Method for extracting equivalent parameters of cavity resonant wireless power transmission system

Abstract

This invention provides a method for extracting equivalent parameters of a cavity resonant wireless power transfer system, comprising the following steps: S1: acquiring system input impedance spectrum data; S2: using the system input impedance spectrum data as input, and obtaining preliminary predicted values ​​of the system equivalent parameters using a trained LSTM regression model; S3: based on the preliminary predicted values ​​of the system equivalent parameters, using a nonlinear fitting algorithm to regression fit and obtain accurate values ​​of the system equivalent parameters; the system equivalent parameters include the cavity internal resistance, cavity equivalent inductance, and mutual inductance between the coil and the resonant cavity. Its advantages are: it can effectively extract equivalent parameters of cavity resonant wireless power transfer systems under all current mainstream modes; it avoids the convergence failure or getting trapped in local optima caused by the NLS algorithm due to the initial values ​​deviating significantly from the true values, and possesses extremely high fitting accuracy and stability.

Description

Technical Field

[0001] This invention relates to wireless power transmission technology, specifically, to a method for extracting equivalent parameters of a cavity resonant wireless power transmission system. Background Technology

[0002] Traditional wireless power transfer methods transmit electrical energy wirelessly directly through two coupled coils or electrode plates, enabling high-power wireless power transfer. However, these systems have significant drawbacks: limited transmission distance, with system efficiency dropping drastically as the distance approaches the diameter of the coils themselves. Furthermore, wireless charging systems using electrode plates, which transfer energy through coupled electric fields, pose risks to biosafety.

[0003] Researchers have proposed adding capacitors to resonant cavities to construct quasi-static resonant cavity wireless power transfer systems, enabling mid-range wireless charging within a room-scale range. The applicant has also proposed developing a quasi-static resonant cavity wireless power transfer system with a more uniform magnetic field and higher resistance to angular deviations through special geometric design and electromagnetic control methods, further improving the system's transmission performance. However, establishing an equivalent mathematical model for quasi-static resonant cavity wireless power transfer systems has been hindered because it is impossible to directly measure the parameters of the cavity itself, and to obtain data on the cavity's resistance, inductance, and mutual inductance between the cavity and the coil. This makes it difficult to establish an equivalent mathematical model. Existing finite element simulation schemes have limited model accuracy due to significant errors compared to the actual system, and cannot simulate parameters such as nodal voltages. Schemes based on reflection impedance analysis inevitably suffer from random errors due to random point selection, resulting in highly unstable mathematical model accuracy. Both of these approaches struggle to accurately establish an equivalent mathematical model for quasi-static resonant cavity wireless power transfer systems. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a method for extracting equivalent parameters of a cavity resonant wireless power transfer system. By using LSTM (Long Short-Term Memory Neural Network) and combining it with a nonlinear fitting (NLS-based) algorithm, the equivalent parameters of the cavity resonant wireless power transfer system are extracted, effectively solving the problem of constructing an accurate equivalent mathematical model of the cavity resonant wireless power transfer system.

[0005] To achieve the above objectives, the specific technical solution adopted by the present invention is as follows:

[0006] A method for extracting equivalent parameters of a cavity resonant wireless power transfer system, the key of which includes the following steps:

[0007] S1: Acquire system input impedance spectrum data;

[0008] S2: Using the system input impedance spectrum data as input, the pre-trained LSTM regression model is used to obtain preliminary predicted values ​​of the system equivalent parameters;

[0009] S3: Based on the preliminary predicted values ​​of the system equivalent parameters, a nonlinear fitting algorithm is used to regress and fit to obtain the accurate values ​​of the system equivalent parameters;

[0010] The system equivalent parameters include the cavity internal resistance, cavity equivalent inductance, and mutual inductance between the coil and the resonant cavity.

[0011] The training set data for the LSTM comes from MATLAB, and the final prediction input is the measured impedance spectrum of the system. The initial value of the nonlinear fitting algorithm is the predicted output of the LSTM, and the fitting target is the measured impedance spectrum of the system. Based on the step regression fitting of the two algorithms, the high-precision equivalent parameters and mathematical model of the system are finally obtained.

