CN109726817A - The WPT system impedance matching methods of genetic algorithm optimization BP neural network - Google Patents
The WPT system impedance matching methods of genetic algorithm optimization BP neural network Download PDFInfo
- Publication number
- CN109726817A CN109726817A CN201811574311.3A CN201811574311A CN109726817A CN 109726817 A CN109726817 A CN 109726817A CN 201811574311 A CN201811574311 A CN 201811574311A CN 109726817 A CN109726817 A CN 109726817A
- Authority
- CN
- China
- Prior art keywords
- neural network
- genetic algorithm
- impedance matching
- wpt system
- layer
- 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.)
- Pending
Links
Landscapes
- Measurement Of Resistance Or Impedance (AREA)
Abstract
The invention discloses a kind of WPT system impedance matching methods of genetic algorithm optimization BP neural network, it is optimized using weight and threshold value of the genetic algorithm to neural network, using after optimization weight and threshold value as the initial weight of the local search of neural network and threshold value, then input signal of the input impedance arrived according to system detection as BP neural network, Optimum Matching capacitance is searched out in search space using the local search ability of neural network, and the impedance matching that variable capacitance realizes WPT system is adjusted by control motor.The beneficial effects of the invention are as follows, easily it is limited to local minimum for existing for neural network algorithm itself, the disadvantages of convergence rate is slow, genetic algorithm is introduced to optimize BP neural network weight and threshold value, the convergence rate for accelerating BP neural network improves the Accuracy and high efficiency of WPT system impedance matching.
Description
Technical field
The invention belongs to wireless power transmission field, especially a kind of radio energy of genetic algorithm optimization BP neural network
Transfer impedance match control method
Background technique
After second industrial revolution, human society just enters Electrification Age, and the transmission of electric energy is mainly led by metal
The point-to-point direct contact of line is transmitted.This " wired " transmission mode brings many problems, for example, due to existing
Friction, aging etc. influence, electric energy be easy in transmission process generate spark, and then influence electrical equipment service life and
Electrical Safety.Increasing overhead transmission line sets up difficulty, maintenance cost, dangerous system while causing city space crowded
Number also gradually rises.In addition, traditional corded power transmission mode is not able to satisfy the needs of some particular applications, such as mine
Deng.In view of many drawbacks of traditional electric energy transmission mode, the wireless transmission of electric energy is pursued, gets rid of conducting wire, the constraint of cable becomes
The direction of new century people's research and probe, then wireless power transmission (Wireless Power Transfer, WPT) technology is answered
It transports and gives birth to.
Wireless power transmission technology is still in the primary stage in the research both domestic and external of many aspects especially application aspect, and one
A major issue is exactly that efficiency of transmission is low, regardless of wireless power transmission technology all suffer from distance or load variation and
The problem of causing electric energy efficiency of transmission to reduce.The efficiency of transmission for improving system must just make system work in the resistance of optimum efficiency
Under anti-, different distance is adapted to from the matched angle Efficiency of load impedance and load is a more potential direction.
In view of the demand adjusted in real time to system, Adaptive impedance matching technology becomes radio energy transmission system and mentions
High system effectiveness, the development trend of optimization system, and the evaluation of matching precision and matching speed as impedance match technique superiority and inferiority
Standard, common impedance matching methods are difficult to meet the requirement to the two simultaneously.
Summary of the invention
The purpose of the present invention is to solve the above problems, devise a kind of WPT of genetic algorithm optimization BP neural network
System impedance matching process.
Realize above-mentioned purpose the technical scheme is that, a kind of WPT system resistance of genetic algorithm optimization BP neural network
Anti- matching process, this method comprises the following steps:
Step 1: suitable input variable of the characteristic quantity as model in WPT system is chosen, to optimize impedance matching ginseng
Number, realizes the impedance matching of system.
Step 2: determining the topological structure of neural network, and initialization, training BP neural network establish WPT system by study
The mapping relations for the input variable and system matches network parameter of uniting.
Step 3: the output error quadratic sum of BP neural network test sample capacitor is inverted as the suitable of genetic algorithm
Response function.
Step 4: carrying out genetic algorithm optimization, initialization population, and the fitness function that BP neural network is obtained carries out anti-
Gemini duplication intersects, mutation process, and final decoding obtains best weight value and threshold value.
Step 5: BP neural network output capacitance value is predicted using weight after optimization and threshold value, then passes through control
Motor makes system reach impedance matching condition come the matching capacitance adjusted in matching network.
Input variable described in the step 1 selects the input impedance Zn (Zn=Re+Im) of WPT system transmitting terminal, optimization
Target component selects two tunable capacitors C1, C2 in matching network, to establish the mapping relations of Re, Im and C1, C2.
Topological structure described in the step 2 selects 2-7-2, i.e. input layer has 2 nodes, and hidden layer has 7 nodes,
Output layer has 2 nodes, shares 28 weights and 9 threshold values.112 groups of data conducts are chosen from 132 groups of inputoutput datas
Training sample, 20 groups of data are as test sample.Hidden layer neuron transmission function uses S type tangent function in BP neural network
Tansig, output layer transmission function use linear function Purelin, the trainlm letter that training function selects convergence rate most fast
Number.
The output error requirement of test sample capacitor described in the step 3 is the smaller the better, that is, takes error sum of squares reciprocal
Fitness function afterwards is the bigger the better.Specific step is as follows for fitness function calculating:
28 weights of BP neural network and 9 threshold values are indicated with W1, W2, B1, B2, wherein W1 indicates that input layer arrives
The weight of hidden layer, the weight of W2 expression hidden layer to output layer, the threshold value of B1 expression input layer to hidden layer, B2 indicate hidden
Threshold value containing layer to output layer.
Calculate input layer to hidden layer output A1,
A1=tansig (W1*p, B1) (1)
P is the input of test sample, i.e., the input impedance detected in WPT system in formula;
Calculate hidden layer to output layer output A2,
A2=purelin (W2*A1, B2) (2)
Error sum of squares SE is calculated, and seeks the adaptive value val of genetic algorithm,
SE=sumsqr (t-A2) (3)
Val=1/SE (4)
T is the output of test sample, i.e., the matching capacitance of matching network in WPT system in formula.
The high chromosome of duplication described in the step 4, intersection, mutation process selection fitness, that is, select test electricity
Hold the lesser chromosome of error, and then finds best initial weights and threshold value, genetic algorithm parameter setting are as follows: population scale 50 is lost
Passage number is 50, crossover probability 0.4, mutation probability 0.1.
Utilize the WPT system impedance matching side of the genetic algorithm optimization BP neural network of technical solution of the present invention production
The disadvantages of method is easily limited to local minimum for existing for neural network algorithm itself, and convergence rate is slow, propose a kind of heredity
The WPT system impedance matching methods of algorithm optimization BP neural network.By introducing genetic algorithm to BP neural network weight and threshold
Value optimizes, and improves the matching speed of WPT system impedance matching, while possessing good matching precision again, has reached excellent
The purpose for changing control, improves the efficiency of transmission of system
Detailed description of the invention
Fig. 1 is genetic algorithm optimization BP neural network flow chart of the invention;
Fig. 2 is impedance matching system structural block diagram of the present invention;
Fig. 3 is optimum individual fitness value;
Fig. 4 is the prediction result of genetic algorithm optimization BP neural network;
Fig. 5 is the matching result of genetic algorithm optimization BP neural network.
Specific embodiment
The present invention is specifically described with reference to the accompanying drawing,
As shown in Fig. 2, WPT system impedance matching controls.Detect the input impedance Zn (Zn=Re+ of WPT system transmitting terminal
Im), and as the input variable of the genetic algorithm optimization BP neural network based on DSP, matched by step motor control
Two tunable capacitors C1, C2 in network, realize the impedance matching of WPT system.
Genetic algorithm optimization BP neural network flow chart is as shown in Figure 1.Neural network algorithm topological structure selects 2-7-2,
I.e. input layer has 2 nodes, and hidden layer has 7 nodes, and output layer has 2 nodes, shares 28 weights and 9 threshold values.From 132
112 groups of data are chosen as training sample in group inputoutput data, and 20 groups of data are as test sample.It is hidden in BP neural network
Neural transferring function containing layer uses S type tangent function Tansig, and output layer transmission function uses linear function Purelin, instruction
Practice the trainlm function that function selects convergence rate most fast.Genetic algorithm optimization obtain the optimal initial weight of BP neural network and
Optimal initial weight and threshold value are assigned to neural network, genetic algorithm parameter setting are as follows: population scale 50, hereditary generation by threshold value
Number is 50, crossover probability 0.4, mutation probability 0.1.
The mapping relations of Re, Im and C1, C2 are established by genetic algorithm optimization BP neural network, are obtained as shown in Figure 3 most
Excellent ideal adaptation angle value, wherein specific step is as follows for fitness function calculating:
28 weights of BP neural network and 9 threshold values are indicated with W1, W2, B1, B2, wherein W1 indicates that input layer arrives
The weight of hidden layer, the weight of W2 expression hidden layer to output layer, the threshold value of B1 expression input layer to hidden layer, B2 indicate hidden
Threshold value containing layer to output layer.
Calculate input layer to hidden layer output A1,
A1=tansig (W1*p, B1) (1)
P is the input of test sample, i.e., the input impedance detected in WPT system in formula;
Calculate hidden layer to output layer output A2,
A2=purelin (W2*A1, B2) (2)
Error sum of squares SE is calculated, and seeks the adaptive value val of genetic algorithm,
SE=sumsqr (t-A2) (3)
Val=1/SE (4)
T is the output of test sample, i.e., the matching capacitance of matching network in WPT system in formula.
Simulation and prediction tunable capacitor C1, C2, as seen from Figure 4, genetic algorithm optimization BP neural network is to WPT system
Prediction effect with parameter is preferable, 20 groups of test samples of simulation run, and the average value for obtaining prediction error is 0.1298, prediction essence
It spends higher.
The initial value that WPT system tunable capacitor C1, C2 is arranged is respectively 0.1439nF, 0.1844nF, detects system at this time
Input impedance be 18.3+j36.64 Ω, the system matches result after genetic algorithm optimization BP neural network as shown in figure 5, through
Input impedance becomes 49.79+j0.008 Ω after crossing impedance matching, and the match time of input resistance is 9 μ s, the matching of input reactance
Time is 7.9 μ s, it can be seen that with higher of the system impedance Matching Model based on genetic algorithm optimization BP neural network
With precision and faster matching speed, and possess good stability, causes to solve output loading or transmission range variation
WPT system impedance mismatching, matching precision it is low, the problems such as matching speed is slow, provides effective method.
Above-mentioned technical proposal only embodies the optimal technical scheme of technical solution of the present invention, those skilled in the art
The principle of the present invention is embodied to some variations that some of them part may be made, belongs to the scope of protection of the present invention it
It is interior.
Claims (5)
1. a kind of WPT system impedance matching methods of genetic algorithm optimization BP neural network, which is characterized in that this method includes such as
Lower step:
Step 1: choosing suitable input variable of the characteristic quantity as model in WPT system, to optimize impedance matching parameter,
The impedance matching of realization system.
Step 2: determining the topological structure of neural network, and it is defeated to establish WPT system by study for initialization, training BP neural network
Enter the mapping relations of variable Yu system matches network parameter.
Step 3: by the inverted fitness as genetic algorithm of the output error quadratic sum of BP neural network test sample capacitor
Function.
Step 4: carrying out genetic algorithm optimization, and initialization population is contaminated the fitness function that BP neural network obtains repeatedly
Colour solid duplication intersects, mutation process, and final decoding obtains best weight value and threshold value.
Step 5: predicting BP neural network output capacitance value using weight after optimization and threshold value, then passes through control motor
Matching capacitance to adjust in matching network makes system reach impedance matching condition.
2. the WPT system impedance matching methods of genetic algorithm optimization BP neural network according to claim 1, feature exist
In input variable described in the step 1 selects the input impedance Zn (Zn=Re+Im) of WPT system transmitting terminal, optimization aim
Two tunable capacitors C1, C2 in parameters selection matching network, to establish the mapping relations of Re, Im and C1, C2.
3. the WPT system impedance matching methods of genetic algorithm optimization BP neural network according to claim 1, feature exist
In topological structure described in the step 2 selects 2-7-2, i.e. input layer has 2 nodes, and hidden layer has 7 nodes, output layer
There are 2 nodes, shares 28 weights and 9 threshold values.112 groups of data are chosen from 132 groups of inputoutput datas as training sample
This, 20 groups of data are as test sample.Hidden layer neuron transmission function uses S type tangent function in BP neural network
Tansig, output layer transmission function use linear function Purelin, the trainlm letter that training function selects convergence rate most fast
Number.
4. the WPT system impedance matching methods of genetic algorithm optimization BP neural network according to claim 1, feature exist
In the output error requirement of test sample capacitor described in the step 3 is the smaller the better, that is, after taking error sum of squares reciprocal
Fitness function is the bigger the better.Specific step is as follows for fitness function calculating:
28 weights of BP neural network and 9 threshold values are indicated with W1, W2, B1, B2, wherein W1 indicate input layer to imply
The weight of layer, W2 indicate hidden layer to the weight of output layer, the threshold value of B1 expression input layer to hidden layer, B2 expression hidden layer
To the threshold value of output layer.
Calculate input layer to hidden layer output A1,
A1=tansig (W1*p, B1) (1)
P is the input of test sample, i.e., the input impedance detected in WPT system in formula;
Calculate hidden layer to output layer output A2,
A2=purelin (W2*A1, B2) (2)
Error sum of squares SE is calculated, and seeks the adaptive value val of genetic algorithm,
SE=sumsqr (t-A2) (3)
Val=1/SE (4)
T is the output of test sample, i.e., the matching capacitance of matching network in WPT system in formula.
5. the WPT system impedance matching methods of genetic algorithm optimization BP neural network according to claim 1, feature exist
In the high chromosome of duplication described in the step 4, intersection, mutation process selection fitness selects testing capacitor mistake
The lesser chromosome of difference, and then best initial weights and threshold value are found, genetic algorithm parameter setting are as follows: population scale 50, hereditary generation
Number is 50, crossover probability 0.4, mutation probability 0.1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811574311.3A CN109726817A (en) | 2018-12-21 | 2018-12-21 | The WPT system impedance matching methods of genetic algorithm optimization BP neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811574311.3A CN109726817A (en) | 2018-12-21 | 2018-12-21 | The WPT system impedance matching methods of genetic algorithm optimization BP neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109726817A true CN109726817A (en) | 2019-05-07 |
Family
ID=66297821
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811574311.3A Pending CN109726817A (en) | 2018-12-21 | 2018-12-21 | The WPT system impedance matching methods of genetic algorithm optimization BP neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109726817A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111310400A (en) * | 2020-02-16 | 2020-06-19 | 苏州浪潮智能科技有限公司 | BP neural network-based capacitance anti-pad optimization method and system |
CN111462092A (en) * | 2020-04-02 | 2020-07-28 | 浙江工业大学 | Vacuum cup surface defect detection method based on deep learning |
CN111481842A (en) * | 2020-04-21 | 2020-08-04 | 重庆邮电大学 | Wearable ultrasonic therapy appearance based on developments match |
CN112330487A (en) * | 2020-11-03 | 2021-02-05 | 河北工业大学 | Photovoltaic power generation short-term power prediction method |
CN112561025A (en) * | 2020-12-09 | 2021-03-26 | 安徽诚越电子科技有限公司 | Method and device for prolonging service life of aluminum electrolytic capacitor |
CN117390368A (en) * | 2023-12-07 | 2024-01-12 | 云南电投绿能科技有限公司 | Lightning probability calculation method, device and equipment for wind turbine and storage medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108490965A (en) * | 2018-04-19 | 2018-09-04 | 东华大学 | Rotor craft attitude control method based on Genetic Algorithm Optimized Neural Network |
CN109002878A (en) * | 2018-07-05 | 2018-12-14 | 淮阴师范学院 | A kind of GA Optimized BP Neural Network |
-
2018
- 2018-12-21 CN CN201811574311.3A patent/CN109726817A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108490965A (en) * | 2018-04-19 | 2018-09-04 | 东华大学 | Rotor craft attitude control method based on Genetic Algorithm Optimized Neural Network |
CN109002878A (en) * | 2018-07-05 | 2018-12-14 | 淮阴师范学院 | A kind of GA Optimized BP Neural Network |
Non-Patent Citations (3)
Title |
---|
范函等: ""一种电力线载波通信自适应阻抗匹配方案"", 《电力系统保护与控制》 * |
郁磊等: "《MATLAB智能算法30个案例分析》", 31 August 2015, 北京航空航天大学出版社 * |
陈严君: ""锂电池阻抗模型参数的BP神经网络预测研究"", 《中国优秀硕士学位论文全文数据库(电子期刊)》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111310400A (en) * | 2020-02-16 | 2020-06-19 | 苏州浪潮智能科技有限公司 | BP neural network-based capacitance anti-pad optimization method and system |
CN111310400B (en) * | 2020-02-16 | 2022-06-07 | 苏州浪潮智能科技有限公司 | BP neural network-based capacitance anti-pad optimization method and system |
CN111462092A (en) * | 2020-04-02 | 2020-07-28 | 浙江工业大学 | Vacuum cup surface defect detection method based on deep learning |
CN111481842A (en) * | 2020-04-21 | 2020-08-04 | 重庆邮电大学 | Wearable ultrasonic therapy appearance based on developments match |
CN112330487A (en) * | 2020-11-03 | 2021-02-05 | 河北工业大学 | Photovoltaic power generation short-term power prediction method |
CN112561025A (en) * | 2020-12-09 | 2021-03-26 | 安徽诚越电子科技有限公司 | Method and device for prolonging service life of aluminum electrolytic capacitor |
CN112561025B (en) * | 2020-12-09 | 2022-10-14 | 安徽诚越电子科技有限公司 | Method and device for prolonging service life of aluminum electrolytic capacitor |
CN117390368A (en) * | 2023-12-07 | 2024-01-12 | 云南电投绿能科技有限公司 | Lightning probability calculation method, device and equipment for wind turbine and storage medium |
CN117390368B (en) * | 2023-12-07 | 2024-03-26 | 云南电投绿能科技有限公司 | Lightning probability calculation method, device and equipment for wind turbine and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109726817A (en) | The WPT system impedance matching methods of genetic algorithm optimization BP neural network | |
CN110929763B (en) | Multi-source data fusion-based mechanical fault diagnosis method for medium-voltage vacuum circuit breaker | |
CN106850254B (en) | Method for identifying key nodes in power communication network | |
CN105388402B (en) | Crosslinked cable Partial Discharge Pattern Recognition Method based on parameter optimization SVM algorithm | |
CN111461922A (en) | Transformer hot spot temperature real-time prediction method based on extreme learning machine | |
CN107168292B (en) | Submarine navigation device circuit failure diagnosis method based on ELM algorithm | |
CN114692505A (en) | Error prediction method and device for capacitor voltage transformer | |
CN108537448A (en) | High iron catenary health state evaluation method based on Set Pair Analysis and evidence theory | |
CN116436152B (en) | Intelligent low-voltage distribution transformer area topology identification method based on characteristic information correlation | |
CN112182961A (en) | Large-scale fading modeling prediction method for wireless network channel of converter station | |
CN111598231A (en) | Cable connector wire temperature prediction based on GA (genetic algorithm) optimized serial BP (back propagation) network | |
CN113095648A (en) | Power distribution network fault power failure risk index evaluation method and system based on data driving | |
CN105207210B (en) | Distribution network optimizing operation method and system | |
CN111628494A (en) | Low-voltage distribution network topology identification method and system based on logistic regression method | |
CN103837764B (en) | Electric energy quality detection method for household solar photovoltaic power generation | |
CN105704741B (en) | A kind of heterogeneous network access selection method | |
CN103123669A (en) | Human body composition analysis method based on genetic algorithm | |
KR20210129893A (en) | System and method for estimating vehicle battery charging time using big data | |
JPH11289664A (en) | Power distribution system control system | |
CN112290887A (en) | BP neural network arc detection model improved based on L-M algorithm | |
CN116702598A (en) | Training method, device, equipment and storage medium for building achievement prediction model | |
CN114838923B (en) | Fault diagnosis model building method and fault diagnosis method for on-load tap-changer | |
CN110348625A (en) | Forecast of Nonlinear Time Series method based on worldlet scales-free network | |
CN114021470B (en) | Relay storage life prediction method based on AMFO algorithm and SVM algorithm | |
CN103513168B (en) | GIS and cable local discharge comprehensive judging method |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190507 |
|
RJ01 | Rejection of invention patent application after publication |