CN104158418B - A kind of photovoltaic network inverter - Google Patents
A kind of photovoltaic network inverter Download PDFInfo
- Publication number
- CN104158418B CN104158418B CN201410408718.4A CN201410408718A CN104158418B CN 104158418 B CN104158418 B CN 104158418B CN 201410408718 A CN201410408718 A CN 201410408718A CN 104158418 B CN104158418 B CN 104158418B
- Authority
- CN
- China
- Prior art keywords
- grid
- moment
- voltage
- output
- input terminus
- 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.)
- Expired - Fee Related
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/50—Photovoltaic [PV] energy
- Y02E10/56—Power conversion systems, e.g. maximum power point trackers
Landscapes
- Inverter Devices (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
A kind of photovoltaic network inverter, it comprises the multiple Miniature inverter one to one of each photovoltaic battery panel with photovoltaic electric station, the reservoir capacitance that each Miniature inverter comprises MPU controller and is connected to successively between photovoltaic battery panel and electrical network, anti exciting converter, power frequency inverter bridge and EMI wave filter, described MPU controller gathers photovoltaic battery panel voltage and electrical network voltage respectively by the first voltage sensor and the 2nd voltage sensor, logical over-current sensor gathers the grid-connected current that EMI wave filter exports, and drive module to control anti exciting converter and power frequency inverter bridge respectively by two. the present invention, while improving generating efficiency, also solves the problem of centralized invertor poor reliability. this device both can realize the linearizing of system together with neural network contrary modeling being combined in PI controller, can play again the robustness of PI controller, thus effectively improves the quality of power supply.
Description
Technical field
The present invention relates to a kind of light adopting neural network contrary modeling to control and lie prostrate grid-connected Miniature inverter system, belong to technical field of power generation.
Background technology
Along with energy dilemma and problem of environmental pollution are day by day serious, solar energy power generating is as cleaning green energy resource, and its cutting-in control technology becomes the focus that countries in the world are paid close attention to and studied.
The requirement that the utilising efficiency of the energy and the quality of power supply are proposed by modern society is more and more higher. Existing grid-connected photovoltaic invertor mostly is centralized invertor, carries out electric energy output by after a large amount of photovoltaic cell series-parallel connection by a high-power inverter. Owing to can not ensure that each photovoltaic cell works in peak power point, there is inefficient shortcoming in this kind of mode. More seriously, the reliability of this kind of mode is poor, if centralized invertor generation fault, connected all photovoltaic cells will be caused to terminate electric energy and export. Therefore, miniatureization of photovoltaic DC-to-AC converter will become development trend from now on. In addition, there is the shortcoming that grid-connected current harmonic wave content is higher in the existing simple PI controller of the many employings of grid-connected photovoltaic invertor. By increase output filtering unit can harmonic inhabitation to a certain extent, but the output impedance but making system presents two rank even high-order LC characteristic, it is easy to cause system unstable. Visible, in order to improve the quality of power supply, it is necessary to seek better to control method.
Summary of the invention
It is an object of the invention to the drawback for prior art, it is provided that a kind of photovoltaic network inverter, effectively to improve reliability and the quality of power supply of photovoltaic generation.
Problem of the present invention realizes with following technical proposals:
A kind of photovoltaic network inverter, formation comprises the multiple Miniature inverter one to one of each photovoltaic battery panel with photovoltaic electric station, the reservoir capacitance that each Miniature inverter comprises MPU controller and is connected to successively between photovoltaic battery panel and electrical network, anti exciting converter, power frequency inverter bridge and EMI wave filter, described MPU controller gathers photovoltaic battery panel voltage and electrical network voltage respectively by the first voltage sensor and the 2nd voltage sensor, logical over-current sensor gathers the grid-connected current that EMI wave filter exports, and drive module to control anti exciting converter and power frequency inverter bridge respectively by two, described MPU controller operates in the following manner:
1. photovoltaic battery panel voltage V is gathered by the first voltage sensorpv, gather electrical network voltage V by the 2nd voltage sensorgrid, logical over-current sensor gathers grid-connected current iac;Using Miniature inverter as object, set up and train corresponding neural network contrary modeling, specific as follows:
A. the inversion model that three layers of BP neural network set up system is chosen, wherein, input layer node number is 10, hidden layer neuron node number is 18, output layer neuron node number is 1, hidden layer neuron transfer function uses hyperbolic tangent function, and output layer neurone transfer function uses S type function;
B. on photovoltaic miniature invertor, service data is gathered;
Single loop PI controller is utilized to carry out grid-connected current control, under making the operating mode that photovoltaic miniature invertor is in rated output, 80% rated output, 60% rated output, 40% rated output, 20% rated output respectively, continuous acquisition N group grid-connected current i under often kind of operating modeac, electrical network voltage Vgrid, anti exciting converter pwm signal dutycycle D, photovoltaic battery panel voltage Vpv, form 5N group service data altogether;
C. the service data structure sample of collection is utilized;
For the k moment, the sample constructed is { Vgrid(k), Vgrid(k-1), Vgrid(k-2), iac(k), iac(k-1), iac(k+1), D (k-1), D (k-2), D (k-3), Vpv(k), D (k) }, wherein, VgridK electrical network voltage that () is the k moment, Vgrid(k-1) it is the electrical network voltage in k-1 moment, Vgrid(k-2) it is the electrical network voltage in k-2 moment, iacK () is the grid-connected current in k moment, iac(k-1) it is the grid-connected current in k-1 moment, iac(k+1) it is the grid-connected current in k+1 moment, the anti exciting converter pwm signal dutycycle that D (k-1) is the k-1 moment, the anti exciting converter pwm signal dutycycle that D (k-2) is the k-2 moment, the anti exciting converter pwm signal dutycycle that D (k-3) is the k-3 moment, VpvK () is the voltage of the photovoltaic battery panel in k moment, the anti exciting converter pwm signal dutycycle that D (k) is the k moment, and 5N group service data forms 5N sample altogether;
D. all the other are N number of as test samples as learning sample to randomly draw the 4N in 5N sampled data, three layers of BP neural network that training is set up; In training process, for the sample in k moment, first input terminus getting BP neural network is the electrical network voltage V in k momentgridK (), the 2nd input terminus is the electrical network voltage V in k-1 momentgrid(k-1), the 3rd input terminus is the electrical network voltage V in k-2 momentgrid(k-2), the 4th input terminus is the grid-connected current i in k momentacK (), the 5th input terminus is the grid-connected current i in k-1 momentac(k-1), the 6th input terminus is the grid-connected current i in k+1 momentac(k+1), 7th input terminus is anti exciting converter pwm signal dutycycle D (k-1) in k-1 moment, 8th input terminus is anti exciting converter pwm signal dutycycle D (k-2) in k-2 moment, 9th input terminus is anti exciting converter pwm signal dutycycle D (k-3) in k-3 moment, and the tenth input terminus is the voltage V of the photovoltaic battery panel in k momentpv(k), anti exciting converter pwm signal dutycycle D (k) that the output of BP neural network is the k moment;
2. utilizing the BP neural network contrary modeling trained to be controlled by Miniature inverter, concrete steps are:
A. utilize the BP neural network contrary modeling trained to carry out the calculating of anti exciting converter pwm signal dutycycle, for the current k moment, get the electrical network voltage V that first input terminus is the k momentgridK (), the 2nd input terminus is the electrical network voltage V in k-1 momentgrid(k-1), the 3rd input terminus is the electrical network voltage V in k-2 momentgrid(k-2), the 4th input terminus is the grid-connected current i in k momentacK (), the 5th input terminus is the grid-connected current i in k-1 momentac(k-1), the 6th input terminus is the grid-connected current set(ting)value i in k+1 moment* ac(k+1), 7th input terminus is anti exciting converter pwm signal dutycycle D (k-1) in k-1 moment, 8th input terminus is anti exciting converter pwm signal dutycycle D (k-2) in k-2 moment, 9th input terminus is anti exciting converter pwm signal dutycycle D (k-3) in k-3 moment, and the tenth input terminus is the voltage V of the photovoltaic battery panel in k momentpvK (), show that the output of BP neural network is d (k);
B. by subsequent time grid-connected current set(ting)value i* ac(k+1) with current k moment grid-connected current iacK the deviation of () sends into PI controller, obtain closed-loop control amount dC(k), namely
Wherein: e (k)=i* ac(k+1)-iac(k), KpRepresent scale-up factor, KiRepresent integral coefficient;
C. by the output d of output d (k) of neural network contrary modeling He PI controllerCAs final anti exciting converter pwm signal dutycycle D (k) after (k) superposition, that is:
D(k)=d(k)+dC(k)。
Above-mentioned photovoltaic network inverter, the training of BP neural network is adopted the error back propagation algorithm becoming learning rate, it is specially: first the weight coefficient in each layer neurone and threshold value are initialized as (-1, + 1) random quantity between, again using input layer 10 neuron output values as each neuronic input of hidden layer, 18 neuron output values of hidden layer are as the single neuronic input of output layer, the single neuronic output value of output layer is network and finally exports, for kth learning sample, calculate the rear actual output of network of the t time trainingWith desired outputError:
And the error of sum square of 4N learning sample is:
Often calculate one time, compare EtWith expectation error EoIf, Et<Eo, then train end condition to meet, training terminates; Otherwise, by EtCarry out backpropagation along connecting path, and adjust weight coefficient and the threshold value of each layer one by one, until Et<EoTill; Then, utilize N number of test samples to be tested by model, calculate sample square errorMSE, if havingMSE<Tr(TrFor a certain fixed threshold), then think that the BP neural network model trained meets accuracy requirement, now record each neuronic weight coefficient and threshold value; Otherwise, again BP neural network model is trained, till it meets accuracy requirement.
Above-mentioned photovoltaic network inverter, described anti exciting converter comprises transformer, the first power tube, diode and the 2nd electric capacity, the primary coil of described transformer connects photovoltaic battery panel voltage through the first power tube, its secondary coil is powered to power frequency inverter bridge through diode, 2nd electric capacity is connected to the direct-flow input end of power frequency inverter bridge, and the grid of the first power tube connects the output terminal of the first driving module.
Above-mentioned photovoltaic network inverter, described EMI wave filter comprises three inductance and six electric capacity, the first end of two coils of the first inductance connects two ac output ends of power frequency inverter bridge respectively, the two ends that their tail end gets access to grid through the 2nd inductance and the 3rd inductance respectively, 3rd electric capacity is attempted by the ac output end of power frequency inverter bridge, one end ground connection of the 4th electric capacity and the 5th electric capacity, the other end connects two ac output ends of power frequency inverter bridge respectively, the two ends of the 6th electric capacity connect the tail end of two coils of the first inductance respectively, one end ground connection of the 7th electric capacity and the 8th electric capacity, the two ends that the other end gets access to grid respectively.
Above-mentioned photovoltaic network inverter, described power frequency inverter bridge is the full control bridge being connected into by four power tubes, and the grid of four power tubes connects the output terminal of the 2nd driving module.
The present invention utilizes multiple Miniature inverter to replace the centralized invertor in conventional photovoltaic power station, makes each photovoltaic battery panel work in peak power point, while improving generating efficiency, also solves the problem of centralized invertor poor reliability. This device both can realize the linearizing of system together with neural network contrary modeling being combined in PI controller, can play again the robustness of PI controller, thus effectively improves the quality of power supply.
Accompanying drawing explanation
Below in conjunction with accompanying drawing, the invention will be further described.
Fig. 1 is the structure block diagram that light lies prostrate grid-connected Miniature inverter device;
Fig. 2 is neural network contrary modeling schematic diagram;
Fig. 3 is single neuronic structural representation;
Fig. 4 is the test result of BP neural network contrary modeling;
Fig. 5 is the program structure diagram of nerve network reverse controller;
Fig. 6 is the grid-connected current curve that light lies prostrate the output of grid-connected Miniature inverter;
Fig. 7 is the grid-connected current graph of errors that light lies prostrate the output of grid-connected Miniature inverter.
In figure, each list of reference numerals is: UT1, the first voltage sensor, UT2, the 2nd voltage sensor, CT, current sense device, D1, diode, Q1～Q5, the first power tube～the 5th power tube, C1～C8, the first electric capacity～the 8th electric capacity, L1～L3, the first inductance～the 3rd inductance, TX1, transformer.
In literary composition, each symbol inventory is: VpvFor photovoltaic battery panel voltage, VgridFor electrical network voltage, iacFor grid-connected current, VgridK electrical network voltage that () is the k moment, D is anti exciting converter pwm signal dutycycle, Vgrid(k-1) it is the electrical network voltage in k-1 moment, Vgrid(k-2) it is the electrical network voltage in k-2 moment, iacK () is the grid-connected current in k moment, iac(k-1) it is the grid-connected current in k-1 moment, iac(k+1) it is the grid-connected current in k+1 moment, the anti exciting converter pwm signal dutycycle that D (k-1) is the k-1 moment, the anti exciting converter pwm signal dutycycle that D (k-2) is the k-2 moment, the anti exciting converter pwm signal dutycycle that D (k-3) is the k-3 moment, VpvK () is the voltage of the photovoltaic battery panel in k moment, the anti exciting converter pwm signal dutycycle that D (k) is the k moment, i* ac(k+1) it is the grid-connected current set(ting)value in k+1 moment,For the actual output of network after t training,For desired output after t training, EoFor expecting error, EtFor the error of sum square of 4N learning sample after t training.
Embodiment
The present invention is made up of multiple Miniature inverter, each Miniature inverter comprises: reservoir capacitance (i.e. the first electric capacity C1 in Fig. 1), anti exciting converter is (by transformer TX1, first power tube Q1, diode D1 and the 2nd electric capacity C2 is formed), power frequency inverter bridge (is made up of the 2nd power tube Q2～the 5th power tube Q5), EMI wave filter is (by the 3rd electric capacity C3～the 8th electric capacity C8, first inductance L the 1～three inductance L 3 is formed), first driving module (i.e. driving module 1 in Fig. 1), 2nd driving module (i.e. driving module 2 in Fig. 1), MPU controller, first voltage sensor UT1, 2nd voltage sensor UT2, current sense device CT, the positive pole of reservoir capacitance is connected with photovoltaic battery panel positive pole, the negative pole ground connection of reservoir capacitance, the negative pole ground connection of the first voltage sensor, the positive pole of the first voltage sensor are connected with photovoltaic battery panel positive pole, the measurement signal output terminal of the first voltage sensor is connected with MPU controller, the input of anti exciting converter is just being held and is being connected with photovoltaic battery panel positive pole, the input negativing ending grounding of anti exciting converter, the output plus terminal of anti exciting converter is connected with power frequency inverter bridge one input terminus, and the output negative terminal of anti exciting converter is connected with another input terminus of power frequency inverter bridge, one output terminal of power frequency inverter bridge is connected with EMI wave filter one input terminus, and another output terminal of power frequency inverter bridge is connected with another input terminus of EMI wave filter, one output terminal of EMI wave filter is connected with the positive pole of current sense device, and another output terminal of EMI wave filter is connected to electrical network, the positive pole of current sense device is connected with an output terminal of EMI wave filter, and the negative pole of current sense device is connected to electrical network, and the measurement signal output terminal of current sense device is connected with MPU controller,The positive pole of the 2nd voltage sensor is connected to the negative pole of current sense device, and the negative pole of the 2nd voltage sensor is connected to electrical network, and the measurement signal output terminal of the 2nd voltage sensor is connected with MPU controller; The input terminus of the first driving module is connected with MPU controller, and the output terminal of the first driving module is connected with the first power tube grid in anti exciting converter; The input terminus of the 2nd driving module is connected with MPU controller, one output terminal of the 2nd driving module is connected with the grid of the 2nd power tube in power frequency inverter bridge, the 5th power tube, and another output terminal of the 2nd driving module is connected with the grid of the 3rd power tube in power frequency inverter bridge, the 4th power tube; MPU controller is connected with the measurement signal output terminal of the first voltage sensor, the 2nd voltage sensor, current sense device, and MPU controller is also connected with the input terminus of the first driving module, the 2nd driving module;
Below in conjunction with the drawings and specific embodiments, the present invention is described further.
In the present embodiment, choosing the 235W photovoltaic battery panel that Ying Li company produces, it in the parameter of standard test condition is: short-circuit current Isc=8.54A, open circuit voltage Voc=37V, peak power point electric current Imax=7.97A, maximum power point voltage Vmax=29.5V。
The Miniature inverter structure block diagram designed for this photovoltaic battery panel is as shown in Figure 1. This device mainly comprises: reservoir capacitance (the first electric capacity) C1, anti exciting converter, power frequency inverter bridge, EMI wave filter, the first driving module, the 2nd drive module, MPU controller, the first voltage sensor UT1, the 2nd voltage sensor UT2, current sense device CT; Wherein the high performance float-point digital signal processor TMS320F28335 of TI company selected by MPU controller, and this digital signal processor is integrated with the function modules such as I/O module, ADC, PWM module, CAN module, UART and SPI module; Reservoir capacitance C1 selects 2200uF chemical capacitor; In anti exciting converter, transformer TX1 adopts NA5814-AL, and the first power tube Q1 selects TK50X15J1, and diode D1 adopts C2D05120E, and the 2nd electric capacity C2 selects 0.47uF polarity free capacitor; 2nd power tube Q2, the 3rd power tube Q3, the 4th power tube Q4, the 5th power tube Q5 select IPB60R190C6, and the first driving module and the 2nd driving module are MCP14E4; In EMI wave filter, the 3rd electric capacity C3, the 6th electric capacity C6 select 0.015uf polarity free capacitor, 4th electric capacity C4, the 5th electric capacity C5, the 7th electric capacity C7, the 8th electric capacity C8 select 4700pF polarity free capacitor, first inductance L 1 selects FE2X03-4-3NL, and the 2nd inductance L 2, the 3rd inductance L 3 select 2100LL-151-H-RC; Voltage sensor adopts MCP6022; Current sense device CT adopts Hall current sensor ACS712ELCTR-058-1;
Fig. 2 is the three layers of BP neural network contrary modeling set up. Input layer node number is 10, and hidden layer neuron node number is 18, and output layer neuron node number is 1.
Fig. 3 is single neuronic structural representation,For neuronic input signal,For neuronic weight coefficient,For neuronic threshold value, then neuronic output is:
Hidden layer neuron transfer function uses hyperbolic tangent function, namely
Wherein:
For transfer function input;
For transfer function exports.
Output layer neurone transfer function uses S type function, namely
(3)
Wherein:
For transfer function input;
For transfer function exports.
Embodiment gathers service data process be specially: utilize single loop PI controller to carry out grid-connected current control, under making the operating mode that photovoltaic miniature invertor is in rated output, 80% rated output, 60% rated output, 40% rated output, 20% rated output respectively, with the sample frequency continuous acquisition N=10000 group grid-connected current i of 114KHz under often kind of operating modeac, electrical network voltage Vgrid, anti exciting converter pwm signal dutycycle D, photovoltaic battery panel voltage Vpv, obtain 50000 groups of service datas altogether, thus form 50000 samples.Randomly drawing 40000 wherein as learning sample, all the other 10000 as test samples.
Embodiment arranges Eo=1.5, threshold value Tr=0.01, after iteration trains 9876 times, Et=1.47, meet end condition Et<Eo, training terminates. Now, model is tested by recycling test samples, and part test result as shown in Figure 4, calculates sample square errorMSE=0.006, hasMSE<Tr, therefore think and meet inversion model accuracy requirement, now record each neuronic weight coefficient and threshold value.
In embodiment, the parameter tuning of PI controller is Kp=0.01, Ki=800; Grid-connected current set(ting)value i* acGiven according to the change of discrete sine rule by digital signal processor.
Write software program according to above-mentioned steps and implant in dsp chip, the control of the nerve network reverse controller of photovoltaic miniature invertor can be realized. Software program comprises system master routine and Neural network inverse control break in service sub-routine, and block diagram is such as Fig. 5.
Fig. 6 is the grid-connected current curve that after using control method of the present invention, photovoltaic miniature invertor exports, and Fig. 7 is grid-connected current deviation curve. Visible, grid-connected current can be good at following the tracks of its set(ting)value, there is less deviation.
Control the effect of method in order to analyze the present invention further, under 20%, 40%, 60%, 80%, 100% rated output exports operating mode, itself and conventional PI control method (are got K respectivelyp=0.01, Ki=800) contrasted, table 1 is the grid-connected current harmonic wave content balance result of two kinds of methods, known the inventive method has good robustness for the change of operating mode, grid-connected current harmonic wave content under different operating mode is significantly less than conventional PI control method, meets the grid-connected harmonic wave content standard of national grid (THD < 5%).
Table 1
。
The above; it is only the present invention's preferably embodiment; but protection scope of the present invention should be not limited thereto; the core technology that any those skilled in the art of being familiar with disclose according to the present invention, the change that can make or replacement, all should be encompassed within protection scope of the present invention.
Claims (5)
1. a photovoltaic network inverter, it is characterized in that, it comprises the multiple Miniature inverter one to one of each photovoltaic battery panel with photovoltaic electric station, the reservoir capacitance that each Miniature inverter comprises MPU controller and is connected to successively between photovoltaic battery panel and electrical network, anti exciting converter, power frequency inverter bridge and EMI wave filter, described MPU controller gathers photovoltaic battery panel voltage and electrical network voltage respectively by the first voltage sensor (UT1) and the 2nd voltage sensor (UT2), logical over-current sensor (CT) gathers the grid-connected current that EMI wave filter exports, and drive module to control anti exciting converter and power frequency inverter bridge respectively by two, described MPU controller operates in the following manner:
1. photovoltaic battery panel voltage V is gathered by the first voltage sensor (UT1)pv, gather electrical network voltage V by the 2nd voltage sensor (UT2)grid, logical over-current sensor (CT) gathers grid-connected current iac; Using Miniature inverter as object, set up and train corresponding neural network contrary modeling, specific as follows:
A. the inversion model that three layers of BP neural network set up system is chosen, wherein, input layer node number is 10, hidden layer neuron node number is 18, output layer neuron node number is 1, hidden layer neuron transfer function uses hyperbolic tangent function, and output layer neurone transfer function uses S type function;
B. on photovoltaic miniature invertor, service data is gathered;
Single loop PI controller is utilized to carry out grid-connected current control, under making the operating mode that photovoltaic miniature invertor is in rated output, 80% rated output, 60% rated output, 40% rated output, 20% rated output respectively, continuous acquisition N group grid-connected current i under often kind of operating modeac, electrical network voltage Vgrid, anti exciting converter pwm signal dutycycle D, photovoltaic battery panel voltage Vpv, form 5N group service data altogether;
C. the service data structure sample of collection is utilized;
For the k moment, the sample constructed is { Vgrid(k), Vgrid(k-1), Vgrid(k-2), iac(k), iac(k-1), iac(k+1), D (k-1), D (k-2), D (k-3), Vpv(k), D (k) }, wherein, VgridK electrical network voltage that () is the k moment, Vgrid(k-1) it is the electrical network voltage in k-1 moment, Vgrid(k-2) it is the electrical network voltage in k-2 moment, iacK () is the grid-connected current in k moment, iac(k-1) it is the grid-connected current in k-1 moment, iac(k+1) it is the grid-connected current in k+1 moment, the anti exciting converter pwm signal dutycycle that D (k-1) is the k-1 moment, the anti exciting converter pwm signal dutycycle that D (k-2) is the k-2 moment, the anti exciting converter pwm signal dutycycle that D (k-3) is the k-3 moment, VpvK () is the voltage of the photovoltaic battery panel in k moment, the anti exciting converter pwm signal dutycycle that D (k) is the k moment, and 5N group service data forms 5N sample altogether;
D. all the other are N number of as test samples as learning sample to randomly draw the 4N in 5N sampled data, three layers of BP neural network that training is set up; In training process, for the sample in k moment, first input terminus getting BP neural network is the electrical network voltage V in k momentgridK (), the 2nd input terminus is the electrical network voltage V in k-1 momentgrid(k-1), the 3rd input terminus is the electrical network voltage V in k-2 momentgrid(k-2), the 4th input terminus is the grid-connected current i in k momentacK (), the 5th input terminus is the grid-connected current i in k-1 momentac(k-1), the 6th input terminus is the grid-connected current i in k+1 momentac(k+1), 7th input terminus is anti exciting converter pwm signal dutycycle D (k-1) in k-1 moment, 8th input terminus is anti exciting converter pwm signal dutycycle D (k-2) in k-2 moment, 9th input terminus is anti exciting converter pwm signal dutycycle D (k-3) in k-3 moment, and the tenth input terminus is the voltage V of the photovoltaic battery panel in k momentpv(k), anti exciting converter pwm signal dutycycle D (k) that the output of BP neural network is the k moment;
2. utilizing the BP neural network contrary modeling trained to be controlled by Miniature inverter, concrete steps are:
A. utilize the BP neural network contrary modeling trained to carry out the calculating of anti exciting converter pwm signal dutycycle, for the current k moment, get the electrical network voltage V that first input terminus is the k momentgridK (), the 2nd input terminus is the electrical network voltage V in k-1 momentgrid(k-1), the 3rd input terminus is the electrical network voltage V in k-2 momentgrid(k-2), the 4th input terminus is the grid-connected current i in k momentacK (), the 5th input terminus is the grid-connected current i in k-1 momentac(k-1), the 6th input terminus is the grid-connected current set(ting)value i in k+1 moment* ac(k+1), 7th input terminus is anti exciting converter pwm signal dutycycle D (k-1) in k-1 moment, 8th input terminus is anti exciting converter pwm signal dutycycle D (k-2) in k-2 moment, 9th input terminus is anti exciting converter pwm signal dutycycle D (k-3) in k-3 moment, and the tenth input terminus is the voltage V of the photovoltaic battery panel in k momentpvK (), show that the output of BP neural network is d (k);
B. by subsequent time grid-connected current set(ting)value i* ac(k+1) with current k moment grid-connected current iacK the deviation of () sends into PI controller, obtain closed-loop control amount dC(k), namely
Wherein: e (k)=i* ac(k+1)-iac(k), KpRepresent scale-up factor, KiRepresent integral coefficient;
C. by the output d of output d (k) of neural network contrary modeling He PI controllerCAs final anti exciting converter pwm signal dutycycle D (k) after (k) superposition, that is:
D (k)=d (k)+dC(k)。
2. a kind of photovoltaic network inverter according to claim 1, it is characterized in that, the training of BP neural network is adopted the error back propagation algorithm becoming learning rate, it is specially: first the weight coefficient in each layer neurone and threshold value are initialized as (-1, + 1) random quantity between, again using input layer 10 neuron output values as each neuronic input of hidden layer, 18 neuron output values of hidden layer are as the single neuronic input of output layer, the single neuronic output value of output layer is network and finally exports, for kth learning sample, calculate the rear actual output of network of the t time trainingWith desired outputError:
And the error of sum square of 4N learning sample is:
Often calculate one time, compare EtWith expectation error EoIf, Et<Eo, then train end condition to meet, training terminates; Otherwise, by EtCarry out backpropagation along connecting path, and adjust weight coefficient and the threshold value of each layer one by one, until Et<EoTill; Then, utilizing N number of test samples to be tested by model, calculating sample square error Δ MSE, if there being Δ MSE < Tr, TrFor a certain fixed threshold, then think that the BP neural network model trained meets accuracy requirement, now record each neuronic weight coefficient and threshold value; Otherwise, again BP neural network model is trained, till it meets accuracy requirement.
3. a kind of photovoltaic network inverter according to claim 1 and 2, it is characterized in that, described anti exciting converter comprises transformer (TX1), first power tube (Q1), diode (D1) and the 2nd electric capacity (C2), the primary coil of described transformer (TX1) connects photovoltaic battery panel voltage through the first power tube (Q1), its secondary coil is powered to power frequency inverter bridge through diode (D1), 2nd electric capacity (C2) is connected to the direct-flow input end of power frequency inverter bridge, the grid of the first power tube (Q1) connects the output terminal of the first driving module.
4. a kind of photovoltaic network inverter according to claim 3, it is characterized in that, described EMI wave filter comprises three inductance and six electric capacity, the first end of two coils of the first inductance (L1) connects two ac output ends of power frequency inverter bridge respectively, the two ends that their tail end gets access to grid through the 2nd inductance (L2) and the 3rd inductance (L3) respectively, 3rd electric capacity (C3) is attempted by the ac output end of power frequency inverter bridge, one end ground connection of the 4th electric capacity (C4) and the 5th electric capacity (C5), the other end connects two ac output ends of power frequency inverter bridge respectively, the two ends of the 6th electric capacity (C6) connect the tail end of two coils of the first inductance (L1) respectively, one end ground connection of the 7th electric capacity (C7) and the 8th electric capacity (C8), the two ends that the other end gets access to grid respectively.
5. a kind of photovoltaic network inverter according to claim 4, is characterized in that, described power frequency inverter bridge is the full control bridge being connected into by four power tubes, and the grid of four power tubes connects the output terminal of the 2nd driving module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410408718.4A CN104158418B (en) | 2014-08-19 | 2014-08-19 | A kind of photovoltaic network inverter |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410408718.4A CN104158418B (en) | 2014-08-19 | 2014-08-19 | A kind of photovoltaic network inverter |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104158418A CN104158418A (en) | 2014-11-19 |
| CN104158418B true CN104158418B (en) | 2016-06-15 |
Family
ID=51883854
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410408718.4A Expired - Fee Related CN104158418B (en) | 2014-08-19 | 2014-08-19 | A kind of photovoltaic network inverter |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104158418B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104965558A (en) * | 2015-05-27 | 2015-10-07 | 华北电力大学（保定） | Photovoltaic power generation system maximum power tracking method and apparatus considering the factor of haze |
| CN106100398A (en) * | 2016-06-06 | 2016-11-09 | 国网青海省电力公司 | A kind of micro-capacitance sensor energy storage inverter control method based on gravitation neutral net |
| CN108075482A (en) * | 2016-11-11 | 2018-05-25 | 中科诺维(北京)科技有限公司 | Permanent magnet direct-drive wind turbine gird-connected inverter device and control method |
| CN111682244B (en) * | 2017-11-08 | 2022-04-29 | 南京晓庄学院 | Control method of energy-saving device of reversible fuel cell stack |
| US11747405B2 (en) * | 2018-02-28 | 2023-09-05 | Robert Bosch Gmbh | System and method for audio and vibration based power distribution equipment condition monitoring |
| CN109889069A (en) * | 2019-03-05 | 2019-06-14 | 淮阴工学院 | A kind of individual event inverter composite control method and system |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7339287B2 (en) * | 2002-06-23 | 2008-03-04 | Powerlynx A/S | Power converter |
| US8269372B2 (en) * | 2008-12-23 | 2012-09-18 | Samsung Electro-Mechanics Co., Ltd. | Photovoltaic and fuel cell hybrid generation system using dual converters and single inverter and method of controlling the same |
| CN102142693A (en) * | 2011-03-10 | 2011-08-03 | 西安交通大学 | Solar photovoltaic grid-connected inversion control system based on field programmable gate array (FPGA) |
| CN103106544B (en) * | 2013-02-01 | 2016-10-26 | 东南大学 | A kind of photovoltaic generation prognoses system based on T-S Fuzzy neutral net |
| CN103441526B (en) * | 2013-08-14 | 2015-04-01 | 华北电力大学（保定） | Grid-connected off-grid small photovoltaic power generation system and control method |
-
2014
- 2014-08-19 CN CN201410408718.4A patent/CN104158418B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN104158418A (en) | 2014-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104158418B (en) | A kind of photovoltaic network inverter | |
| CN109659980B (en) | Energy management optimization method for traction power supply system integrating hybrid energy storage and photovoltaic device | |
| CN112103946A (en) | Particle swarm algorithm-based microgrid energy storage optimization configuration method | |
| CN104701871A (en) | Wind, light and water-containing multi-source complementary micro-grid hybrid energy storage capacity optimal proportion method | |
| CN104638678A (en) | Multi-fan grid-connected resonance analysis method | |
| CN105976055A (en) | Distributed photovoltaic-energy storage system (PV-BES) output optimization and capacity allocation method counting power loss | |
| CN103441526B (en) | Grid-connected off-grid small photovoltaic power generation system and control method | |
| CN109659973B (en) | Distributed power supply planning method based on improved direct current power flow algorithm | |
| CN102142693A (en) | Solar photovoltaic grid-connected inversion control system based on field programmable gate array (FPGA) | |
| WO2023060815A1 (en) | Energy storage capacity optimization configuration method for improving reliability of power distribution network | |
| CN107276110B (en) | Multi-infeed direct current transmission receiving end alternating current system equivalent evaluation method | |
| CN107666155A (en) | System Stochastic Stability Analysis method of providing multiple forms of energy to complement each other based on Markov model | |
| CN103809650B (en) | A kind of equivalent modeling method of photovoltaic generating system | |
| CN104333034B (en) | Grid-connected Miniature inverter based on SVMs Adverse control | |
| CN111222097A (en) | Method for evaluating reliability and analyzing sensitivity of photovoltaic direct-current boosting and collecting system | |
| CN108460228B (en) | Wind power plant equivalence method based on multi-objective optimization algorithm | |
| Wu et al. | Equivalent modeling method for regional decentralized photovoltaic clusters based on cluster analysis | |
| CN106160003B (en) | Method of electric energy metering system of grid-connected wind-solar combined power generation system | |
| CN105896613B (en) | A kind of micro-capacitance sensor distribution finite-time control method for considering communication time lag | |
| CN104967353B (en) | A kind of off-network type wind power generation inverter | |
| Rashid et al. | An RES-based grid connected electric vehicle charging station for Bangladesh | |
| CN104269883B (en) | A kind of photovoltaic generating system equivalence method based on real-time digital simulator | |
| CN112595932B (en) | Monopole fault line selection method suitable for medium-voltage direct-current power distribution network | |
| CN205565860U (en) | Mixed energy storage converter of power battery and super capacitor | |
| Fritz | Validation of solar pv inverter harmonics behaviour at different power levels in a test network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160615 Termination date: 20180819 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |