CN111987779B - MC-WPT system load and mutual inductance identification model, method and system based on TensorFlow - Google Patents

MC-WPT system load and mutual inductance identification model, method and system based on TensorFlow Download PDF

Info

Publication number
CN111987779B
CN111987779B CN202010895878.1A CN202010895878A CN111987779B CN 111987779 B CN111987779 B CN 111987779B CN 202010895878 A CN202010895878 A CN 202010895878A CN 111987779 B CN111987779 B CN 111987779B
Authority
CN
China
Prior art keywords
mutual inductance
wpt system
load
matrix
model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202010895878.1A
Other languages
Chinese (zh)
Other versions
CN111987779A (en
Inventor
苏玉刚
阳剑
王智慧
孙跃
戴欣
唐春森
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University
Original Assignee
Chongqing University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chongqing University filed Critical Chongqing University
Priority to CN202010895878.1A priority Critical patent/CN111987779B/en
Publication of CN111987779A publication Critical patent/CN111987779A/en
Application granted granted Critical
Publication of CN111987779B publication Critical patent/CN111987779B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J7/00—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries
    • H02J7/02—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries for charging batteries from ac mains by converters
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J50/00—Circuit arrangements or systems for wireless supply or distribution of electric power
    • H02J50/10—Circuit arrangements or systems for wireless supply or distribution of electric power using inductive coupling
    • H02J50/12—Circuit arrangements or systems for wireless supply or distribution of electric power using inductive coupling of the resonant type
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J50/00—Circuit arrangements or systems for wireless supply or distribution of electric power
    • H02J50/90—Circuit arrangements or systems for wireless supply or distribution of electric power involving detection or optimisation of position, e.g. alignment

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Power Engineering (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Feedback Control In General (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention relates to the technical field of MC-WPT, and particularly discloses a load and mutual inductance identification model, method and system of an MC-WPT system based on TensorFlow. On the whole, the invention can realize the on-line simultaneous recognition of the load and the mutual inductance by off-line training the model and guiding the trained model into the microcontroller, has high recognition speed and high precision, is beneficial to the real-time control of the system, has lower cost, is easy to realize and is beneficial to the engineering popularization and application.

