CN112583139B - Frequency tracking method of WPT system based on fuzzy RBF neural network - Google Patents

Frequency tracking method of WPT system based on fuzzy RBF neural network Download PDF

Info

Publication number
CN112583139B
CN112583139B CN202011502059.2A CN202011502059A CN112583139B CN 112583139 B CN112583139 B CN 112583139B CN 202011502059 A CN202011502059 A CN 202011502059A CN 112583139 B CN112583139 B CN 112583139B
Authority
CN
China
Prior art keywords
neural network
fuzzy
rbf neural
frequency
resonant
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
CN202011502059.2A
Other languages
Chinese (zh)
Other versions
CN112583139A (en
Inventor
冯宏伟
黄麟
刘媛媛
单正娅
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.)
Hefei Tanzhen Testing Technology Co ltd
Original Assignee
Wuxi Institute of Technology
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 Wuxi Institute of Technology filed Critical Wuxi Institute of Technology
Priority to CN202011502059.2A priority Critical patent/CN112583139B/en
Publication of CN112583139A publication Critical patent/CN112583139A/en
Application granted granted Critical
Publication of CN112583139B publication Critical patent/CN112583139B/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
    • H02J50/00—Circuit arrangements or systems for wireless supply or distribution of electric power
    • H02J50/20—Circuit arrangements or systems for wireless supply or distribution of electric power using microwaves or radio frequency waves
    • 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

Landscapes

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

Abstract

The invention discloses a frequency tracking method of a WPT system based on a fuzzy RBF neural network, which comprises an MCR-WPT system main circuit and a frequency tracking control circuit, wherein the frequency tracking control circuit is used for detecting and collecting the phase difference and the phase difference change rate of resonant voltage and resonant current of a transmitting end in the MCR-WPT system main circuit, and a PID controller utilizes the RBF neural network to adjust the control quantity of the frequency of an H bridge inverter in real time so as to realize the self-adaptive tracking of the resonant working frequency of the transmitting end, so that the resonant voltage and the resonant current of the transmitting end keep the same frequency and the same phase at the moment, namely the phase difference of the resonant voltage and the resonant current of the transmitting end is kept at 0 degree, thereby ensuring that the resonant working frequency of the system keeps consistent with the inherent resonant frequency of the system, and providing a guarantee for realizing higher transmission efficiency of a wireless electric energy transmission system.

Description

Frequency tracking method of WPT system based on fuzzy RBF neural network
Technical Field
The invention relates to the technical field of wireless charging systems, in particular to a frequency tracking method of a WPT system based on a fuzzy RBF neural network.
Background
At present, a magnetic coupling resonance type wireless power transmission (Magnetic Coupled Resonance Wireless Power Transfer, MCR-WPT) system has the advantages of long transmission distance, high transmission efficiency and the like in a near-field WPT technology, and is widely applied to the fields of wireless charging of electric automobiles, health monitoring, embedded devices and the like, so that the magnetic coupling resonance type wireless power transmission system becomes a research hot spot in the wireless charging field. However, environmental temperature, operating conditions, coil size, surface effects, and the like may cause magnetic flux and current variations of the resonant coil. These variations may cause a variation in the actual operating resonant frequency, resulting in a rapid drop in transmission efficiency, so maintaining the MCR-WPT system operating at the resonant frequency is one of the key technologies to improve transmission efficiency.
In order to ensure that the MCR-WPT system works at a resonance frequency point, the control method mainly comprises three aspects of coil topology optimization, dynamic compensation tuning and frequency tracking. Compared with the other two methods, the frequency tracking control is widely applied to the system due to easy implementation and quick response.
However, most of the current automatic frequency tracking methods of the WPT system realize linear adjustment of the working frequency by detecting the phase difference between the resonant voltage and the current. However, MCR-WPT systems often have a steep course at the moment of frequency mismatch or resonance frequency drift, and this nonlinear variation is not particularly model-referenced.
Disclosure of Invention
Aiming at the technical defects, the invention establishes a frequency tracking closed-loop system model based on a fuzzy RBF neural network by analyzing the influence of control parameters on the system performance, performs nonlinear frequency tracking in real time, realizes fast response and high-precision frequency tracking control on the basis of improving the transmission efficiency of an MCR-WPT system, and proves that the frequency control algorithm enhances the self-adaptive tracking capacity of the working resonant frequency of the WPT system, greatly improves the transmission efficiency of the system and provides an important reference for the efficiency optimization design of the WPT system.
In order to solve the technical problems, the invention adopts the following technical scheme:
the invention provides a WPT system based on a fuzzy RBF neural network, which comprises an MCR-WPT system main circuit and a frequency tracking control circuit, wherein the frequency tracking control circuit comprises a voltage and current acquisition circuit, a zero crossing detection circuit, a digital phase discriminator, a fuzzy controller, a PWM generator, a PID controller and an H bridge inversion driving circuit;
the voltage and current acquisition circuit is used for completing resonance voltage of the transmitting end of the main circuit of the MCR-WPT system
Figure SMS_1
And resonance current->
Figure SMS_2
Is detected;
the zero-crossing detection circuit detects the voltage of the main circuit of the MCR-WPT system
Figure SMS_3
And resonance current->
Figure SMS_4
Square wave signal converted into same frequency and same phase>
Figure SMS_5
and />
Figure SMS_6
；
Digital phase detector outputs a signal
Figure SMS_7
and />
Figure SMS_8
Phase comparison is carried out, resulting in a phase difference +.>
Figure SMS_9
Rate of change of phase difference
Figure SMS_10
；
The fuzzy controller adjusts parameters of the PID controller based on the fuzzy RBF neural network, and when the PID controller works, the fuzzy RBF neural network adjusts parameters of the PID controller according to the parameters of the PID controller
Figure SMS_11
and />
Figure SMS_12
The PID parameters are adjusted in real time, and the phase difference is maintained at 0 degrees;
the PWM generator adjusts the frequency according to the PID parameters to generate and output the PID parameters
Figure SMS_13
The same-frequency and same-phase H-bridge driving logic signals pass through the H-bridge inversion driving circuit to realize the switching control of MOSFET tubes in the H-bridge inversion driving circuit.
The frequency tracking method of the WPT system based on the fuzzy RBF neural network comprises the following steps:
s1: constructing a wireless power transmission system model based on an RBF neural network, forming a resonant network, and carrying out real-time self-adaptive tracking on the resonant working frequency of a transmitting end through a fuzzy RBF neural network;
s2: initializing a fuzzy RBF neural network, randomly acquiring a sample in a training set, and selecting the center of a membership function in a fuzzification layer of the RBF neural network