[0012] Optionally, the electrical parameters of the transmitting coil and the receiving coil are determined by electrical measurements, including the self-inductance, internal resistance, and resonant capacitance of the transmitting coil, as well as the self-inductance, internal resistance, and resonant capacitance of the receiving coil.

[0013] Alternatively, if the transmitting coil and resonant cavity are equivalent to a dual-coil mutual inductance model, then the input impedance at the transmitting coil end is:

[0014]

[0015] Where: Z1 represents the input impedance of the transmitting coil, R1 represents the internal resistance of the transmitting coil, and R q M represents the internal resistance of the resonant cavity, ω represents the system's operating angular frequency, and M represents the internal resistance of the cavity. 1q L represents the mutual inductance between the transmitting coil and the resonant cavity. q C represents the cavity equivalent inductance of the resonant cavity. q Let J represent the resonant capacitance of the resonant cavity, j represent the imaginary part, L1 represent the self-inductance of the transmitting coil, and C1 represent the resonant capacitance of the transmitting coil.

[0016] Alternatively, if the receiving coil and resonant cavity are equivalent to a dual-coil mutual inductance model, then the input impedance at the receiving coil end is:

[0017]

[0018] Where Z2 represents the input impedance of the receiving coil, R2 represents the internal resistance of the receiving coil, and R q M represents the internal resistance of the resonant cavity, ω represents the system's operating angular frequency, and M represents the internal resistance of the cavity. 2q L represents the mutual inductance between the receiving coil and the resonant cavity. q C represents the cavity equivalent inductance of the resonant cavity.q L1 represents the resonant capacitance of the resonant cavity, j represents the imaginary part, L2 represents the self-inductance of the receiving coil, and C2 represents the resonant capacitance of the receiving coil.

[0019] Optionally, the hidden layer shape of the LSTM regression model is 4×128, and the number of layers is 3. The input training data during model training includes frequency, real part of input impedance, imaginary part of input impedance, phase of input impedance, and label, wherein the label corresponds to the internal resistance of the cavity, the equivalent inductance of the cavity, and the mutual inductance between the coil and the resonant cavity.

[0020] Optionally, the nonlinear fitting algorithm uses minimizing the sum of squared residuals as its core algorithm.

[0021] Alternatively, training data for the LSTM regression model can be generated using circuit simulation or mathematical analysis software.

[0022] Optionally, the bandwidth of the training data is set to 40–60 kHz.

[0023] Optionally, depending on the system's operating mode, a distributed approach is adopted to extract the system's equivalent parameters, ensuring that only one coil is extracted each time.

[0024] The so-called step-by-step parameter extraction first excludes the receiving coil, then measures the system input impedance containing only the transmitting coil and the cavity, and finally obtains the equivalent parameter, which is the cavity resistance R. q Inductor L q The mutual inductance value M between the transmitting coil and the cavity 1q Then, excluding the transmitting coil, the system input impedance, consisting only of the receiving coil and the cavity, is measured. The final equivalent parameter obtained is the cavity resistance R. q Inductor L q The mutual inductance value M between the receiving coil and the cavity 2q .

[0025] Optionally, the system operating modes include PI mode (no conductive electrode mode), O-QSCR mode (open cavity mode) and UMPI mode (uniform magnetic field no conductive electrode mode).

[0026] Based on the above design, the present invention has the following advantages:

[0027] (1) The present invention proposes an equivalent parameter extraction method for cavity resonant wireless power transmission system, which uses LSTM combined with nonlinear fitting algorithm to effectively extract the equivalent parameters of cavity resonant wireless power transmission system under all current mainstream modes.

[0028] (2) Since the LSTM network algorithm is used to make preliminary predictions of the equivalent parameters of the system, the predicted output is used as the initial input for nonlinear fitting, which avoids the failure of the NLS algorithm to converge or get stuck in a local optimum due to the initial value being seriously deviated from the true value.

[0029] (3) Since this method is to fit a segment of spectrum data as a whole, the nonlinear fitting algorithm performs error analysis on each point in the measurement frequency band, avoiding random errors caused by random point selection in theoretical calculations. Therefore, it has extremely high fitting accuracy and stability.

[0030] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0031] The accompanying drawings of this invention are described below.

[0032] Figure 1 The diagram shows the mainstream operating modes of existing cavity resonant wireless power transfer systems, including magnetic field strength, direction and current direction, where (a) is PI mode, (b) is O-QSCR mode and (c) is UMPI mode.

[0033] Figure 2 This is a schematic diagram of the main structure of the system in a specific embodiment of the present invention;

[0034] Figure 3 This is a flowchart illustrating the parameter extraction process in a specific embodiment of the present invention;

[0035] Figure 4 for Figure 1 The system equivalent circuit topology diagrams corresponding to the three operating modes; where: R L For the load resistor, (a) PI mode, (b) O-QSCR mode, (c) UMPI mode;

[0036] Figure 5 This is a schematic diagram of the measurement points in the experimental verification process of this invention;

[0037] Figure 6 This is a comparison curve of the output voltage and transmission efficiency of the experimental measurement and mathematical model in this invention. Detailed Implementation

[0038] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0039] like Figure 1 As shown, the equivalent parameter extraction method for a cavity resonant wireless power transfer system provided in this embodiment is applicable to the three existing mainstream operating modes, but the target extraction object is not limited to... Figure 1 The resonant cavity body used in the extraction method provided in this example is made of aluminum, with capacitors connecting the walls. Under the current path on the resonant cavity surface, it can be considered as one or two decoupled series RLC circuits. In this embodiment, the resonant cavity is an octagonal prism, with a 1:1 ratio between the width of the side opening and the width of the side cavity wall. In this embodiment, the number and position of the transmitting coils differ in the three different modes; please refer to the relevant documentation for details. Figure 1 The location of the "transmitting coil" is specified. In this embodiment, the transmitting coil is a helical square planar coil, and the receiving coil is a helical square planar coil. The receiving coil can be located at any position within the resonant cavity. Figure 1 The receiving coil is placed at the center of the cavity, where the optimal angular offset characteristics can be obtained.

[0040] In this embodiment, a cavity structure is used to complete the operation and experimental verification of the three modes. The structural parameters are: h = 285mm, w = 200mm, l = 400mm, corresponding to... Figure 2 As shown.

[0041] Figure 3 This is a flowchart of the parameter extraction method of the present invention, which includes the following steps:

[0042] S1: Acquire system input impedance spectrum data. The impedance data here can come from actual measurements.

[0043] S2: Using the system input impedance spectrum data as input, the pre-trained LSTM regression model is used to obtain preliminary predicted values ​​of the system equivalent parameters;

[0044] S3: Based on the preliminary predicted values ​​of the system's equivalent parameters, a nonlinear fitting algorithm is used to regress and fit to obtain the accurate values ​​of the system's equivalent parameters; the nonlinear fitting (NLS) algorithm uses minimizing the sum of squared residuals as the core algorithm, specifically using Python: scipy.optimize.curve_fit(NLS).

[0045] The system equivalent parameters include the cavity internal resistance, cavity equivalent inductance, and mutual inductance between the coil and the resonant cavity.

[0046] In practical implementation, the first step is to generate the LSTM network training set data using circuit simulation or mathematical analysis software (such as MATLAB). The target topology network is then established in the software, and several sets of circuit parameters are randomly generated. This patent uses an SS (series-series) topology as an example to simulate the circuit model when the transmitting or receiving coil is coupled to the cavity.

[0047] According to the skin effect, the system resistance changes with frequency. However, the training set obtained by simulation cannot accurately reproduce this feature. In order to minimize this influence, this embodiment minimizes the bandwidth as much as possible, but needs to retain the complete characteristics of the impedance spectrum. Therefore, the bandwidth of the training set data is set to 40-60kHz.

