Metallic foreign object detection method and system for multi-parameter WPT system based on neural network

Abstract

The application relates to the technical field of magnetic coupling wireless power transmission, and particularly discloses a multi-parameter WPT system metal foreign matter detection method and system based on a neural network, which can uniquely determine a one-to-one relationship between different working states of a WPT system based on different combinations of M identification parameters, construct a neural network model and a data set according to the M identification parameters corresponding to N working states, and train and test the neural network model by using the constructed data set, so that the obtained neural network model can output a corresponding working state for any group of M identification parameters, thereby determining whether metal foreign matter is contained. The application fuses multi-parameters of the WPT system, studies metal foreign matter detection under different working conditions of the system by using the neural network model, can judge the change of working conditions, and can identify metal foreign matter of a certain size at a certain position.

Description

Technical Field

[0001] This invention relates to the field of magnetic coupling wireless power transfer technology, and in particular to a method and system for detecting metallic foreign objects in a multi-parameter WPT system based on a neural network. Background Technology

[0002] Magnetic-coupled resonant wireless power transfer (MCR-WPT) generates high-frequency electromagnetic fields in space, making it susceptible to the influence of metallic materials. Metallic materials generally have good electrical conductivity, but the magnetic permeability varies considerably among different metals. Based on their magnetic permeability, metals can be classified into non-magnetic metals (such as copper, aluminum, and gold) and magnetic metals (such as iron, nickel, cobalt, and manganese). The main effect of non-magnetic metals on the system is the eddy current effect, while magnetic metals enhance both the eddy current and magnetic effects in a magnetic field.

[0003] In high-frequency electromagnetic fields, the eddy current and magnetic effects of non-magnetic metals can alter the equivalent parameters of a system, affecting system performance and creating safety hazards. Currently, most research on the detection of non-magnetic metallic foreign objects operates under ideal conditions, such as constant input power and coupling mechanism positions, involving system modeling, parameter and performance analysis, and system optimization design. However, in most cases, the position of the coupling mechanism, the magnitude of the input voltage, or the load are not constant, and the system's operating conditions may change. Directly using power loss or the equivalent quality factor method for calculation may misinterpret changes in system operating conditions as the presence of non-magnetic metallic foreign objects. Summary of the Invention

[0004] This invention provides a method and system for detecting metallic foreign objects in a multi-parameter WPT system based on neural networks. The technical problem it solves is: how to accurately detect the presence of metallic foreign objects, especially non-magnetic metallic foreign objects, in a WPT system.

[0005] To address the above technical problems, this invention provides a method for detecting metallic foreign objects using a multi-parameter WPT system based on neural networks, comprising the following steps:

[0006] Determine the architecture and design parameters of the WPT system;

[0007] A neural network model is constructed, which has M input nodes and N output nodes. The M input nodes are used to input the M identification parameters of the WPT system, and the N output nodes are used to output the probabilities of the N working states of the WPT system.

[0008] Construct a dataset of M identification parameters corresponding to N operating states of the WPT system using a defined WPT system;

[0009] The constructed neural network model is trained and tested using the constructed dataset;

[0010] Input a set of M identification parameters of the WPT system into the neural network model that has been trained and tested, and the neural network model outputs the probabilities of N working states;

[0011] Determine if there are any metallic foreign objects based on the label of the working state with the highest probability.

[0012] Furthermore, M equals 2, and the M identification parameters of the WPT system are the ratio Q of U2 to U1. e and power loss factor P eqLoss =P Loss / P OUT U1 is the equivalent inverter output square wave voltage source, U2 is the effective voltage value across the transmitting coil and the equivalent internal resistance of the circuit, and P Loss The power loss caused by the metallic foreign object, P OUT This represents the system's output power.

[0013] Furthermore, the power loss P caused by the metallic foreign object Loss Equal to the total energy P of the transmitting coil PT Subtract the total energy P of the receiving coil PR The total energy P of the transmitting coil PT Equal to the input power P before the transmitter drive system in Subtract the system loss P at the transmitting end PTLoss The total energy P of the receiving coil PR Equal to the output power P after the rectifier circuit at the receiving end OUT Adding the system loss P at the receiving end PRLoss .

[0014] Furthermore, the transmitting end of the WPT system includes a DC power supply, a full-bridge inverter, a transmitting end series compensation capacitor, and a transmitting coil connected in sequence; the receiving end of the WPT system includes a receiving coil, a receiving end series compensation capacitor, a rectifier filter circuit, and a load connected in sequence.

[0015] System loss P at the transmitter PTLoss equal I p R p R represents the current and internal resistance of the transmitting coil, respectively. dson This represents the on-resistance of the switching transistor in the full-bridge inverter;

[0016] The system loss P at the receiver PRLoss equal I s R s U represents the current and internal resistance of the receiving coil, respectively. d This represents the forward voltage drop of the diode in the rectifier filter circuit.

[0017] Furthermore, N=3, and the N working state packages are: the system is working normally, the system has a metal foreign object, and the system's working conditions have changed, corresponding to three different labels.

[0018] Furthermore, when the system is working normally, Q e Fixed, P eqLoss Approximately 0;

[0019] When a metallic foreign object is present in the system, Q e The slope of P should not be lower than the preset slope. eqLoss Greater than 0;

[0020] When the system's operating conditions change, Q e P eqLoss The changes are different from those when the system is working normally or when there are metallic foreign objects in the system.

[0021] Furthermore, the neural network model is a BP neural network.

[0022] The present invention also provides a multi-parameter WPT system metal foreign object detection system based on neural networks, which is applied to the aforementioned multi-parameter WPT system metal foreign object detection method based on neural networks. The key features are: a model generation module, and a transmitter detection module and a receiver detection module respectively connected to the transmitter and receiver of the WPT system.

[0023] During the training and testing phases, the transmitter detection module is used to collect information from the transmitter under N operating states, and the receiver detection module is used to collect information from the receiver under N operating states and send it to the transmitter. The transmitter detection module calculates M identification parameters of the WPT system based on the information collected under the N operating states and sends them to the model generation module. The model generation module is used to construct a neural network model, construct a dataset based on the M identification parameters corresponding to the N operating states, and train and test the neural network model using the constructed dataset.

[0024] During the application phase, the transmitter detection module is used to collect information from the transmitter, and the receiver detection module is used to collect information from the receiver and send it to the transmitter. The transmitter detection module calculates M identification parameters based on the collected information and sends them to the model generation module. The model generation module outputs the probabilities of N working states. The transmitter detection module determines whether there is a metal foreign object based on the label of the working state with the highest probability.

[0025] This invention provides a method and system for detecting metal foreign objects in a multi-parameter WPT system based on neural networks. It uniquely determines the one-to-one relationship between different operating states of the WPT system based on different combinations of M identification parameters. A neural network model and dataset are constructed according to the M identification parameters corresponding to N operating states. The constructed dataset is used to train and test the neural network model. The resulting neural network model can output the corresponding operating state for any set of M identification parameters, thereby determining whether a metal foreign object is present. This invention integrates multiple parameters of the WPT system and uses a neural network model to study metal foreign object detection under different operating conditions. It can judge changes in operating conditions and identify metal foreign objects of a certain size at a certain location. Attached Figure Description

[0026] Figure 1 This is a model diagram of a magnetic coupling mechanism containing a metallic foreign object provided in an embodiment of the present invention;

[0027] Figure 2 These are magnetic flux density mode distribution diagrams of the coupling mechanism with and without foreign objects, as provided in the embodiments of the present invention.

[0028] Figure 3 This is a diagram showing the relationship between the mutual inductance and coil spacing between the transmitting and receiving coils when there are no foreign objects and when there are metallic foreign objects, provided in an embodiment of the present invention.

[0029] Figure 4 This is a diagram showing the change in mutual inductance between the transmitting and receiving coils when the position of a metallic foreign object is changed, provided in an embodiment of the present invention.

[0030] Figure 5 This is a diagram showing the change in mutual inductance between the transmitting and receiving coils when the metallic foreign material is changed, provided in an embodiment of the present invention.

[0031] Figure 6 This is the topological equivalent circuit diagram of the foreign object-free SS-type WPT system provided in the embodiments of the present invention;

[0032] Figure 7 This is the topological equivalent circuit diagram of an SS-type WPT system with non-magnetic metallic foreign matter provided in an embodiment of the present invention;

[0033] Figure 8 This is a power transmission loss analysis diagram of a magnetically coupled wireless power transmission system provided in an embodiment of the present invention;

[0034] Figure 9 This is a diagram showing the changes in system parameters when the position of the coupling mechanism changes, as provided in an embodiment of the present invention.

[0035] Figure 10 This is a diagram showing the changes in system parameters when the DC input voltage changes, as provided in an embodiment of the present invention.

[0036] Figure 11 This is a graph showing the changes in system parameters under varying loads, provided in an embodiment of the present invention.

[0037] Figure 12 This is a diagram showing the change of system parameters when the mutual inductance of the metal-foreign coupling changes, as provided in an embodiment of the present invention.

[0038] Figure 13 This is a graph showing the change of system parameters when the equivalent internal resistance of a metallic foreign object changes, as provided in an embodiment of the present invention.

[0039] Figure 14 This is a diagram showing the changes in system parameters when the equivalent self-inductance of a metallic foreign object changes, as provided in an embodiment of the present invention.

[0040] Figure 15 This is an example diagram of the BP network prediction results provided in an embodiment of the present invention;

[0041] Figure 16 This is a flowchart of a multi-parameter WPT system metal foreign object detection method based on neural networks provided in an embodiment of the present invention. Detailed Implementation

[0042] The embodiments of the present invention are described in detail below with reference to the accompanying drawings. The embodiments are given for illustrative purposes only and should not be construed as limiting the present invention. The accompanying drawings are for reference and illustration only and do not constitute a limitation on the scope of patent protection of the present invention, because many changes can be made to the present invention without departing from the spirit and scope of the present invention.

[0043] The influence of a non-magnetic metallic foreign object on the magnetic field of the coupling mechanism was studied using the finite element simulation software COMSOL. Both the transmitting and receiving coils were made of copper Litz wire wound into a circular-circular symmetrical structure. The Litz wire had an outer diameter of 1.9 mm, an inner diameter of 80 mm, an outer diameter of 140 mm, and 14 turns. The non-magnetic metallic foreign object (hereinafter referred to as the metallic foreign object) was a cube (copper) with a side length of 1 cm. Figure 1 As shown.

[0044] When a high-frequency excitation is applied to the transmitting coil, the magnetic flux density modulus diagrams at a distance of 5 cm are as follows: (Diagrams showing magnetic flux density with and without a foreign object, and with a metallic foreign object (copper)). Figure 2 As shown. Figure 2 (a) does not contain foreign matter, and the magnetic flux distribution between its coils is relatively uniform. Figure 2 In (b), the middle copper block blocks part of the magnetic flux, and eddy currents are induced in the metal, resulting in losses.

[0045] The influence of nonferromagnetic metallic foreign objects on the mutual inductance of coils was analyzed. The foreign object material was fixed as copper, and the coil spacing was gradually increased while keeping the foreign object always centered within the coils. The graph of mutual inductance as a function of coil spacing is shown below. Figure 3As shown, M and Mfod represent the mutual inductance between the transmitting and receiving coils when there is no foreign object and when there is a foreign object, respectively. Figure 3 It can be seen that as the distance between the coils increases, the mutual inductance exhibits a nonlinear and sharp decreasing trend. Furthermore, in models containing metallic foreign objects, the magnetic field coupling between the foreign object and the coil weakens the coupling between the coils at the same distance, resulting in a greater reduction in mutual inductance.

[0046] This study analyzes the influence of different positions and types of nonferromagnetic metallic foreign objects on the magnetic field of a coupling mechanism. Keeping the position of the coupling mechanism constant, the position of the metallic foreign object is changed to investigate its effect on the magnetic field. Figure 4 As shown. Figure 4 This indicates that the closer the metallic foreign object is to the transmitting coil, the stronger the coupling with the magnetic field of the transmitting coil, and the smaller the mutual inductance between the two coils, meaning that the coupling between the metallic foreign object and the receiving coil is weaker.

[0047] Keeping the size and position of the metallic foreign object constant, the type of metallic foreign object (copper, lead, tin) is changed to study the effect of different types of foreign objects on the magnetic field, such as... Figure 5 As shown, Mcu, Mni, and Msn represent the mutual inductance between the transmitting and receiving coils under the influence of copper, lead, and tin, respectively. Figure 5 This indicates that different non-ferromagnetic metallic foreign objects have different effects on the magnetic field. Because the equivalent inductance and internal resistance of the metallic foreign object circuit are different, they will have different effects on the magnetic field even at the same location.

[0048] Considering an SS-type magnetically coupled wireless power transfer system without foreign objects, its equivalent circuit is as follows: Figure 6 As shown, where L p and L s R represents the self-inductance of the transmitting coil and the receiving coil, respectively. p and R s C represents the internal resistance of the transmitting coil and the receiving coil, respectively. p and C s I represents the capacitors connected in series with the transmitting and receiving coils, respectively. p and I s U represents the current in the transmitting coil and the receiving coil, respectively. s R represents the equivalent AC source of the front-end DC power supply and the full-bridge inverter circuit. eq Let M represent the equivalent load resistance of the rectifier filter circuit and the load, and M represent the mutual inductance between the transmitting and receiving coils. According to Kirchhoff's Voltage Law, the KVL equations for the transmitting and receiving circuits are:

[0049]

[0050] w represents the system's operating angular frequency.

[0051] The effective voltage U2 across the transmitting coil and the equivalent internal resistance of the circuit in the resonant cavity is:

[0052]

[0053] The equivalent inverter output square wave voltage source U1 is:

[0054]

[0055] Define the equivalent Q value Q e The ratio of U2 to U1 can be obtained from equations (1)-(3):

[0056]

[0057] As can be seen from equation (4), in an SS-type magnetically coupled wireless power transmission system, once the system parameters such as the transmitting coil, receiving coil, compensation capacitor, mutual inductance, and load resistance are determined, the equivalent quality factor Q... e It is a fixed value, meaning that the equivalent quality factor is only related to the system parameters.

[0058] In the equivalent circuit of a nonferromagnetic metallic foreign object, when a nonferromagnetic conductive medium appears between or around the coupling mechanism, it can be equivalent to an eddy current calculation loop of an inductor and a resistor connected in series. R m The internal resistance of the eddy current calculation loop is given by L, which represents the energy loss caused by the eddy current. m M represents the equivalent inductance of the eddy current calculation loop, indicating the influence of the eddy current magnetic field on the magnetic field of the resonant coil. pm and M sm These represent the mutual inductance between the metallic foreign object and the transmitting and receiving coils, respectively. Figure 7 As shown.

[0059] Similarly, the KVL equations for the three loops are:

[0060]

[0061] Z m Let Q be the equivalent impedance of the circuit with the metallic foreign object. Similarly, define the equivalent quality factor Q. e for:

[0062]

[0063] The parameters a, b, c, and d, which are defined to simplify the formula, are as follows:

[0064]

[0065] From equations (6) and (7), it can be seen that when the inherent parameters of the system are determined, the equivalent quality factor Q e Similarly, only the parameter R in the equivalent circuit of the metallic foreign object...m L m M pm and M sm Related to this. When other system conditions remain unchanged, and a metallic foreign object is present near the coupling mechanism, the equivalent quality factor Q... e The value will change, and the degree of change depends on the size, type, and location of the metallic foreign object. Therefore, the equivalent quality factor Q can be... e As an indicator to measure whether there are non-ferromagnetic metallic foreign objects in the system.

[0066] Figure 8 This is a power transfer diagram for an SS-type magnetically coupled wireless power transfer system. Where P... PT The total energy of the transmitting coil is equal to the input power P before the transmitting drive system. in Subtract the system loss P at the transmitting end PTLoss P PR The total energy of the receiving coil is equal to the output power P after the rectifier circuit at the receiving end. OUT Adding the system loss P at the receiving end PRLoss The power loss P caused by the metallic foreign object Loss Satisfy the following formula:

[0067] P Loss =P PT -P PR (8)

[0068] Theoretically, if P PTLoss and P PRLoss It can be accurately measured under the condition of no foreign objects, P PT It should be equal to P PR That is, P Loss The value equals 0. If a metallic foreign object exists between the two coils, the foreign object will induce eddy currents in the magnetic field, resulting in eddy current losses, which will cause P to decrease. Loss Much greater than 0, thus allowing us to calculate P Loss The value is used to detect the presence of metallic foreign objects.

[0069] When a non-ferromagnetic metallic foreign object is present in an SS-type magnetically coupled wireless power transfer system, its equivalent circuit model is as follows: Figure 7 As shown. According to the definition of the power loss method and equation (5), the total energy P of the transmitting coil can be calculated respectively. PT The total energy P of the receiving coil PR :

[0070]

[0071] Where R dson and U d R represents the on-resistance of the full-bridge inverter switch and the on-state voltage drop of the full-bridge rectifier diodes, respectively. LR is the load resistance. eq The equivalent load as seen from the input port of the rectifier bridge satisfies:

[0072]

[0073] Similarly, looking backward from the input port of the full-bridge inverter, the equivalent impedance of the subsequent stage of the full-bridge inverter is a conjugate relationship with the equivalent load impedance of the full-bridge rectifier. That is:

[0074]

[0075] As can be seen from equation (5), the eddy currents generated by the metallic foreign object in the magnetic field will form a reflection impedance at the transmitting and receiving ends, thereby affecting the transmitting and receiving coil current I. p and I s Then, the power loss P is calculated according to equations (8) and (9). Loss This allows us to determine whether any metallic foreign objects are present.

[0076] In most cases, the position of the coupling mechanism, the magnitude of the input voltage, or the load size are not constant, and the system operating conditions may change. Equations (6) and (9) show that, regardless of whether the power loss method or the equivalent quality factor method is used, changes in coil mutual inductance, load, or input voltage will also cause power loss P. Loss A value greater than 0 or equivalent quality factor Q e When changes occur, directly using power loss or the equivalent quality factor method for calculation may cause the system to misjudge the presence of metallic foreign objects due to changes in operating conditions. Therefore, it is very important to analyze the changes in system parameters when operating conditions change.

[0077] Based on the previous analysis, when there is a non-ferromagnetic metallic foreign object in the system, it can be equivalent to an eddy current calculation circuit with an inductor and a resistor connected in series. In the simulation, the metallic foreign object is represented by an equivalent inductor 2 connected in series with an equivalent resistor, which couples with the transmitting coil 1 and the receiving coil 3. If the foreign object is not considered, the series circuit is set as an open circuit.

[0078] Based on the modeling and analysis of the coupling mechanism above, the parameters selected for normal operation of the computing system are shown in Table 1 below.

[0079] Table 1 System Normal Operating Parameters

[0080]

[0081] Using an input voltage of 10V, a load resistance of 10Ω, a coil spacing of 5cm (i.e., a mutual inductance of 4.2uH) as the standard operating conditions, the simulated power loss P is obtained. Loss The power loss factor P is 1.79W. eqLoss The equivalent quality factor Q is 0.021. eIt is 19.03.

[0082] In practical operation of wireless power transfer systems, the position of the coupling mechanism, the input voltage, or the equivalent load resistance often change. Although these changes affect the system power loss P... Loss Because there are no metal foreign objects between the coils, the power loss remains essentially unchanged. However, this ignores the energy leakage during the magnetic field transfer between the transmitting and receiving coils. When the two coils are far apart, the system power loss P increases. Loss The current will increase sharply due to magnetic field energy leakage. Similarly, when the load is too large, it will cause the transmitting coil current I to increase. p The current I in the receiving coil increases sharply. s The power loss P of the system decreases sharply. According to equations (8) and (9), the power loss P of the system decreases sharply. Loss The power loss will also increase. Only when the distance between the two coils is appropriate and the load is suitable will the power loss under the above three operating conditions approach zero. Although the system power loss P will increase when the distance between the two coils is large or the load is large. Loss It will be much greater than 0, but at this time the system's output power P OUT If the value is large, then P Loss With P OUT The ratio P eqLoss As a power loss factor, that is:

[0083]

[0084] The power loss factor can be used to measure the presence or absence of metallic foreign objects in a system. When the system is free of foreign objects, even if the coupling mechanism is far apart and the load resistance is high, the power loss factor P will still be high. eqLoss It still approaches 0; however, when metallic foreign objects are present, the power loss factor P eqLoss It will still be greater than 0.

[0085] For an SS-type topology wireless power transfer system, when the coupling mechanism becomes closer (i.e., the mutual inductance M between the coils increases), the current I in the transmitting and receiving coils will increase. p and I s Both decrease, and from equation (9), we can know that the total energy P of the transmitting coil at this time is... PT The total energy P of the receiving coil PR Both will decrease. Since there are no losses due to foreign objects in the transmitting and receiving coils, and if magnetic field energy leakage is not considered, the system power loss P Loss Approximately 0. Since the coupling between the metallic foreign object and the receiving coil is weak, to simplify the model, the electromagnetic coupling M between the metallic foreign object and the receiving end coupling mechanism is ignored. sm From equations (6) and (7), it can be seen that when the mutual inductance increases, both c and d increase, and the equivalent quality factor Q... e The power loss P decreases. Similarly, when the coupling mechanism becomes farther away or shifts, i.e., when the mutual inductance between the coils decreases, the system power loss P...Loss Approaching 0, but the equivalent quality factor Q e Increase.

[0086] Despite power loss P Loss Within a certain range, the magnetic leakage rate approaches zero regardless of the position of the coupling mechanism. However, if the distance between the coupling mechanisms is too great, the magnetic leakage will increase, leading to a power loss P. Loss Much greater than 0, but the power loss factor P eqLoss It still approaches 0. By setting the metal foreign object circuit to an open circuit and keeping other parameters constant, the position change of the coupling mechanism is simulated by adjusting the mutual inductance of the coupling mechanism, and the equivalent quality factor Q is obtained. e Power loss P Loss and power loss factor P eqLoss The results of these three parameters are as follows Figure 9 As shown. To facilitate observation of trends, Figure 9 Medium power loss factor P eqLoss Multiply the value by 100, and the unit is %. Figure 9 This indicates that the equivalent quality factor Q e The power loss P increases as mutual inductance decreases. Loss The power loss factor P approaches 0W when the coupling mechanisms are close together (i.e., when the mutual inductance is high), but exceeds 0W when the coupling mechanisms are far apart. To prevent system misjudgment, the power loss factor P proposed in this example... eqLoss It still approaches 0 across the entire distance range.

[0087] When the input voltage changes (1-50V), it can be seen from equations (5) and (9) that the transmitting power P of the transmitting coil is... PT With the transmitting power P of the receiving coil PR Each term in the expression contains U in 2, therefore P PT With P PR Both increase, neglecting magnetic field energy leakage, the system power loss P between the transmitting and receiving coils Loss Approaching 0. Equations (6) and (7) show that the equivalent quality factor Q e It is independent of the input voltage magnitude. Simulation results are as follows: Figure 10 As shown, Q e It remains essentially constant and consistent with the value under standard operating conditions, P Loss When the input voltage is too high, a significant deviation will occur due to the high power of the transmitter and receiver. This is why the power loss method for detecting metallic foreign objects is only applicable to low-power systems. However, the power loss factor P... eqLoss If the value is still close to 0, this indicator can be used to measure whether there are metallic foreign objects in the system.

[0088] Similarly, when the equivalent load resistance R eqWhen the impedance increases within a certain range (1-100Ω), the system power loss P between the transmitting and receiving coils decreases because there is no loss due to metallic foreign objects. Loss The value remains basically unchanged, but the equivalent quality factor Q... e Increase, and vice versa. Simulation results are as follows: Figure 11 As shown, Q e P increases gradually with increasing load. Loss Under light load, it approaches 0, but under heavy load, it deviates significantly due to the higher power, although the power loss factor P... eqLoss It is still close to 0, and this indicator can be used to measure whether there are metallic foreign objects in the system.

[0089] When a non-ferromagnetic metallic foreign object is present in the system, its equivalent foreign object circuit coupling inductance, equivalent internal resistance, and equivalent self-inductance may change due to various factors. According to the magnetic simulation above, the position of the metallic foreign object from the transmitting coil affects the coupling inductance of the foreign object circuit, while the type and size of the metallic foreign object affect the equivalent internal resistance and equivalent self-inductance of the foreign object circuit. Figure 12 , Figure 13 , Figure 14 These are the equivalent quality factor Q when the mutual inductance of the foreign circuit coupling, the equivalent internal resistance, and the equivalent self-inductance change. e Power loss P Loss and power loss factor P eqLoss The closer the metallic foreign object is to the transmitting coil, the greater the mutual inductance of the foreign object circuit, and the greater the Q value. e The lower the value, the better. eqLoss The larger the value, the greater the impact of the foreign object circuit's equivalent internal resistance and equivalent self-inductance on the heat loss and magnetic loss generated by eddy currents, respectively. The presence of a metallic foreign object increases Q. e All of them are different from the values ​​under normal working conditions, and P eqLoss Relatively large. This can be determined based on the equivalent quality factor Q. e and power loss factor P eqLoss Assess whether the system contains non-ferromagnetic metallic foreign objects.

[0090] The larger the foreign object, the greater its equivalent internal resistance and self-inductance, and the greater its impact on system coupling. Therefore, Q e The more obvious the change, the more accurate the identification by the system parameter method.

[0091] The preceding analysis shows that, when the operating conditions remain unchanged, the power loss factor P can be set based on the condition of no foreign objects. eqLoss and equivalent quality factor Q e The threshold is used to determine the presence of foreign objects. However, when operating conditions change, the power loss factor P... eqLoss and equivalent quality factor Q e Changes may occur, leading to misjudgments, but these changes have a certain trend, which can be summarized in Table 2 below based on the analysis above.

[0092] Table 2. Parameter variation trends under different working conditions

[0093]

[0094] By obtaining a large amount of experimental data on system parameters under normal operating conditions, the presence of foreign objects in the system, and changes in operating conditions, a neural network (in this example, a backpropagation (BP) neural network) can be trained on these sample data to continuously adjust the network weights and thresholds, causing the error function to decrease along the negative gradient direction and approach the desired output. During training, the input layer data is the power loss factor P. eqLoss and equivalent quality factor Q e The system assigns a label to each data set indicating its operating state: 0 for normal system operation, 1 for the presence of a non-ferromagnetic metallic foreign object, and 2 for a change in operating conditions. The output layer represents the probability of each operating state. After training, the measured system parameter Q is used as the input. e With P eqLoss The data was fed into the model for prediction, and the results were as follows: Figure 15 As shown, Q was measured. e With P eqLoss When the values ​​are 22.65 and 0.7795 respectively, the predicted label is 0, indicating no metallic foreign object; Q e With P eqLoss The predicted label was 1 when the values ​​were 20.62 and 54.776, respectively, indicating the presence of a metallic foreign object, which is consistent with the simulation results.

[0095] Based on the above analysis, embodiments of the present invention provide a method for detecting metallic foreign objects using a multi-parameter WPT system based on a neural network, such as... Figure 16 As shown, it includes the following steps:

[0096] S1. Determine the architecture of the WPT system (e.g., ...) Figure 6 ) and its design parameters (as shown in Table 1);

[0097] S2. Construct a neural network model (this example uses a BP neural network model), with M input nodes and N output nodes. The M input nodes are used to input the identification parameters of M types of WPT systems (in this example, M=2, referring to the power loss factor P). eqLoss and equivalent quality factor Q e N output nodes are used to output the probabilities of N working states (in this example N=3, which refer to the system working normally, the presence of non-ferromagnetic metal foreign objects, and the change of working conditions, respectively, with corresponding labels 0, 1, and 2).

[0098] S3. Construct a dataset of M identification parameters corresponding to the N working states of the WPT system using a defined WPT system;

[0099] S4. Use the constructed dataset to train and test the constructed neural network model (the training objective is to minimize the loss between the test value and the true value);

[0100] S5. Input a set of identification parameters for the WPT system (a set of power loss factors P) eqLoss and equivalent quality factor Q e Once the neural network model has been trained and tested, it outputs the probabilities of N working states.

[0101] S6. Determine whether there is a metal foreign object based on the label of the working state with the highest probability. If the label is 1, then it is determined that there is a metal foreign object.

[0102] This example also includes the construction of a prototype platform for metal foreign object detection for experimental verification. Power is supplied by a DC input power supply, with an electronic load instrument serving as the output load. Since the output power information needs to be collected and transmitted to the transmitter, wireless communication between the transmitter and receiver is required. The prototype uses 2.4GHz WIFI communication to transmit information such as the receiver's output power to the transmitter controller. The transmitter calculates the equivalent quality factor Q based on the voltage ratio of U2 to U1 in equation (4). e And calculate the power loss factor P based on the loss information transmitted back from the receiving end. eqLoss Based on the fitting function trained by the BP neural network, the system determines whether the operating conditions have changed or whether there are foreign metal objects, and displays this information on the screen in real time.

[0103] Under normal operating conditions, without foreign objects, the power loss factor P is calculated by detecting the voltage at the transmitting and receiving ends and the transmitted and received power. eqLoss With Q e Afterwards, the screen displays the calculated data, showing "Non-FOD" when there are no foreign objects. Under normal operating conditions, the load resistance is set to 10Ω. A corresponding circular coil is wound according to the magnetic circuit simulation described above. When the coil distance is 5cm, the mutual inductance is 4.2uH, and the voltage gain G is M / L. f =1.91. Input voltage 3V, theoretical output voltage is 5.73V.

[0104] Without introducing foreign objects, the system can identify and display different operating conditions by varying the load, input voltage, and coil distance. Changing the load and coil distance will affect the equivalent quality factor Q. e and power loss P Loss However, the power loss factor P eqLoss The numbers are still low.

[0105] Introduce 1cm respectively 3When non-ferromagnetic metallic foreign objects, such as copper and tin, are placed in the center of the two coils, the system can detect the metallic foreign objects and display "FOD Detected" on the screen.

[0106] Experimental results show that, after introducing metallic foreign substances copper and tin, in addition to the equivalent quality factor Q e In addition to the reduction, the power loss factor P is also affected by the eddy current losses induced in the metallic foreign object. eqLoss The numerical value increases significantly, indicating that the system can detect and identify foreign objects when they are located in the center of the coil. Furthermore, the system also has a certain detection effect on other non-ferromagnetic metals such as titanium, aluminum, and zinc.

[0107] It should also be noted that the neural network used in this example is not limited to the BP neural network; any other neural network with the same function as the BP neural network can be used.

[0108] Based on the above-mentioned method for detecting metal foreign objects in a multi-parameter WPT system based on neural networks, the present invention also provides a multi-parameter WPT system for detecting metal foreign objects based on neural networks, which includes a model generation module, and a transmitter detection module and a receiver detection module respectively connected to the transmitter and receiver of the WPT system.

[0109] During the training and testing phases, the transmitter detection module is used to collect information from the transmitter under N operating conditions, and the receiver detection module is used to collect information from the receiver under N operating conditions and send it to the transmitter. The transmitter detection module calculates M identification parameters of the WPT system based on the information collected under N operating conditions and sends it to the model generation module. The model generation module is used to build a neural network model, construct a dataset based on the M identification parameters corresponding to the N operating conditions, and use the constructed dataset to train and test the neural network model.

[0110] During the application phase, the transmitter detection module is used to collect information from the transmitter, and the receiver detection module is used to collect information from the receiver and send it to the transmitter. The transmitter detection module calculates M identification parameters based on the collected information and sends them to the model generation module. The model generation module outputs the probabilities of N working states. The transmitter detection module determines whether there is a metal foreign object based on the label of the working state with the highest probability.

[0111] In summary, the multi-parameter WPT system metal foreign object detection method and system based on neural networks provided in this invention can uniquely determine the one-to-one relationship between different operating states of the WPT system based on different combinations of M identification parameters. A neural network model and dataset are constructed according to the M identification parameters corresponding to N operating states. The constructed dataset is used to train and test the neural network model. The resulting neural network model can output the corresponding operating state for any set of M identification parameters, thereby determining whether a metal foreign object is present. This invention integrates multiple parameters of the WPT system and uses a neural network model to study metal foreign object detection under different operating conditions, enabling it to judge changes in operating conditions and identify metal foreign objects of a certain size at a certain location.

[0112] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for detecting metallic foreign objects using a multi-parameter WPT system based on neural networks, characterized in that, Including the following steps: Determine the architecture and design parameters of the WPT system; The transmitting end of the WPT system includes a DC power supply, a full-bridge inverter, a transmitting end series compensation capacitor, and a transmitting coil connected in sequence; the receiving end of the WPT system includes a receiving coil, a receiving end series compensation capacitor, a rectifier filter circuit, and a load connected in sequence. System loss at the transmitter P PTLoss equal , I p , R p These represent the current and internal resistance of the transmitting coil, respectively. R dson This represents the on-resistance of the switching transistor in the full-bridge inverter; System loss at the receiving end P PRLoss equal , I s , R s These represent the current and internal resistance of the receiving coil, respectively. U d This represents the forward voltage drop of the diode in the rectifier filter circuit; A neural network model is constructed, which has M input nodes and N output nodes. The M input nodes are used to input the M identification parameters of the WPT system, and the N output nodes are used to output the probabilities of the N working states of the WPT system. M equals 2, and the M identification parameters of the WPT system are: U 2 and U The ratio of 1 Q e and power loss factor P eqLoss = P Loss / P OUT , U 1 is an equivalent inverter output square wave voltage source. U 2 represents the effective value of the voltage across the equivalent internal resistance of the transmitting coil and the circuit. P Loss Power loss caused by metallic foreign objects. P OUT The system's output power; Construct a dataset of M identification parameters corresponding to N operating states of the WPT system using a defined WPT system; The constructed neural network model is trained and tested using the constructed dataset; By measuring experimental data of system parameters under normal operating conditions, the presence of non-ferromagnetic metallic foreign objects, and altered operating conditions, the neural network model is trained on this experimental data to continuously adjust network weights and thresholds, causing the error function to decrease along the negative gradient direction and approach the desired output. During training, the input layer data is... P eqLoss and Q e The system records the working status of each set of data with a label: 0 indicates that the system is working normally, 1 indicates that there are non-ferromagnetic metal foreign objects in the system, and 2 indicates that the system's working conditions have changed. The output layer is the probability of each working status. After training is completed, the measured values ​​will be... P eqLoss and Q e The data is fed into the neural network model for prediction. Input a set of M identification parameters of the WPT system into the neural network model that has been trained and tested, and the neural network model outputs the probabilities of N working states; Determine if there are any metallic foreign objects based on the label of the working state with the highest probability.

2. The method for detecting metallic foreign objects using a multi-parameter WPT system based on a neural network according to claim 1, characterized in that: Power loss due to metallic foreign objects P Loss Equal to the total energy of the transmitting coil P PT Subtract the total energy of the receiving coil P PR Total energy of the transmitting coil P PT Equal to the input power before the transmitter drive system P in Subtract system losses at the transmitting end P PTLoss The total energy of the receiving coil P PR Equal to the output power after the rectifier circuit at the receiving end P OUT In addition to the system loss at the receiving end P PRLoss .

3. The method for detecting metallic foreign objects using a multi-parameter WPT system based on a neural network according to claim 1, characterized in that: When the system is working normally, Q e fixed, P eqLoss =0; When a metallic foreign object is present in the system, Q e The slope should not be lower than the preset slope. P eqLoss Greater than 0; When the system's operating conditions change, Q e , P eqLoss The changes are different from those when the system is working normally or when there are metallic foreign objects in the system.

4. The method for detecting metallic foreign objects in a multi-parameter WPT system based on a neural network according to claim 3, characterized in that: The neural network model is a BP neural network.

5. A neural network-based multi-parameter WPT system for detecting metallic foreign objects, used to implement the neural network-based multi-parameter WPT system for detecting metallic foreign objects as described in any one of claims 1 to 4, characterized in that: It includes a model generation module, as well as a transmitter detection module and a receiver detection module that connect to the transmitter and receiver of the WPT system, respectively; During the training and testing phases, the transmitter detection module is used to collect information from the transmitter under N operating states, and the receiver detection module is used to collect information from the receiver under N operating states and send it to the transmitter. The transmitter detection module calculates M identification parameters of the WPT system based on the information collected under the N operating states and sends them to the model generation module. The model generation module is used to construct a neural network model, construct a dataset based on the M identification parameters corresponding to the N operating states, and train and test the neural network model using the constructed dataset. During the application phase, the transmitter detection module is used to collect information from the transmitter, and the receiver detection module is used to collect information from the receiver and send it to the transmitter. The transmitter detection module calculates M identification parameters based on the collected information and sends them to the model generation module. The model generation module outputs the probabilities of N working states. The transmitter detection module determines whether there is a metal foreign object based on the label of the working state with the highest probability.

Images