CN114547872A - Wireless charging system parameter optimization method based on Taguchi algorithm - Google Patents

Abstract

The invention discloses a wireless charging system parameter optimization method based on a Tiankou algorithm, and belongs to the technical field of wireless charging. The method comprises the following steps: selecting proper optimization parameters and optimization targets, determining the value range of each variable without cross coupling between the optimization parameters, screening out the value points of each variable, and equating the value points to a plurality of horizontal numbers; constructing an orthogonal table by adopting a Latin square and matrix transformation mode, and if no completely proper orthogonal table exists, reconstructing the orthogonal table by adopting a deletion method, a quasi-horizontal method, a combination method or a parallel method; processing the experimental result of the orthogonal table, and determining the influence direction and significance of each parameter on the experimental result; and developing further simulation according to the determined value range, continuously applying the orthogonal method, or performing an exhaustion method under the condition of less data points to finally obtain the parameter combination meeting the requirements. The method has the advantages of small calculated amount, easiness in jumping out of a local minimum value, independence of an accurate mathematical model and the like.

Description

Wireless charging system parameter optimization method based on Taguchi algorithm

Technical Field

The invention relates to a wireless charging system parameter optimization method based on a Taguchi algorithm, and belongs to the technical field of wireless charging.

Background

The parameter optimization refers to selecting reasonable parameters to enable the design target to reach an optimal value under the condition of meeting a series of relevant condition constraints. Common parameter optimization algorithms include precision-finding and approximation algorithms. The accurate optimization method mainly comprises linear programming, dynamic programming, integer programming, a branch-and-bound method and the like; the approximate optimization method mainly comprises a hill climbing method, a greedy method, a genetic algorithm, an orthogonal experiment method and the like. Because the precise optimization method usually needs a more precise mathematical model, the calculation complexity is higher, the method is suitable for solving a small-scale problem, and the method is often not practical in engineering. In the approximation algorithm, heuristic algorithms such as a hill climbing method and a greedy method also need relatively accurate mathematical models, and the step length and the direction of the next step are judged by experience, so that the problem of falling into a local minimum exists. Although this problem can be alleviated by designing a tabu table, it results in a large increase in the amount of calculation, and the design of the tabu table relies on a mathematical model, thereby making it difficult to play a role in parameter optimization of practical engineering problems.

Disclosure of Invention

The invention aims to provide a wireless charging system parameter optimization method based on a Taguchi algorithm, so as to solve the technical problem to be solved by the application of the invention.

A wireless charging system parameter optimization method based on a Taguchi algorithm comprises the following steps:

s100, selecting appropriate optimization parameters and optimization targets, determining the value range of each variable without cross coupling among the optimization parameters, screening out the value points of each variable, and enabling the value points to be equivalent to a plurality of horizontal numbers;

s200, constructing an orthogonal table by adopting a Latin square and matrix transformation mode, and if the orthogonal table is not completely suitable, reconstructing the orthogonal table by adopting a deletion method, a quasi-horizontal method, a combination method or a parallel method;

s300, processing the experiment result of the orthogonal table, and determining the influence direction and significance of each parameter on the experiment result;

s400, on the basis of reducing the optimal range of each parameter, further simulation is developed according to the determined value range, the orthogonal method is continuously applied, and an exhaustion method can be performed under the condition of less data points, so that a parameter combination meeting the requirements is finally obtained.

Further, in S300, the method specifically includes the following steps:

s310, generating an orthogonal experiment table according to the Latin square;

s320, testing the data in the orthogonal test table;

s330, performing data regression processing, and analyzing the correlation between variables and results;

s340, analyzing whether partial correlation exists among the variables, and if yes, executing S350; otherwise, S370 is performed;

s350, judging whether a proper value area can be selected or not, and if so, executing S370; otherwise, executing S360;

s360, regenerating the orthogonal table and returning to S320;

and S370, screening the value area which possibly meets the requirement.

Further, in S400, the method specifically includes the following steps:

s410, simulating point by point in a value area;

s420, judging whether value combinations meeting requirements exist or not, and if yes, ending the process; otherwise, go to S430;

and S430, reselecting the value area, and returning to S410.

The invention has the following beneficial effects: the invention discloses a wireless charging system parameter optimization method based on a Taguchi algorithm, which comprises the steps of carrying out a test by maximally including 'uniformly dispersed and neatly comparable' representative points in a parameter value range in a mode of pre-specifying an orthogonal table, mastering the condition of a comprehensive test on the basis of analyzing the test result of the part, and exploring the influence trend of each factor on the test result so as to determine a parameter combination meeting the requirements. Compared with an accurate optimization method and an approximate algorithm, the Taguchi algorithm has the advantages of small calculated amount, easiness in jumping out of a local minimum value, independence on an accurate mathematical model and the like. Meanwhile, the experimental design based on the Taguchi algorithm is an efficient, rapid and economic experimental method based on the orthogonal table, and provides favorable support for system parameter optimization.

The invention can effectively reduce the number of parameter combinations to be simulated, thereby greatly saving time and computing resources, and the construction method of the orthogonal standard is mature, so the method has the advantages of convenience and quickness.

Drawings

FIG. 1 is a flow chart of a Taguchi algorithm;

FIG. 2 is a schematic view of a magnetic coupling mechanism;

FIG. 3 is a scatter plot of NB0 and pianyi.

Detailed Description

The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Due to the structural complexity of the wireless charging system, the influence of each parameter is difficult to be accurately modeled, so that the application of an accurate algorithm and a heuristic algorithm partially dependent on modeling is limited. In the case of intelligent algorithms (e.g., genetic algorithms, ant colony algorithms, etc.) that rely on accurate mathematical models, these algorithms require numerous and multiple experiments without accurate mathematical models, and are difficult to be well applied in the simulation of wireless charging systems where a single experiment takes a long time. Meanwhile, as the parameters of the wireless charging system are more and the parameter value range is wider, if an exhaustion method is used for carrying out experiments, the test times are too many and the calculation complexity is too large, so that the wireless charging system is difficult to be applied to engineering; if a step method is used for experiments, the method has the problems of sensitivity to step change, dependence on user experience and the like.

Aiming at the problems, the invention provides a Tiankou algorithm. Because the selected orthogonal table is derived by pure mathematics and is designed only for ensuring that each factor can participate in the experiment, the scheme has no dependence on an accurate model; because the parameters only comprise the number of the factors and the horizontal number of each factor when the orthogonal table is constructed, the method reduces the risks of low coverage rate and redundant test cases caused by artificial test to the maximum extent, and has the advantages of saving time, controlling the number of the test cases, ensuring the coverage rate of the test cases and the like.

Referring to fig. 1, a method for optimizing parameters of a wireless charging system based on a tian kou algorithm includes the following steps:

s100, selecting appropriate optimization parameters and optimization targets, determining the value range of each variable without cross coupling among the optimization parameters, screening out the value points of each variable, and enabling the value points to be equivalent to a plurality of horizontal numbers;

s200, constructing an orthogonal table according to a plurality of equivalent horizontal numbers by adopting a Latin square and matrix transformation mode, and if the orthogonal table is not completely suitable, reconstructing the orthogonal table by adopting a deletion method, a quasi-horizontal method, a combination method or a parallel method;

s300, processing the experiment result of the orthogonal table, and determining the influence direction and significance of each parameter on the experiment result;

s400, on the basis of reducing the optimal range of each parameter, further simulation is developed according to the determined value range, the orthogonal method is continuously applied, and an exhaustion method can be performed under the condition of less data points, so that a parameter combination meeting the requirements is finally obtained.

Specifically, in S100, the optimization parameters are selected according to actual situations, and are usually the number of turns of the coil, the transmission distance of the wireless charging system, the position relationship between the coil and the coil, the turn-to-turn distance inside the coil, and the size and position of the magnetic core. As in the embodiments described below, the parameter to be optimized is the number of turns of 4 coils.

The value ranges of the variables should be constructed according to actual conditions. For example, in the optimization of the wireless charging system, the actual conditions are the external dimensions, the diameter of the cable, the transmission distance and the like.

Further, in S300, the method specifically includes the following steps:

s310, generating an orthogonal experiment table according to the Latin square;

s320, testing the data in the orthogonal test table;

s330, performing data regression processing, and analyzing the correlation between variables and results;

s340, analyzing whether partial correlation exists among the variables, and if yes, executing S350; otherwise, S370 is performed;

s350, judging whether a proper value area can be selected or not, and if so, executing S370; otherwise, executing S360;

s360, regenerating the orthogonal table and returning to S320;

and S370, screening the value area which possibly meets the requirement.

Further, in S400, the method specifically includes the following steps:

s410, simulating point by point in a value area;

s420, judging whether value combinations meeting requirements exist or not, and if yes, ending the process; otherwise, go to S430;

and S430, reselecting the value area, and returning to S410.

The following is a specific embodiment of the present invention:

as shown in fig. 2, the technical advantage of the tian kou algorithm proposed by the present invention is verified by taking the optimization of the offset performance of the magnetic coupling mechanism in the wireless charging system as an example.

Optimizing parameters: the number of turns NA of the transmitting end coil A and 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 are all 1-19. Optimizing the target: the mutual inductance value is kept about 11 muH, and the mutual inductance value fluctuation is not more than +/-5% within the horizontal offset range of 0-90 mm. Each set of experiment includes 10 times of simulation (0-90 mm, 10mm is step length), and the anti-offset capability and the minimum mutual inductance value in the range are analyzed. Each parameter is divided into 9 horizontal numbers representing 2 turns, 4 turns, …, 18 turns, limited by computer performance and orthogonal table generation capability. Construction of orthogonal tables (L) using Latin squares and matrix transforms₈₁(9¹⁰) 81 experiments covering 10 levels of 9, deleting useless variables (6) and deletingRepeating the rows to obtain the orthogonal table (L)₈₁(9⁴) As shown in orthogonal design table 1:

orthogonal design Table 1

Orthogonal design Table 1

Orthogonal design Table 1

TABLE 1

In orthogonal design Table 1, factor 1 corresponds to the number of levels of NA, factor 2 corresponds to the number of levels of NB, factor 3 corresponds to the number of levels of NA0, and factor 4 corresponds to the number of levels of NB 0. Taking the first set of experiments as an example, the horizontal numbers are 6, 2, 9, and 9 respectively, and according to the corresponding relationship between the horizontal numbers and the number of turns, the horizontal numbers represent the number of turns of the transmitting terminal NA, the transmitting terminal NB, the receiving terminal NA0, and the receiving terminal NB0 are 12, 4, 18, and 18 respectively. Each set of parameter combinations was tested separately, a total of 81 tests were required, and the results are summarized in table 2:

the data in table 2 are arranged in descending order of the reciprocal of the rate of change of the mutual inductance values. Pinanyi represents the reciprocal of the mutual inductance change rate, Hugan represents the mutual inductance sequence set to 1 when the minimum mutual inductance is greater than 11 muh, and otherwise set to 0. Since there is no interaction between the four variables (NA, NA0, NB0), the four variables can be treated independently in the experiment. In order to find out the direction and significance of the influence of four variables on a target, the invention adopts hierarchical regression to analyze data and draw the following conclusion: to achieve better offset resistance and higher mutual inductance, larger NA and NA0 are selected, NB should be slightly smaller, and NB0 needs to be further discussed.

As can be seen from fig. 3: NB and NB0 have stronger anti-deviation performance when the middle level number is taken as a value (4-6, corresponding to 8-12 turns in practice).

In conclusion, the multi-objective optimization fully considers the directionality and the significance of the influence of the variables on the result. Since both NA and NA0 have a positive effect on both parameters. When further optimizing, the values of the two parameters are proper larger, and although NB has obvious negative influence on the offset resistance, the NB has low significance and has little influence on the mutual inductance value, and the NB is proper. Similarly, the value of NB0 should be moderate. Considering the directionality and the significance of the influence of the variables on the result, and combining a scatter diagram to perform multi-objective optimization, the following two groups of more reasonable value ranges are obtained:

a first group: NA: 14-17, NA 0: 14-17, NB: 7-9, NB 0: 9-11; (NA and NA0 are as large as possible, with NB and NB0 taking intermediate values)

Second group: NA, NA 0: 13-15, NB 0: 8 to 12. (taking into account that a larger NA may cause too little influence on NB and NB0 and thus poor anti-drift performance, NA and NA0 take relatively small values (but still within a larger range of values) to expand the range of NB and NB0 values)

Because data points in the interval are few, each combination is simulated by using an exhaustion method, and finally, some parameter combinations with better anti-offset performance are obtained as follows:

considering that the mutual inductance value is about 11 muH, the invention selects two groups of parameters (13, 14, 12, 12) and (14, 13, 11, 12) to carry out comparative experiments. In terms of speed, with CPUi79700@4.5GHzAnd ANSYS version 2021R2 for example, the time taken to calculate the data for the orthogonal table is about 6 hours (10 sets of data analysis offset performance are calculated per point) plus 180+225 data points in screening the data for a total of about 36 hours. If the Taguchi algorithm is not adopted, each variable is scanned according to the step length of 2, the total number of data points is about 6561, and the simulation by the computer takes about 486 hours. Obviously, the optimal value range of the variable is screened by adopting the Taguchi algorithm, so that the calculation time and the calculation load are effectively saved.

Claims (3)

1. A wireless charging system parameter optimization method based on Taguchi algorithm is characterized by comprising the following steps:

s100, selecting appropriate optimization parameters and optimization targets, wherein the optimization parameters are not cross-coupled, the value range of each variable is determined, the value points of each variable are screened out and are equivalent to a plurality of horizontal numbers;

s200, constructing an orthogonal table by adopting a Latin square and matrix transformation mode, and if the orthogonal table is not completely suitable, reconstructing the orthogonal table by adopting a deletion method, a quasi-horizontal method, a combination method or a parallel method;

s300, processing the experiment result of the orthogonal table, and determining the influence direction and significance of each parameter on the experiment result;

s400, on the basis of reducing the optimal range of each parameter, further simulation is developed according to the determined value range, the orthogonal method is continuously applied, and an exhaustion method can be performed under the condition of less data points, so that a parameter combination meeting the requirements is finally obtained.

2. The method for optimizing parameters of a wireless charging system based on the Taguchi algorithm according to claim 1, wherein in S300, the method specifically comprises the following steps:

s310, generating an orthogonal experiment table according to the Latin square;

s320, testing the data in the orthogonal test table;

s330, performing data regression processing, and analyzing the correlation between variables and results;

s340, analyzing whether partial correlation exists among the variables, and if yes, executing S350; otherwise, go to S370;

s350, judging whether a proper value area can be selected or not, and if so, executing S370; otherwise, executing S360;

s360, regenerating the orthogonal table and returning to S320;

and S370, screening the value area which possibly meets the requirement.

3. The method for optimizing parameters of a wireless charging system based on the Taguchi algorithm according to claim 2, wherein in S400, the method specifically comprises the following steps:

s410, point-by-point simulation is carried out in a value area;

s420, judging whether value combinations meeting requirements exist or not, and if yes, ending the process; otherwise, go to S430;

and S430, reselecting the value area, and returning to S410.

Images