Figure SMS_14
Width->
Figure SMS_15
And calculate the rule applicability in the modeling layer +.>
Figure SMS_16
And fuzzy membership->
Figure SMS_17
An initial value;
wherein the sample is
Figure SMS_18
，/>
Figure SMS_19
Represents a transpose of X;
wherein the blurring layer comprises h classes, forming h nodes, each node corresponding to a blurring rule,
Figure SMS_20
；
Figure SMS_21
representing the%>
Figure SMS_22
Features are about>
Figure SMS_23
The%>
Figure SMS_24
Fuzzy membership of the fuzzy subsets; wherein,
Figure SMS_25
s3: obtaining the resonance current value of the system transmitting end through sampling
Figure SMS_26
And the resonance voltage value of the transmitting terminal +.>
Figure SMS_27
，
Figure SMS_28
Indicate->
Figure SMS_29
Acquiring discrete values for times, calculating the phase difference of the system>
Figure SMS_30
And rate of change of phase difference
Figure SMS_31
；
S4: determining performance index functions of fuzzy RBF neural networks
Figure SMS_32
；
S5: fuzzy rule fitness in samples
Figure SMS_33
If +.>
Figure SMS_34
Adding a corresponding +.f. to the fuzzification layer of RBF neural network>
Figure SMS_35
Node of the fuzzy rule, the characteristic component of the corresponding dimension of the sample is taken as the center of the membership function +.>
Figure SMS_36
Width->
Figure SMS_37
The weight values in the corresponding fuzzy rules are all initialized to 0 for the preset positive number;
if not, entering the next step;
s6: if there is a rule
Figure SMS_38
And rule->
Figure SMS_39
If the formula (20) is satisfied, merging the two fuzzy rules, otherwise, entering the next parameter learning;
Figure SMS_40
（20）；
wherein, the middle part
Figure SMS_41
Is a preset value;
s7: the RBF neural network performs parameter learning and calculates the center of membership function in the RBF neural network
Figure SMS_42
Width->
Figure SMS_43
And network weight coefficient->
Figure SMS_44
；
S8: according to the formula (15), three parameters of the PID controller output by the RBF neural network output layer are calculated
Figure SMS_45
，
Figure SMS_46
and />
Figure SMS_47
；/>
Figure SMS_48
；/>
Figure SMS_49
；/>
Figure SMS_50
；（15）
in the formula ：
Figure SMS_51
input for RBF neural network output layer, < ->
Figure SMS_52
Is->
Figure SMS_53
A connection weight matrix with the output layer,
Figure SMS_54
respectively corresponding to 3 parameters->
Figure SMS_55
，/>
Figure SMS_56
and />
Figure SMS_57
；
The fuzzy RBF neural network PID controller calculates the frequency modulation signal increment
Figure SMS_58
The method comprises the following steps:
Figure SMS_59
（22）；
s9: according to S8, a control signal capable of keeping the resonance working frequency in the resonance network consistent with the natural resonance frequency of the system is obtained
Figure SMS_60
, wherein ：
Figure SMS_61
（23）；
will be
Figure SMS_62
Added to a resonant network formed by modeling a system to ensure a master circuit of an MCR-WPT systemThe H-bridge inverter in (2) is operated at resonance point, phase difference is +.>
Figure SMS_63
Maintain zero;
s10: order the
Figure SMS_64
Then return to step S2 for the next cycle +.>
Figure SMS_65
Detection and frequency control signal->
Figure SMS_66
And adjusting to form real-time detection and adjustment.
Preferably, in step S2, the blurring layers are discretized into [ -5, respectively, using a k-means clustering algorithm]Between (a) and (b)
Figure SMS_67
Class, this->
Figure SMS_68
Class formation->
Figure SMS_69
Each node has two gaussian membership functions.
Preferably, in step S6, the merging of the two rules is performed according to the following formula:
Figure SMS_70
（21）
wherein ,
Figure SMS_71
default value representing the network weight coefficient, +.>
Figure SMS_72
。
Preferably, in step S7, the membership function parameters and the network weight coefficients in the RBF neural network are learned according to the following formula:
Figure SMS_73
（17）；
Figure SMS_74
（18）；
Figure SMS_75
（19）；
wherein ,
Figure SMS_76
the iteration step number of the network; />
Figure SMS_77
For learning rate +.>
Figure SMS_78
Is to learn momentum factors, and->
Figure SMS_79
，
Figure SMS_80
。/>
Preferably, in step S8, the input of the output layer
Figure SMS_81
Is->
Figure SMS_82
And normalized fitness->
Figure SMS_83
Is a linear combination of (a):
Figure SMS_84
（14）
wherein ,
Figure SMS_85
normalized layer of RBF neural network, normalized +.>
Figure SMS_86
The connection weight of the input of the fuzzification layer and the output layer of the fuzzification layer serving as the RBF neural network.
Preferably, the RBF neural network is normalized by means of barycentric defuzzification:
Figure SMS_87
（12）；
Figure SMS_88
（8）
wherein ,
Figure SMS_89
。
the invention has the beneficial effects that:
according to the invention, a frequency tracking closed-loop system model based on a fuzzy RBF neural network is established by analyzing the influence of control parameters on the system performance, a fuzzy self-adaptive controller for frequency tracking control is designed, real-time nonlinear frequency tracking is realized, quick response and high-precision frequency tracking control are realized on the basis of improving the transmission efficiency of an MCR-WPT system, the self-adaptive tracking capacity of the working resonant frequency of the MCR-WPT system is enhanced, the resonant voltage and the resonant current of a transmitting end are kept in the same frequency and the same phase at the moment, namely, the phase difference of the resonant voltage and the resonant current is kept at 0 degree, so that the resonant working frequency of the system is kept consistent with the inherent resonant frequency of the system, a guarantee is provided for realizing higher transmission efficiency of a wireless electric energy transmission system, and an important reference is provided for the optimal design of the efficiency of the MCR-WPT system.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a system block diagram of the present invention;
FIG. 2 is a diagram of a main circuit of a string topology of an MCR-WPT system;
fig. 3 is a frequency tracking block diagram of a fuzzy RBF neural network.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Examples
As shown in fig. 1 and fig. 2, the present invention provides a WPT system based on a fuzzy RBF neural network, the system including an MCR-WPT system main circuit and a frequency tracking control circuit, in particular:
with reference to fig. 2, the main circuit of the mcr-WPT system adopts a typical double-coil string topology main circuit structure, and mainly comprises an H-bridge inverter, a transmitting end series resonant circuit, a receiving end series resonant circuit and a full-bridge rectifying circuit; wherein,
Figure SMS_98
is a direct current power supply->
Figure SMS_91
For power supply filter capacitor, field effect transistor +.>
Figure SMS_94
Constitute H bridge inverter, ">
Figure SMS_93
For transmitting coil inductance, < >>
Figure SMS_96
For receiving coil inductance>
Figure SMS_97
、/>
Figure SMS_101
For the resonance compensation capacitance corresponding to the transmitting end and the receiving end, M is the mutual inductance between the transmitting coil and the receiving coil, and the magnetic coupling tightness of the two coils is determined by the coupling coefficient +.>
Figure SMS_99
Indicating (I)>
Figure SMS_103
、/>
Figure SMS_90
High-frequency resonant currents of the transmitting side and the receiving side, respectively,>
Figure SMS_95
、/>
Figure SMS_100
rectifier diode for parasitic resistance of transmitting circuit and receiving circuit>
Figure SMS_104
To form a full bridge rectifier, ">
Figure SMS_102
The filter capacitor of rectifier bridge is used to charge and discharge the output voltage>
Figure SMS_105
Tends to be smooth; />
Figure SMS_92
Is the load side equivalent resistance;
referring to fig. 1, the frequency tracking control circuit includes a voltage and current acquisition circuit, a zero crossing detection circuit, a digital phase discriminator, a fuzzy controller, a PWM generator, a PID controller and an H-bridge inverter driving circuit;
the voltage and current acquisition circuit is used for completing resonance voltage of the transmitting end of the main circuit of the MCR-WPT system
Figure SMS_106
And resonance current->
Figure SMS_113
Is detected; the zero-crossing detection circuit is used for detecting the resonance voltage of the main circuit of the MCR-WPT system>
Figure SMS_116
And resonance current->
Figure SMS_109
Square wave signal converted into same frequency and same phase>
Figure SMS_111
and />
Figure SMS_115
The method comprises the steps of carrying out a first treatment on the surface of the Digital phase detector signals->
Figure SMS_117
and />
Figure SMS_107
Phase comparison is carried out, resulting in a phase difference +.>
Figure SMS_110
Rate of change in phase difference->
Figure SMS_114
The method comprises the steps of carrying out a first treatment on the surface of the The fuzzy controller adjusts parameters of the PID controller based on the fuzzy RBF neural network, and when the PID controller works, the fuzzy RBF neural network is controlled according to +.>
Figure SMS_118
and />
Figure SMS_108
The PID parameters are adjusted in real time, and the phase difference is maintained at 0 degrees; PWM generator frequency adjustment based on PID parametersGenerate and->
Figure SMS_112
The same-frequency and same-phase H-bridge driving logic signals realize the switch control of MOSFET tubes in the H-bridge inversion driving circuit through the H-bridge inversion driving circuit, so that the working frequency of the H-bridge inverter is at a resonance point.
Further, in connection with fig. 2, to facilitate analysis of frequency mismatch characteristics, the SS topology of fig. 2 is analyzed, wherein
Figure SMS_119
，/>
Figure SMS_120
,/>
Figure SMS_121
，
The equivalent model column loop equation is obtained according to kirchhoff's voltage law:
Figure SMS_122
wherein
Figure SMS_123
and />
Figure SMS_124
For the equivalent impedance of the transmitting end and the receiving end, the two satisfy:
Figure SMS_125
for ease of analysis, the transmitter coil and receiver coil are chosen to be identical in structure, i.e
Figure SMS_126
,/>
Figure SMS_127
And->
Figure SMS_128
；
Based on formulas (1) and (2), the current values on both sides can be calculated as:
Figure SMS_129
wherein
Figure SMS_130
Is the inverter angular frequency, input power of MCR-WPT system +.>
Figure SMS_131
And output power->
Figure SMS_132
The method can be calculated as follows:
Figure SMS_133
Figure SMS_134
for inputting the power supply voltage->
Figure SMS_135
Is effective.
According to electromagnetic resonance conditions, the defined detuning rate is:
Figure SMS_136
when (when)
Figure SMS_137
When (I)>
Figure SMS_138
，/>
Figure SMS_139
For the resonance angular frequency, the resonance network is in a resonance state, and the coil loop is pureResistance; when->
Figure SMS_140
When (I)>
Figure SMS_141
The resonant network is in an overresonance state, and the loop is inductive; when->
Figure SMS_142
In the time-course of which the first and second contact surfaces,
Figure SMS_143
the resonant network is in an under-resonant state, and the coil loop is capacitive;
from (4) and (5), the transmission efficiency can be calculated
Figure SMS_144
Figure SMS_145
As can be seen from the formula (6), when the resonant network is in a resonant state, the equivalent impedance of the coil loop is minimum, and the energy in the coil can realize the transmission with the highest transmission efficiency; in a non-resonance state, the larger the detuning rate is, the transmission efficiency of the system is greatly reduced; because the MCR-WPT system adopts a series resonance structure, the current of the transmitting end is a sine signal, the voltage is a square wave signal, and the self-adaptive frequency tracking is carried out on the resonance current of the transmitting end in real time by adopting a fuzzy control method.
Specific:
for the WPT system, the structure and parameters in the model can be changed at any time due to the influence of factors such as transmission distance, coil offset, load change and the like, and if the control is performed by adopting a conventional PID strategy with unchanged parameters, the ideal frequency tracking effect is difficult to realize; the structure of the fusion fuzzy RBF neural network to be adopted in the invention is shown in figure 3 on the basis of the conventional PID frequency tracking control;
the fuzzy RBF neural network comprises an input layer, a fuzzification layer (i.e. hidden layer), a normalization layer and an output layer;
input layer:
the node number of the input layer of the fuzzy RBF neural network is 2, and the node number is respectively the frequency phase difference
Figure SMS_146
And rate of change of phase difference->
Figure SMS_147
I.e. input sample +.>
Figure SMS_148
, wherein />
Figure SMS_149
，/>
Figure SMS_150
Representing the transpose of X (matrix transpose of index, one row to one column).
Hidden layer:
the hidden layers are also called fuzzification layers, which are discretized into [ -5,5 ] respectively using a k-means clustering algorithm]Between (a) and (b)
Figure SMS_151
Class, this->
Figure SMS_152
Class formation->
Figure SMS_153
Each node corresponds to a fuzzy rule, and each node has two Gaussian membership functions; this is->
Figure SMS_154
Class (/ -)>
Figure SMS_155
) Cluster center of->
Figure SMS_156
Initial center parameters as each Gaussian membership function of the hidden layer;
Figure SMS_157
（7）
in the formula
Figure SMS_158
Representing the%>
Figure SMS_159
Features are about>
Figure SMS_160
The%>
Figure SMS_161
Fuzzy membership of the fuzzy subsets, +.>
Figure SMS_162
、
Figure SMS_163
Respectively representing the center and the width of the Gaussian membership function;
hidden layer (L)
Figure SMS_164
The output value of the individual node, i.e. +.>
Figure SMS_165
Applicability of the bar rule->
Figure SMS_166
The fuzzy rule fitness calculation method based on the Markov distance to replace the traditional Euclidean distance is adopted, namely:
Figure SMS_167
（8）
the fitting effect of the model is better by adopting the method of self-adaptive modification of the width parameters of different membership functions, and the formula (8) can be expressed as:
Figure SMS_168
（9）
Figure SMS_169
（10）
of the formula (I)
Figure SMS_170
Representing input samples and hidden layer +.>
Figure SMS_171
The mahalanobis distance of the individual nodes,
Figure SMS_172
expressed as:
Figure SMS_173
（11）
in the formula (11)
Figure SMS_174
Represents->
Figure SMS_175
The>
Figure SMS_176
Membership function width for each fuzzy subset.
Normalization layer:
although the traditional standard RBF fuzzy neural network has good performance in training, the generalization capability in the test is not high, and the RBF fuzzy neural network after normalization can effectively improve the generalization capability of the model. The invention adopts a gravity center method defuzzification mode to normalize the network after fusing the T-S fuzzy model;
Figure SMS_177
（12）
normalized to
Figure SMS_178
The connection weight of the input of the fuzzification layer and the output layer of the fuzzification layer serving as the RBF neural network.
Output layer:
the output layer comprises
Figure SMS_179
Strip blurring rule and +.>
Figure SMS_180
The hidden layer nodes are in one-to-one correspondence and are +.>
Figure SMS_181
Output of bar fuzzy rule generation>
Figure SMS_182
The method is calculated by reasoning the following rules:
if
Figure SMS_183
then/>
Figure SMS_184
(13)
wherein ,
Figure SMS_185
represents->
Figure SMS_186
The +.>
Figure SMS_187
Fuzzy subset->
Figure SMS_188
Is real number, < >>
Figure SMS_189
；
Input of output layer
Figure SMS_190
Is->
Figure SMS_191
And normalized fitness->
Figure SMS_192
Is a linear combination of (a):
Figure SMS_193
（14）
the output layer mainly outputs 3 parameters of the PID controller
Figure SMS_194
，/>
Figure SMS_195
and />
Figure SMS_196
The activation function is selected as:
Figure SMS_197
；/>
Figure SMS_198
；/>
Figure SMS_199
（15）
in the formula ：
Figure SMS_200
is->
Figure SMS_201
Connection weight matrix with output layer, +.>
Figure SMS_202
Respectively corresponding to 3 parameters->
Figure SMS_203
，/>
Figure SMS_204
and />
Figure SMS_205
。
Referring to fig. 1 to 3, the frequency tracking method for the system is specifically as follows:
s1: constructing a wireless power transmission system model based on an RBF neural network to form a resonant network, and carrying out real-time self-adaptive tracking on the resonant working frequency of a transmitting end through a fuzzy RBF neural network so as to ensure that the resonant current and the resonant voltage of the transmitting end of the system keep the same frequency and the same direction at the moment;
s2: initializing a fuzzy RBF neural network, randomly acquiring a sample in a training set, and selecting the center of a membership function in a fuzzification layer of the RBF neural network
Figure SMS_206
Width->
Figure SMS_207
And calculate the rule applicability in the modeling layer +.>
Figure SMS_208
And fuzzy membership->
Figure SMS_209
An initial value;
s3: obtaining the resonance current value of the system transmitting end through sampling
Figure SMS_210
And the resonance voltage value of the transmitting terminal +.>
Figure SMS_211
，
Figure SMS_212
Indicate->
Figure SMS_213
Acquiring discrete values for times, calculating the phase difference of the system>
Figure SMS_214
And rate of change of phase difference
Figure SMS_215
, wherein ,/>
Figure SMS_216
；
S4: in order to find the optimal weight, the output of the fuzzy RBF neural network is closest to the expected output value, namely the minimum error is reached, and a cost function is defined as
Figure SMS_217
Namely, determining a performance index function of the fuzzy RBF neural network;
s5: fuzzy rule fitness in samples
Figure SMS_218
If +.>
Figure SMS_219
Adding a corresponding +.f. to the fuzzification layer of RBF neural network>
Figure SMS_220
Node of the fuzzy rule, the characteristic component of the corresponding dimension of the sample is taken as the center of the membership function +.>
Figure SMS_221
Width->
Figure SMS_222
The weight values in the corresponding fuzzy rules are all initialized to 0 for the preset positive number;
wherein the function is
Figure SMS_223
Indicating whether the number in the given point set reaches a given maximum value, if not, entering the next step;
s6: if there is a rule
Figure SMS_224
Rule for regulating sumThen->
Figure SMS_225
If the formula (20) is satisfied, merging the two fuzzy rules, otherwise, entering the next parameter learning; />
Figure SMS_226
（20）；
Wherein, the middle part
Figure SMS_227
Is a preset value;
if the two rules are combined, the method is carried out according to the following formula:
Figure SMS_228
（21）
wherein ,
Figure SMS_229
default value representing the network weight coefficient, +.>
Figure SMS_230
；
S7: the RBF neural network performs parameter learning and calculates the center of membership function in the RBF neural network
Figure SMS_231
Width->
Figure SMS_232
And network weight coefficient->
Figure SMS_233
And learns according to the following formula:
Figure SMS_234
（17）；
Figure SMS_235
（18）；
Figure SMS_236
（19）；
wherein ,
Figure SMS_237
the iteration step number of the network; />
Figure SMS_238
For learning rate +.>
Figure SMS_239
Is to learn momentum factors, and->
Figure SMS_240
，
Figure SMS_241
。
S8: according to the formula (15), three parameters of the PID controller output by the RBF neural network output layer are calculated
Figure SMS_242
，
Figure SMS_243
and />
Figure SMS_244
；
The fuzzy RBF neural network PID controller calculates the frequency modulation signal increment
Figure SMS_245
The method comprises the following steps:
Figure SMS_246
（22）；
s9: according to S8, a control signal capable of keeping the resonance working frequency in the resonance network consistent with the natural resonance frequency of the system is obtained
Figure SMS_247
, wherein ：
Figure SMS_248
（23）；
will be
Figure SMS_249
Adding a PWM generator to drive the H-bridge inverter to ensure that the H-bridge inverter in the main circuit of the MCR-WPT system works at a resonance point and the phase difference is +.>
Figure SMS_250
Maintain zero;
s10: order the
Figure SMS_251
Then return to step S2 for the next cycle +.>
Figure SMS_252
Detection and frequency control signal->
Figure SMS_253
And adjusting to form real-time detection and adjustment. />
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims (6)

1. The frequency tracking method of the WPT system based on the fuzzy RBF neural network is characterized in that the system comprises an MCR-WPT system main circuit and a frequency tracking control circuit, wherein the frequency tracking control circuit comprises a voltage and current acquisition circuit, a zero crossing detection circuit, a digital phase discriminator, a fuzzy controller, a PWM generator, a PID controller and an H bridge inversion driving circuit;
wherein, the voltage and current acquisition circuit completes the transmitting end of the main circuit of the MCR-WPT systemResonant voltage of (2)
Figure QLYQS_1
And resonant current
Figure QLYQS_2
Is detected;
the zero-crossing detection circuit detects the resonant voltage of the transmitting end of the main circuit of the MCR-WPT system
Figure QLYQS_3
And resonance current->
Figure QLYQS_4
Square wave signal converted into same frequency and same phase>
Figure QLYQS_5
and />
Figure QLYQS_6
；
Digital phase detector outputs a signal
Figure QLYQS_7
and />
Figure QLYQS_8
Phase comparison is performed to generate a phase difference +.>
Figure QLYQS_9
And rate of change of phase difference->
Figure QLYQS_10
；
The fuzzy controller adjusts parameters of the PID controller based on the fuzzy RBF neural network, and when the PID controller works, the fuzzy RBF neural network adjusts parameters of the PID controller according to the parameters of the PID controller
Figure QLYQS_11
and />
Figure QLYQS_12
The PID parameters are adjusted in real time, and the phase difference is maintained at 0 degrees;
the PWM generator adjusts the frequency according to the PID parameters to generate and output the PID parameters
Figure QLYQS_13
The same-frequency and same-phase H-bridge driving logic signals realize the switch control of MOSFET tubes in an H-bridge inverter in a main circuit of an MCR-WPT system through an H-bridge inverter driving circuit;
the method comprises the following steps:
s1: constructing a wireless power transmission system model based on an RBF neural network, forming a resonant network, and carrying out real-time self-adaptive tracking on the resonant working frequency of a transmitting end through a fuzzy RBF neural network;
s2: initializing a fuzzy RBF neural network, randomly acquiring a sample in a training set, and selecting the center of a membership function in a fuzzification layer of the RBF neural network
Figure QLYQS_14
Width->
Figure QLYQS_15
And calculate the rule applicability in the modeling layer +.>
Figure QLYQS_16
And fuzzy membership->
Figure QLYQS_17
An initial value;
wherein the sample is
Figure QLYQS_18
，/>
Figure QLYQS_19
Represents a transpose of X;
wherein the blurring layer comprises h classes, forming h nodes, each node corresponding to a blurring rule,
Figure QLYQS_20
；/>
Figure QLYQS_21
representing the%>
Figure QLYQS_22
Features are about>
Figure QLYQS_23
The%>
Figure QLYQS_24
Fuzzy membership of the fuzzy subsets; wherein (1)>
Figure QLYQS_25
；
S3: obtaining the resonance current value of the system transmitting end through sampling
Figure QLYQS_26
And the resonance voltage value of the transmitting terminal +.>
Figure QLYQS_27
Calculating the phase difference ∈>
Figure QLYQS_28
And rate of change of phase difference->
Figure QLYQS_29
；
S4: determining performance index functions of fuzzy RBF neural networks
Figure QLYQS_30
；
S5: fuzzy rule fitness in samples
Figure QLYQS_31
Compared to a preset threshold delta,if->
Figure QLYQS_32
Adding a corresponding +.f. to the fuzzification layer of RBF neural network>
Figure QLYQS_33
Node of the fuzzy rule, the characteristic component of the corresponding dimension of the sample is taken as the center of the membership function +.>
Figure QLYQS_34
Width->
Figure QLYQS_35
The weight values in the corresponding fuzzy rules are all initialized to 0 for the preset positive number;
if not, entering the next step;
s6: if there is a rule
Figure QLYQS_36
And rule->
Figure QLYQS_37
If the formula (20) is satisfied, merging the two fuzzy rules, otherwise, entering the next parameter learning;
Figure QLYQS_38
（20）；/>
wherein, the middle part
Figure QLYQS_39
Is a preset value;
s7: the RBF neural network performs parameter learning and calculates the center of membership function in the RBF neural network
Figure QLYQS_40
Width->
Figure QLYQS_41
And network weight coefficient->
Figure QLYQS_42
；
S8: according to the formula (15), three parameters of the PID controller output by the RBF neural network output layer are calculated
Figure QLYQS_43
，/>
Figure QLYQS_44
And
Figure QLYQS_45
；
Figure QLYQS_46
；/>
Figure QLYQS_47
；/>
Figure QLYQS_48
；（15）
in the formula ：
Figure QLYQS_49
input for RBF neural network output layer, < ->
Figure QLYQS_50
Is->
Figure QLYQS_51
A connection weight matrix with the output layer,
Figure QLYQS_52
respectively corresponding to 3 parameters->
Figure QLYQS_53
，/>
Figure QLYQS_54
and />
Figure QLYQS_55
；
The fuzzy RBF neural network PID controller calculates the frequency modulation signal increment
Figure QLYQS_56
The method comprises the following steps:
Figure QLYQS_57
（22）；
s9: according to S8, a control signal capable of keeping the resonance working frequency in the resonance network consistent with the natural resonance frequency of the system is obtained
Figure QLYQS_58
, wherein ：
Figure QLYQS_59
（23）；
will be
Figure QLYQS_60
Adding the phase difference into a resonance network formed by constructing a system model to ensure that an H-bridge inverter in a main circuit of the MCR-WPT system works at a resonance point and the phase difference is +.>
Figure QLYQS_61
Maintain zero;
s10: order the
Figure QLYQS_62
Then return to step S2 for the next cycle +.>
Figure QLYQS_63
Detection and frequency control signal->
Figure QLYQS_64
And adjusting to form real-time detection and adjustment.
2. The frequency tracking method of WPT system based on fuzzy RBF neural network as claimed in claim 1, wherein in step S2, the fuzzy layers are discretized into [ -5, respectively, using k-means clustering algorithm]Between (a) and (b)
Figure QLYQS_65
Class, this->
Figure QLYQS_66
Class formation->
Figure QLYQS_67
Each node has two gaussian membership functions.
3. The frequency tracking method of WPT system based on fuzzy RBF neural network as claimed in claim 1, wherein in step S6, the combining of two rules is performed according to the following formula:
Figure QLYQS_68
（21）
wherein ,
Figure QLYQS_69
。
4. the frequency tracking method of WPT system based on fuzzy RBF neural network as claimed in claim 1, wherein membership function parameters and network weight coefficients in the RBF neural network are learned as follows in step S7:
Figure QLYQS_70
（17）；
Figure QLYQS_71
（18）；
Figure QLYQS_72
（19）；
wherein ,
Figure QLYQS_73
the iteration step number of the network; />
Figure QLYQS_74
For learning rate +.>
Figure QLYQS_75
Is to learn momentum factors, and->
Figure QLYQS_76
，/>
Figure QLYQS_77
。
5. The frequency tracking method of WPT system based on fuzzy RBF neural network as claimed in claim 1, wherein in step S8, the input of the output layer
Figure QLYQS_78
Is->
Figure QLYQS_79
And normalized fitness->
Figure QLYQS_80
Is a linear combination of (a):
Figure QLYQS_81
（14）
wherein ,
Figure QLYQS_82
normalized layer of RBF neural network, normalized +.>
Figure QLYQS_83
The connection weight of the input of the fuzzification layer and the output layer of the fuzzification layer serving as the RBF neural network.
6. The frequency tracking method of WPT system based on fuzzy RBF neural network as claimed in claim 5, wherein the RBF neural network is normalized by means of barycenter defuzzification:
Figure QLYQS_84
（12）；
Figure QLYQS_85
（8）
wherein ,
Figure QLYQS_86
。/>
CN202011502059.2A 2020-12-17 2020-12-17 Frequency tracking method of WPT system based on fuzzy RBF neural network Active CN112583139B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011502059.2A CN112583139B (en) 2020-12-17 2020-12-17 Frequency tracking method of WPT system based on fuzzy RBF neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011502059.2A CN112583139B (en) 2020-12-17 2020-12-17 Frequency tracking method of WPT system based on fuzzy RBF neural network

Publications (2)

Publication Number Publication Date
CN112583139A CN112583139A (en) 2021-03-30
CN112583139B true CN112583139B (en) 2023-06-09

Family

ID=75136379

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011502059.2A Active CN112583139B (en) 2020-12-17 2020-12-17 Frequency tracking method of WPT system based on fuzzy RBF neural network

Country Status (1)

Country Link
CN (1) CN112583139B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113555873B (en) * 2021-07-29 2022-03-18 山西大学 Load frequency control method of multi-region interconnected power system under denial of service attack

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1731779A (en) * 2005-08-05 2006-02-08 西安电子科技大学 Wavelet packet multi-carrier spread-spectrum system and control method based on neural net equalizer
CN106891763A (en) * 2017-02-27 2017-06-27 中国石油大学(华东) Wireless charging system for electric automobile frequency tracking method
CN108885229A (en) * 2016-04-06 2018-11-23 田建龙 The dynamic detection and compensation technique of wireless power transmission and related system resonance frequency
CN112039520A (en) * 2020-09-08 2020-12-04 西安建筑科技大学 Digital phase-locked loop based on fuzzy RBF self-adaptive control

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
ES2549652B1 (en) * 2014-04-24 2016-10-07 Alstom Transporte, S.A. Method and system to automatically detect faults in a rotary axis

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1731779A (en) * 2005-08-05 2006-02-08 西安电子科技大学 Wavelet packet multi-carrier spread-spectrum system and control method based on neural net equalizer
CN108885229A (en) * 2016-04-06 2018-11-23 田建龙 The dynamic detection and compensation technique of wireless power transmission and related system resonance frequency
CN106891763A (en) * 2017-02-27 2017-06-27 中国石油大学(华东) Wireless charging system for electric automobile frequency tracking method
CN112039520A (en) * 2020-09-08 2020-12-04 西安建筑科技大学 Digital phase-locked loop based on fuzzy RBF self-adaptive control

Also Published As

Publication number Publication date
CN112583139A (en) 2021-03-30

Similar Documents

Publication Publication Date Title
US11223349B2 (en) Joint control method with variable ZVS angles for dynamic efficiency optimization in wireless power charging for electric vehicles
CN111555612B (en) Magnetic coupling resonant wireless energy transfer maximum efficiency tracking method based on constant output voltage
CN111106676A (en) Magnetic coupling mechanism parameter multi-objective optimization method of LCC-S type MC-WPT system
CN112583139B (en) Frequency tracking method of WPT system based on fuzzy RBF neural network
Hsu et al. Fuzzy logic-based directional full-range tuning control of wireless power pickups
CN109921504A (en) Vehicle-mounted mixed energy storage system and its non linear robust adaptive power control method
CN109149923B (en) APFC control system for estimating control rate on line
CN108649804B (en) Direct current transformer system based on magnetic coupling wireless power transmission
Zhang et al. Optimal triple-phase-shift controller design of isolated bidirectional DC-DC converter based on ant colony algorithm and BP neural network
CN111082540B (en) Multi-tap wireless power transmission system and efficiency optimization method thereof
Chow et al. Use of transmitter-side electrical information to estimate system parameters of wireless inductive links
Tang et al. Deep reinforcement learning aided variable-frequency triple-phase-shift control for dual-active-bridge converter
Ouyang et al. Wireless battery charging control for electric vehicles: a user‐involved approach
CN110429835A (en) A kind of RBFNN segmentation on-line optimization Passive Shape Control system and method based on LCL filtering
Zhou et al. Multi-objective optimization for 10-kW rated power dynamic wireless charging systems of electric vehicles
Hou et al. Modified fuzzy neural network control using sliding mode technique for power quality improvement system with experimental verification
CN113193663B (en) Load and mutual inductance dual-parameter identification method for magnetic coupling wireless power transmission system
Hang et al. Switched reluctance motors direct torque control research based on rbf neural network and fuzzy adaptive PID controller
CN111308247A (en) WPT system harmonic judgment method based on unilateral detection current phase change rate detection
NL2022589B1 (en) Continuous control of a contactless electrical energy transfer system
Hsu et al. Fuzzy based directional tuning controller for a wireless power pick-up
CN112671115A (en) WPT (wavelet packet transform) system based on fuzzy self-adaptive PI (proportional-integral) and frequency tracking method
Liu et al. An Integrated Optimization Design Method of Single-Phase PV Inverter Based on Machine Learning
CN115133666A (en) Bilateral capacitor array WPT system and adaptive critical coupling coefficient adjusting method
Cao et al. A study on modeling of wireless charging system based on fuzzy logic control

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
TR01 Transfer of patent right

Effective date of registration: 20231020

Address after: Room 505, Private Technology Entrepreneurship Service Center, No. 605, Mount Huangshan Road, High tech Zone, Hefei City, Anhui Province, 230000

Patentee after: Hefei Tanzhen Testing Technology Co.,Ltd.

Address before: 214121 No. 1600 West high wave road, Jiangsu, Wuxi

Patentee before: WUXI INSTITUTE OF TECHNOLOGY

TR01 Transfer of patent right