Description

MC-WPT system load and mutual inductance identification model, method and system based on TensorFlow
Technical Field
The invention relates to the technical field of MC-WPT (magnetic field coupling wireless power transmission), in particular to a model, a method and a system for recognizing load and mutual inductance of an MC-WPT system based on TensorFlow.
Background
With the development of economic society, the flexibility of mobile electric equipment, such as rail trains, mobile hoisting equipment, household appliances, rotating machinery and other equipment, can be influenced by using the traditional lead for power supply, the potential safety hazard of power utilization can be increased under certain special environments, and the challenge is brought to the practical application of engineering. The Wireless Power Transfer (WPT) technology provides a safe, environment-friendly, convenient and easy-to-maintain Power supply mode, and the Power supply mode is concerned and researched by numerous scholars at home and abroad, so that the novel Power supply mode is promoted to be continuously developed. Among them, the Magnetic Coupling Wireless Power Transfer (MC-WPT) technology is one of the most interesting technologies at present, and is gradually popularized and applied in the fields of electric vehicles, household appliances, aerospace, underwater equipment Power supply, and the like.
In some practical applications of the MC-WPT system, for example, in an electric vehicle wireless charging/power supplying system, a change in mutual inductance value will be caused by a change in a relative position between an energy transmitting terminal and an energy receiving terminal of the system, and a change in system load will be caused by different energy receiving devices, which all cause a change in load and mutual inductance when the electric vehicle is wirelessly charged/powered, thereby affecting system energy transmission efficiency and power transmission capability. Therefore, in the MC-WPT system, when the load and the mutual inductance change, the system needs to adjust the control mode of the energy transmitting terminal according to the current situation, so as to realize the optimal efficiency tracking or the constant voltage output of the system, and the identification of the load and the mutual inductance parameter is the key problem.
The current load and mutual inductance identification is mainly carried out by establishing an identification model based on the steady-state characteristics of a system or adding an additional circuit for auxiliary identification, and the load and mutual inductance identification is carried out by using methods such as a genetic algorithm. In the existing load and mutual inductance identification method, part of methods can only carry out single-parameter identification on the load or the mutual inductance; the problem that the online identification speed is slow exists in part of methods, and real-time control over the system is not facilitated; the partial method has the problems of high implementation cost, low precision and the like.
Disclosure of Invention
The invention provides a model, a method and a system for recognizing load and mutual inductance of an MC-WPT system based on TensorFlow, which solve the technical problems that: the load and mutual inductance identification method of the MC-WPT system at present cannot realize double-parameter identification at the same time, and is high in identification speed, low in cost and high in identification precision.
In order to solve the technical problems, the invention provides a MC-WPT system load and mutual inductance identification model based on TensorFlow, which comprises the following generation steps:
s1, constructing a full-connection neural network model based on a TensorFlow framework;
s2, establishing COSMOL and Simulink simulation models of the MC-WPT system to obtain a plurality of groups of input current values and coil transmission distance simulation data of the MC-WPT system, and dividing the simulation data into a training set and a test set;
s3, inputting the training set into the fully-connected neural network model for model training, and continuously optimizing parameters in the fully-connected neural network model according to a training error value;
and S4, when the training error rate of the fully-connected neural network model is lower than a preset error rate, finishing training to obtain the MC-WPT system load and mutual inductance identification model after training.
Preferably, the fully-connected neural network model comprises an input layer, an output layer and 1 st to N hidden layers which are sequentially and fully connected between the input layer and the output layer, wherein N is more than or equal to 1; the 1 st to N hidden layers are provided with k nodes, and k is more than or equal to 2.
Preferably, the nonlinear activation function of the 1 st to N th hidden layers uses a Sigmoid activation function in a tensrflow framework.
Preferably, the parameters optimized in step S3 include a 1 st weight matrix and a 1 st bias matrix acting on the 1 st hidden layer, and a 2 nd weight matrix and a 2 nd bias matrix acting on the 2 nd hidden layer, up to an nth weight matrix and an nth bias matrix acting on the nth hidden layer, and an N +1 th weight matrix and an N +1 th bias matrix acting on the output layer.
Preferably, N is 3 and k is 10.
Preferably, the preset error rate is set to not more than 2%.
Based on the MC-WPT system load and mutual inductance identification model, the invention also provides a MC-WPT system load and mutual inductance identification method, which comprises the following steps:
x1., detecting the input current value and the coil transmission distance of the current MC-WPT system;
x2. inputting the current input current value and the coil transmission distance into the MC-WPT system load and mutual inductance identification model, and calculating to obtain corresponding load value and mutual inductance value.
Further, in the step X2, the equation for calculating the load and mutual inductance identification model of the MC-WPT system is as follows:
Figure GDA0002681688770000031
wherein l1Represents an intermediate variable of the 1 st hidden layer, [ h I ]m]Representing a matrix consisting of the inter-coil transmission distances and the input current values of the MC-WPT system,
Figure GDA0002681688770000038
and
Figure GDA0002681688770000039
respectively representing the 1 st weight matrix and the 1 st bias matrix, L1Represents a pair of1Substituting each element into the activation function
Figure GDA0002681688770000032
Carrying out operation to obtain a hidden layer output matrix;
l2an intermediate variable representing the 2 nd hidden layer,
Figure GDA0002681688770000033
and
Figure GDA0002681688770000034
respectively representing said 2 nd weight matrix and said 2 nd bias matrix, L2Represents a pair of2Substituting each element into the activation function
Figure GDA0002681688770000035
Carrying out operation to obtain a hidden layer output matrix;
……；
lNan intermediate variable representing the Nth hidden layer,
Figure GDA0002681688770000036
and
Figure GDA0002681688770000037
respectively representing the Nth weight matrix and the Nth bias matrix, LNRepresents a pair ofNSubstituting each element into the activation function
Figure GDA0002681688770000041
Carrying out operation to obtain a hidden layer output matrix;
lN+1an intermediate variable representing the output layer,
Figure GDA0002681688770000042
and
Figure GDA0002681688770000043
respectively representing the N +1 th weight matrix and the N +1 th bias matrix, M, ReqRespectively representing the mutual inductance value and the load value output by the output layer.
The invention also provides a TensorFlow-based MC-WPT system load and mutual inductance identification system, which comprises a controller, a current detection module and a distance measurement module, wherein the current detection module and the distance measurement module are connected with the controller; the current detection module is used for detecting an input current value of an LCC circuit topology at a transmitting end in the MC-WPT system and sending the input current value to the controller; the distance measuring module is used for detecting the transmission distance between a transmitting coil and a receiving coil in the MC-WPT system and sending the transmission distance to the controller; the controller is used for installing the MC-WPT system load and mutual inductance identification model and calculating a load value and a mutual inductance value under a corresponding input current value and a corresponding transmission distance according to the MC-WPT system load and mutual inductance identification method.
Preferably, the current detection module is a hall sensor, and the distance measurement module is an infrared distance measurement sensor.
The invention provides a TensorFlow-based MC-WPT system load and mutual inductance identification model, which is based on a TensorFlow deep learning framework and adopts a neural network model, so that the load and mutual inductance identification problem of an MC-WPT system are equivalent to the solving problem of a nonlinear equation, the solving problem is further converted into a deep learning nonlinear fitting problem, a training set is adopted to train the model for ten thousand times, and finally the MC-WPT system load and mutual inductance identification model with high identification speed and high precision is obtained;
the invention also provides a TensorFlow-based MC-WPT system load and mutual inductance identification method and a TensorFlow-based MC-WPT system load and mutual inductance identification system.
On the whole, the invention can realize the on-line simultaneous recognition of the load and the mutual inductance by off-line training the model and guiding the trained model into the microcontroller, has high recognition speed and high precision, is beneficial to the real-time control of the system, has lower cost, is easy to realize and is beneficial to the engineering popularization and application.
Drawings
Fig. 1 is a main circuit topology diagram of a dual LCC type MC-WPT system provided by an embodiment of the present invention;
fig. 2 is a first equivalent circuit diagram of a dual LCC type MC-WPT system according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a coupling mechanism in a dual LCC MC-WPT system according to an embodiment of the present invention;
fig. 4 is an equivalent circuit diagram of a receiving end of a dual LCC MC-WPT system according to an embodiment of the present invention;
fig. 5 is a second equivalent circuit diagram of a dual LCC type MC-WPT system provided by an embodiment of the present invention;
fig. 6 is a network structure diagram of a load and mutual inductance identification model of the MC-WPT system provided in embodiment 1 of the present invention;
fig. 7 is a simulation model diagram of a coupling mechanism of a dual LCC MC-WPT system according to embodiment 1 of the present invention;
FIG. 8 is a three-dimensional graph of the relationship between d, h and M provided in embodiment 1 of the present invention;
FIG. 9 is a Simulink simulation model of a dual LCC type MC-WPT system provided in embodiment 1 of the present invention;
FIG. 10 is a graph of training error versus training times provided in example 1 of the present invention;
fig. 11 is a simulation diagram of a load and mutual inductance identification model of a dual-LCC MC-WPT system based on a tensrflow framework according to embodiment 3 of the present invention.
Detailed Description
The embodiments of the present invention will be described in detail below with reference to the accompanying drawings, which are given solely for the purpose of illustration and are not to be construed as limitations of the invention, including the drawings which are incorporated herein by reference and for illustration only and are not to be construed as limitations of the invention, since many variations thereof are possible without departing from the spirit and scope of the invention.
The embodiment aims at mutual loading and mutual inductance identification of the MC-WPT system, and firstly needs to know which MC-WPT system is and why mutual loading and mutual inductance of the MC-WPT system needs to be identified.
1. Overview of the System
The present embodiment is described by taking a dual LCC type MC-WPT system as an example, but is not limited to the LCC type MC-WPT system.
The double-LCC type MC-WPT system has the characteristics of symmetrical structures of a transmitting end and a receiving end, easiness in tuning of a compensation topology network, independence of output current and load and the like, is good in offset resistance, and is widely applied to wireless charging systems of electric automobiles and the like.
FIG. 1 shows a main circuit topology of a dual LCC MC-WPT system, wherein a system input power supply E is arranged at a system transmitting enddcThe power supply can be directly provided by a direct current power supply or obtained by rectifying and filtering an alternating current power grid. Switch tube S1-S4And a high-frequency inverter circuit is formed, high-frequency alternating current is output, and energy is transmitted to a receiving end by a transmitting coil after the high-frequency alternating current passes through the LCC type resonance compensation network. At a system receiving end, after the receiving coil picks up the electric energy, the electric energy is rectified and filtered into a load R through an LCC resonance compensation networkLAnd (5) supplying power.
At the transmitting end of the system, a DC power supply EdcThe high-frequency alternating current is output after passing through the high-frequency inverter circuit and can be equivalent to a high-frequency alternating current voltage source
Figure GDA0002681688770000061
Its effective value UmAnd EdcThe relationship is as follows:
Figure GDA0002681688770000062
at the receiving end of the system, a rectification filtering link and a load RLWhen in parallel connection, the resistance can be equivalent to a load resistance ReqThe relationship is given as follows:
Figure GDA0002681688770000063
therefore, the dual LCC type MC-WPT system main circuit topology shown in fig. 1 can be further equivalent to the system equivalent circuit shown in fig. 2, in which
Figure GDA0002681688770000064
High-frequency AC voltage source L for output of high-frequency inverter circuit at transmitting endpFor transmitting end transmitting coil self-inductance, LsFor self-inductance of the receiving coil at the receiving end, RpFor transmitting coil internal resistance, RsFor receiving internal resistance of the coil, Lf1、Cf1And CpForm a transmitting end compensation network, Lf2、Cf2And CsForming a receiver-side compensation network, ReqIs the load equivalent resistance. M represents the mutual inductance between the transmitting side coil and the receiving side coil.
2. System modeling
On the premise that the planes of the transmitting coil and the receiving coil are parallel, the mutual inductance value M is only related to the transmission distance h (which may be simply referred to as transmission distance h) and the offset distance d between the coils after the number of turns and the geometric dimension of the coils are determined, as shown in fig. 3.
The relationship of the mutual inductance values M and d, h can be described by an implicit function as:
M＝g(h，d) (3)
the detection of the offset distance h is relatively complex to implement and can be replaced by detecting other parameters, taking into account the need to identify the load, sinceThis requires detecting the current value of the system transmitter circuit to identify the load and mutual inductance. In the circuit system, the impedance analysis is first performed on the receiving end circuit, as shown in fig. 4, the receiving end picks up the voltage as
Figure GDA0002681688770000071
The receiving end circuit impedance expression is as follows:
Z2＝(Req+ZLf2)//ZCf2+ZCs+ZLs+Rs (4)
wherein the content of the first and second substances,
Figure GDA00026816887700000710
ω is the system resonant frequency, and the overall impedance of the receiving end circuit can be further obtained according to equation (4):
Figure GDA0002681688770000072
because the load and mutual inductance identification of the system is mainly analyzed by detecting related parameters from the transmitting end, the impedance analysis is carried out on the transmitting end circuit, and the influence of the receiving end circuit on the transmitting end circuit can be equivalent to impedance:
Figure GDA0002681688770000073
as can be seen from equation (6), the mutual inductance and the influence of load change on the system are reflected in the change of the reflection impedance to the transmitting end. Therefore, in order to further study the influence of the change of the reflection impedance on the system, the main circuit of the dual LCC type MC-WPT system needs to be equivalent to an equivalent circuit model as shown in fig. 4. Wherein the content of the first and second substances,
Figure GDA0002681688770000074
represents the input current, ImRepresents its effective value, ZinIs the input impedance of the system.
As can be seen from FIG. 5, the input impedance and the effective value of the currentImThe following relationship holds:
Figure GDA0002681688770000075
wherein the total input impedance Z of the systeminComprises the following steps:
Figure GDA0002681688770000076
bringing formula (6) into formula (8) and taking
Figure GDA0002681688770000077
Figure GDA0002681688770000078
Can obtain Zin：
Figure GDA0002681688770000079
The load equivalent resistance R can be obtained by bringing the formula (9) into the formula (7)eqAnd mutual inductance M and input current ImThe relation of (A) is as follows:
Figure GDA0002681688770000081
in equation (10), the effective value of the system input current I can be seenmLoaded equivalent resistance ReqAnd mutual inductance M, the combination of equation (3) can be obtained:
(M，Req)＝f(Im，h) (11)
therefore, the load and mutual inductance identification can be carried out by detecting the effective value I of the input currentmAnd a transmission distance h. The identification of load and mutual inductance is further transformed to solve the problem of the f-function in the non-linear equation (11). In this embodiment, a neural network model is established based on a TensorFlow deep learning framework, and the neural network model is performed in a nonlinear function fitting mannerAnd (4) approximate solving of the f function.
Example 1
The embodiment provides a MC-WPT system load and mutual inductance identification model based on the dual LCC type MC-WPT system, and the generation steps include:
s1, constructing a full-connection neural network model based on a TensorFlow framework;
s2, establishing COSMOL and Simulink simulation models of the MC-WPT system to obtain a plurality of groups of input current values and coil transmission distance simulation data of the MC-WPT system, and dividing the simulation data into a training set and a test set;
s3, inputting the training set into the fully-connected neural network model for model training, and continuously optimizing parameters in the fully-connected neural network model according to a training error value;
and S4, when the training error rate of the fully-connected neural network model is lower than a preset error rate, finishing training to obtain the MC-WPT system load and mutual inductance identification model after training.
The fully-connected neural network model comprises an input layer, an output layer and 1 st to N hidden layers which are sequentially and fully connected between the input layer and the output layer, wherein N is more than or equal to 1; the 1 st to N hidden layers are provided with k nodes, and k is more than or equal to 2; the output layer has two fixed nodes.
Wherein, the nonlinear activation function of the 1 st to N hidden layers uses a Sigmoid activation function in a TensorFlow frame. The optimized parameters of step S3 include the 1 st weight matrix and the 1 st bias matrix acting on the 1 st hidden layer, and the 2 nd weight matrix and the 2 nd bias matrix acting on the 2 nd hidden layer, up to the nth weight matrix and the nth bias matrix acting on the nth hidden layer, and the N +1 th weight matrix and the N +1 th bias matrix acting on the output layer.
In step S4, the preset error rate is set to not more than 2%. In other embodiments, this may be determined based on actual demand, such as 3%.
In the present embodiment, N is preferably 3, and k is preferably 10, and as shown in fig. 6, the 1 st to N-th hidden layers are represented as hidden layers 1 to 3 in fig. 6. In other embodiments, N, k may be determined based on actual needs.
The hidden layer 1 can be described by the formula:
Figure GDA0002681688770000091
wherein
Figure GDA0002681688770000092
And
Figure GDA0002681688770000093
respectively representing a weight matrix and a bias matrix of the neural network;
Figure GDA0002681688770000094
will l1Substituting the Simgioid activation function results in the output of hidden layer 1 as:
Figure GDA0002681688770000095
by analogy, l can be obtained2，L2，l3，L3，l4And further obtaining the load and mutual inductance value:
Figure GDA0002681688770000096
after the model is established, the weight matrix and the bias matrix in the model need to be determined
Figure GDA0002681688770000097
Figure GDA0002681688770000098
So that the output of the neural network model is as close to the actual value as possible (step S3).In order to determine the parameters, a batch of training data is required to train the model, so as to determine the parameters of the model, the present embodiment obtains the data required by training the model by using the simulation software to build the system simulation model (step S2).
Firstly, a coupling mechanism simulation model is established in COMSOL multi-physics field simulation software as shown in FIG. 7, and simulation parameters are shown in Table 1.
TABLE 1 simulation model parameters of coupling mechanism of dual LCC type MC-WPT system
Figure GDA0002681688770000099
Figure GDA0002681688770000101
The coupling mechanism simulation is carried out in COMSOL multi-physical field simulation software, and the self-inductance of the transmitting coil and the self-inductance of the receiving coil are both 540 uH. After parametric scan simulation is performed, 30 groups of h and M data are obtained, and from the simulation data, it can be seen that after the number of turns and the geometric size of the coil are determined, a three-dimensional graph of the relationship between the transmission distance h and the offset distance d and the mutual inductance value M is shown in fig. 8.
After the simulation data of the coupling mechanism is acquired, a double LCC type MC-WPT simulation model shown in FIG. 9 is established in Simulink, and the single simulation time is set to 0.02s, at which time the system is already in steady-state operation. The transmission distance h and mutual inductance value M in the model are set by using h and M data sets simulated by COMSOL software, and the rest parameter settings in the model are shown in Table 2.
TABLE 2 Dual LCC MC-WPT systems simulation Primary parameters
Parameter(s) Numerical value Parameter(s) Numerical value
System frequency f 79KHz Transmitting terminal compensation inductance 67uH
Transmitting coil inductance Lp 534uH Receiving end compensation inductance 67uH
Inductance L of receiving coils 534uH Load Req 10-100Ω
Transmitting terminal compensation capacitor Cp 8.69nF DC power supply Edc 360V
Receiving end compensation capacitor Cs 8.69nF
In the simulation model shown in fig. 9, the simulation model is run, and automatic simulation is performed by writing M files, each time the transmission distance h andload equivalent resistance ReqObtaining the system input current value ImAnd the mutual inductance value M, 410 sets of simulation model data were thus obtained.
When the TensorFlow load and mutual inductance recognition model is trained, 370 groups of data are randomly selected to serve as a training set, 40 groups of data serve as a testing set, and the testing set is only used for testing the training effect of the model and does not participate in model training. Parameters in the TensorFlow Optimizer Adam Optimizer optimization model are used according to the training error values by continuously inputting training set data into the model until the training error values fall to a satisfactory error value (which may also be expressed as an error rate). As shown in fig. 10, after 10000 times of training, the error value of the model training has been reduced to a very small value, at this time, the recognition precision of the training set mutual inductance M is 99%, the recognition precision of the testing set mutual inductance M reaches 99.5%, and the training set load R iseqRecognition accuracy 98%, test set load ReqThe recognition precision is 99%, and the model training is completed.
The embodiment of the invention provides a TensorFlow-based MC-WPT system load and mutual inductance identification model, which is based on a TensorFlow deep learning framework and adopts a neural network model, so that the load and mutual inductance identification problem of an MC-WPT system are equivalent to the solving problem of a nonlinear equation, the solving problem is further converted into a deep learning nonlinear fitting problem, a training set is adopted to train the model for ten thousand times, and finally the MC-WPT system load and mutual inductance identification model with high identification speed and high precision is obtained.
Example 2
Based on the MC-WPT system load and mutual inductance identification model in the embodiment 1, the embodiment of the invention provides a MC-WPT system load and mutual inductance identification method, which comprises the following steps:
x1., detecting the input current value and the coil transmission distance of the current MC-WPT system;
x2. inputting the current input current value and the coil transmission distance into the MC-WPT system load and mutual inductance identification model, and calculating to obtain corresponding load value and mutual inductance value.
In the step X2, the equation for calculating the MC-WPT system load and the mutual inductance identification model is as follows:
Figure GDA0002681688770000111
wherein l1Represents an intermediate variable of the 1 st hidden layer, [ h I ]m]Representing a matrix consisting of the inter-coil transmission distances and the input current values of the MC-WPT system,
Figure GDA00026816887700001212
and
Figure GDA00026816887700001213
respectively representing the 1 st weight matrix and the 1 st bias matrix, L1Represents a pair of1Substituting each element into the activation function
Figure GDA0002681688770000121
Carrying out operation to obtain a hidden layer output matrix;
l2an intermediate variable representing the 2 nd hidden layer,
Figure GDA0002681688770000122
and
Figure GDA0002681688770000123
respectively representing said 2 nd weight matrix and said 2 nd bias matrix, L2Represents a pair of2Substituting each element into the activation function
Figure GDA0002681688770000124
Carrying out operation to obtain a hidden layer output matrix;
……；
lNan intermediate variable representing the Nth hidden layer,
Figure GDA0002681688770000125
and
Figure GDA0002681688770000126
respectively represent the Nth weightThe weight matrix and the Nth bias matrix, LNRepresents a pair ofNSubstituting each element into the activation function
Figure GDA0002681688770000127
Carrying out operation to obtain a hidden layer output matrix;
lN+1an intermediate variable representing the output layer,
Figure GDA0002681688770000128
and
Figure GDA0002681688770000129
respectively representing the N +1 th weight matrix and the N +1 th bias matrix, M, ReqRespectively representing the mutual inductance value and the load value output by the output layer.
Similarly, N is 3 and k is 10 in this embodiment. After the model training of example 1 is completed, the weight matrix and the bias matrix are derived
Figure GDA00026816887700001210
When load and mutual inductance are identified on line, the current value I is input through a detection systemmAnd a transmission distance h, wherein the derived value is calculated by substituting the formula (16), so that the load and mutual inductance can be identified in a microcontroller.
Figure GDA00026816887700001211
Example 3
The embodiment of the invention provides a TensorFlow-based MC-WPT system load and mutual inductance identification system, which comprises a controller, a current detection module and a distance measurement module, wherein the current detection module and the distance measurement module are connected with the controller; the current detection module is used for detecting an input current value I of an LCC circuit topology at a transmitting end in the MC-WPT systemmAnd sending to the controller; the distance measurement module is used for detecting the transmission distance h between the transmitting coil and the receiving coil in the MC-WPT system and sending the transmission distance h to the controller; controller for installing MC-WPT system load andand (3) calculating a load value and a mutual inductance value under a corresponding input current value and a transmission distance according to the MC-WPT system load and mutual inductance identification method in the embodiment 2.
Preferably, the current detection module is a hall sensor. The distance measurement module is an infrared distance measurement sensor and is arranged in the center of a transmitting coil of the coupling mechanism.
In order to verify the feasibility and effectiveness of the system (including the identification model and method thereof), the Simulink simulation verification model shown in fig. 11 is established based on the dual LCC type MC-WPT system and the load and mutual inductance identification model based on the tensrflow. The model mainly comprises the following parts: a main circuit of a double LCC type MC-WPT system; a signal generating module of the inverter circuit; transmitting end LCC topology network inductance Lf1Current value ImAnd a transmission distance h acquisition unit, in simulation, a current value ImThe detection module detects the transmission distance h in real time, and the transmission distance h is automatically input and changed through programming; and identifying a model algorithm unit. The system simulation parameters are consistent with table 2.
In the simulation verification model, the load and mutual inductance identification model arithmetic formula (16) based on the TensorFlow frame is packaged in the identification model arithmetic unit in FIG. 11 by writing M language, the time consumption of the identification algorithm for one-time operation is about 25us, and 10 groups of loads and mutual inductance values are randomly set and are brought into the simulation verification model to obtain the identification result and the relative error, which are shown in Table 3.
TABLE 3 LCC type MC-WPT system load and mutual inductance identification result
Figure GDA0002681688770000131
Figure GDA0002681688770000141
It can be seen from table 3 that the maximum relative error of the load and mutual inductance identification is only 0.5% and 0.34%, and the identification result obtained by the tensrflow load and mutual inductance identification model is very close to the set value.
In summary, the present embodiment provides a model, a method, and a system for identifying load and mutual inductance of a dual LCC type MC-WPT system based on tensrflow, where the model guides a trained model into a microcontroller through an offline training model, and then can identify load and mutual inductance simultaneously on line, and the model has the advantages of fast identification speed, high accuracy, benefit for real-time control of the system, low cost, easy implementation, and benefit for engineering popularization and application.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.

Claims (3)

1. The MC-WPT system load and mutual inductance identification method based on TensorFlow is characterized by comprising the following steps of:
x1., detecting the input current value and the coil transmission distance of the current MC-WPT system;
x2., inputting the current input current value and the coil transmission distance into the MC-WPT system load and mutual inductance identification model, and calculating to obtain a corresponding load value and a corresponding mutual inductance value;
the MC-WPT system load and mutual inductance identification model generation method comprises the following steps:
s1, constructing a full-connection neural network model based on a TensorFlow framework; the fully-connected neural network model comprises an input layer, an output layer and 1 st to N hidden layers which are sequentially and fully connected between the input layer and the output layer, wherein N is more than or equal to 1; the 1 st to N hidden layers are provided with k nodes, and k is more than or equal to 2;
s2, establishing COSMOL and Simulink simulation models of the MC-WPT system to obtain a plurality of groups of input current values and coil transmission distance simulation data of the MC-WPT system, and dividing the simulation data into a training set and a test set;
s3, inputting the training set into the fully-connected neural network model for model training, and continuously optimizing parameters in the fully-connected neural network model according to a training error value; the optimized parameters comprise a 1 st weight matrix and a 1 st bias matrix acting on the 1 st hidden layer, a 2 nd weight matrix and a 2 nd bias matrix acting on the 2 nd hidden layer until an Nth weight matrix and an Nth bias matrix acting on the Nth hidden layer, and an N +1 th weight matrix and an N +1 th bias matrix acting on the output layer; the nonlinear activation function of the 1 st to N hidden layers uses a Sigmoid activation function in a TensorFlow frame;
s4, when the training error rate of the fully-connected neural network model is lower than a preset error rate, finishing training to obtain a trained MC-WPT system load and mutual inductance identification model;
the MC-WPT system load and mutual inductance identification model is calculated according to the following formula:
Figure FDA0003249361950000021
wherein l1Represents an intermediate variable of the 1 st hidden layer, [ h I ]m]Representing a matrix consisting of the inter-coil transmission distances and the input current values of the MC-WPT system,
Figure FDA0003249361950000022
and
Figure FDA0003249361950000023
respectively representing the 1 st weight matrix and the 1 st bias matrix, L1Represents a pair of1Substituting each element into the activation function
Figure FDA0003249361950000024
Carrying out operation to obtain a hidden layer output matrix;
l2an intermediate variable representing the 2 nd hidden layer,
Figure FDA0003249361950000025
and
Figure FDA0003249361950000026
respectively representing said 2 nd weight matrix and said 2 nd bias matrix, L2Represents a pair of2Substituting each element into the activation function
Figure FDA0003249361950000027
Carrying out operation to obtain a hidden layer output matrix;
......；
lNan intermediate variable representing the Nth hidden layer,
Figure FDA0003249361950000028
and
Figure FDA0003249361950000029
respectively representing the Nth weight matrix and the Nth bias matrix, LNRepresents a pair ofNSubstituting each element into the activation function
Figure FDA00032493619500000210
Carrying out operation to obtain a hidden layer output matrix;
lN+1an intermediate variable representing the output layer,
Figure FDA00032493619500000211
and
Figure FDA00032493619500000212
respectively representing the N +1 th weight matrix and the N +1 th bias matrix, M, ReqRespectively representing the mutual inductance value and the load value output by the output layer.
2. MC-WPT system load and mutual inductance identification system based on TensorFlow, its characterized in that: the device comprises a controller, a current detection module and a distance measurement module, wherein the current detection module and the distance measurement module are connected with the controller; the current detection module is used for detecting an input current value of an LCC circuit topology at a transmitting end in the MC-WPT system and sending the input current value to the controller; the distance measuring module is used for detecting the transmission distance between a transmitting coil and a receiving coil in the MC-WPT system and sending the transmission distance to the controller; the controller is used for installing an MC-WPT system load and mutual inductance identification model, and calculating a load value and a mutual inductance value under a corresponding input current value and a transmission distance according to the MC-WPT system load and mutual inductance identification method of claim 1.
3. The TensorFlow-based MC-WPT system load and mutual inductance identification system according to claim 2, wherein: the current detection module is a Hall sensor, and the distance measurement module is an infrared distance measurement sensor.
CN202010895878.1A 2020-08-31 2020-08-31 MC-WPT system load and mutual inductance identification model, method and system based on TensorFlow Active CN111987779B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010895878.1A CN111987779B (en) 2020-08-31 2020-08-31 MC-WPT system load and mutual inductance identification model, method and system based on TensorFlow

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010895878.1A CN111987779B (en) 2020-08-31 2020-08-31 MC-WPT system load and mutual inductance identification model, method and system based on TensorFlow

Publications (2)

Publication Number Publication Date
CN111987779A CN111987779A (en) 2020-11-24
CN111987779B true CN111987779B (en) 2021-10-15

Family

ID=73440584

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010895878.1A Active CN111987779B (en) 2020-08-31 2020-08-31 MC-WPT system load and mutual inductance identification model, method and system based on TensorFlow

Country Status (1)

Country Link
CN (1) CN111987779B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115425768B (en) * 2022-07-21 2023-04-07 广西电网有限责任公司电力科学研究院 PyTorch-based LCC-S type WPT system load and self-inductance identification method and system

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105141046A (en) * 2015-10-26 2015-12-09 重庆大学 SP type ICPT system load and mutual inductance identification method

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9882675B2 (en) * 2013-08-16 2018-01-30 Origin Wireless, Inc. Time-reversal wireless systems having asymmetric architecture
CN108337919B (en) * 2015-06-01 2021-10-29 香港大学 Fast method for identifying coil misalignment/mutual coupling in wireless charging systems
CN107526899B (en) * 2017-09-12 2020-03-10 重庆大学 Load and mutual inductance identification method of magnetic coupling WPT system based on genetic algorithm
CN109004842A (en) * 2018-08-01 2018-12-14 西南交通大学 A kind of radio energy transmission system and method that mutual inductance and load can be recognized from primary side
CN110867973B (en) * 2018-08-08 2020-09-25 哈尔滨工业大学 Static-dynamic magnetic coupling wireless power transmission system online or offline mutual inductance identification method
CN110112841A (en) * 2019-05-28 2019-08-09 中国电力科学研究院有限公司 A kind of load of wireless charging system and mutual inductance parameter identification method and system
CN110350672B (en) * 2019-07-08 2023-06-20 华南理工大学 Half-bridge inversion type magnetic resonance wireless charging system modeling and control method
CN111144549A (en) * 2019-12-23 2020-05-12 深圳供电局有限公司 Microgrid attack identification method based on convolutional neural network and microgrid coordination controller

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105141046A (en) * 2015-10-26 2015-12-09 重庆大学 SP type ICPT system load and mutual inductance identification method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于遗传算法的SS型磁耦合WPT系统负载与互感识别方法;苏玉刚等;《电工技术学报》;20180930;第33卷(第18期);全文 *

Also Published As

Publication number Publication date
CN111987779A (en) 2020-11-24

Similar Documents

Publication Publication Date Title
CN108494073B (en) Metal foreign matter detection method and magnetic coupling resonant wireless charging device
EP3304688B1 (en) Fast method for identifying coil misalignment/mutualcoupling in wireless charging systems
CN102969801B (en) Current type wireless power supply system load identification method
Yin et al. A systematic approach for load monitoring and power control in wireless power transfer systems without any direct output measurement
CN106787233A (en) Cable tunnel inspection robot and charging method with wireless charging device
CN206283301U (en) Cable tunnel inspection robot wireless charging system based on overlapping four transmitting coils
CN108695996A (en) Radio energy transmission system sliding-mode control based on Buck links
CN111987779B (en) MC-WPT system load and mutual inductance identification model, method and system based on TensorFlow
CN107526899A (en) The load of magnetic coupling WPT system and mutual inductance recognition methods based on genetic algorithm
CN106295976A (en) IPT system electromagnetic mechanism method of evaluating performance based on analytic hierarchy process (AHP) and fuzzy overall evaluation
CN107677905A (en) A kind of off-board charging machine detecting device of electric automobile
Tan et al. Study of constant current-constant voltage output wireless charging system based on compound topologies
CN112937319B (en) Multi-frequency resonance compensation-based multi-channel wireless charging system for intelligent mobile restaurant
Moghaddami et al. Sensorless electric vehicle detection in inductive charging stations using self-tuning controllers
CN116054424A (en) Dynamic mutual inductance online evaluation method based on wireless power transmission system
CN108879869A (en) Wireless charging system primary-side-control method and its realization system based on load characteristic
Shen et al. Modeless prediction of variable coupling effect for multiple-load wireless power transfer
Hu et al. Optimal design of electromagnetic coupling mechanism for ICPT system
CN113437810A (en) WPT system impedance matching method based on GA-BP algorithm
Christen et al. Wireless power transfer system for linear drives
CN115425768B (en) PyTorch-based LCC-S type WPT system load and self-inductance identification method and system
Kovacova et al. Laboratory model of an electric vehicle charger based on the principle of energy transfer by air
Kim et al. Predicting Receiver Characteristics without Sensors in an LC–LC Tuned Wireless Power Transfer System Using Machine Learning
Zheng et al. ANALYSIS OF MUTUAL INDUCTANCE BETWEEN TRANSMITTER AND RECEIVER COILS IN WIRELESS POWER TRANSFER SYSTEM OF ELECTRIC VEHICLE.
Huang et al. Novel Coil Detection Method for Dynamic Wireless Power Transfer for Electric Vehicles

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant