Particle swarm optimization method for embedded iterative E-type saturated inductor of high-frequency converter

Abstract

The invention discloses a particle swarm optimization method for an embedded iterative E-type saturated inductor of a high-frequency converter, and provides an innovative scheme integrating nonlinear analysis and intelligent optimization of a magnetic circuit in order to solve the problems of large inductance value calculation deviation and difficulty in multi-objective optimization in the traditional linear magnetic circuit design. A nonlinear iterative algorithm is embedded into a particle swarm optimization algorithm, and on the basis of a nonlinear relationship between magnetic core magnetic flux density and magnetic conductivity, a magnetic resistance model is utilized to perform loop iteration to approach a real magnetic flux density value, so that the problem of accurate modeling of a magnetic circuit under a local saturation effect is solved; and meanwhile, a multi-target fitness function is constructed by taking geometric parameters and electrical parameters of the E-like magnetic core as optimization variables, and parameter collaborative optimization is realized through a particle swarm algorithm. According to the method, the modeling limitation of the traditional design on the nonlinear characteristic of the magnetic circuit is broken through, the inductance value calculation precision and the multi-index collaboration are improved, and the high-precision design of the saturation inductance can be realized without depending on complex electromagnetic field simulation.

Description

Technical Field

[0001] The present invention relates to the field of high-frequency power converters, and in particular to a particle swarm optimization method for high-frequency converters with built-in iterative E-type saturated inductors. Background Art

[0002] In the design and application of high-frequency power converters, the saturated inductor is a core component whose performance directly affects the converter's efficiency, stability, and reliability. Traditional linear magnetic circuit design methods are based on idealized assumptions and ignore the strong nonlinear relationship between the magnetic flux density and permeability of the core material. As a result, in actual operation, especially when the core enters a state of partial saturation, the calculated inductance value deviation increases significantly, making it impossible to accurately predict the dynamic characteristics of the saturated inductor under different current conditions. This modeling flaw makes the designed saturated inductor prone to performance instability under rated current or overload current, such as inductance value jumps and abnormal core loss. This seriously restricts the miniaturization and efficiency of high-frequency converters.

[0003] On the other hand, the structural characteristics of E-type saturated inductors dictate that their optimization design involves multi-dimensional parameter coupling, including core geometry and electrical parameters, and must simultaneously meet multi-objective optimization requirements such as inductance accuracy, core volume, and overload characteristics. Traditional design methods typically employ trial-and-error or single-objective optimization strategies, which struggle to achieve coordinated parameter optimization in complex scenarios with multiple variables and indicators. These methods often rely on time-consuming and computationally intensive electromagnetic field simulations for repeated debugging, resulting in lengthy design cycles, high costs, and difficulty in achieving a globally optimal solution.

[0004] The rapid development of new energy technologies and power electronics has placed higher demands on the design of saturated inductors for high-frequency converters: not only must the inductance be stable over a wide current range, but core volume must also be minimized and losses optimized. The limitations of traditional linear magnetic circuit modeling and single-objective optimization methods are becoming increasingly prominent. Breaking through the bottleneck of accurately modeling the nonlinear characteristics of magnetic circuits and achieving intelligent optimization of multi-objective parameters have become key technical challenges that need to be addressed in the field of high-frequency power converters. Summary of the Invention

[0005] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art. The present invention proposes an embedded iterative E-type saturated inductor particle swarm optimization method for high-frequency converters. By embedding a nonlinear iterative algorithm into the particle swarm optimization algorithm, the true value is approximated based on the nonlinear relationship, thereby solving the problem of accurate modeling of the local saturated magnetic circuit. With the geometry and electrical parameters of the E-type core as variables, a multi-objective fitness function is constructed to achieve parameter collaborative optimization, breaking through the limitations of traditional magnetic circuit nonlinear modeling, improving the inductance calculation accuracy and multi-index synergy, and achieving high-precision design without complex simulation.

[0006] The technical solution to achieve the purpose of the present invention is:

[0007] A particle swarm optimization method for an embedded iterative E-type saturated inductor of a high-frequency converter is characterized by comprising the following steps:

[0008] Step S1, constructing a magnetoresistance model of an E-type saturated inductor, wherein the magnetoresistance model is based on a nonlinear relationship between magnetic flux density and magnetic permeability of a magnetic core;

[0009] Step S2, solving the true value of the magnetic flux density of the reluctance model by a nonlinear iterative algorithm;

[0010] Step S3: embedding the iterative algorithm into the particle swarm optimization algorithm, taking the core structure parameters and electrical parameters as optimization variables, and constructing a multi-objective fitness function for optimization design.

[0011] Furthermore, the step S1 includes:

[0012] Step S11, initialize known parameters: set vacuum permeability μ0, rated current I Lf_f , overload current I Lf_o , rated inductance value L f_f , Rated center column magnetic flux density B center_f , the rated current is the non-saturated current, and the overload current is the saturated current;

[0013] Step S12, define optimization variables: set core optimization parameters, which include geometric parameters: magnetic plate length l up 、l up1 , magnetic plate cross-sectional area A up , the cross-sectional area of ​​the middle column A center , side column length l edge , cross-sectional area of ​​side column A edge ;Electrical parameters: Number of turns of side column N edge ;

[0014] Step S13, initializing the particle swarm: setting the maximum number of iterations G and the number of populations N, and setting parameter boundaries and speed boundaries for each optimization variable;

[0015] Step S14: Calculate the initial parameters of the center column:

[0016] Number of turns of center column N center :

[0017]

[0018] Length of the middle column air gap l gap :

[0019]

[0020] Middle column magnetic resistance Z center :

[0021]

[0022] Under rated current, the middle column magnetomotive force F centerf :

[0023] F centerf =N center I Lf_f

[0024] At rated current, the side column magnetomotive force F edgef :

[0025] F edgef =N edge I Lf_f .

[0026] Furthermore, the step S2 includes:

[0027] Step S21, preset initial value: preset initial value of magnetic flux density of each part under non-saturated current: magnetic flux density of side column B edge_f , magnetic flux density B at the center of the top plate up_f , magnetic flux density at the top plate edge B up1_f ;

[0028] Step S22, calculate the magnetic permeability: Based on the magnetic flux density B-magnetic permeability μ curve, determine the magnetic permeability μ of each part Bedge_f 、

[0029]

[0030] Step S23, performing iterative solution;

[0031] Step S24: Calculate the inductance value.

[0032] Furthermore, the iterative solution of step S23 specifically includes:

[0033] Reluctance calculation:

[0034] Side column magnetoresistance Z under non-saturation current edgef The calculation formula is

[0035]

[0036] Top plate center magnetic resistance Z under non-saturation current upf The calculation formula is

[0037]

[0038] Top plate edge magnetoresistance Z under non-saturation current up1f The calculation formula is

[0039]

[0040] Magnetic flux solution:

[0041] Calculation of the magnetic flux of the side column under non-saturated current based on the magnetoresistance model Top plate magnetic flux under non-saturation current Column magnetic flux in non-saturated current

[0042] Magnetic flux density calculation:

[0043] Side column magnetic flux density B under non-saturated current nedge_f The calculation formula is

[0044]

[0045] Magnetic flux density B at the center of the top plate under non-saturated current nup_f The calculation formula is

[0046]

[0047] Magnetic flux density B at the top plate edge under non-saturated current nup1_f The calculation formula is

[0048]

[0049] Deviation calculation:

[0050] Side column magnetic flux density deviation ΔB under non-saturation current edge_f The calculation formula is

[0051] ΔB edge_f =B nedge_f -B edge_f

[0052] Magnetic flux density deviation ΔB at the center of the top plate under non-saturated current up_f The calculation formula is ΔB up_f =B nup_f -B up_f

[0053] Magnetic flux density ΔB at the top plate edge under non-saturated current up1_f The calculation formula is

[0054] ΔB up1_f =B nup1_f -B up1_f

[0055] The preset value is adjusted according to the deviation until or when the maximum number of iterations k is reached.

[0056] Furthermore, the inductance calculation in step S24 specifically includes:

[0057] Non-saturated current mid-column winding inductance L centerf :

[0058]

[0059] Non-saturated current side winding inductance L edgef :

[0060]

[0061] The process of solving the saturated current inductance value is consistent with the process of solving the unsaturated current inductance value. Further, the step S3 needs to construct a fitness function, specifically including the following steps: Core volume evaluation item:

[0062]

[0063] Inductance variation range evaluation items:

[0064] ΔL=|L centerf -L centero |

[0065] Rated center column inductance offset error evaluation items:

[0066] ΔL centerf =|L centerf -L f_f |

[0067] Fitness function:

[0068] f x =αV L +βΔL+γΔL centerf

[0069] For each iteration number G, the core parameters corresponding to the particle position are input into the iterative algorithm to obtain the high non-saturated current inductance value; the particle speed and position are updated to drive the population to converge to the optimal solution.

[0070] Furthermore, the magnetic core structure of the E-type saturated inductor is:

[0071] The four-pillar or three-pillar E-shaped magnetic core consists of an upper magnetic plate, a lower magnetic plate, a middle pillar and side pillars, with only the middle pillar having an air gap.

[0072] Furthermore, the magnetic resistance model includes a center column magnetic resistance, a side column magnetic resistance and a magnetic plate magnetic resistance, which are connected in parallel or in series to form a magnetic circuit.

[0073] Furthermore, the particle swarm optimization algorithm updates the particle positions based on the standard model without the need for additional formulas.

[0074] Compared with the prior art, the present invention adopts the above technical solution and has the following beneficial effects:

[0075] (1) This invention embeds a nonlinear iterative algorithm into a particle swarm optimization algorithm, constructs a reluctance model based on the nonlinear relationship between the core's magnetic flux density and permeability, and approximates the true magnetic flux density value through cyclic iteration. This effectively solves the problem of accurate modeling of the magnetic circuit under local saturation effects. Compared with traditional linear magnetic circuit design methods, this method can more realistically reflect the nonlinear characteristics of the core under different current conditions, significantly improves the accuracy of inductance calculation, and provides a more reliable theoretical basis for the design of saturation inductance in high-frequency converters.

[0076] (2) The multi-objective optimization mechanism adopted in the present invention can find the optimal balance point between key indicators such as core volume and inductance performance, which not only meets the miniaturization requirements of high-frequency converters, but also ensures the stable operation of the inductor under rated current and overload current, thereby improving the overall performance of the design scheme.

[0077] (3) This invention achieves high-precision design of saturated inductors without relying on complex electromagnetic field simulations, simplifying the design process and reducing reliance on specialized simulation tools and high-computing resources. Through standardized iterative algorithms and particle swarm optimization processes, designers can more efficiently optimize parameters, shortening the R&D cycle while reducing the costs associated with repeated simulation and debugging, thus possessing significant engineering application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 This is a flow chart of the particle swarm optimization method for high-frequency converter embedded iterative E-type saturated inductor proposed by the present invention;

[0079] Figure 2 It is a schematic diagram of a four-pillar E-type core;

[0080] Figure 3 It is a schematic diagram of a three-pillar E-type core;

[0081] Figure 4 Optimal variable diagram for the core;

[0082] Figure 5 This is a diagram of the four-pillar E-type magnetic core reluctance model;

[0083] Figure 6 It is the number of iterations k of the embedded nonlinear iterative solution method and the rated current diagram of the magnetic flux density difference Bn-B curve;

[0084] Figure 7 It is the number of iterations k of the embedded nonlinear iterative solution method and the overload current diagram of the magnetic flux density difference Bn-B curve;

[0085] Figure 8 This is the convergence result diagram of particle length space and cross-sectional area space in the first iteration of particle swarm optimization;

[0086] Figure 9 This is the convergence result diagram of particle length space and cross-sectional area space at the 300th iteration of particle swarm optimization;

[0087] Figure 10 The particle length space l in the first iteration of particle swarm optimization edge 、l up 、l up1 and cross-sectional area space A up 、A center 、A edge The convergence result diagram of ;

[0088] Figure 11 The particle length space l at the 300th iteration of particle swarm optimization edge 、l up 、l up1 and cross-sectional area space A up 、A center 、A edge The convergence result diagram of ;

[0089] Figure 12 This is the ANSYS simulation diagram of the change in inductance before and after load shedding. DETAILED DESCRIPTION

[0090] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0091] A high-frequency converter embedded iterative E-type saturated inductor particle swarm optimization method, the process is as follows Figure 1 As shown, it is characterized in that it includes the following steps:

[0092] Step S1, constructing a magnetoresistance model of an E-type saturated inductor, wherein the magnetoresistance model is based on a nonlinear relationship between magnetic flux density and magnetic permeability of a magnetic core;

[0093] Step S2, solving the true value of the magnetic flux density of the reluctance model by a nonlinear iterative algorithm;

[0094] Step S3: embedding the iterative algorithm into the particle swarm optimization algorithm, taking the core structure parameters and electrical parameters as optimization variables, and constructing a multi-objective fitness function for optimization design.

[0095] Furthermore, the step S1 includes:

[0096] Step S11, initialize known parameters: set vacuum permeability μ0, rated current I Lf_f, overload current I Lf_o , rated inductance value L f_f , Rated center column magnetic flux density B center_f ;

[0097] Step S12, define optimization variables: set core optimization parameters: including geometric parameters - magnetic plate length l up 、l up1 , magnetic plate cross-sectional area A up , the cross-sectional area of ​​the middle column A center , side column length l edge , cross-sectional area of ​​side column A edge ;Electrical parameters - number of turns of side column N edge ;

[0098] Step S13, initializing the particle swarm: setting the maximum number of iterations G and the number of populations N, and setting parameter boundaries and speed boundaries for each optimization variable;

[0099] Step S14: Calculate the initial parameters of the center column:

[0100] Number of turns of center column N center :

[0101]

[0102] Length of the middle column air gap l gap :

[0103]

[0104] Middle column magnetic resistance Z center :

[0105]

[0106] Middle column magnetomotive force F center :

[0107] F centerf =N center I Lf_f

[0108] Side column magnetomotive force F edge :

[0109] F edgef =N edge I Lf_f

[0110] Furthermore, the step S2 includes:

[0111] Step S21, preset initial value: preset initial value B of magnetic flux density of each part under non-saturated current edge_f 、B up_f 、B up1_f ;

[0112] Step S22, calculate the magnetic permeability: Based on the magnetic flux density B-magnetic permeability μ curve, determine the magnetic permeability μ of each part Bedge_f 、

[0113]

[0114] Step S23, performing iterative solution;

[0115] Step S24: Calculate the inductance value.

[0116] Furthermore, the iterative solution of step S23 specifically includes:

[0117] Reluctance calculation:

[0118] Magnetic flux solution: calculated based on the magnetic resistance model

[0119] Calculated magnetic flux density:

[0120] Deviation calculation: ΔB = Bn-B, adjust the preset value according to the deviation until or reach the maximum number of iterations k.

[0121] Furthermore, the inductance calculation in step S24 specifically includes:

[0122] Non-saturated current center column inductance:

[0123] Non-saturated current side column inductance:

[0124] The process of solving the saturated current inductance value is consistent with the process of solving the non-saturated current inductance value.

[0125] Furthermore, step S3 includes:

[0126] Constructing a fitness function

[0127] Core volume evaluation items:

[0128]

[0129] Inductance variation range evaluation items:

[0130] ΔL=|L centerf -L centero |

[0131] Rated center column inductance offset error evaluation items:

[0132] ΔL centerf =|L centerf -L f_f |

[0133] Fitness function: f x =αV L +βΔL+γΔL centerf

[0134] For each iteration number G, the core parameters corresponding to the particle position are input into the iterative algorithm to obtain the high non-saturated current inductance value; the particle speed and position are updated to drive the population to converge to the optimal solution.

[0135] Furthermore, the magnetic core structure of the E-type saturated inductor is:

[0136] The four-pillar or three-pillar E-shaped magnetic core consists of an upper magnetic plate, a lower magnetic plate, a middle pillar and side pillars. Only the middle pillar has an air gap. The structural diagram is shown in the figure below. Figure 2 、 3 As shown. Take the four-pillar E-type magnetic core as an example, the magnetic plate length l up 、l up1 , magnetic plate cross-sectional area A up , the cross-sectional area of ​​the middle column A center , side column length l edge , cross-sectional area of ​​side column A edge , number of side column turns N edge As optimization variables. The core volume and inductance variation range are considered in the fitness function. This method can realize the saturation inductance optimization design under multiple optimization variables and multiple optimization objectives, such as Figure 4 As shown; the four-pillar E-type magnetic core reluctance model is as follows Figure 5 shown.

[0137] Furthermore, the magnetic resistance model includes a center column magnetic resistance, a side column magnetic resistance and a magnetic plate magnetic resistance, which are connected in parallel or in series to form a magnetic circuit.

[0138] Furthermore, the particle swarm optimization algorithm updates the particle positions based on the standard model without the need for additional formulas.

[0139] Example:

[0140] In order to verify the validity of the patent of this invention, a simulation verification is performed using a set of Buck operating parameters in Table 1 and a set of particle swarm parameters in Table 2.

[0141] Table 1B UCK Working parameters

[0142]

[0143] Table 2 Particle swarm parameters

[0144]

[0145] Figure 6 、 7The approximation process of the true value of magnetic flux density using the embedded nonlinear iterative solution method is given.

[0146] Within 3000 iterations, the calculation error of magnetic flux density at rated current was reduced to 0.01 T. Within 1200 iterations, the calculation error of magnetic flux density at overload current was reduced to 0.01 T.

[0147] Figure 8 、 9 Given the particle swarm optimization, the particle length space l edge 、l up 、l up1 Convergence results after the first to 300th iterations.

[0148] Figure 10 、 11 A is the particle cross-sectional area up 、A center 、A edge Convergence results plot after the first to 300th iterations.

[0149] from Figures 8 to 11 It can be concluded that after the 300th iteration, 82% of the particles are concentrated in 0.5% of the space, and the fitness function fx is the maximum value.

[0150] Table 2 shows the results of particle swarm optimization. centero is the inductance of the middle column winding under overload current (saturation current), L edgero It is the inductance value of the side column winding under overload current (saturation current).

[0151] Table 2 Particle swarm optimization results

[0152]

[0153] The optimized parameters are imported into ANSYS simulation for verification. The changes of inductance before and after load shedding are as follows: Figure 12 shown.

[0154] Simulation results show that the total inductance is 28.84 μH at 15 A and 19.72 μH at 30 A, consistent with the optimization results. The simulated core loss is 0.3 W.

[0155] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A particle swarm optimization method for high-frequency converter embedded iterative E-type saturated inductor, characterized by: The following steps are involved: Step S1, constructing a magnetoresistance model of an E-type saturated inductor, wherein the magnetoresistance model is based on a nonlinear relationship between magnetic flux density and magnetic permeability of a magnetic core; Step S2, solving the true value of the magnetic flux density of the reluctance model by a nonlinear iterative algorithm; Step S3: embedding the iterative algorithm into the particle swarm optimization algorithm, taking the core structure parameters and electrical parameters as optimization variables, and constructing a multi-objective fitness function for optimization design.

2. The method for iterative particle swarm optimization of a high-frequency converter with an embedded E-type saturated inductor according to claim 1, characterized in that: The step S1 comprises: Step S11, initialize known parameters: set vacuum permeability μ0, rated current I Lf_f , overload current I Lf_o , rated inductance value L f_f , Rated center column magnetic flux density B center_f , the rated current is the non-saturated current, and the overload current is the saturated current; Step S12, define optimization variables: set core optimization parameters, which include geometric parameters: magnetic plate length l up 、l up1 , magnetic plate cross-sectional area A up , the cross-sectional area of ​​the middle column A center , side column length l edge , cross-sectional area of ​​side column A edge ;Electrical parameters: Number of turns of side column N edge ; Step S13, initializing the particle swarm: setting the maximum number of iterations G and the number of populations N, and setting parameter boundaries and speed boundaries for each optimization variable; Step S14: Calculate the initial parameters of the center column: Number of turns of center column N center : Length of the middle column air gap l gap : Middle column magnetic resistance Z center : Under rated current, the middle column magnetomotive force F centerf : F centerf =N center ·I Lf_f At rated current, the side column magnetomotive force F edgef : F edgef =N edge ·I Lf_f 。 3. The method for iterative particle swarm optimization of a high-frequency converter with an embedded E-type saturated inductor according to claim 1, characterized in that: The step S2 comprises: Step S21, preset initial value: preset initial value of magnetic flux density of each part under non-saturated current: magnetic flux density of side column B edge_f , magnetic flux density B at the center of the top plate up_f , magnetic flux density at the top plate edge B up1_f ; Step S22, calculate magnetic permeability: Based on the magnetic flux density B-magnetic permeability μ curve, determine the magnetic permeability of each part Step S23, performing iterative solution; Step S24: Calculate the inductance value.

4. The method for iterative particle swarm optimization of a high-frequency converter with an embedded E-type saturated inductor according to claim 3, characterized in that: The iterative solution of step S23 specifically includes: Reluctance calculation: Side column magnetoresistance Z under non-saturation current edgef The calculation formula is Top plate center magnetic resistance Z under non-saturation current upf The calculation formula is Top plate edge magnetoresistance Z under non-saturation current up1f The calculation formula is Magnetic flux solution: Calculation of the magnetic flux of the side column under non-saturated current based on the magnetoresistance model Top plate magnetic flux under non-saturation current Column magnetic flux in non-saturated current Magnetic flux density calculation: Side column magnetic flux density B under non-saturated current nedge_f The calculation formula is Magnetic flux density B at the center of the top plate under non-saturated current nup_f The calculation formula is Magnetic flux density B at the top plate edge under non-saturated current nup1_f The calculation formula is Deviation calculation: Side column magnetic flux density deviation ΔB under non-saturation current edge_f The calculation formula is ΔB edge_f =B nedge_f -B edge_f Magnetic flux density deviation ΔB at the center of the top plate under non-saturated current up_f The calculation formula is ΔB up_f =B nup_f -B up_f Magnetic flux density ΔB at the top plate edge under non-saturated current up1_f The calculation formula is ΔB up1_f =B nup1_f -B up1_f The preset value is adjusted according to the deviation until or when the maximum number of iterations k is reached.

5. The method for iterative particle swarm optimization of a high-frequency converter with an embedded E-type saturated inductor according to claim 3, characterized in that: The inductance calculation in step S24 specifically includes: Non-saturated current mid-column winding inductance L centerf : Non-saturated current side winding inductance L edgef : The process of solving the saturated current inductance value is consistent with the process of solving the non-saturated current inductance value.

6. The method for iterative particle swarm optimization of a high-frequency converter with an embedded E-type saturated inductor according to claim 1, characterized in that: The step S3 needs to construct a fitness function, specifically including the following steps: Core volume evaluation items: Inductance variation range evaluation items: ΔL=|L centerf -L centero | Rated center column inductance offset error evaluation items: ΔL centerf =|L centerf -L f_f | Fitness function: f x =αV L +βΔL+γΔL centerf For each iteration number G, the core parameters corresponding to the particle position are input into the iterative algorithm to obtain the high non-saturated current inductance value; the particle speed and position are updated to drive the population to converge to the optimal solution.

7. The method for iterative particle swarm optimization of a high-frequency converter with an embedded E-type saturated inductor according to claim 1, characterized in that: The core structure of the E-type saturated inductor is: The four-pillar or three-pillar E-shaped magnetic core consists of an upper magnetic plate, a lower magnetic plate, a middle pillar and side pillars, with only the middle pillar having an air gap.

8. The method for iterative particle swarm optimization of a high-frequency converter with an embedded E-type saturated inductor according to claim 1, characterized in that: The magnetic resistance model includes a center column magnetic resistance, a side column magnetic resistance and a magnetic plate magnetic resistance, which are connected in parallel or in series to form a magnetic circuit.

9. The method for iterative particle swarm optimization of a high-frequency converter with an embedded E-type saturated inductor according to claim 1, characterized in that: The particle swarm algorithm updates the particle positions based on the standard model without the need for additional formulas.

Images