0045] Where n is the number of units in the VCSEL array, is the steady-state junction temperature of the j-th VCSEL unit predicted by the BP neural network.

[0046] In the optimization design of the oxidation aperture of the VCSEL array, as the integration of the array increases, the number of units continues to increase. The traditional single-tube VCSEL oxidation aperture design ideas and the method of manually designing the oxidation aperture of the VCSEL unit will no longer be practical. The present invention adopts BP neural network and genetic algorithm to realize automatic oxidation aperture design of the VCSEL array, and the complex and cumbersome optimization process is handed over to the computer iteration, which greatly saves manpower and time costs, and can obtain the optimal oxidation aperture design method. At the same time, the present invention directly selects the oxidation aperture information of the VCSEL array to encode the chromosome, which can realize the optimization of the VCSEL array with any number of units, and solves the problem of difficulty in manual design caused by the increase in the number of VCSEL units. In addition, during the iterative process of the genetic algorithm, the fitness of each chromosome is calculated by predicting the temperature through the BP neural network model, and the lengthy simulation process of the thermoelectric feedback model is replaced by a fast and accurate calculation method, making the genetic algorithm iteration and the optimization results of VCSEL more efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The objects and advantages of the present invention may be further understood from the following description taken in conjunction with the accompanying drawings. In these drawings:

[0048] Figure 1 An example of a flow chart of a VCSEL array oxidation aperture intelligent optimization design method proposed by the present invention is provided;

[0049] Figure 2 The chromosome genetic process diagram of the genetic algorithm of the present invention is illustrated;

[0050] Figure 3 An example of a conventional hexagonal uniformly oxidized aperture VCSEL array structure is shown;

[0051] Figure 4 An example of an array structure diagram after optimizing the oxidation aperture according to an embodiment of the present invention is shown;

[0052] Figure 5 The temperature distribution diagram of a conventional hexagonal uniformly oxidized aperture VCSEL array is illustrated;

[0053] Figure 6An example of array temperature distribution after optimizing the oxidation aperture according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0054] In order to more clearly demonstrate the purpose, technical solutions and advantages of the present invention, the contents of the present invention are described in detail with reference to the accompanying drawings, taking a hexagonal VCSEL array of 61 units as an example. It should be pointed out that the embodiments described herein are only a part of the present invention and not all of them. The present invention is applicable to the optimization of oxidation apertures of VCSEL arrays of any shape and any number of units. On the technical basis of the present invention, any other implementation methods that can be obtained by any technician in this field without creative labor, as well as improvements or deformations made without departing from the technical principles of the present invention, all belong to the protection scope of the present invention.

[0055] The present invention proposes a VCSEL array oxidation aperture intelligent optimization design method, the process is as follows Figure 1 As shown, the following steps are included:

[0056] Step 1: Select the oxidation aperture R1, R2, R3, ..., R of each unit in the hexagonal VCSEL array to be optimized with 61 units. 61 As the input variable of the neural network; the steady-state junction temperature of each unit in the VCSEL array As the output variable of BP neural network;

[0057] Step 2: The allowed variation range of the oxide aperture of each unit in the VCSEL array (R min ~R max ) is a random generation of N = 500 groups of VCSEL array unit oxidation aperture sample values ​​within 5 μm to 10 μm, and each group of samples contains 61 oxidation aperture values ​​R i =5+rand()·5, where rand() represents a random number between 0 and 1;

[0058] Step 3: Sequentially calculate the 61 oxidation pore size values ​​R in each group of samples. i Import the VCSEL array thermoelectric feedback model to simulate the temperature distribution and obtain the corresponding steady-state junction temperature T i (i=1,2,3,...,61), thereby generating a data set for the BP neural network;

[0059] Step 4: Use the data set obtained in step 3 to train and test the BP neural network, wherein 80% of the samples in the data set, totaling 400 groups, are randomly selected as the training data of the neural network, and the remaining 20%, totaling 100 groups, are used as the test data, until the error between the predicted value of the neural network and the sample value is limited to the allowable error range of 1K, and the construction of the BP neural network is completed; in this embodiment, after the BP neural network is trained, the mean square error MSE is 2.4346e-03, and the training effect is good;

[0060] Step 5: Let the oxide aperture of the jth VCSEL unit be R j , M=100 chromosomes are generated within the allowed variation range of 5 μm to 10 μm of the oxidation aperture of each unit in the VCSEL array, such as Figure 2 As shown in (a), chromosome encoding is performed, where each chromosome contains 61 oxidation pore genes that satisfy: R j =5+rand()·5, where rand() represents a random number between 0 and 1, and each chromosome represents an oxidation aperture optimization method, constituting the initial population;

[0061] Step 6: Import each chromosome in the population into the BP neural network constructed in step 4 to obtain the steady-state junction temperature of each unit. Constructing the fitness function L that reflects the temperature uniformity of the VCSEL array p , that is, using temperature variance as the evaluation criterion to calculate the fitness value of the chromosome, the fitness function of the pth chromosome is:

[0062]

[0063] Wherein, n is the total number of VCSEL array units, and in this embodiment, n is 61. is the steady-state junction temperature of the j-th VCSEL unit predicted by the BP neural network.

[0064] Step 7: According to the fitness value, the population is selected based on the reverse ranking. When the chromosome fitness value is smaller, it means that its temperature variance is smaller and the uniformity is better. The top 100 chromosomes are selected for inheritance;

[0065] Step 8: Determine whether the genetic algorithm has reached the maximum number of iterations of 1000 times. If so, execute step 12; otherwise, execute step 9;

[0066] Step 9: For all chromosomes selected in step 7, Figure 2 As shown in (b), a one-point crossover operation is performed, where two chromosomes are selected for crossover in the order of crossover probability, the crossover point is a random integer from 2 to 60, and the new chromosome generated after the crossover is added to the population, with a crossover probability of 0.5;

[0067] Step 10: For all chromosomes in step 9, perform mutation operations based on the oxidation aperture gene point, such as Figure 2 As shown in (c), a chromosome is selected with a mutation probability of 0.1 to perform random oxidation aperture gene mutation, and the mutated new chromosome is added to the population;

[0068] Step 11: Generate offspring chromosome population, return to step 6, import BP neural network, calculate fitness value, and start the next round of genetic algorithm iterative optimization. The parameters of the example genetic algorithm are shown in the following table:

[0069] parameter Numeric The number of unit oxidation pores to be optimized is the number of genes n 61 Chromosome number M 100 Crossover probability 0.5 Mutation probability 0.1 <![CDATA[Minimum oxidation aperture R min > 5μm <![CDATA[Maximum oxidation aperture R max > 10μm Gene and variant manipulation range 5μm~10μm Maximum number of iterations 1000

[0070] Step 12: The genetic algorithm iteration is completed and the chromosome ranked first is extracted;

[0071] Step 13: Decode the chromosome to obtain the optimal VCSEL array oxidation aperture size information, such as Figure 4 As shown;

[0072] Step 14: Import the optimal VCSEL array oxidation aperture size information into the thermoelectric feedback model to simulate the temperature distribution;

[0073] Step 15: Get the optimized VCSEL array steady-state temperature distribution as Figure 6 As shown, the optimization is finished.

[0074] For optimization comparison, the present invention compares the conventional equidistant non-optimized hexagonal VCSEL array with equal oxidation apertures of each unit with the optimized VCSEL array, and keeps the sum of the oxidation aperture areas of each unit of the VCSEL array before and after optimization constant, and obtains the non-optimized uniform oxidation aperture of 8.15 μm. Figure 3 As shown, the center-to-center distance between adjacent units of the conventional hexagonal VCSEL array is 52 μm, the radius of the oxidation hole and the light exit hole of all units is 8.15 μm, the radius of the VCSEL table is 18.15 μm, the substrate size is 260 μm, and the temperature distribution is simulated by the thermoelectric feedback model. In addition, the neural network training and test data sets in step 3 and step 14 in the embodiment of the present invention are also calculated and obtained by this method, and the steps are as follows:

[0075] (1) Obtain the uniform oxidation aperture information or oxidation aperture sample values ​​of the unoptimized hexagonal VCSEL array and the structural parameters including the VCSEL array shape, the center-to-center distance between adjacent units, and the substrate size, and use ANSYS to establish the VCSEL array structure model.

[0076] (2) Set the injection current of the VCSEL array to 427 mA and the ambient temperature to 300 K;

[0077] (3) Simulate the temperature distribution of the VCSEL array and extract the peak junction temperature T M0 ;

[0078] (4) Use MATLAB to calculate the heat flux density of each unit in the VCSEL array;

[0079] (5) Extract the heat flux density in (4) and load it into the VCSEL array. Use ANSYS to simulate the temperature distribution of the VCSEL array and extract the peak junction temperature T M1 ;

[0080] (6) Judgement |T M1 -T M0 |Is it less than 1K? If it is less than 1K, execute step (7); otherwise, make T M0 =T M1 , return to step (4);

[0081] (7) Output the steady-state junction temperature T1, T2, T3, ..., T of each unit in the VCSEL array n .

[0082] In step 4 described in the embodiment of the present invention, constructing the BP neural network model includes the following steps:

[0083] (1) A three-layer BP neural network structure consisting of an input layer, a hidden layer, and an output layer was established. The number of neurons in the input layer was 61, and the input variables were R1, R2, R3, ..., R 61 , the number of neurons in the output layer is 61, and the output variable is The number of neurons in the hidden layer is 20;

[0084] (2) The tansig function is used as the activation function of the hidden layer, and the purelin linear function is used as the activation function of the output layer:

[0085]

[0086] purelin(x)=x

[0087] (3) Use the mean square error function to calculate the error of the BP neural network:

[0088]

[0089] in, is the steady-state junction temperature predicted by the BP neural network, T i is the sample steady-state junction temperature corresponding to the test data in the data set;

[0090] (4) Use the gradient descent method trainlm function to train the BP neural network.

[0091] Through the above steps, we obtain the temperature distribution diagram of the conventional hexagonal VCSEL array and the temperature distribution diagram after optimization in the embodiment of the present invention, which are respectively Figure 5 and Figure 6 As shown. It can be found that before optimization, the temperature of the central unit of the VCSEL array is high and the temperature distribution is uneven. After optimizing the oxidation aperture, the temperature distribution is greatly improved. At the same time, the temperature uniformity (variance) before optimization is 22.8374, and the temperature uniformity after optimization is 12.3553. By comparison, the embodiment of the present invention improves the temperature uniformity by 45.90% compared with the conventional array, and the light output power is increased by 2mW. In summary, after the optimization of the present invention, the thermal problem of the VCSEL array is effectively improved, thereby improving the overall performance of the device, and illustrating the effectiveness of the present invention.

Claims

1. A VCSEL array oxidation aperture intelligent optimization design method, characterized in that the steps include: Step 1: Select the oxidation aperture R corresponding to each unit in the VCSEL array with n units to be optimized i (i=1,2,3,...,n) as the input variables of the BP neural network; the steady-state junction temperature of each unit in the VCSEL array As the output variable of BP neural network; Step 2: The allowed variation range of the oxide aperture of each unit in the VCSEL array (R min ~R max ) randomly generates N groups of VCSEL array unit oxidation aperture sample values, each group of samples contains n oxidation aperture values ​​R i =R min +rand()·(R max -R min ), where rand() represents a random number between 0 and 1; Step 3: Set the n oxidation pore size values ​​R in each group of samples i The temperature distribution is simulated in the VCSEL array thermoelectric feedback model to obtain the corresponding steady-state junction temperature T i (i=1,2,3,...,n), thereby generating a data set for the BP neural network; Step 4: Use the data set obtained in step 3 to train and test the BP neural network, wherein 80% of the samples in the data set are randomly selected as the training data of the neural network, and the remaining 20% ​​are used as the test data, until the error between the predicted value of the neural network and the sample value is limited to the allowable error range of 1K, and the construction of the BP neural network is completed; Step 5: Let the oxide aperture of the jth VCSEL unit be R j , the allowed variation range of the oxide aperture of each unit in the VCSEL array (R min ~R max ) generates M chromosomes, the number of which is between 100 and 200, and each chromosome contains n oxidation pore genes that satisfy: R j =R min +rand()·(R max -R min ), where rand() represents a random number between 0 and 1, and each chromosome represents an oxidation aperture optimization method, constituting the initial population; Step 6: Import each chromosome in the population into the BP neural network constructed in step 4 to obtain the steady-state junction temperature of each unit. Constructing the fitness function L that reflects the temperature uniformity of the VCSEL array p , that is, using temperature variance as the evaluation criterion to calculate the fitness value of the chromosome; Step 7: Perform a selection operation on the population based on the reverse ranking according to the fitness value, and select the top M chromosomes for inheritance; Step 8: Determine whether the genetic algorithm has reached the maximum number of iterations. If so, execute step 12; otherwise, execute step 9. The number of iterations is between 500 and 1000. Step 9: For all chromosomes selected in step 7, perform a one-point crossover operation, select two chromosomes for crossover in the order of crossover probability, the crossover point is a random integer from 2 to n-1, and add the new chromosome generated after crossover to the population, the crossover probability is between 0.4 and 0.6; Step 10: For all chromosomes in step 9, a mutation operation based on the oxidation aperture gene point is performed, one chromosome is selected according to the mutation probability for random oxidation aperture gene mutation, and the mutated new chromosome is added to the population, and the mutation probability is between 0.01 and 0.1; Step 11: Generate a population of offspring chromosomes, return to step 6, import the BP neural network to calculate the fitness value of the chromosome, and start the next round of genetic algorithm iterative optimization; Step 12: The genetic algorithm iteration is completed and the chromosome ranked first is extracted; Step 13: Decode the chromosome to obtain the optimal VCSEL array oxidation aperture size information; Step 14: Import the optimal VCSEL array oxidation aperture size information into the thermoelectric feedback model to simulate the temperature distribution; Step 15: The optimized steady-state temperature distribution of the VCSEL array is obtained, and the optimization is completed.

2. A VCSEL array oxidation aperture intelligent optimization design method according to claim 1, characterized in that: The step 3 of simulating temperature distribution using the thermoelectric feedback model includes the following steps: (1) Using ANSYS to establish a VCSEL array structure model based on the oxidation aperture sample value or the oxidation aperture information corresponding to the optimal chromosome and structural parameters including VCSEL array shape, center-to-center distance between adjacent units, and substrate size; (2) Setting the operating conditions of the VCSEL array including injection current and ambient temperature; (3) Simulate the temperature distribution of the VCSEL array and extract the peak junction temperature T M0 ; (4) Use MATLAB to calculate the heat flux density of each unit in the VCSEL array; (5) Extract the heat flux density in (4) and load it into the VCSEL array. Use ANSYS to simulate the temperature distribution of the VCSEL array and extract the peak junction temperature T M1 ; (6) Judgement |T M1 -T M0 |Is it less than 1K? If it is less than 1K, execute step (7); otherwise, make T M0 =T M1 , return to step (4); (7) Output the steady-state junction temperature T1, T2, T3, ..., T of each unit in the VCSEL array n .

3. A VCSEL array oxidation aperture intelligent optimization design method according to claim 1, characterized in that: The step 4 of constructing the BP neural network model includes the following steps: (1) Establish a three-layer BP neural network structure including input layer, hidden layer and output layer, where the number of neurons in the input layer is n, and the input variables are R1, R2, R3, ..., R n ; The number of neurons in the output layer is n, and the output variable is The number of neurons in the hidden layer is q is a constant (q∈[1,10]); (2) The tansig function is used as the activation function of the hidden layer, and the purelin linear function is used as the activation function of the output layer: purelin(x)=x (3) Use the mean square error function to calculate the error of the BP neural network: in, is the steady-state junction temperature predicted by the BP neural network, T i is the sample steady-state junction temperature corresponding to the test data in the data set; (4) Use the gradient descent method trainlm function to train the BP neural network.

4. A VCSEL array oxidation aperture intelligent optimization design method according to claim 1, characterized in that: The fitness function L in step 6 P , select the VCSEL array temperature uniformity as the evaluation criterion, use the variance of the VCSEL array temperature to calculate, the fitness function of the pth chromosome is: Where n is the number of units in the VCSEL array, is the steady-state junction temperature of the j-th VCSEL unit predicted by the BP neural network.

Images