[0048] In this embodiment, the model is generated using MATLAB mathematical analysis. For the X or Y direction, based on the current flow direction of the resonant cavity in a specific operating mode, the resonant cavity and the capacitor on the cavity wall are constructed as a series LC equivalent circuit model. Considering the resistance value of the cavity wall, and based on the impedance characteristics of the SS topology network, the input impedance at the transmitting coil end is:

[0049]

[0050] Where: Z1 represents the input impedance of the transmitting coil, R1 represents the internal resistance of the transmitting coil, and R q M represents the internal resistance of the resonant cavity, ω represents the system's operating angular frequency, and M represents the internal resistance of the cavity. 1q L represents the mutual inductance between the transmitting coil and the resonant cavity. q C represents the cavity equivalent inductance of the resonant cavity. q Let J represent the resonant capacitance of the resonant cavity, j represent the imaginary part, L1 represent the self-inductance of the transmitting coil, and C1 represent the resonant capacitance of the transmitting coil.

[0051] Furthermore, if the receiving coil and resonant cavity are equivalent to a dual-coil mutual inductance model, then the input impedance at the receiving coil end is:

[0052]

[0053] Where Z2 represents the input impedance of the receiving coil, R2 represents the internal resistance of the receiving coil, and R q M represents the internal resistance of the resonant cavity, ω represents the system's operating angular frequency, and M represents the internal resistance of the cavity. 2q L represents the mutual inductance between the receiving coil and the resonant cavity. q C represents the cavity equivalent inductance of the resonant cavity. q L1 represents the resonant capacitance of the resonant cavity, j represents the imaginary part, L2 represents the self-inductance of the receiving coil, and C2 represents the resonant capacitance of the receiving coil.

[0054] Taking the transmitting coil as an example, R in the equation qL q and M 1q The parameter is unknown; all other parameters can be directly measured. Using MATLAB, 100,000 sets of R values ​​are randomly generated within a reasonable range. q L q and M 1q The corresponding impedance spectrum data is obtained, and the data is preprocessed to generate frequency, real part of impedance, imaginary part of impedance, impedance phase, and tag (R). q L q and M 1q ), and package them into a training set.

[0055] In the LSTM network framework, the data channels are set to 4 channels (frequency, real part of impedance, imaginary part of impedance, and impedance phase), and data normalization is added to improve model training efficiency. Gaussian noise layers are turned off to avoid data pollution. The hidden layer shape is set to 4×128, the number of layers is set to 3, and the epoch is set to 50 to avoid overfitting. The weight parameters are saved after training.

[0056] Measure the input impedance of the actual system, including the real and imaginary parts of the impedance. The measured bandwidth must be consistent with the training set. Modify the shape of the measured data to match the unlabeled training set data. This data is used as input to the trained LSTM network model to initially obtain the system impedance R. q L q and M 1q The R obtained at this time q L q and M 1q The precision of the parameters does not yet meet the requirements for accurate modeling.

[0057] The coarse R from the LSTM network q L q and M 1q The initial input values ​​are used as the input values ​​for a nonlinear fitting algorithm based on NLS to perform regression fitting and obtain a high-precision R-value. q L q and M 1q The target curve for NLS nonlinear fitting is derived from the actual measured impedance, and the fitting objective equation is:

[0058]

[0059] R in the equation q L q and M 1q Three fitting target parameters.

[0060] Figure 4 These are the equivalent circuit topologies of the overall system under different modes. Figure 4(a), (b), and (c) are the equivalent circuit topologies of PI, O-QSCR, and UMPI modes, respectively. The topologies of PI and O-QSCR are the same, while UMPI is a two-phase system. However, the parameter measurement method of UMPI mode is the same as the former two, but when measuring the parameters between the transmitting coil and the cavity, the parameters of the two transmitting coils need to be extracted separately, that is, to ensure that there is only one coil in the system during each parameter extraction process.

[0061] Figure 5 This is a schematic diagram of the experimental measurement. The dashed line in the diagram represents the placement trajectory of the receiving coil. Figure 5 Tests were conducted at the test points shown, and the results were obtained. Figure 6 The experimental measurements and the comparison curves of output voltage and transmission efficiency of the mathematical model are shown. The mathematical model is established using Simulink, and the parameters are assigned to the Simulink model according to the parameter extraction method proposed in this invention. Figure 6 The data shows that the model has a very high accuracy (>95%) in predicting the output voltage and transmission efficiency of the actual system.

[0062] The equivalent parameter extraction method for cavity resonant wireless power transfer systems proposed in this invention, based on LSTM and nonlinear fitting algorithm, is a general equivalent parameter extraction method. It has high efficiency in parameter extraction for circuit models with known topology but unknown parameters. Especially for cavity resonant wireless power transfer systems, this method can effectively extract all key target parameters, which is of great significance for parameter optimization analysis and system design of cavity resonant wireless power transfer systems.

[0063] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for extracting equivalent parameters of a cavity resonant wireless power transfer system, characterized in that, The method comprises the following steps: S1: collecting system input impedance spectrum data; S2: using the trained LSTM regression model to obtain system equivalent parameter preliminary prediction values by taking the system input impedance spectrum data as input; S3: obtaining system equivalent parameter accurate values by using a nonlinear fitting algorithm to regress and fit based on the system equivalent parameter preliminary prediction values; The system equivalent parameters include the cavity internal resistance of the resonant cavity, the cavity equivalent inductance, and the mutual inductance value between the coil and the resonant cavity.

2. The cavity resonant wireless power transfer system equivalent parameter extraction method of claim 1, wherein, The electrical parameters of the transmitting coil and the receiving coil are determined by electrical measurement, including the self-inductance value of the transmitting coil, the internal resistance value of the transmitting coil, the resonant capacitance value of the transmitting coil, the self-inductance value of the receiving coil, the internal resistance value of the receiving coil, and the resonant capacitance value of the receiving coil.

3. The equivalent parameter extraction method of the cavity resonant wireless power transmission system according to claim 2, characterized in that: The transmitting coil and the resonant cavity are equivalent to a double-coil mutual inductance model, and the input impedance at the transmitting coil end is: wherein: Z1 represents the input impedance of the transmitting coil end, R1 represents the internal resistance value of the transmitting coil, R q represents the cavity internal resistance of the resonant cavity, ω represents the system operating angular frequency, M 1q represents the mutual inductance between the transmitting coil and the resonant cavity, L q represents the cavity equivalent inductance of the resonant cavity, C q represents the resonant capacitance of the resonant cavity, j represents the imaginary part, L1 represents the self-inductance value of the transmitting coil, and C1 represents the resonant capacitance value of the transmitting coil.

4. The equivalent parameter extraction method of the cavity resonant wireless power transmission system according to claim 2, characterized in that: The receiving coil and the resonant cavity are equivalent to a double-coil mutual inductance model, and the input impedance at the receiving coil end is: wherein Z2 represents a receiving coil end input impedance, R2 represents a receiving coil internal resistance value, R q represents a cavity internal resistance of a resonant cavity, ω represents a system operating angular frequency, M 2q represents a mutual inductance between a receiving coil and a resonant cavity, L q represents a cavity equivalent inductance of a resonant cavity, C q represents a resonant capacitance of a resonant cavity, j represents an imaginary part, L2 represents a receiving coil self-inductance value, C2 represents a receiving coil resonant capacitance value.

5. The method of claim 1-4, wherein: The LSTM regression model has a hidden layer shape of 4x128 and 3 layers, and the input training data during model training includes frequency, input impedance real part, input impedance imaginary part, input impedance phase, and label, wherein the label corresponds to the internal resistance of the cavity, the equivalent inductance of the cavity, and the mutual inductance value between the coil and the resonant cavity.

6. The method of claim 5, wherein: The nonlinear fitting algorithm uses the minimization of residual sum of squares as the core algorithm.

7. The method of claim 5, wherein: The training data of the LSTM regression model is generated by circuit simulation or mathematical analysis software.

8. The cavity resonant wireless power transfer system equivalent parameter extraction method of claim 7, wherein: The frequency band width of the training data is set to 40-60 kHz.

9. The method of claim 1, wherein: According to the system working mode, the distributed mode is adopted to extract the system equivalent parameters, and the principle of only one coil is extracted at a time is ensured.

10. The cavity resonant wireless power transfer system equivalent parameter extraction method of claim 9, wherein: The system working modes include PI mode, O-QSCR mode, and UMPI mode.

Images