A wireless charging system parameter optimization method based on Taguchi algorithm

Abstract

The present invention discloses a wireless charging system parameter optimization method based on Taguchi algorithm, which belongs to the field of wireless charging technology. The method comprises: selecting appropriate optimization parameters and optimization targets, ensuring no cross-coupling between the optimization parameters and determining the value range of each variable, screening out the value points of each variable, and converting them into multiple levels; constructing an orthogonal table by Latin square and matrix transformation, and if there is no completely suitable orthogonal table, selecting a deletion method, a quasi-level method, a combination method, or a parallel method to transform the orthogonal table; processing the experimental results of the orthogonal table, and determining the direction and significance of the influence of each parameter on the experimental results; carrying out further simulation according to the determined value range, continuing to apply the orthogonal method, and also carrying out an exhaustive method when there are fewer data points, to finally obtain a parameter combination that meets the requirements. The present invention has the advantages of low computational complexity, easy to jump out of the local minimum, and no reliance on precise mathematical models.

Description

Technical Field

[0001] The present invention relates to a wireless charging system parameter optimization method based on Taguchi algorithm, and belongs to the technical field of wireless charging. Background Art

[0002] Parameter optimization involves selecting appropriate parameters to achieve the optimal value of a design objective while satisfying a series of relevant constraints. Common parameter optimization algorithms include exact optimization methods and approximate algorithms. Exact optimization methods primarily include linear programming, dynamic programming, integer programming, and branch-and-bound methods; approximate optimization methods primarily include hill climbing, greedy algorithms, genetic algorithms, and orthogonal experimental methods. Exact optimization methods typically require relatively precise mathematical models and are computationally complex, making them suitable for solving small-scale problems and often impractical in engineering. Among approximate algorithms, heuristic algorithms such as hill climbing and greedy algorithms also require relatively precise mathematical models and rely on experience to determine the next step size and direction, which can lead to the problem of being trapped in local minima. While this problem can be alleviated by designing tabu tables, this significantly increases the computational effort. Furthermore, the design of tabu tables relies on mathematical models, making them difficult to use in parameter optimization for practical engineering problems. Summary of the Invention

[0003] The purpose of the present invention is to propose a wireless charging system parameter optimization method based on Taguchi algorithm to solve the technical problem to be solved by the present invention application.

[0004] A wireless charging system parameter optimization method based on Taguchi algorithm, the wireless charging system parameter optimization method comprising the following steps:

[0005] S100, selecting appropriate optimization parameters and optimization targets, with no cross-coupling between the optimization parameters and determining the value range of each variable, screening the value points of each variable, and converting them into multiple level numbers;

[0006] S200, constructing an orthogonal array using Latin square and matrix transformation. If there is no completely suitable orthogonal array, modifying the orthogonal array using deletion method, quasi-horizontal method, combination method or parallel method;

[0007] S300, processing the experimental results of the orthogonal array to determine the direction and significance of the influence of each parameter on the experimental results;

[0008] S400. On the basis of narrowing the optimal range of each parameter, further simulation is carried out according to the determined value range, and the orthogonal method is continued to be applied. The exhaustive method can also be used when there are fewer data points, and finally a parameter combination that meets the requirements is obtained.

[0009] Furthermore, in S300, the following steps are specifically included:

[0010] S310, generating an orthogonal experiment table according to the Latin square;

[0011] S320, conducting an experiment on the data in the orthogonal experiment table;

[0012] S330, perform data regression processing to analyze the correlation between variables and results;

[0013] S340, analyzing whether there is partial correlation between the variables, if so, executing S350; otherwise, executing S370;

[0014] S350, determine whether a suitable value range can still be selected, if so, execute S370; otherwise, execute S360;

[0015] S360, regenerate the orthogonal array and return to S320;

[0016] S370. Filter the value range that may meet the requirements.

[0017] Furthermore, in S400, the following steps are specifically included:

[0018] S410, point-by-point simulation within the value region;

[0019] S420: Determine whether there is a value combination that meets the requirements. If so, the process ends; otherwise, execute S430;

[0020] S430. Reselect the value range and return to S410.

[0021] The present invention has the following beneficial effects: The present invention provides a method for optimizing wireless charging system parameters based on the Taguchi algorithm. By pre-specifying an orthogonal table, the method maximizes the inclusion of representative points in the parameter value range that are "evenly dispersed, neatly comparable" for testing. The results of this part of the test are analyzed to understand the overall test results, explore the influence trend of each factor on the experimental results, and thus determine a parameter combination that meets the requirements. Compared with precise optimization methods and approximate algorithms, the Taguchi algorithm has the advantages of low computational complexity, easy escape from local minima, and no reliance on precise mathematical models. At the same time, the experimental design based on the Taguchi algorithm is an efficient, fast, and economical experimental method based on an orthogonal table, providing favorable support for system parameter optimization.

[0022] Furthermore, the use of the present invention can effectively reduce the number of parameter combinations that need to be simulated, thereby significantly saving time and computing resources. Furthermore, since the construction method of the orthogonal scale is relatively mature, the use of this method has the advantages of being convenient and fast. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is the flowchart of Taguchi algorithm;

[0024] Figure 2 Schematic diagram of magnetic coupling mechanism;

[0025] Figure 3 This is a scatter plot of NB0 and pianyi. DETAILED DESCRIPTION

[0026] 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. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0027] Due to the structural complexity of wireless charging systems, it is difficult to accurately model the impact of various parameters, which limits the application of precise algorithms and heuristic algorithms that partially rely on modeling. As for intelligent algorithms that rely on precise mathematical models (such as genetic algorithms and ant colony algorithms), since these algorithms require a large number of experiments in the absence of precise mathematical models, they are difficult to apply effectively in the simulation of wireless charging systems, where a single experiment takes a long time. At the same time, since wireless charging systems have many parameters and a wide range of parameter values, if exhaustive experiments are used, the number of trials is too large and the computational complexity is too high, making it difficult to apply in engineering. If the step size method is used for experiments, this method has problems such as sensitivity to step size changes and reliance on user experience.

[0028] To address these issues, the present invention proposes the Taguchi algorithm. Because the orthogonal array it selects is purely mathematically derived and designed solely to ensure that every factor can participate in the experiment, the scheme is independent of the precise model. Furthermore, since the parameters used in constructing the orthogonal array include only the number of factors and the number of levels for each factor, this method minimizes the risks of low coverage and redundant test cases caused by manual testing, saving time, limiting the number of test cases, and ensuring test case coverage.

[0029] Reference Figure 1 As shown, a wireless charging system parameter optimization method based on Taguchi algorithm includes the following steps:

[0030] S100, selecting appropriate optimization parameters and optimization targets, with no cross-coupling between the optimization parameters and determining the value range of each variable, screening the value points of each variable, and converting them into multiple level numbers;

[0031] S200, using Latin square and matrix transformation to construct an orthogonal array based on multiple equivalent levels. If there is no completely suitable orthogonal array, select a deletion method, a quasi-level method, a combination method, or a parallel method to transform the orthogonal array;

[0032] S300, processing the experimental results of the orthogonal array to determine the direction and significance of the influence of each parameter on the experimental results;

[0033] S400. On the basis of narrowing the optimal range of each parameter, further simulation is carried out according to the determined value range, and the orthogonal method is continued to be applied. The exhaustive method can also be used when there are fewer data points, and finally a parameter combination that meets the requirements is obtained.

[0034] Specifically, in S100 , the optimization parameters are selected based on actual conditions, typically the number of coil turns, the transmission distance of the wireless charging system, the positional relationship between coils, the turn spacing within the coils, and the size and position of the magnetic core. For example, in the following embodiment, the parameters to be optimized are the number of turns of the four coils.

[0035] The value ranges of each variable should be constructed based on actual conditions. For example, in the optimization of wireless charging systems, the actual conditions are the external dimensions, cable diameter, and transmission distance.

[0036] Furthermore, in S300, the following steps are specifically included:

[0037] S310, generating an orthogonal experiment table according to the Latin square;

[0038] S320, conducting an experiment on the data in the orthogonal experiment table;

[0039] S330, perform data regression processing to analyze the correlation between variables and results;

[0040] S340, analyzing whether there is partial correlation between the variables, if so, executing S350; otherwise, executing S370;

[0041] S350, determine whether a suitable value range can still be selected, if so, execute S370; otherwise, execute S360;

[0042] S360, regenerate the orthogonal array and return to S320;

[0043] S370. Filter the value range that may meet the requirements.

[0044] Furthermore, in S400, the following steps are specifically included:

[0045] S410, point-by-point simulation within the value region;

[0046] S420: Determine whether there is a value combination that meets the requirements. If so, the process ends; otherwise, execute S430;

[0047] S430. Reselect the value range and return to S410.

[0048] The following is a specific embodiment of the present invention:

[0049] like Figure 2 As shown in FIG, the technical advantages of the Taguchi algorithm proposed in the present invention are verified by taking the offset performance optimization of the magnetic coupling mechanism in the wireless charging system as an example.

[0050] Optimization parameters: the number of turns of the transmitting coil A is NA, the number of turns of the receiving coil A is NA0; the number of turns of the transmitting coil B is NB, the number of turns of the receiving coil B is NB0, and the value range is 1 to 19. Optimization goal: the mutual inductance value is kept at about 11μH, and the fluctuation of the mutual inductance value within the horizontal offset range of 0 to 90mm is no more than ±5%. Each set of experiments includes 10 simulations (0 to 90mm, 10mm as a step size) to analyze its anti-offset capability and the minimum mutual inductance value within the range. Due to the limitations of computer performance and orthogonal table generation capabilities, each parameter is divided into 9 horizontal numbers, representing 2 turns, 4 turns, ..., 18 turns. Use Latin square and matrix transformation to construct the orthogonal table (L 81 (9 10 )), 81 groups of experiments covering 10 9-level numbers, deleting useless variables (6) and deleting duplicate rows, and finally obtaining the orthogonal array (L 81 (9 4 ))As shown in the orthogonal design table 1:

[0051] Orthogonal design table 1

[0052]

[0053] Orthogonal design table 1

[0054]

[0055] Orthogonal design table 1

[0056]

[0057] Table 1

[0058] In the orthogonal design table 1, factor 1 corresponds to the number of NA levels, factor 2 corresponds to the number of NB levels, factor 3 corresponds to the number of NA0 levels, and factor 4 corresponds to the number of NB0 levels. Taking the first set of experiments as an example, the levels were 6, 2, 9, and 9, respectively. Based on the correspondence between the level number and the number of turns, the levels represent the number of turns for the transmitter NA, the transmitter NB, the receiver NA0, and the receiver NB0, respectively: 12, 4, 18, and 18. Each parameter combination was tested separately, requiring a total of 81 experiments, the results of which are summarized in Table 2:

[0059]

[0060]

[0061]

[0062] The data in Table 2 are arranged in descending order according to the reciprocal of the mutual inductance change rate. Pianyi represents the reciprocal of the mutual inductance change rate, and Hugan represents that the mutual inductance column is set to 1 when the minimum mutual inductance is greater than 11μH, otherwise it is set to 0. Since there is no interaction between the above four variables (NA, NA0, NB, NB0), the four variables can be processed independently in the experiment. In order to find out the direction and significance of the influence of the four variables on the target, the present invention uses hierarchical regression to analyze the data and draws the following conclusions: In order to achieve better anti-offset capability and higher mutual inductance values, larger NA and NA0 are selected, NB should be slightly smaller, and the selection of NB0 needs further discussion.

[0063]

[0064] Depend on Figure 3 It can be seen that NB and NB0 have stronger anti-offset performance when the number of levels is intermediate (4 to 6, corresponding to 8 to 12 turns in practice).

[0065] In summary, the directionality and significance of the variables' impact on the results should be fully considered during multi-objective optimization. Since NA and NA0 have a positive impact on both parameters, the values ​​of these two parameters should be appropriately larger during further optimization. Although NB has a significant negative impact on anti-offset performance, its significance is low and its impact on mutual inductance is not significant, so it should be moderately valued. Similarly, the value of NB0 should be moderate. Taking into account the directionality and significance of the variables' impact on the results mentioned above, and combining it with the scatter plot for multi-objective optimization, the following two sets of relatively reasonable value ranges are obtained:

[0066] Group 1: NA: 14-17, NA0: 14-17, NB: 7-9, NB0: 9-11; (NA and NA0 should be as large as possible, and NB and NB0 should take the middle value)

[0067] Group 2: NA, NA0: 13-15, NB, NB0: 8-12. (Considering that a large NA may cause the influence of NB and NB0 to be too small, resulting in poor anti-skew performance, NA and NA0 are set to relatively small values ​​(but still within the larger value range), expanding the range of NB and NB0 values.)

[0068] Since there are few data points in the interval, an exhaustive method is used to simulate each combination, and finally some parameter combinations with better anti-drift performance are obtained as follows:

[0069]

[0070] Considering that the mutual inductance is about 11μH, the present invention selects two sets of parameters (13, 14, 12, 12) and (14, 13, 11, 12) to conduct comparative experiments. i79700@4.5GHz Taking ANSYS version 2021R2 as an example, calculating the data required for the orthogonal array takes approximately 6 hours (calculating 10 sets of data for each point to analyze offset performance), plus filtering the data for 180 + 225 points, for a total of approximately 36 hours. Without the Taguchi algorithm, scanning each variable with a step size of 2, the total number of data points is approximately 6,561, and the simulation time on this computer takes approximately 486 hours. Clearly, using the Taguchi algorithm to filter the optimal range of variables effectively saves computational time and load.

Claims

1. A wireless charging system parameter optimization method based on Taguchi algorithm, characterized in that: The wireless charging system parameter optimization method comprises the following steps: S100, selecting appropriate optimization parameters and optimization objectives, wherein there is no cross-coupling between the optimization parameters and the value range of each variable is determined, the value points of each variable are screened out, and the values ​​are equivalent to multiple horizontal numbers. The optimization parameters include the number of turns NA of the transmitting end coil A, the number of turns NA0 of the receiving end coil A, the number of turns NB of the transmitting end coil B, and the number of turns NB0 of the receiving end coil B, all of which have a value range of 1 to 19; the optimization objective is to maintain the mutual inductance value at approximately 11 μH, and the mutual inductance value fluctuation within the horizontal offset range of 0 to 90 mm is no more than ±5%; S200, constructing an orthogonal array using Latin square and matrix transformation. If there is no completely suitable orthogonal array, modifying the orthogonal array using deletion method, quasi-horizontal method, combination method or parallel method; S300, process the experimental results of the orthogonal array to determine the direction and significance of the influence of each parameter on the experimental results, In S300, the following steps are specifically included: S310, generating an orthogonal experiment table according to the Latin square; S320, conducting an experiment on the data in the orthogonal experiment table; S330, perform data regression processing to analyze the correlation between variables and results; S340, analyzing whether there is partial correlation between the variables, if so, executing S350; otherwise, executing S370; S350, determine whether a suitable value range can still be selected, if so, execute S370; otherwise, execute S360; S360, regenerate the orthogonal array and return to S320; S370, screening the value range that may meet the requirements; S400. On the basis of narrowing the optimal range of each parameter, further simulation is carried out according to the determined value range, and the orthogonal method is continued to be applied. The exhaustive method can also be used when there are fewer data points, and finally a parameter combination that meets the requirements is obtained.

2. The wireless charging system parameter optimization method based on Taguchi algorithm according to claim 1, characterized in that: In S400, the following steps are specifically included: S410, point-by-point simulation within the value region; S420: Determine whether there is a value combination that meets the requirements. If so, the process ends; otherwise, execute S430; S430. Reselect the value range and return to S410.

Images