CN108663570A - Current harmonics analysis method based on trigonometric function neural network - Google Patents
Current harmonics analysis method based on trigonometric function neural network Download PDFInfo
- Publication number
- CN108663570A CN108663570A CN201810222379.9A CN201810222379A CN108663570A CN 108663570 A CN108663570 A CN 108663570A CN 201810222379 A CN201810222379 A CN 201810222379A CN 108663570 A CN108663570 A CN 108663570A
- Authority
- CN
- China
- Prior art keywords
- neural network
- trigonometric function
- current
- weight
- electric vehicle
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R23/00—Arrangements for measuring frequencies; Arrangements for analysing frequency spectra
- G01R23/16—Spectrum analysis; Fourier analysis
Abstract
The present invention relates to a kind of current harmonics analysis methods based on trigonometric function neural network to realize the frequency analysis to electric vehicle charging current for the harmonic wave that electric vehicle charging generates electric system.It converts electric vehicle charging current to the trigonometric function indicated with weight, constructs trigonometric function neural network, by the forward recursion of trigonometric function neural network, obtain the output current containing each harmonic component；By output current compared with input current, the two difference uses inverse iteration of the method for negative gradient descent method by trigonometric function neural network, the optimal weights of trigonometric function neural network is sought, to obtain the accurate estimation of electric vehicle charging current harmonic parameters.This method has faster convergence property and better noise tolerance.
Description
Technical field
The present invention relates to a kind of electric power detection technique, more particularly to a kind of current harmonics based on trigonometric function neural network
Analysis method.
Background technology
Traditional charger has the shortcomings that high Harmonics of Input of low order, low power factor, uncontrollable charged state, right
The control of battery current is limited.Moreover, low frequency and the high Harmonics of Input of low power factor do not meet IEC1000-
3-2 or IEEE519 harmonic standards, uncontrollable charged state will shorten battery life.
In order to mitigate harmonic pollution, it is primary to carry out accurate measurements and analysis to harmonic components, and compensation technique is answered
For correction signal, or terminate power transmission.Therefore, in order to reinforce Detecting Power Harmonicies, to harmonic parameters such as size and phase
Estimation is especially important.It is developed so far, power system harmonic measurement field makes significant progress, the estimation of most common of which
The method of harmonic component is to be based on Fast Fourier Transform (FFT) (FFT), but due to pointed stake barrage and spectrum leakage, FFT is filling
Have certain limitation in the practical application of electric car charging current frequency analysis, it will usually cause harmonic current, phase,
The evaluated error of frequency etc..Artificial neural network (ANN) had attracted extensive concern in past 20 years, by ANN and harmonic wave
It organically combines to realize and the accurate of harmonic wave is estimated as people's focus of attention, due to its calculating speed and robustness,
Each component of current harmonics can rapidly and accurately be traced into.But how by electric vehicle charging current and artificial neural network
Network is combined, and it is current a big difficulty accurately to find out the amplitude of harmonic current and phase.
Invention content
The problem of the present invention be directed to electric car charger faults itself or anomalous effects electric power qualities proposes
A kind of current harmonics analysis method based on trigonometric function neural network carries out the frequency analysis of electric vehicle charging current,
The analysis to electric current each harmonic parameter can be realized rapidly.
The technical scheme is that：A kind of current harmonics analysis method based on trigonometric function neural network, will be electronic
Automobile charging current is converted into the trigonometric function indicated with weight, trigonometric function neural network is constructed, by seeking trigonometric function
The optimal weights of neural network obtain the accurate estimation of electric vehicle charging current harmonic parameters, specifically include following steps：
1) trigonometric function neural network is constructed
Electric vehicle charging current is expressed as with trigonometric function：
Wherein：I (t) indicates the charging current of electric vehicle；idc(t) DC component of electric vehicle charging current is indicated；N
Indicate electric vehicle charging stream overtone order；ω0Indicate the angular frequency w of fundametal compomentjIndicate jth primary current harmonic parameters, wN+j
Indicate N+j primary current harmonic parameters；
Neural network input layer is n to training dataset { x, y }, x, y ∈ R1*nMatrix, wherein x=(x (1), x
(2) ... ..., x (i) ... ... x (n)), y=(y (1), y (2) ... ..., y (i) ... ... y (n)), x (i)=t (i) is to adopt for i-th
The sampling point corresponding time；Y (i) is the corresponding current instantaneous value of ith sample point；
Neural network hidden layer trigonometric functionActivation primitive is constructed, 2N+1 hidden neuron is shared, design is hidden
Hide the activation primitive of layerFor：
Neural network output layer is usedIt indicates：
Wherein：Indicate the output current of ith sample time corresponding trigonometric function Neural Networks Representation；w0It indicates
Weight of the hidden layer the 1st between neuron and output layer；wjIt indicates jth primary current harmonic parameters, while also illustrating that hidden layer
Weight between j-th of neuron and output layer；wN+jIt indicates N+j primary current harmonic parameters, while also illustrating that hidden layer N+
Weight between j neuron and output layer；
2) utility function and method of negative gradient descent method update weight solve analysis current harmonics：
Design learning rule performance function is：
The output current of neural network that estimation obtains will be calculated to make comparisons with input current, when being unsatisfactory for target error,
That is desired setting value e is not achieved in e (w)objWhen, update weight with negative gradient iterative method；
Based on method of negative gradient descent method, the weight iterative formula of design triangle Function Neural Network is：
W (k+1)=w (k)-η PT(Pw(k)-y)
Wherein：
Wherein：K=1,2 ..., itermax, indicate kth time iteration, itermaxFor iterations；W (k) indicates kth time repeatedly
For corresponding weight；W (k+1) indicates the corresponding weight of+1 iteration of kth；η indicates learning rate；If making η ＞ 0 and sufficiently small,
The weight of trigonometric function neural network can be made to converge on optimal weights by iteration；
By solving the optimal weight vector of neural network, DC component and jth can be estimated by weight iterative formula
The amplitude and phase angle of subharmonic：
Wherein：AdcIndicate the amplitude of DC component；AjIndicate the amplitude of jth subharmonic；φjIndicate the phase of jth subharmonic
Position.
The beneficial effects of the present invention are：The present invention is based on the current harmonics analysis methods of trigonometric function neural network, lead to
The mode for crossing method of negative gradient descent updates weight, to obtain the accurate harmonic wave estimation of current or voltage, can effectively calculate harmonic wave
Component.This method provides the solution of simple possible for harmonic wave estimation.
Description of the drawings
Fig. 1 is that the present invention is based on the structure charts of trigonometric function neural network；
Fig. 2 is that the present invention is based on the flow charts of the harmonic analysis method of trigonometric function neural network；
Fig. 3 a are the property of trigonometric function neural network of the present invention and FFT when analyzing electric vehicle charging current fundamental voltage amplitude
Figure can be compared；
Fig. 3 b are trigonometric function neural network of the present invention and FFT when analyzing electric vehicle charging current triple-frequency harmonics amplitude
Performance compare figure；
Fig. 4 is the property of trigonometric function neural network of the present invention and FFT when analyzing electric vehicle charging current fundamental wave phase angle
Figure can be compared.
Specific implementation mode
Current harmonics analysis method based on trigonometric function neural network：Mainly convert electric vehicle charging current to
The trigonometric function indicated with weight constructs trigonometric function neural network, by the forward recursion of trigonometric function neural network, obtains
Output current containing each harmonic component；By output current compared with input current, the two difference is passed through using method of negative gradient descent method
The inverse iteration of trigonometric function neural network seeks the optimal weights of trigonometric function neural network, using optimal weights can quickly,
Accurate estimating system harmonic component and relevant parameter, robustness is good, can realize the analysis to electric current each harmonic parameter rapidly.
Electric vehicle charging current can be expressed as the sum of each harmonic component with amplitude and phase：
Wherein：I (t) indicates the charging current of electric vehicle；idc(t) DC component of electric vehicle charging current is indicated；N
Indicate electric vehicle charging stream overtone order；ω0Indicate the angular frequency of fundametal compoment；AjIndicate the amplitude of jth subharmonic；φjTable
Show the phase of jth subharmonic.
According to triangular equation, definitionφj=tan-1[wj/wN+j], then formula (1) can be written as：
Wherein：wjIndicate jth primary current harmonic parameters, wN+jIndicate N+j primary current harmonic parameters.
Structure chart as shown in Figure 1 based on trigonometric function neural network, input layer are n to training dataset { x, y }, x, y
∈R1*nMatrix.Wherein x=(x (1), x (2) ... ..., x (i) ... ... (n)), y=(y (1), y (2) ... ..., y (i) ... ... y
(n)), x (i)=t (i) is the ith sample point corresponding time；Y (i) is the corresponding current instantaneous value of ith sample point.
Hidden layer trigonometric functionActivation primitive is constructed, 2N+1 hidden neuron is shared, designs swashing for hidden layer
Function livingFor：
Output layer is usedIt indicates：
Wherein：Indicate the output current of ith sample time corresponding trigonometric function Neural Networks Representation；w0It indicates
Weight of the hidden layer the 1st between neuron and output layer；wjIt indicates jth primary current harmonic parameters, while also illustrating that hidden layer
Weight between j-th of neuron and output layer；wN+jIt indicates N+j primary current harmonic parameters, while also illustrating that hidden layer N+
Weight between j neuron and output layer.
Design learning rule performance function is：
Based on method of negative gradient descent method, the weight iterative formula of design triangle Function Neural Network is：
W (k+1)=w (k)-η PT(Pw(k)-y) (6)
Wherein：
Wherein：K=1,2 ..., itermax, indicate kth time iteration, itermaxFor iterations；W (k) indicates kth time repeatedly
For corresponding weight；W (k+1) indicates the corresponding weight of+1 iteration of kth；η indicates learning rate.If making η ＞ 0 and sufficiently small,
The weight of trigonometric function neural network can be made to converge on optimal weights by iteration.
By solving the optimal weight vector of neural network, DC component and jth subharmonic can be estimated by formula (6)
Amplitude and phase angle：
Wherein：AdcIndicate the amplitude of DC component；AjIndicate the amplitude of jth subharmonic；φjIndicate the phase of jth subharmonic
Position.
The flow chart of harmonic analysis method based on trigonometric function neural network is as shown in Fig. 2, will calculate what estimation obtained
The output current of neural network is made comparisons with input current, and when being unsatisfactory for target error, i.e. desired setting value e is not achieved in e (w)obj
When, weight is updated with negative gradient iterative method, actual current is accurately and rapidly estimated in realization.
In order to verify the correctness and validity of trigonometric function neural network analysis harmonic wave, to electric current of the reality containing harmonic wave
Signal carries out simulation study by MATLAB/Simulink.Input data is the actual value of electric system electric current, and research institute examines
The each harmonic component of current signal of worry is shown in Table 1.Input current signal is calculated by table 1 and formula (1).
Table 1
| Overtone order | Amplitude | Phase | Order | Amplitude | Phase |
| 1 | 3.00 | -23.10 | 6 | 0.00 | - |
| 2 | 0.03 | 115.60 | 7 | 003 | -31.80 |
| 3 | 0.15 | 59.30 | 8 | 0.00 | - |
| 4 | 0.01 | 52.40 | 9 | 0.01 | -63.70 |
| 5 | 0.04 | 123.80 | 10 | - | - |
When emulation, ω0100 π rad/s are set as, learning rate η is 0.01, and harmonic wave highest times N=9, each iteration is to adopt
The f=1000Hz samplings of sample frequency, sample 10 points (n=10), target error e altogetherobjIt is set as 0.01.
To prove the superiority of trigonometric function neural network analysis harmonic wave, itself and Fourier decomposition (FFT) are emulated
Comparison, simulation result is as shown in Fig. 3 a, 3b, 4.ANN and FFT can track the amplitude and phase of fundamental wave and other each components
Angle, FFT need at least one period 0.02s that can just calculate corresponding amplitude and phase angle, and ANN only needs half period 0.01s
Corresponding amplitude and phase angle can be traced into, this will be to monitoring and quickly protecting beneficial in real time.The lower waveform of FFT controls is compared,
ANN can rapidly trace into each component of actual current signal.
Claims (1)
1. a kind of current harmonics analysis method based on trigonometric function neural network, which is characterized in that
It converts electric vehicle charging current to the trigonometric function indicated with weight, trigonometric function neural network is constructed, by asking
The optimal weights of trigonometric function neural network are taken, the accurate estimation of electric vehicle charging current harmonic parameters is obtained, specifically includes
Following steps：
1) trigonometric function neural network is constructed
Electric vehicle charging current is expressed as with trigonometric function：
Wherein：I (t) indicates the charging current of electric vehicle；idc(t) DC component of electric vehicle charging current is indicated；N is indicated
Electric vehicle charging stream overtone order；ω0Indicate the angular frequency w of fundametal compomentjIndicate jth primary current harmonic parameters, wN+jIt indicates
N+j primary current harmonic parameters；
Neural network input layer is n to training dataset { x, y }, x, y ∈ R1*nMatrix, wherein x=(x (1), x (2) ...,
X (i) ... x (n)), y=(y (1), y (2) ..., y (i), ... y (n)), x (i)=t (i) is ith sample
The point corresponding time；Y (i) is the corresponding current instantaneous value of ith sample point；
Neural network hidden layer trigonometric functionActivation primitive is constructed, 2N+1 hidden neuron is shared, designs hidden layer
Activation primitiveFor：
Neural network output layer is usedIt indicates：
Wherein：Indicate the output current of ith sample time corresponding trigonometric function Neural Networks Representation；w0It indicates to hide
Weight of the 1st, the layer between neuron and output layer；wjIt indicates jth primary current harmonic parameters, while also illustrating that hidden layer j-th
Weight between neuron and output layer；wN+jIt indicates N+j primary current harmonic parameters, while also illustrating that the N+j god of hidden layer
Through the weight between member and output layer；
2) utility function and method of negative gradient descent method update weight solve analysis current harmonics：
Design learning rule performance function is：
The output current of neural network that estimation obtains will be calculated to make comparisons with input current, when being unsatisfactory for target error, i.e. e
(w) desired setting value e is not achievedobjWhen, update weight with negative gradient iterative method；
Based on method of negative gradient descent method, the weight iterative formula of design triangle Function Neural Network is：
W (k+1)=w (k)-η PT(Pw(k)-y)
Wherein：
Wherein：K=1,2 ..., itermax, indicate kth time iteration, itermaxFor iterations；W (k) indicates kth time iteration pair
The weight answered；W (k+1) indicates the corresponding weight of+1 iteration of kth；η indicates learning rate；If making η ＞ 0 and sufficiently small, can lead to
Crossing iteration makes the weight of trigonometric function neural network converge on optimal weights；
By solving the optimal weight vector of neural network, DC component can be estimated by weight iterative formula and jth time is humorous
The amplitude and phase angle of wave：
Wherein：AdcIndicate the amplitude of DC component；AjIndicate the amplitude of jth subharmonic；φjIndicate the phase of jth subharmonic.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810222379.9A CN108663570B (en) | 2018-03-15 | 2018-03-15 | Current harmonic analysis method based on trigonometric function neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810222379.9A CN108663570B (en) | 2018-03-15 | 2018-03-15 | Current harmonic analysis method based on trigonometric function neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108663570A true CN108663570A (en) | 2018-10-16 |
| CN108663570B CN108663570B (en) | 2023-05-23 |
Family
ID=63785241
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810222379.9A Active CN108663570B (en) | 2018-03-15 | 2018-03-15 | Current harmonic analysis method based on trigonometric function neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108663570B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109800520A (en) * | 2019-01-25 | 2019-05-24 | 国网浙江省电力有限公司湖州供电公司 | A kind of electric automobile charging station Harmonic Modeling method neural network based |
| CN112117947A (en) * | 2020-09-30 | 2020-12-22 | 桂林电子科技大学 | SRM torque ripple suppression control system and method based on current injection method |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101216512A (en) * | 2007-12-29 | 2008-07-09 | 湖南大学 | Non-sine periodic signal real time high precision detection method |
| CN102338827A (en) * | 2011-06-10 | 2012-02-01 | 中国矿业大学 | Method for analyzing electric network waveform distortions and automatically monitoring electric power harmonic parameters |
| CN102353839A (en) * | 2011-07-18 | 2012-02-15 | 华北电力大学（保定） | Electric power system harmonics analysis method based on multilayered feedforward neural network |
| CN102998527A (en) * | 2012-11-26 | 2013-03-27 | 上海电力学院 | Pass band type fundamental wave, harmonic wave and direct current component detection method |
| CN103383413A (en) * | 2013-07-09 | 2013-11-06 | 温州大学 | Real-time harmonic detection method based on direct weight determination method |
| CN103424621A (en) * | 2013-08-20 | 2013-12-04 | 江苏大学 | Artificial neural network detecting method of harmonic current |
| CN104777356A (en) * | 2015-03-10 | 2015-07-15 | 三峡大学 | Neural-network-based real-time high-accuracy harmonic detection method |
-
2018
- 2018-03-15 CN CN201810222379.9A patent/CN108663570B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101216512A (en) * | 2007-12-29 | 2008-07-09 | 湖南大学 | Non-sine periodic signal real time high precision detection method |
| CN102338827A (en) * | 2011-06-10 | 2012-02-01 | 中国矿业大学 | Method for analyzing electric network waveform distortions and automatically monitoring electric power harmonic parameters |
| CN102353839A (en) * | 2011-07-18 | 2012-02-15 | 华北电力大学（保定） | Electric power system harmonics analysis method based on multilayered feedforward neural network |
| CN102998527A (en) * | 2012-11-26 | 2013-03-27 | 上海电力学院 | Pass band type fundamental wave, harmonic wave and direct current component detection method |
| CN103383413A (en) * | 2013-07-09 | 2013-11-06 | 温州大学 | Real-time harmonic detection method based on direct weight determination method |
| CN103424621A (en) * | 2013-08-20 | 2013-12-04 | 江苏大学 | Artificial neural network detecting method of harmonic current |
| CN104777356A (en) * | 2015-03-10 | 2015-07-15 | 三峡大学 | Neural-network-based real-time high-accuracy harmonic detection method |
Non-Patent Citations (3)
| Title |
|---|
| 操吴兵等: "基于分段迭代的电力谐波神经网络分析方法", 《电测与仪表》 * |
| 陈忠华 等: "《基于三角函数神经网络的电动汽车充电电流谐波分析方法》", 《上海电力学院学报》 * |
| 雷锡社等: "基于神经网络的高精度电力系统谐波分析", 《电测与仪表》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109800520A (en) * | 2019-01-25 | 2019-05-24 | 国网浙江省电力有限公司湖州供电公司 | A kind of electric automobile charging station Harmonic Modeling method neural network based |
| CN112117947A (en) * | 2020-09-30 | 2020-12-22 | 桂林电子科技大学 | SRM torque ripple suppression control system and method based on current injection method |
| CN112117947B (en) * | 2020-09-30 | 2022-03-11 | 桂林电子科技大学 | SRM torque ripple suppression control system and method based on current injection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108663570B (en) | 2023-05-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hu et al. | Lithium-ion battery modeling and parameter identification based on fractional theory | |
| Chan et al. | The available capacity computation model based on artificial neural network for lead–acid batteries in electric vehicles | |
| Li et al. | A new carbon price prediction model | |
| CN106683122A (en) | Particle filtering method based on Gaussian mixture model and variational Bayes | |
| CN103728535A (en) | Extra-high-voltage direct-current transmission line fault location method based on wavelet transformation transient state energy spectrum | |
| CN108257044A (en) | A kind of non-intrusion type load decomposition method based on steady-state current model | |
| CN108448585B (en) | A kind of electric network swim equation solution method of linearization based on data-driven | |
| CN108663570A (en) | Current harmonics analysis method based on trigonometric function neural network | |
| CN111709350B (en) | Low-frequency oscillation modal parameter identification method and system based on FCM clustering | |
| Yang et al. | Multi-step prediction of strong earthquake ground motions and seismic responses of SDOF systems based on EMD-ELM method | |
| CN115201563A (en) | Multi-harmonic source positioning method and system based on joint entropy | |
| Qin et al. | State of Charge estimation of lithium-ion power battery based on online parameter identification method and BP neural network | |
| CN110163148A (en) | A kind of electric car DC charging distorted signal self-adaptive identification method | |
| Dash et al. | Accurate tracking of harmonic signals in VSC-HVDC systems using PSO based unscented transformation | |
| CN114139778A (en) | Wind turbine generator power prediction modeling method and device | |
| CN107167658B (en) | A kind of jamproof electric system fundamental frequency of high-precision and Method for Phase Difference Measurement | |
| CN111639463B (en) | XGboost algorithm-based frequency characteristic prediction method for power system after disturbance | |
| CN106546824B (en) | The system harmonic impedance estimation method of power grid containing photovoltaic returned based on variable coefficient | |
| Shen et al. | A novel neural network approach for power system low frequency oscillation mode identification | |
| Li et al. | Improvement of model predictive control for single‐phase inverters by using sinusoidal signal observers | |
| Wu et al. | A new method for identifying a fault in T-connected lines based on multiscale S-transform energy entropy and an extreme learning machine | |
| CN105572472B (en) | The frequency measurement method and system of distributed generation resource environment | |
| Yin et al. | Multi-step Prediction Algorithm of Traffic Flow Chaotic Time Series Based on Volterra Neural Network. | |
| CN112821392A (en) | Static voltage stability boundary calculation method based on all-pure embedding method | |
| Yuan et al. | Analysis of power grid harmonics with wavelet network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Yu Rongjiang Inventor after: Wang Xinyue Inventor after: Zhang Fan Inventor after: Jin Jiao Inventor after: Zhu Yijia Inventor after: Shen Guoheng Inventor after: Luo Jinsheng Inventor after: Chen Zhonghua Inventor after: Wang Yufei Inventor after: Hu Chengang Inventor after: Chen Pan Inventor after: Xue Hua Inventor after: Xu Xiuzhen Inventor after: Chen Bing Inventor before: Yu Rongjiang Inventor before: Wang Xinyue Inventor before: Zhang Fan Inventor before: Jin Jiao Inventor before: Zhu Yijia Inventor before: Shen Guoheng Inventor before: Luo Jinsheng Inventor before: Chen Zhonghua Inventor before: Wang Yufei Inventor before: Hu Chengang Inventor before: Chen Pan Inventor before: Xue Hua Inventor before: Xu Xiuzhen Inventor before: Chen Bing |
|
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |