Optimization design method for structural parameters of heavy oil burner

Abstract

The invention provides an optimization design method for structural parameters of a heavy oil burner, which comprises the following steps of: firstly, simulating the atomization characteristics of heavy oil, then establishing a heavy oil burner simulation model and verifying the reliability of the model; on the basis, key structure parameters influencing the performance of the heavy oil burner are determined and serve as design variables for experimental design, all the schemes are simulated, and a neural network is trained through simulation results; next, a multi-objective optimization model of heavy oil burner structure parameters is established, optimization is carried out by utilizing a genetic algorithm, multi-objective decision of a Pareto solution set is carried out according to user preferences and a grey correlation method, and optimal structure parameters are determined; and finally, testing the operation performance of the heavy oil burner, re-optimizing if the requirements are not met, and otherwise, finishing the design of the heavy oil burner. Based on a numerical simulation method and a small amount of test data, the optimal structural parameters of the heavy oil burner can be determined, and development time and capital cost can be reduced.

Description

Technical Field

[0001] The invention relates to the field of burner design, and in particular to an optimization design method for structural parameters of a heavy oil burner. Background Art

[0002] Heavy oil burner is a common combustion equipment, which has been widely used in industries such as glass melting furnace and asphalt mixing plant. However, in actual application, heavy oil burner often has problems such as long combustion flame, burning materials, high NOx concentration in tail gas and low combustion efficiency. Therefore, how to reasonably design the burner to effectively improve the comprehensive performance of the burner has become a technical problem that needs to be solved urgently.

[0003] Since the heavy oil atomization combustion process is extremely complex, the current heavy oil burner structure design mainly adopts the empirical method. However, this method is very dependent on the designer's own engineering experience and is not very versatile. In addition, the burner structure parameters obtained by design are often not the best solution, and there is still a lot of room for improvement. In addition, a few heavy oil burners are designed using the experimental method, but this method is costly and has a long cycle. Therefore, it is urgent to develop a new method for the structural design of heavy oil burners to shorten the design cycle and cost of the burner.

[0004] Numerical simulation technology not only has high simulation accuracy and reliability, but also can grasp the detailed information of the entire combustion process, thereby effectively shortening the experimental cycle and R&D costs. At present, numerical simulation has become a heavy oil means for the structural design of heavy oil burners. However, the computer resources required for three-dimensional numerical simulation lead to a long cycle for the design of heavy oil burners. BP neural network has the ability to self-learn and approximate any nonlinear function, has high prediction accuracy, can reduce the amount of three-dimensional simulation calculations, and speed up the optimization calculation speed.

[0005] At the same time, the actual atomization characteristics of the fuel are not considered in the existing heavy oil burner simulation design, which can easily lead to the simulation results deviating from the actual situation. In addition, the existing heavy oil burner design often only considers the combustion efficiency, while ignoring performance indicators such as pollutant concentration and flame length, and does not adopt optimization algorithms. It is not only difficult to obtain the global optimal structure of the burner, but also cannot take into account the other performance indicators of the burner. To this end, an optimization design method for the structural parameters of heavy oil burners is proposed to solve the above problems existing in the existing heavy oil burner structural design. Summary of the invention

[0006] In order to solve the above technical problems, the present invention proposes an optimization design method for the structural parameters of a heavy oil burner, which reduces the calculation amount of the heavy oil burner, reduces the design cycle and cost, and improves the comprehensive performance of the burner.

[0007] The technical solution of the present invention to solve the above problem is: a method for optimizing the structural parameters of a heavy oil burner, comprising the following steps:

[0008] Step 1: According to the initial structural parameters of the heavy oil burner, a three-dimensional simulation model of heavy oil nozzle atomization is established, and the atomization characteristics of heavy oil under different loads are simulated and analyzed to obtain the particle size distribution characteristics of fuel atomization under different loads;

[0009] Step 2: Conduct experimental tests on the heavy oil burner prototype to obtain experimental measurement values ​​of the combustion performance of the heavy oil burner under different working conditions;

[0010] Step 3: Establish a three-dimensional simulation model of the combustion characteristics of the heavy oil burner, perform CFD simulation on the combustion characteristics of the heavy oil burner under the same working conditions as the experiment, and compare the results with the experimental results to verify the reliability of the simulation model;

[0011] Step 4: Select the main structural parameters of the heavy oil burner to design a heavy oil burner simulation scheme, establish simulation models for each scheme and conduct simulation analysis to obtain simulation results of the combustion characteristics of the heavy oil burner, and then determine the main structural parameters that affect the performance of the heavy oil burner through the Taguchi method;

[0012] Step 5: Use the random optimization Latin hypercube design method to determine the test scheme of the heavy oil burner, establish the simulation model of each scheme and carry out simulation analysis; use the simulation results to train the neural network and obtain the artificial neural network prediction model of the heavy oil burner performance;

[0013] Step 6: Establish a multi-objective optimization model for the performance of heavy oil burners, use genetic algorithms to perform multi-objective optimization on the structural parameters of heavy oil burners, obtain the optimal Pareto solution set, and perform multi-objective decision-making based on user preferences and grey correlation method to determine the optimal structural parameters of heavy oil burners;

[0014] Step 7: According to the optimization results, a prototype is manufactured and tested. If the performance of the heavy oil burner meets the design requirements, the design is completed. Otherwise, the number of training samples of the neural network is increased and the optimization calculation is continued.

[0015] According to the above-mentioned method for optimizing the structural parameters of a heavy oil burner, it is characterized in that the simulation of the atomization characteristics of the heavy oil burner in step 1 comprises the following steps:

[0016] Step M1: According to the geometric structural parameters of the heavy oil nozzle, a three-dimensional simulation model including the nozzle and the atomization space is established, and the size of the atomization space is required to be more than 30 times the size of the nozzle; then the mesh for numerical simulation is divided, the nozzle part adopts a tetrahedral mesh, and the y+ value near the wall is less than 1, the atomization space adopts a hexahedral mesh, and the atomization core area is encrypted;

[0017] Step M2: Import the nozzle atomization simulation grid into the computational fluid dynamics software FLUENT, set the nozzle heavy oil flow and temperature, atomizer inlet pressure, atomization space outlet pressure, and set the atomization and breakup model parameters according to the heavy oil atomization method and fuel physical properties; on this basis, use finite differences to iteratively calculate the heavy oil flow atomization process, and wait for the results to converge to obtain the simulation results of the oil droplet size distribution characteristics.

[0018] According to the above-mentioned method for optimizing the structural parameters of a heavy oil burner, it is characterized in that the experimental test of the heavy oil burner in step 2 comprises the following steps:

[0019] Step N1: Install the heavy oil burner prototype on the burner performance test bench, adjust the frequency of the blower and the oil pump, evenly arrange the temperature measuring thermocouples at a spacing of 1m in the length direction of the drum, use the combustion efficiency analyzer to obtain the test values ​​of the concentrations of components such as O2, CO2, CO, and NO in the flue gas at the outlet of the experimental system under the load W1 of the burner, and record the temperature value of each measuring point. During the test, the sampling heads of the combustion efficiency analyzer and the thermocouple are located at the center of the channel;

[0020] Step N2: Adjust the fan and oil pump frequency of the burner performance test bench to increase the combustion load of the burner and obtain 3-5 experimental data of flue gas composition and temperature under different loads;

[0021] Step N3: Step N2 is repeated 2-3 times until experimental test values ​​of flue gas composition and temperature corresponding to 3-5 different load values ​​are obtained, and the average value of several test results is taken as the final experimental result.

[0022] According to the above-mentioned method for optimizing the structural parameters of a heavy oil burner, it is characterized in that the simulation analysis of the combustion characteristics of the heavy oil burner in steps 3, 4 and 5 includes the following steps:

[0023] Step O1: According to the structural parameters of the heavy oil burner, a three-dimensional graph of the heavy oil burner and the fluid domain of the external combustion area is drawn, and the size of the external combustion area is more than three times the size of the burner; then, the fluid domain is divided into meshes for numerical simulation, wherein the heavy oil burner area adopts a tetrahedral mesh, the external combustion area adopts a hexahedral mesh, and the mesh of the central combustion area is encrypted;

[0024] Step O2: Import the grid for heavy oil burner simulation into the computational fluid dynamics software Fluent, and use the heavy oil particle size distribution obtained by the simulation in step 1 as the particle phase boundary of the combustion simulation; set the model air inlet velocity, burner area outlet pressure, flue gas radiation heat transfer model and combustion model, iteratively calculate the combustion and flow heat transfer process in the fluid domain, and read the simulation results after the results converge; modify the boundary conditions of the combustion model according to the remaining experimental conditions and carry out numerical simulation to obtain the simulation results under different experimental conditions;

[0025] Step O3: After the simulation is completed, the combustion NOx generation model is set according to the elemental composition of heavy oil, and then the NOx generation characteristics of the combustion process are simulated to obtain the NOx concentration in the flue gas under different loads. The simulated flue gas concentration and temperature under different combustion loads are compared with the experimental test values. If the difference between the two is less than 10%, it indicates that the simulation model has good reliability; if the deviation is greater than 10%, adjust the setting parameters and repeat the above steps; when verifying the NOx concentration, the NOx concentration in the flue gas at the outlet needs to be corrected to the standard state (oxygen content 6%), and the correction calculation formula is:

[0026]

[0027] Where: NO is the corrected NOx concentration, NO sim represents the simulated NOx concentration, and O2 represents the simulated average O2 concentration at the outlet section.

[0028] According to the above-mentioned optimization design method for structural parameters of a heavy oil burner, it is characterized in that the Box-Behnken method is used in step 4 to design an orthogonal experimental scheme, mainly considering parameters such as the number of nozzle holes, the diameter of the flame stabilizing disk, the number of swirl plates, the installation angle of the swirl plates, and the relative position of the nozzle; the performance of the heavy oil burner mainly considers indicators such as NOx concentration in the exhaust gas under standard conditions, flame length and combustion efficiency; wherein the flame length L is defined as the distance between the nozzle outlet and the axial front end of the flame front, and the flame front is evaluated by the chemical equivalent mixture fraction f, and its expression is:

[0029]

[0030] Where: Φ is the chemical equivalent ratio. When Φ=1, the heavy oil and air react completely. air is the mass of air, m fuel is the quality of heavy oil;

[0031] The calculation formula of the range of the main structural parameters of heavy oil burner for different performance indicators can be expressed as:

[0032]

[0033]

[0034] For combustion efficiency:

[0035]

[0036] For flame length and outlet NOx concentration

[0037]

[0038] Where: R is the range value, S / N represents the signal-to-noise ratio, J represents the number of factors with the same value at the same level; subscript k is the factor number, i is the number of factor levels, p is the performance indicator number, and n is the test plan number.

[0039] According to the above-mentioned optimization design method of structural parameters of a heavy oil burner, it is characterized in that in the step 5, structural parameters that have a great influence on the performance of the heavy oil burner are selected as design variables of the test scheme, and no less than 70 groups of experimental schemes are designed; the design variables are used as the input of the neural network, the flame length, combustion efficiency and outlet NOx concentration are used as the output of the neural network, and tansig is selected as the transfer function of the intermediate hidden layer; the GA genetic algorithm is used to optimize the weights and thresholds of the BP neural network, and 80% of the simulation results of the design scheme are randomly selected as the training set, and the remaining 20% ​​are the test set, and a GA-BP neural network prediction model for the combustion performance of the heavy oil burner is established, and the relative error (E r ) and the coefficient of determination (R) are used to quantitatively evaluate the prediction accuracy of the model:

[0040]

[0041]

[0042] In the formula, y i is the predicted value, t i is the experimental data, and are the average values ​​of the predicted data and experimental data, respectively;

[0043] If the maximum relative error between the predicted value and the simulation result of the GA-BP neural network is less than 5% and the deviation between the R value and 1 is less than 2%, the model is considered to have high accuracy and can be used for heavy oil burner performance prediction.

[0044] According to the above-mentioned method for optimizing the structural parameters of a heavy oil burner, it is characterized in that the multi-objective optimization of the structural parameters of the heavy oil burner in step 6 comprises the following steps:

[0045] Step P1: Taking the main structural parameters of the heavy oil burner as design variables, and taking the shortest flame length, the lowest outlet NOx concentration and the highest combustion efficiency as the goals, a multi-objective optimization model of the heavy oil burner is established using the trained GA-BP neural network model, and its expression is as follows:

[0046]

[0047] The constraint condition C is set mainly considering the limitation of structure size, which can be expressed as:

[0048]

[0049] Step P2: Use Matlab software to write a multi-objective optimization program based on genetic algorithm, set the genetic generation, population size and crossover mutation factor, carry out multi-objective optimization design of heavy oil burner structural parameters, and obtain the Pareto solution set.

[0050] Step P3: Determine the weight of the objective function according to user preferences, and then use the grey correlation method to make multi-objective decisions on the Pareto solution set to obtain the optimal solution for the optimization design; the expression of grey correlation is as follows:

[0051]

[0052]

[0053] For combustion efficiency η:

[0054]

[0055] For outlet NOx and flame length L:

[0056]

[0057] P on =[max P m1 … max P mN ]=[P o1 … P oN ] (15)

[0058] Where w is the weight value, which is determined by user preference; ρ is the resolution coefficient; X is the performance index; P on is the maximum value of each column and is taken as the ideal solution; n is the number of evaluation objects; m is the number of evaluation indicators.

[0059] According to the calculation results, GRD m The point with the highest value is selected as the optimal point, and the design variable corresponding to this point is the optimal structural solution for the heavy oil burner.

[0060] The beneficial effects of the present invention are:

[0061] (1) The empirical method used in the design of heavy oil burners in the present invention has poor versatility, while the experimental method has shortcomings such as high cost and long cycle. The performance of heavy oil burners under different structural parameters is obtained mainly through numerical simulation methods. Only a small number of burner performance tests are required, which can avoid the high cost of a large number of experiments. Therefore, the economic cost and time cost of this method are low and the versatility is strong.

[0062] (2) The present invention uses the simulation analysis results of heavy oil atomization characteristics as the input conditions for heavy oil combustion characteristics simulation, which helps to improve the reliability of the simulation results and make the simulation results more consistent with the actual situation; at the same time, the structural design of the heavy oil burner combines and considers multiple performance indicators at the same time, so that the design results balance the conflicts between different performance indicators and improve the comprehensive performance of the heavy oil burner.

[0063] (3) The present invention utilizes the GA-BP neural network prediction method to realize the rapid prediction of the heavy oil burner performance under different structural parameters, and combines it with the genetic optimization algorithm to realize the multi-objective optimization of the heavy oil burner structural parameters. This method is objective, scientific, easy to operate, and helps to obtain the global optimal solution of the heavy oil burner structural parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 Flow chart of the heavy oil burner result parameter optimization design method of the present invention

[0065] Figure 2 Comparison of GA-BP neural network prediction and experimental values ​​of NOx concentration from heavy oil burners Figure 3 Grey correlation degree of each solution individual in Pareto solution set DETAILED DESCRIPTION

[0066] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0067] See attached Figure 1 The flowchart shown in the figure introduces the basic process of a method for optimizing the structural parameters of a heavy oil burner of the present invention. The method is carried out according to the following steps:

[0068] The main geometric structural parameters of a heavy oil burner prototype are: number of nozzles 13, flame stabilizing disk diameter 50mm, number of swirl plates 20, swirl plate installation angle 45°; the burner uses compressed air as the atomizer and realizes the atomization of heavy oil through a Y-type nozzle, with a maximum load of 40MW and an operating load range of 0.2-1.1; the Y-type nozzle outlet diameter is 4mm, the heavy oil and air inlet diameters are 3mm, the fuel flow rate is 66g / s, the fuel temperature is 100℃, and the air inlet pressure is 0.6MPa.

[0069] A three-dimensional model was established based on the structural parameters of the heavy oil nozzle, and then the mesh for numerical simulation was divided. The wall y+ of the nozzle was less than 1, and the hexahedral mesh of the atomization core area was encrypted. According to the mesh independence verification, the overall number of meshes was 1.45 million. The nozzle atomization simulation mesh was imported into Fluent to simulate the heavy oil atomization characteristics. The maximum particle size of the oil droplets under the design conditions was 114μm, the minimum particle size was 42.5μm, and the average particle size of the Sotel was 60.5μm. Then, according to the burner load range, the boundary conditions of the atomization simulation setting were modified, and the physical properties under no less than 8 load conditions were simulated and analyzed to obtain the distribution characteristics of the oil droplet size under different loads.

[0070] The heavy oil burner prototype was experimentally tested to obtain the experimental measurement values ​​of the flue gas composition at the outlet and the temperature of each test point corresponding to the three operating points; the experimental test was repeated twice, and the average value of the two test results was taken as the experimental test value of the heavy oil burner. The results are shown in Table 1.

[0071] Table 1 Component test values ​​of heavy oil burner under different loads in the original design

[0072] load 0.2 0.25 0.3 0.35 0.4 <![CDATA[O2%]]> 15.9 14.3 12.9 12.1 10.8 CO / ppm 336 164 268 684 1369 NOx / ppm 135 213 243 279 271 <![CDATA[T1 / K]]> 1073 1081 1082 1085 1090 <![CDATA[T2 / K]]> 1027 1043 1062 1071 1102 <![CDATA[T3 / K]]> 1134 1165 1217 1288 1324 <![CDATA[T4 / K]]> 1488 1505 1538 1562 1597 <![CDATA[T5 / K]]> 1562 1610 1638 1670 1713

[0073] According to the structural parameters of the heavy oil burner prototype, a three-dimensional simulation model of the burner combustion characteristics is established. The particle size distribution of heavy oil atomization under the corresponding working conditions is loaded into the simulation model through the DPM model. The DO radiation model and the non-premixed combustion model based on the probability density function are used. The same boundary conditions under the experimental working conditions are used to perform CFD simulation on the heavy oil combustion characteristics to obtain the simulation results under various experimental working conditions. The simulation results under different experimental loads are shown in Table 2:

[0074] Table 2 Component simulation values ​​of heavy oil burner prototype under different loads

[0075]

[0076] According to the comparison between the simulation results and the experimental test values, the deviation between the two is less than 10%, indicating that the simulation model has good reliability and the setting method of the model can be used to simulate the combustion characteristics of heavy oil burners.

[0077] The number of burner nozzles, the diameter of the flame stabilizing disk, the number of swirl plates, the installation angle of the swirl plates, the relative position of the nozzles and other structural parameters were selected as factors for the orthogonal experiment. Each factor took 3 levels, and 16 simulation schemes were designed using the Box-Behnken method. For these 16 design schemes, the same simulation model was used to carry out numerical simulation calculations, and the results are shown in Table 3.

[0078] Table 3 Burner performance under each design scheme

[0079] plan Flame length / m Combustion efficiency / % Export NO / ppm 1 5.6 98.8 358 2 7.1 93.1 278 3 6.4 91.1 272 4 5.5 98.2 248 5 5.4 97.8 252 6 7.8 97.0 274 7 5.3 96.8 240 8 6.8 90.3 302 9 5.4 96.3 268 10 6 95.8 254 11 5.8 97.7 270 12 5.8 96.8 250 13 5.2 98.5 260 14 6.0 98.3 252 15 4.9 98.3 238 16 5.3 96.4 246

[0080] The range of main structural parameters for different performance indicators was calculated based on the simulation results, and the results are shown in Table 4.

[0081] Table 4 Range of main structural parameters of burner

[0082]

[0083] According to the calculation results, the relative position of the nozzle has the smallest effect on the burner performance, while the flame stabilizing disk has the greatest direct effect on the flame length, the number of nozzle holes has the greatest effect on the outlet CO concentration, and the swirl vane installation angle has the greatest effect on the outlet NO. Since the relative position of the nozzle has the smallest effect, the relative position of the nozzle can be determined empirically in the design, and this parameter can be omitted as an optimization variable.

[0084] Taking the flame stabilizing disk diameter, the number of swirl plates, the installation angle of the swirl plates and the number of nozzles as design variables, 70 test schemes were designed using the random optimization Latin hypercube design method. Simulation models of these schemes were established and simulation analysis was carried out to obtain the performance of the heavy oil burner under each scheme.

[0085] The four parameters of flame stabilizing disk diameter, number of swirl plates, swirl plate installation angle, and number of nozzles are used as input layers, flame length, outlet NOx concentration, and combustion efficiency are used as output layers, and the middle layer is a hidden layer with tansig selected as the transfer function; 56 of the 70 test schemes are randomly selected as training sets to train the GA-BP neural network, and the results of the remaining 14 schemes are used as test sets to verify the reliability of the neural network. The prediction accuracy of the GA-BP neural network for combustion efficiency of the training set and test set is shown in the attached figure. Figure 2 As shown. It can be seen that the GA-BP neural network model can accurately predict the combustion efficiency of the heavy oil burner, with a maximum deviation of less than 3%. Similarly, for outlet NOx and flame length, the maximum prediction deviation of the GA-BP neural network is also less than 4%, which shows that the GA-BP neural network can accurately predict the combustion performance of the heavy oil burner under different working conditions.

[0086] The diameter of the flame stabilizing disk d and the number of swirl plates N x 、Swirl blade installation angle α x 、Number of nozzles p As the decision variable, a multi-objective optimization model of heavy oil burner structural parameters is established with the shortest flame length, the lowest outlet NOx concentration and the highest combustion efficiency as the goal. Its expression is as follows:

[0087]

[0088] Considering the limitation of structure size, the following constraints are set:

[0089] Combined with the trained GA-BP neural network model, the genetic optimization calculation program was written using Matlab software. During the optimization process, the population size was set to 50, the genetic generation was 100, and the crossover mutation factor was set to 0.9. The multi-objective optimization design of the structural parameters of the heavy oil burner was carried out using the written program, and a Pareto optimization solution set with 50 individual solutions was obtained.

[0090] According to the user's preference, the weight of outlet NOx is set to 0.4, the weight of combustion efficiency is set to 0.35, and the weight of flame length is set to 0.25. The grey correlation method is used to make multi-objective decisions on the optimized Pareto solution set, and the grey correlation degree of each point is calculated. The results are shown in the attached Figure 3 . According to the calculation results, the grey correlation degree of the 34th solution in the Pareto solution set is the highest. This point is selected as the optimal design scheme for the heavy oil burner. The corresponding parameters are: flame stabilizing disk diameter 590mm, number of swirl plates 22, swirl plate installation angle 42.6°, number of nozzles 10. After optimization, the flame length of the burner is 4.6m, the combustion efficiency is 97.8%, and the outlet NO concentration is 232ppm. Compared with the original design scheme, that is, Scheme 3 in Table 3, the performance of the heavy oil burner is significantly improved.

[0091] The present embodiment provides an optimization design method for the structural parameters of a heavy oil burner. In view of the shortcomings of the empirical method, such as poor versatility, and the experimental method, such as high cost and long cycle, the performance of the heavy oil burner under different structural parameters is obtained mainly through a numerical simulation method. Only a small number of burner performance tests are required, which can avoid the high cost of a large number of experiments. At the same time, based on the simulation results of the heavy oil atomization characteristics, a simulation analysis of the heavy oil combustion characteristics is carried out to improve the reliability of the simulation results. In addition, the GA-BP neural network prediction method is combined with the genetic optimization algorithm to carry out the optimization design of the structural parameters of the heavy oil burner, so that the design scheme comprehensively considers multiple performance indicators of the heavy oil burner. This method has low economic and time costs, strong versatility, is objective and scientific, and is easy to operate.

[0092] The above is only a specific implementation scheme of the patent of the present invention and is not intended to limit the present invention. Any modifications and improvements made based on the principles of this invention are included in the protection scope of the present invention.

Claims

1. A method for optimizing the structural parameters of a heavy oil burner, characterized in that: The following steps are involved: Step 1: According to the initial structural parameters of the heavy oil burner, a three-dimensional simulation model of heavy oil nozzle atomization is established, and the atomization characteristics of heavy oil under different loads are simulated and analyzed to obtain the particle size distribution characteristics of fuel atomization under different loads; Step 2: Conduct experimental tests on the heavy oil burner prototype to obtain experimental measurement values ​​of the combustion performance of the heavy oil burner under different working conditions; Step 3: Establish a three-dimensional simulation model of the combustion characteristics of the heavy oil burner, perform CFD simulation on the combustion characteristics of the heavy oil burner under the same working conditions as the experiment, and compare the results with the experimental results to verify the reliability of the simulation model; Step 4: Select the main structural parameters of the heavy oil burner to design a heavy oil burner simulation scheme, establish simulation models for each scheme and conduct simulation analysis to obtain simulation results of the combustion characteristics of the heavy oil burner, and then determine the main structural parameters that affect the performance of the heavy oil burner through the Taguchi method; Step 5: Use the random optimization Latin hypercube design method to determine the test scheme of the heavy oil burner, establish the simulation model of each scheme and carry out simulation analysis; use the simulation results to train the neural network and obtain the artificial neural network prediction model of the heavy oil burner performance; Step 6: Establish a multi-objective optimization model for the performance of heavy oil burners, use genetic algorithms to perform multi-objective optimization on the structural parameters of heavy oil burners, obtain the optimal Pareto solution set, and perform multi-objective decision-making based on user preferences and grey correlation method to determine the optimal structural parameters of heavy oil burners; Step 7: According to the optimization results, a prototype is manufactured and tested. If the performance of the heavy oil burner meets the design requirements, the design is completed. Otherwise, the number of training samples of the neural network is increased and the optimization calculation is continued.

2. The method for optimizing the structural parameters of a heavy oil burner according to claim 1, characterized in that: The simulation of the atomization characteristics of the heavy oil burner in step 1 includes the following steps: Step M1: According to the geometric structural parameters of the heavy oil nozzle, a three-dimensional simulation model including the nozzle and the atomization space is established, and the size of the atomization space is required to be more than 30 times the size of the nozzle; then the mesh for numerical simulation is divided, the nozzle part adopts a tetrahedral mesh, and the y+ value near the wall is less than 1, the atomization space adopts a hexahedral mesh, and the atomization core area is encrypted; Step M2: Import the nozzle atomization simulation grid into the computational fluid dynamics software FLUENT, set the nozzle heavy oil flow and temperature, atomizer inlet pressure, atomization space outlet pressure, and set the atomization and breakup model parameters according to the heavy oil atomization method and fuel physical properties; on this basis, use finite differences to iteratively calculate the heavy oil flow atomization process, and wait for the results to converge to obtain the simulation results of the oil droplet size distribution characteristics.

3. The method for optimizing the structural parameters of a heavy oil burner according to claim 1, characterized in that: The heavy oil burner experimental test in step 2 includes the following steps: Step N1: Install the heavy oil burner prototype on the burner performance test bench, adjust the frequency of the blower and the oil pump, evenly arrange the temperature measuring thermocouples at a spacing of 1m in the length direction of the drum, use the combustion efficiency analyzer to obtain the test values ​​of the concentrations of components such as O2, CO2, CO, and NO in the flue gas at the outlet of the experimental system under the load W1 of the burner, and record the temperature value of each measuring point. During the test, the sampling heads of the combustion efficiency analyzer and the thermocouple are located at the center of the channel; Step N2: Adjust the fan and oil pump frequency of the burner performance test bench to increase the combustion load of the burner and obtain 3-5 experimental data of flue gas composition and temperature under different loads; Step N3: Step N2 is repeated 2-3 times until experimental test values ​​of flue gas composition and temperature corresponding to 3-5 different load values ​​are obtained, and the average value of several test results is taken as the final experimental result.

4. The method for optimizing the structural parameters of a heavy oil burner according to claim 1, characterized in that: The simulation analysis of the combustion characteristics of the heavy oil burner in steps 3, 4 and 5 includes the following steps: Step O1: According to the structural parameters of the heavy oil burner, a three-dimensional graph of the heavy oil burner and the fluid domain of the external combustion area is drawn, and the size of the external combustion area is more than three times the size of the burner; then, the fluid domain is divided into meshes for numerical simulation, wherein the heavy oil burner area adopts a tetrahedral mesh, the external combustion area adopts a hexahedral mesh, and the mesh of the central combustion area is encrypted; Step O2: Import the grid for heavy oil burner simulation into the computational fluid dynamics software Fluent, and use the heavy oil particle size distribution obtained by the simulation in step 1 as the particle phase boundary of the combustion simulation; set the model air inlet velocity, burner area outlet pressure, flue gas radiation heat transfer model and combustion model, iteratively calculate the combustion and flow heat transfer process in the fluid domain, and read the simulation results after the results converge; modify the boundary conditions of the combustion model according to the remaining experimental conditions and carry out numerical simulation to obtain the simulation results under different experimental conditions; Step O3: After the simulation is completed, the combustion NOx generation model is set according to the elemental composition of heavy oil, and then the NOx generation characteristics of the combustion process are simulated to obtain the NOx concentration in the flue gas under different loads. The simulated flue gas concentration and temperature under different combustion loads are compared with the experimental test values. If the difference between the two is less than 10%, it indicates that the simulation model has good reliability; if the deviation is greater than 10%, adjust the setting parameters and repeat the above steps; when verifying the NOx concentration, the NOx concentration in the flue gas at the outlet needs to be corrected to the standard state (oxygen content 6%), and the correction calculation formula is: Where: NO is the corrected NOx concentration, NO sim represents the simulated NOx concentration, and O2 represents the simulated average O2 concentration at the outlet section.

5. The method for optimizing the structural parameters of a heavy oil burner according to claim 1, characterized in that: In step 4, the Box-Behnken method is used to design the orthogonal experimental scheme, mainly considering parameters such as the number of nozzle holes, the diameter of the flame stabilizing disk, the number of swirl plates, the installation angle of the swirl plates, and the relative position of the nozzle; the performance of the heavy oil burner mainly considers indicators such as the NOx concentration in the exhaust gas under standard conditions, the flame length, and the combustion efficiency; wherein the flame length L is defined as the distance between the nozzle outlet and the axial front end of the flame front, and the flame front is evaluated by the chemical equivalent mixture fraction f, and its expression is: Where: Φ is the chemical equivalent ratio. When Φ=1, the heavy oil and air react completely. air is the mass of air, m fuel is the quality of heavy oil; the range calculation formula of the main structural parameters of heavy oil burner for different performance indicators can be expressed as: For combustion efficiency: For flame length and outlet NOx concentration: Where: R is the range value, S / N represents the signal-to-noise ratio, J represents the number of factors with the same value at the same level; subscript k is the factor number, i is the number of factor levels, p is the performance indicator number, and n is the test plan number.

6. The method for optimizing the structural parameters of a heavy oil burner according to claim 1, characterized in that: In the step 5, the structural parameters that have a great influence on the performance of the heavy oil burner are selected as the design variables of the test scheme, and no less than 70 groups of experimental schemes are designed; the design variables are used as the input of the neural network, the flame length, the combustion efficiency and the outlet NOx concentration are used as the output of the neural network, and tansig is selected as the transfer function of the intermediate hidden layer; the weights and thresholds of the BP neural network are optimized by using the GA genetic algorithm, and 80% of the simulation results of the design scheme are randomly selected as the training set, and the remaining 20% ​​are the test set, and the GA-BP neural network prediction model for the combustion performance of the heavy oil burner is established, and the relative error (E r ) and the coefficient of determination (R) are used to quantitatively evaluate the prediction accuracy of the model: In the formula, y i is the predicted value, t i is the experimental data, and are the average values ​​of the predicted data and experimental data, respectively; If the maximum relative error between the predicted value and the simulation result of the GA-BP neural network is less than 5% and the deviation between the R value and 1 is less than 2%, the model is considered to have high accuracy and can be used for heavy oil burner performance prediction.

7. The method for optimizing the structural parameters of a heavy oil burner according to claim 1, characterized in that: The multi-objective optimization of the heavy oil burner structural parameters in step 6 includes the following steps: Step P1: Taking the main structural parameters of the heavy oil burner as design variables, and taking the shortest flame length, the lowest outlet NOx concentration and the highest combustion efficiency as the goals, a multi-objective optimization model of the heavy oil burner is established using the trained GA-BP neural network model, and its expression is as follows: The constraint condition C is set mainly considering the limitation of structure size, which can be expressed as: Step P2: Use Matlab software to write a multi-objective optimization program based on genetic algorithm, set the genetic generation, population size and crossover mutation factor, carry out multi-objective optimization design of heavy oil burner structural parameters, and obtain the Pareto solution set; Step P3: Determine the weight of the objective function according to user preferences, and then use the grey correlation method to make multi-objective decisions on the Pareto solution set to obtain the optimal solution for the optimization design; the expression of grey correlation is as follows: For combustion efficiency η: For outlet NOx and flame length L: P on =[maxP m1 ...maxP mN ]=[P o1 …P oN ] (15) Where w is the weight value, which is determined by user preference; ρ is the resolution coefficient; X is the performance index; P on is the maximum value of each column and is taken as the ideal solution; n is the number of evaluation objects; m is the number of evaluation indicators; According to the calculation results, the point with the highest GRDm value is selected as the optimal point, and the design variable corresponding to this point is the optimal structural solution of the heavy oil burner.

Images