Wireless charging system online parameter identification method and system based on BP neural network

Abstract

A wireless power transfer (WPT) online parameter identification method based on a BP neural network comprises the steps of designing a multi-level parameter combination table of a mutual inductor M and a load RL in combination with an actual wireless charging system circuit topology, and performing model simulation calculation according to parameter design in the parameter combination table to obtain system state data; in combination with simulation data, establishing and training a BP neural network by using a neural network function in numerical simulation software to obtain a WPT system parameter online identification model; and required state quantities are acquired from an actual system and are input into the recognition model for calculation, so that real-time parameter recognition of the WPT is completed. According to the method, additional communication equipment or auxiliary circuits do not need to be added, the design difficulty and cost are low, and the influence on the working state of the system is low. The method has the advantages that a complex circuit equation does not need to be calculated, errors are corrected through circuit actual measurement data, parameter identification errors are effectively reduced, and the requirement for WPT online parameter identification is met.

Description

Technical Field

[0001] The present invention belongs to the technical field of parameter identification, and more specifically, relates to an online parameter identification method and system for a wireless charging system based on a BP neural network. Background Art

[0002] Wireless Power Transfer (WPT) uses coupling coils to transmit energy through a large air gap, which is free from the constraints of cables compared to traditional conductive charging. WPT technology has the following advantages: 1. It is not easy to generate electric sparks or leakage in harsh environments, and has relatively few safety hazards; 2. The electrical equipment has a large range of movement when drawing power, and is flexible and beautiful; 3. It can power micro medical devices implanted in the human body and other machines that require special environments to operate. At present, domestic and foreign research institutions have conducted a lot of research on WPT technology in the fields of wearable smart devices, medical care, wireless charging of electric vehicles, wireless charging of mobile phones, smart homes, etc., making WPT technology widely used.

[0003] The application scope of WPT system is very wide, which means that it may be affected by various environmental factors during use, resulting in changes in circuit parameters, which in turn reduces the system transmission power and efficiency, and even damages system components. For example, in the actual wireless power transmission process, random changes in factors such as the degree of coil alignment and the width of the interval between coils will cause the coupling coefficient between coils to fluctuate; the internal resistance of the battery will also change due to factors such as temperature and charging level. In actual applications, the most common parameter changes faced by WPT systems are changes in mutual inductance and load resistance.

[0004] These problems will cause the WPT system to deviate from the rated working state, increase energy loss, and affect the safety of the system. In order to return the WPT system to the rated working state, it is necessary to develop a suitable state control method based on its current specific state parameters. Therefore, conducting research on WPT system parameter identification is crucial to improving the stability and transmission efficiency of the system.

[0005] In order to correctly adjust the working state of the WPT system when the system parameters are offset, ensure the safety and reliability of the system, and reduce energy loss, it is necessary to identify the current system state. The existing parameter identification methods mainly include the following: 1. Parameter identification method based on unilateral communication

[0006] Unilateral communication refers to the use of a communication device to transmit the signal measured by the sensor at the receiving end of the WPT system back to the transmitting end of the WPT system. Because all the input energy of the receiving end device comes from the coupling with the transmitting end, except for a small amount of circuit loss, the remaining output energy is transferred to the load, so its working state changes are strongly correlated with the fluctuations of the mutual inductance value and the load resistance value. Transmitting the electrical measurement signal from the receiving end of the WPT system back to its transmitting end can more easily identify its parameters. However, the additional communication device increases power consumption and reduces system efficiency, and the magnetic coupling mechanism will generate electromagnetic interference to the communication link, resulting in bit errors and frame loss problems.

[0007] 2. Parameter identification method based on external auxiliary circuit

[0008] An additional auxiliary circuit refers to a circuit with specific functions that is added to the original WPT system circuit structure. The additional auxiliary circuit can regulate the WPT system to make it show different working states. In different working states, the effects of mutual inductance and load resistance on the system state will also show different characteristics, and their parameter values ​​can be calculated. However, additional auxiliary control circuits need to be added, and the system design is complicated.

[0009] 3. Parameter identification method based on optimization algorithm

[0010] An optimization algorithm is an algorithm that obtains the optimal solution based on certain operation rules through iterative calculation. The parameter identification problem is converted into an optimization problem that minimizes the error between the observable electrical quantity and its theoretical calculation quantity based on the estimated parameter value. The optimization algorithm is used to search for the optimal parameter value in the possible solution set, which is the result of parameter identification. However, the error between the actual WPT system and the mathematical model is not considered, so the parameter identification value will also have errors. Summary of the invention

[0011] In order to solve the deficiencies in the prior art, the present invention provides an online parameter identification method and system for a wireless charging system based on a BP neural network.

[0012] The present invention adopts the following technical solution.

[0013] The first aspect of the present invention proposes an online parameter identification method for a wireless charging system using a BP neural network, comprising the following contents:

[0014] According to the actual wireless charging WPT system to be identified or the design drawing of the wireless charging WPT system, a simulation model of the WPT system is established;

[0015] Design of circuit mutual inductance M and load R in WPT system simulation model LA multi-level parameter combination table, wherein the multi-level parameter combination table is a table used to represent different parameters and their respective level combinations in experimental design;

[0016] According to the parameter design in the parameter combination table, the constructed simulation model is simulated and calculated to obtain the corresponding circuit mutual inductance M and load R L Relevant system status data;

[0017] Based on the corresponding system status data, a BP neural network is established and trained to obtain an online identification model of WPT system parameters;

[0018] The required state quantities are collected from the actual system and input into the online identification model of WPT system parameters for calculation, so as to obtain the mutual inductance and load resistance of WPT, the real-time parameters of the wireless charging system.

[0019] Preferably, the WPT system simulation model includes three parts: circuit, output and control; wherein the structure of the circuit part is the same as the circuit of the actual wireless charging WPT system or the design diagram of the wireless charging WPT system, the simulation element uses a non-ideal model, and its electrical parameters are set to the measured values ​​or rated values ​​of the actual element parameters, and the output part includes a virtual instrument module VM and AM and an RMS module, and the output part is used to collect and calculate the circuit mutual inductance M and the load R L System status data, the control part includes powergui module and pulse generator module. The control part uses powergui module to control the simulation status of the system, uses pulse generator module to output square wave of inverter operating frequency, and controls the inverter operating frequency in combination with inverter module.

[0020] Preferably, the mutual inductance M and the load R L The multi-level parameter combination table contains the combination parameters including circuit mutual inductance M and load R L , where the mutual inductance M ranges from 7 to 42uH, with an interval of 1.4uH between each level, and a total of 26 levels, load R L The value range of the mutual inductance M is 5 to 100Ω, with an interval of 2.5Ω between each level, and a total of 39 levels. Each level of the mutual inductance M is related to the load R L Each level of is combined to form a multi-level parameter combination table, which includes a total of 1014 parameter combinations.

[0021] Preferably, the system status data is the true effective value U of the output voltage of the inverter. in , the true effective value of the inverter output current I p , the ratio of the fundamental RMS value to the true RMS value of the inverter output voltage λ uinThe ratio of the fundamental RMS value to the true RMS value of the inverter output current λ ip .

[0022] Preferably, the simulation model constructed according to the parameter design in the parameter combination table is simulated and calculated to obtain the corresponding system state data, specifically: in the working state of the WPT system simulation model, the virtual instrument modules VM and AM in the WPT system simulation model are set as voltage sensors and current sensors respectively, and the voltage signal sequence {u i} and the current signal sequence {i i}, and then use the RMS module to calculate its true effective value and fundamental effective value, and then calculate the ratio of the two.

[0023] Preferably, the set frequency is 512 times the inverter operating frequency.

[0024] Preferably, the true effective value of the inverter output voltage and current, as well as the ratio of the fundamental effective value to the true effective value of the two, are calculated using the following calculation formula:

[0025]

[0026] Where m is the number of signal samples in a complete cycle, and ω is the frequency of the discrete Fourier transform target amplitude, which is equal to the operating frequency of the inverter.

[0027] Preferably, the input parameter of the BP neural network is the output voltage u of the inverter. in With the output current i p True effective value U in with I p , and the inverter output voltage u in With the output current i p The ratio of the fundamental RMS value to the true RMS value λ uin and λ ip , the output parameters are the mutual inductance M and the load R L , the number of input layer nodes is 4, the number of output layer nodes is 2, and the number of hidden layer nodes l is determined by the following empirical formula:

[0028]

[0029] Among them, p is the number of input layer nodes, q is the number of output layer nodes, and α is a constant set between 1 and 10.

[0030] Preferably, the error function of the BP neural network adopts mean square error MSE, Where n′ is the number of samples, y j is the true value of the jth sample, is the predicted value of the jth sample. The weights and thresholds are adjusted layer by layer based on the Levenberg-Marquardt algorithm until the error of the network output is reduced to the set threshold or the preset number of training times is reached.

[0031] The second aspect of the present invention proposes an online parameter identification system for a wireless charging system using a BP neural network using the parameter identification method described in the first aspect of the present invention, comprising a WPT simulation model construction module, a parameter combination table generation module, a simulation module, a BP neural network construction module, a state quantity acquisition module, and a parameter identification module, characterized in that:

[0032] WPT simulation model building module: used to build a simulation model of the WPT system according to the actual wireless charging WPT system to be identified or the design drawing of the wireless charging WPT system;

[0033] Parameter combination table generation module: used to design the circuit mutual inductance M and load R in the WPT system simulation model L A multi-level parameter combination table, wherein the multi-level parameter combination table is a table used to represent different parameters and their respective level combinations in experimental design;

[0034] Simulation module: used to simulate and calculate the constructed simulation model according to the parameter design in the parameter combination table, and obtain the corresponding circuit mutual inductance M and load R L System status data;

[0035] BP neural network building module: Based on the corresponding system status data, the BP neural network is established and trained to obtain the online identification model of WPT system parameters;

[0036] State quantity acquisition module: used to collect the required state quantities from the actual system;

[0037] Parameter identification module: used to input the state quantity collected by the state quantity acquisition module into the WPT system parameter online identification model for calculation, and obtain the real-time parameters of the wireless charging system, namely the mutual inductance and load resistance of WPT.

[0038] The beneficial effects of the present invention are that, compared with the prior art, (1) true effective value and harmonic factor are used as inputs for identifying the mutual inductance value and load resistance value of the WPT system, no additional communication equipment or auxiliary circuits are added, the design difficulty and cost are low, and the impact on the working state of the system is also low.

[0039] (2) Design a multi-level parameter combination table of mutual inductance and load, simulate the model according to the parameter design of the parameter combination table, and systematically show how the different levels of each factor in the experiment are combined with each other, while ensuring the representativeness and economy of the experiment.

[0040] (3) BP neural network is used to complete parameter identification, which does not require the calculation of complex circuit equations, and the error can be corrected using the actual circuit measured data, so the possibility of error is low. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is the circuit topology of the WPT system;

[0042] Figure 2 is the inverter output voltage waveform;

[0043] Figure 3 It is a BP neural network structure;

[0044] Figure 4 A flow chart of an online parameter identification method for a wireless charging system;

[0045] Figure 5 It is the Simulink simulation model of the WPT system. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only embodiments of a part of the present invention, rather than all embodiments. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the protection scope of the present invention.

[0047] like Figure 1 As shown in Figure 1, the WPT system consists of a DC voltage source, a full-bridge inverter, an SS compensation topology, a magnetic coupling mechanism, and a load. dc Provides power input for the entire system; switch tube S 1 ~S 4 To form a full-bridge inverter, two sets of switch tubes (S 1 , S 4 ) and (S 2 , S 3 ) take turns to conduct and output an approximate square wave voltage u in At the transmitting end, the transmitting coil L p With series compensation capacitor C p A series resonant circuit is formed. At the receiving end, the energy receiving coil L s With series compensation capacitor C s A series resonant circuit is formed and energy is transferred to the load R L . Where M is the coupling coil L p , L s The mutual inductance between p The transmitting coil Lp With compensation capacitor C p The series equivalent resistance, R s For receiving coil L s With compensation capacitor C s The series equivalent resistance. R e For load R L With R s The equivalent series load, R e =R s +R L .

[0048] In order to reduce the reactive power of the power input and improve the energy transmission capacity of the system, the operating frequency ω of the inverter will be adjusted to the inherent resonant frequency ω of the transmitting and receiving circuits of the system. 0 The same, that is

[0049]

[0050] Because the voltage signal u output by the inverter in The approximate square wave contains a large number of high-order harmonic signals, so the circuit state of the actual WPT system cannot be simply expressed by the circuit formula under the resonant state. In addition, the inverter in reality is not composed of ideal components, and its output voltage waveform will also be affected by the nonlinear load in the circuit, resulting in more distortion. Figure 2 It is a comparison of the distortion of the inverter output voltage waveform under different parameter settings.

[0051] Therefore, the inverter output voltage u in (t) can be expressed in the following Fourier form:

[0052]

[0053] Among them, U m_n is the peak value of the nth harmonic of the full-bridge inverter output power.

[0054] Depend on Figure 1 We can get the relationship:

[0055]

[0056] Among them, I p_n and U p_n are the nth harmonic current and voltage phasors of the system inverter output respectively; Z n is the system input impedance under the inverter output nth harmonic voltage excitation, and its expression is:

[0057]

[0058] Because the frequency of high-order harmonics is greater than the natural resonant frequency of the system, the input impedance of the WPT system is not completely resistive under the action of high-order harmonics, which causes a phase difference between the output voltage of the inverter and the high-order harmonics of the output current. p (t) can be expressed as:

[0059]

[0060] Among them, I m_n is the peak value of the nth harmonic of the inverter output current; It is the phase difference between the nth harmonic of the inverter output current and voltage.

[0061] From the above analysis, it can be concluded that the inverter output voltage u in And the output current i p The states of the high-order harmonic components contained in the WPT system circuit are related to the mutual inductance M and the load R L The mutual inductance M and the load R can be measured by the state parameters of the high-order harmonic components. L To characterize.

[0062] Embodiment 1 of the present invention provides a method for online parameter identification of a wireless charging WPT system, such as Figure 4 As shown, the specific steps include:

[0063] According to the actual wireless charging WPT system to be identified or the design drawing of the wireless charging WPT system, a simulation model of the WPT system is established.

[0064] Preferably, if Figure 5 As shown in the figure, the WPT system simulation model includes three parts: circuit, output and control; the structure of the circuit part is similar to that of the actual wireless charging WPT system or Figure 1 The circuit of the design diagram of the wireless charging WPT system shown in the figure is the same. The simulation components use non-ideal models, and their electrical parameters are set to the measured values ​​or rated values ​​of the actual component parameters. The output part includes virtual instrument modules VM and AM and RMS modules. The output part is used to collect and calculate the circuit mutual inductance M and load R L System status data, the control part includes powergui module and pulse generator module. The control part uses powergui module to control the simulation status of the system, uses pulse generator module to output square wave of inverter operating frequency, and controls the inverter operating frequency in combination with inverter module.

[0065] It should be noted that the simulation model was made using the Specialized Power Systems library in Simulink, and the control part used the powergui module to control the simulation state of the system. Specifically, this example is a discrete system simulation, the simulation step is set to 1 / (512*85000)s, and the operating frequency of the inverter is a square wave of 85kHz.

[0066] Design of circuit mutual inductance M and load R in WPT system simulation model L A multi-level parameter combination table, wherein the multi-level parameter combination table is a table used to represent different parameters and their respective level combinations in experimental design;

[0067] It should be noted that the multi-level parameter combination table is a table used to represent different parameters and their level combinations in experimental design. It can systematically show how the different levels of each factor in the experiment are combined with each other for experimental testing. It can ensure that the parameter combination in the experiment is both representative and economical, because not all combinations need to be tested.

[0068] Preferably, in this experiment, the combined parameters only include the circuit mutual inductance M and the load R L The number of parameters is relatively small, so the parameter combination table is all possible combinations between the two parameters. Among them, the value range of M is 7 to 42uH, and the interval between each level is 1.4uH, which is divided into 26 levels in total. L The value range of is 5 to 100Ω, each level is separated by 2.5Ω, and there are 39 levels in total. The parameter combination table includes 1014 parameter combinations in total.

[0069] According to the parameter design in the parameter combination table, the constructed simulation model is simulated and calculated to obtain the corresponding circuit mutual inductance M and load R L Relevant system status data;

[0070] Preferably, the system status data is the true effective value U of the output voltage of the inverter in , the true effective value of the inverter output current I p , the ratio of the fundamental RMS value to the true RMS value of the inverter output voltage λ uin The ratio of the fundamental RMS value to the true RMS value of the inverter output current λ ip ;

[0071] Preferably, the simulation model constructed according to the parameter design in the parameter combination table is simulated and calculated to obtain the corresponding system state data, specifically: in the working state of the WPT system simulation model, the virtual instrument modules VM and AM in the WPT system simulation model are set as voltage sensors and current sensors respectively, and the voltage signal sequence {u i} and the current signal sequence {i i}, and then use the RMS module to calculate its true effective value and fundamental effective value, and then calculate the ratio of the two.

[0072] Preferably, the set frequency is 512 times the inverter operating frequency ω.

[0073] Preferably, the true effective value of the inverter output voltage and current, as well as the ratio of the fundamental effective value to the true effective value of the two, are calculated using the following calculation formula:

[0074]

[0075] Where m is the number of signal samples in a complete cycle, and ω is the frequency of the discrete Fourier transform target amplitude, which is equal to the operating frequency of the inverter.

[0076] Based on the corresponding system status data, a BP neural network is established and trained to obtain an online identification model of WPT system parameters;

[0077] Preferably, the input parameter of the BP neural network is the output voltage u of the inverter. in With the output current i p True effective value U in with I p , and the inverter output voltage u in With the output current i p The ratio of the fundamental RMS value to the true RMS value λ uin and λ ip , the output parameters are M and R L , the number of input layer nodes is 4, the number of output layer nodes is 2, and the number of hidden layer nodes l is determined by the following empirical formula:

[0078]

[0079] Among them, p is the number of input layer nodes, q is the number of output layer nodes, and α is a constant set between 1 and 10. In this example, α is taken as the maximum value 10, and the value of l is 12.

[0080] Preferably, if Figure 3As shown in the figure, the calculation process of the BP neural network is that the input data enters the network from the input layer, passes through each hidden layer layer by layer, and then is transferred to the output layer to obtain the output data. Each node of the BP neural network is connected to all the nodes of the adjacent layer, so its hidden layer nodes have input and output, while the input layer nodes only have output, and the output layer nodes only have input. The connection between the nodes of the adjacent layers has a weight w ij , each layer except the input layer has a threshold b j , then the input value of each layer except the input layer is the weighted sum of all nodes in the previous layer plus the threshold, that is, I j =∑w ij O i +b j , the hidden layer nodes have a nonlinear activation function (tanh function in this real-time example), and the output is obtained by performing a nonlinear transformation on the input, that is, O i ′=tanh(I j ).

[0081] The training process of BP neural network is the process of updating its weights and thresholds. First, the weights and thresholds are randomly initialized, and then sample data is input into the BP neural network. The predicted output obtained by forward propagation is compared with the sample output, and the mean square error (MSE) is calculated. Where n′ is the number of samples, y j is the true value of the jth sample, is the predicted value of the jth sample, and then the error is back-propagated from the output layer to the input layer, and the weights and thresholds are adjusted layer by layer based on a specific optimization method (Levenberg-Marquardt algorithm in this real-time example). The forward propagation and back-propagation process are repeated until the error of the network output is reduced to an acceptable level, or the preset number of training times is reached (the preset number of training times in this example is 1000 times).

[0082] The required state quantities are collected from the actual system and input into the online identification model of WPT system parameters for calculation, so as to obtain the mutual inductance and load resistance of WPT, the real-time parameters of the wireless charging system.

[0083] In this real-time example, the average relative error of mutual inductance parameter identification is 2.3%, the maximum relative error is 8.1%, the average relative error of load resistance parameter identification is 3.7%, the maximum relative error is 12.1%, which has practical value.

[0084] Embodiment 2 of the present invention proposes an online parameter identification system of a wireless charging system using a BP neural network using the parameter identification method described in Embodiment 1 of the present invention, comprising a WPT simulation model construction module, a parameter combination table generation module, a simulation module, a BP neural network construction module, a state quantity acquisition module, and a parameter identification module, characterized in that:

[0085] WPT simulation model building module: used to build a simulation model of the WPT system according to the actual wireless charging WPT system to be identified or the design drawing of the wireless charging WPT system;

[0086] Parameter combination table generation module: used to design the circuit mutual inductance M and load R in the WPT system simulation model L A multi-level parameter combination table, wherein the multi-level parameter combination table is a table used to represent different parameters and their respective level combinations in experimental design;

[0087] Simulation module: used to simulate and calculate the constructed simulation model according to the parameter design in the parameter combination table, and obtain the corresponding circuit mutual inductance M and load R L System status data;

[0088] BP neural network building module: Based on the corresponding system status data, the BP neural network is established and trained to obtain the online identification model of WPT system parameters;

[0089] State quantity acquisition module: used to collect the required state quantities from the actual system;

[0090] Parameter identification module: used to input the state quantity collected by the state quantity acquisition module into the WPT system parameter online identification model for calculation, and obtain the real-time parameters of the wireless charging system, namely the mutual inductance and load resistance of WPT.

[0091] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0092] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0093] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0094] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A BP neural network wireless charging system online parameter identification method, characterized in that: Contains the following: According to the actual wireless charging WPT system to be identified or the design drawing of the wireless charging WPT system, a simulation model of the WPT system is established; Design of circuit mutual inductance M and load R in WPT system simulation model L A multi-level parameter combination table, wherein the multi-level parameter combination table is a table used to represent different parameters and their respective level combinations in experimental design; According to the parameter design in the parameter combination table, the constructed simulation model is simulated and calculated to obtain the corresponding circuit mutual inductance M and load R L Relevant system status data; Based on the corresponding system status data, a BP neural network is established and trained to obtain an online identification model of WPT system parameters; The required state quantities are collected from the actual system and input into the online identification model of WPT system parameters for calculation, so as to obtain the mutual inductance and load resistance of WPT, the real-time parameters of the wireless charging system.

2. The method for online parameter identification of a wireless charging system using a BP neural network according to claim 1, characterized in that: The WPT system simulation model includes three parts: circuit, output and control. The structure of the circuit part is the same as the circuit of the actual wireless charging WPT system or the design diagram of the wireless charging WPT system. The simulation components use non-ideal models, and their electrical parameters are set to the measured values ​​or rated values ​​of the actual component parameters. The output part includes virtual instrument modules VM and AM and RMS module. The output part is used to collect and calculate the circuit mutual inductance M and load R L The control part uses the powergui module to control the simulation state of the system, and uses the pulse generator module to output the square wave of the inverter operating frequency, and combines with the inverting module to control the inverter operating frequency.

3. The method for online parameter identification of a wireless charging system using a BP neural network according to claim 1, characterized in that: The mutual inductance M and the load R L The multi-level parameter combination table contains the combination parameters including circuit mutual inductance M and load R L , where the mutual inductance M ranges from 7 to 42uH, with an interval of 1.4uH between each level, and a total of 26 levels, load R L The value range of the mutual inductance M is 5 to 100Ω, with an interval of 2.5Ω between each level, and a total of 39 levels. Each level of the mutual inductance M is related to the load R L Each level of is combined to form a multi-level parameter combination table, which includes a total of 1014 parameter combinations.

4. The method for online parameter identification of a wireless charging system using a BP neural network according to claim 1, characterized in that: The system status data is the true effective value U of the inverter output voltage in , the true effective value of the inverter output current I p , the ratio of the fundamental RMS value to the true RMS value of the inverter output voltage λ uin The ratio of the fundamental RMS value to the true RMS value of the inverter output current λ ip .

5. The method for online parameter identification of a wireless charging system using a BP neural network according to claim 4 is characterized in that: The simulation model constructed according to the parameter design in the parameter combination table is simulated and calculated to obtain the corresponding system status data, specifically: in the working state of the WPT system simulation model, the virtual instrument modules VM and AM in the WPT system simulation model are set as voltage sensors and current sensors respectively, and the voltage signal sequence {u i } and the current signal sequence {i i }, and then use the RMS module to calculate its true effective value and fundamental effective value, and then calculate the ratio of the two.

6. The method for online parameter identification of a wireless charging system using a BP neural network according to claim 4 is characterized in that: The set frequency is 512 times the inverter operating frequency.

7. The method for online parameter identification of a wireless charging system using a BP neural network according to claim 6 is characterized in that: Calculate the true RMS value of the inverter output voltage and current, as well as the ratio of their fundamental RMS value to the true RMS value, using the following calculation formula: Where m is the number of signal samples in a complete cycle, and ω is the operating frequency of the inverter.

8. The method for online parameter identification of a wireless charging system using a BP neural network according to claim 4 is characterized in that: The input parameter of the BP neural network is the output voltage u of the inverter in With the output current i p True effective value U in with I p , and the inverter output voltage u in With the output current i p The ratio of the fundamental effective value to the true effective value λ uin and λ ip , the output parameters are the mutual inductance M and the load R L , the number of input layer nodes is 4, the number of output layer nodes is 2, and the number of hidden layer nodes l is determined by the following empirical formula: Among them, p is the number of input layer nodes, q is the number of output layer nodes, and α is a constant set between 1 and 10.

9. The method for online parameter identification of a wireless charging system using a BP neural network according to claim 8, characterized in that: The error function of the BP neural network adopts the mean square error MSE. Where n′ is the number of samples, y j is the true value of the jth sample, is the predicted value of the jth sample. The weights and thresholds are adjusted layer by layer based on the Levenberg-Marquardt algorithm until the error of the network output is reduced to the set threshold or the preset number of training times is reached.

10. An online parameter identification system for a wireless charging system using a BP neural network using the parameter identification method according to any one of claims 1 to 9, comprising a WPT simulation model building module, a parameter combination table generation module, a simulation module, a BP neural network building module, a state quantity acquisition module, and a parameter identification module, characterized in that: WPT simulation model building module: used to build a simulation model of the WPT system according to the actual wireless charging WPT system to be identified or the design drawing of the wireless charging WPT system; Parameter combination table generation module: used to design the circuit mutual inductance M and load R in the WPT system simulation model L A multi-level parameter combination table, wherein the multi-level parameter combination table is a table used to represent different parameters and their respective level combinations in experimental design; Simulation module: used to simulate and calculate the constructed simulation model according to the parameter design in the parameter combination table, and obtain the corresponding circuit mutual inductance M and load R L System status data; BP neural network building module: Based on the corresponding system status data, the BP neural network is established and trained to obtain the online identification model of WPT system parameters; State quantity acquisition module: used to collect the required state quantities from the actual system; Parameter identification module: It is used to input the state quantity collected by the state quantity acquisition module into the WPT system parameter online identification model for calculation, and obtain the real-time parameters of the wireless charging system, namely the mutual inductance and load resistance of WPT.

Images