CN112398131B - UPQC control system based on time lag compensation and control method thereof - Google Patents
UPQC control system based on time lag compensation and control method thereof Download PDFInfo
- Publication number
- CN112398131B CN112398131B CN202011285909.8A CN202011285909A CN112398131B CN 112398131 B CN112398131 B CN 112398131B CN 202011285909 A CN202011285909 A CN 202011285909A CN 112398131 B CN112398131 B CN 112398131B
- Authority
- CN
- China
- Prior art keywords
- current
- module
- harmonic
- voltage
- phase
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims description 49
- 238000013528 artificial neural network Methods 0.000 claims abstract description 80
- 238000001514 detection method Methods 0.000 claims abstract description 73
- 238000004364 calculation method Methods 0.000 claims abstract description 34
- 238000001914 filtration Methods 0.000 claims abstract description 23
- 230000009466 transformation Effects 0.000 claims description 75
- 238000006243 chemical reaction Methods 0.000 claims description 12
- 239000000284 extract Substances 0.000 claims description 12
- 238000003064 k means clustering Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 230000001537 neural effect Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 abstract description 5
- 230000000694 effects Effects 0.000 abstract description 4
- 230000006872 improvement Effects 0.000 abstract description 2
- 239000011159 matrix material Substances 0.000 description 10
- 238000010586 diagram Methods 0.000 description 4
- 238000000844 transformation Methods 0.000 description 4
- 238000005070 sampling Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/01—Arrangements for reducing harmonics or ripples
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02M—APPARATUS FOR CONVERSION BETWEEN AC AND AC, BETWEEN AC AND DC, OR BETWEEN DC AND DC, AND FOR USE WITH MAINS OR SIMILAR POWER SUPPLY SYSTEMS; CONVERSION OF DC OR AC INPUT POWER INTO SURGE OUTPUT POWER; CONTROL OR REGULATION THEREOF
- H02M1/00—Details of apparatus for conversion
- H02M1/12—Arrangements for reducing harmonics from ac input or output
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02M—APPARATUS FOR CONVERSION BETWEEN AC AND AC, BETWEEN AC AND DC, OR BETWEEN DC AND DC, AND FOR USE WITH MAINS OR SIMILAR POWER SUPPLY SYSTEMS; CONVERSION OF DC OR AC INPUT POWER INTO SURGE OUTPUT POWER; CONTROL OR REGULATION THEREOF
- H02M5/00—Conversion of ac power input into ac power output, e.g. for change of voltage, for change of frequency, for change of number of phases
- H02M5/02—Conversion of ac power input into ac power output, e.g. for change of voltage, for change of frequency, for change of number of phases without intermediate conversion into dc
- H02M5/04—Conversion of ac power input into ac power output, e.g. for change of voltage, for change of frequency, for change of number of phases without intermediate conversion into dc by static converters
- H02M5/10—Conversion of ac power input into ac power output, e.g. for change of voltage, for change of frequency, for change of number of phases without intermediate conversion into dc by static converters using transformers
-
- 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
- Y02E40/00—Technologies for an efficient electrical power generation, transmission or distribution
- Y02E40/40—Arrangements for reducing harmonics
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Supply And Distribution Of Alternating Current (AREA)
- Inverter Devices (AREA)
Abstract
The UPQC control system based on time lag compensation comprises a power grid, and further comprises a series transformer, a load, a first filtering module, a second filtering module, a voltage inverter, a current inverter, a harmonic voltage detection module, a UPQC compensation calculation module, a harmonic current detection module, a harmonic prediction neural network module, a three-phase voltage current beat, a three-phase voltage previous n beat, a three-phase current beat, a three-phase current previous n beat, a voltage driving circuit module, a current driving circuit module, a voltage sensor, a current sensor and a parallel transformer; the UPQC system based on time lag compensation has the advantages of high detection precision, no influence of distorted voltage on detection effect, capability of compensating time lag of signal extraction and improvement of signal compensation instantaneity.
Description
Technical Field
The invention relates to the technical field of micro-grid harmonic governance, and discloses a hysteresis error prediction method applied to unified power quality modulators (Unified power quality controller, UPQC) and a harmonic compensation control system thereof.
Background
The UPQC has the advantages of comprehensively compensating the problems of voltage sag, harmonic and current three-phase imbalance, reactive power, harmonic and the like, harmonic signal extraction is generally carried out by adopting an ip-iq algorithm based on instantaneous reactive power at present, the UPQC has the advantages of high detection precision and no influence of distorted voltage on detection effect, but the delay of at least one sampling period exists in the extraction of command signals due to the fact that the delay exceeds 100 mu s in the multiple coordinate transformation links in the harmonic extraction process. The characteristics of large fluctuation of harmonic current and quick dynamic response aggravate the influence of hysteresis deviation on harmonic compensation precision.
Disclosure of Invention
The invention aims to:
in order to solve the problems, the invention adds a harmonic prediction neural network module in the UPQC system to train the UPQC system, and predicts the hysteresis deviation of harmonic voltage and current by using the harmonic prediction neural network module, so as to improve the traditional UPQC control system and improve the harmonic compensation precision.
The technical scheme is as follows:
the UPQC system based on time lag compensation comprises a power grid (1), and is characterized in that: the system further comprises a series transformer (2), a load (3), a first filtering module (4), a second filtering module (5), a voltage inverter (6), a current inverter (7), a harmonic voltage detection module (8), a UPQC compensation calculation module (9), a harmonic current detection module (10), a harmonic prediction neural network module (11), a three-phase voltage current beat (12), a three-phase voltage previous beat (13), a three-phase voltage previous n beat (14), a three-phase current beat (15), a three-phase current previous beat (16), a three-phase current previous n beat (17), a voltage driving circuit module (18), a current driving circuit module (19), a voltage sensor (20), a current sensor (21) and a parallel transformer (22);
the power grid (1) is connected to the series transformer (2), and the series transformer (2) is connected with the load (3) again to finish the power supply of the power grid (1) to the load (3);
the load (3) is connected with the harmonic current detection module (10) through the current sensor (21), and the harmonic current detection module (10) is connected to the harmonic prediction neural network module (11) and the UPQC compensation calculation module (9);
the harmonic prediction neural network module (11) is connected with the UPQC compensation calculation module (9), and the UPQC compensation calculation module (9) is connected with the first voltage driving circuit (18) and the second voltage driving circuit (19);
the voltage driving circuit (18) is connected with the voltage inverter (6), and the voltage inverter (6) is connected with the series transformer (2) through the first filtering module (4);
the current driving circuit module (19) is connected with the current inverter (7), the current inverter (7) is connected with the parallel transformer (22) through the second filtering module (5), and the parallel transformer (22) is connected with the load (3);
the power grid (1) is also connected with a voltage sensor (20), the voltage sensor (20) is connected with a harmonic voltage detection module (8), and the harmonic voltage detection module (8) is connected with a harmonic prediction neural network module (11) and a UPQC compensation calculation module (9).
The voltage sensor (20) receives a voltage signal of a power grid end and outputs the voltage signal to the harmonic voltage detection module (8);
the current sensor (21) receives a current signal at the end of the load (3) and outputs the current signal to the harmonic current detection module (10);
the harmonic prediction neural network module (11) extracts a current beat (12) of the three-phase voltage from the harmonic voltage detection module (8) to a lag error compensation signal of the harmonic voltage calculated by n beats (14) before the three-phase voltage, calculates a harmonic voltage signal through the UPQC compensation calculation module (9) and transmits the harmonic voltage signal to the voltage driving circuit (18);
the harmonic prediction neural network module (11) extracts a three-phase current beat (15) from the harmonic current detection module (10) to a three-phase current front n beat (17), and a corresponding hysteresis error compensation signal of the calculated harmonic current calculates a harmonic current signal through the UPQC compensation calculation module (9) and sends the harmonic current signal to the current drive circuit (19);
the UPQC compensation calculation module (9) receives the current beat (12) voltage detection signal output by the harmonic voltage detection module (8) and the current beat current detection signal output by the harmonic current detection module (10), calculates corresponding current beat voltage compensation signals and current beat current compensation signals, and sends the current beat voltage compensation signals and the current beat current compensation signals to the voltage driving circuit module (18) and the current driving circuit module (19) respectively; (the current beat voltage compensation signal is sent to the voltage driving circuit module (18), the current beat current compensation signal is sent to the current driving circuit module (19))
The voltage driving circuit module (18) receives the current beat voltage compensation signal and the harmonic voltage signal to generate a driving voltage signal, the driving voltage signal is sent to the voltage inverter (6), filtered by the first filtering module (4), sent to the series transformer (2) and then output to the load (3);
the second driving circuit module (19) receives the current beat current compensation signal and the harmonic current signal to generate a driving current signal, the driving current signal is sent to the current inverter (7), filtered by the second filtering module (5), and then sent to the parallel transformer (22) to be output to the load (3).
The harmonic prediction neural network module (11) includes a first harmonic prediction neural network module (88) and a second harmonic prediction neural network module (116)
The first harmonic prediction neural network module (88) receives the current beats of the three-phase a, b and c power grid voltages, the previous beats of the three-phase a, b and c power grid voltages until the previous n beats of the three-phase a, b and c power grid voltages, and the first harmonic prediction neural network module (88) calculates a, b and c three-phase harmonic voltage hysteresis error compensation signals;
the second harmonic prediction neural network module (116) receives the current beats of the a, b and c three-phase load currents, the previous beats of the a, b and c three-phase load currents until the previous n beats of the a, b and c three-phase load currents, and the harmonic prediction neural network module (116) calculates a, b and c three-phase harmonic current lag error compensation signals.
The method comprises a UPQC control system harmonic voltage method and a harmonic current detection method based on time lag compensation;
the harmonic voltage method comprises the following steps: the participation module comprises: a first phase-locked loop, PLL, (81), a first triangular transformation module (82), a first Clark transformation module (83), a first Park transformation module (84), a first low pass filter (85), a first Park inverse transformation module (86), and a first Clark inverse transformation module (87); (these are circuit block components of UPQC compensation calculation block (9))
The first phase-locked loop PLL (81) receives a-phase power grid voltage and extracts a rotation angle of the voltage for the next various coordinate transformations;
the first triangular transformation module (82) transforms the rotation angle extracted by the first phase-locked loop (81) into a trigonometric function value for the first Park transformation module (84) and the first Park inverse transformation module (86);
the harmonic current detection method comprises the following steps: the participation module comprises: comprises a second phase-locked loop PLL (119), a second triangular transformation module (110), a second Clark transformation module (111), a second Park transformation module (112), a second low-pass filter (113), a second Park inverse transformation module (114) and a second Clark inverse transformation module (115);
the second phase-locked loop PLL (119) receives the a-phase grid current and extracts the rotation angle of the current for the next various coordinate transformations;
the second triangular transformation module (110) transforms the rotation angle extracted by the second phase-locked loop (PLL) (119) into a trigonometric function value for the second Park transformation module (112) and the second Park inverse transformation module (114).
The control method implemented by the UPQC system based on time lag compensation is characterized by comprising the following steps of:
the harmonic voltage detection method of the UPQC control system based on time lag compensation comprises the following steps: distorted mains voltage u a 、u b 、u c Transformed by a first Clark transformation module (83) into u in an alpha beta coordinate system α 、u β The corresponding transformation matrix is present; u (u) α 、u β Then transformed to u under dq coordinate system by a first Park transformation module (84) p 、u q The corresponding transformation matrix is present; u (u) p 、u q Obtaining a direct current component by a first low pass filter (85)Then the fundamental wave active voltage u is obtained through a first Park inverse conversion module (86) af 、u bf 、u cf The method comprises the steps of carrying out a first treatment on the surface of the Three-phase network terminal voltage u a 、u b 、u c And fundamental wave active voltage u af 、u bf 、u cf The difference is made to obtain a command voltage signal u ah 、u bh 、u ch The method comprises the steps of carrying out a first treatment on the surface of the The first harmonic prediction neural network module (88) sends out a harmonic voltage hysteresis error compensation signal u aph 、u bph 、u cph Compensation signal u with the current beat voltage ah 、u bh 、u ch The sum is the harmonic voltage signal +.>
The harmonic current detection method of the UPQC control system based on time lag compensation is as follows: three-phase load side current i La 、i Lb 、i Lc Transformed into i in the alpha beta coordinate system by a second Clark transformation module (111) α 、i β The corresponding transformation matrix is present; i.e α 、i β Then the i is transformed to the dq coordinate system by a second Park transformation module (112) p 、i q The corresponding transformation matrix is present; i.e p 、i q Obtaining a DC component by a second low-pass filter (113)Then the fundamental wave active current i is obtained through a second Park inverse conversion module (114) af 、i bf 、i cf The method comprises the steps of carrying out a first treatment on the surface of the Distorted load current i La 、i Lb 、i Lc And then the fundamental wave active current i af 、i bf 、i cf The difference is made to obtain a compensation signal i of the current beat current ah 、i bh 、i ch The method comprises the steps of carrying out a first treatment on the surface of the The second harmonic prediction neural network module (116) sends out a harmonic current lag error compensation signal i aph 、i bph 、i cph Compensating signal i with current ah 、i bh 、i ch The sum is the harmonic current signal +.>
The harmonic prediction neural network module is divided into 3 layers: an input layer (201), an hidden layer (202) and an output layer (203); wherein the input layer (201) contains 3 x 2 x n input signals, which are respectively the current beat, the previous beat and the previous n beats of a, b and c three-phase power grid voltages and a, b and c three-phase load currents; the output layer (203) comprises 6 signals, namely an a, b and c three-phase harmonic voltage hysteresis error compensation signal and a harmonic current hysteresis error compensation signal; the hidden layer (202) of the harmonic prediction neural network module adopts a radial base vector, and the radial base vector is H= [ H ] 1 ,h 2 ,…,h m ] T Wherein h is m Is a Gaussian basis function:
the I & is an European norm; the central vector of the neural element of the harmonic prediction neural network module is C= [ C ] 1 ,c 2 ,L,c m ] T The method comprises the steps of carrying out a first treatment on the surface of the The base width vector of the harmonic prediction neural network module is B= [ B ] 1 ,b 2 ,L,b m ] T The method comprises the steps of carrying out a first treatment on the surface of the The weight vector of the harmonic prediction neural network module is as follows: w= [ W ] 1 ,w 2 ,L,w m ] T The method comprises the steps of carrying out a first treatment on the surface of the The output of the network at time k is: f (x) k )=WH=w 1 h 1 +w 2 h 2 +L+w m h m 。
Training of the first harmonic prediction neural network module (88) and the second harmonic prediction neural network module (116) is divided into two phases, wherein the first phase adopts a K-means clustering algorithm: selecting the voltage value and the current value of the first three beats, respectively calculating Euclidean distance input to each clustering center, distributing the Euclidean distance to each clustering center according to the principle of nearest distance, calculating the average value of data in each clustering center, taking the average value as a new clustering center, redistributing input data, and repeating the processes until the position of the clustering center is not changed; and in the second stage, the output weight is determined by adopting a least square method.
The advantages and effects are that:
the system comprises a series transformer, a parallel transformer, a voltage inverter and a filtering module thereof, a current inverter and a filtering module thereof, a harmonic voltage detection link, a harmonic current detection link, a UPQC compensation calculation module, a harmonic prediction neural network module, a voltage and current driving circuit module, a voltage sensor, a current sensor, a power grid and a load;
the power grid is connected to the series transformer and the parallel transformer, and the voltage output by the power grid and the harmonic voltage compensation voltage output by the series transformer are added in series and sent to the load; the current output by the power grid and the harmonic current compensation current output by the parallel transformer are added in parallel, and the power grid output supplies power to the load after series-parallel double compensation;
the voltage sensor detects voltage signals of the power grid end in real time, and the current sensor detects current signals of the load end in real time;
the harmonic voltage detection link and the harmonic current detection link are connected to the voltage sensor and the current sensor so as to acquire sampling values of voltage at the power grid end and current at the load end in real time, store the three-phase voltage and current in the first beat and the first two beats and the first n beats by a storage method, and output the three-phase voltage and current in the first beat and the first n beats to the UPQC compensation calculation module and the harmonic prediction neural network module;
the UPQC compensation calculation module receives current beat voltage and current detection signals output by a harmonic voltage detection link and a harmonic current detection link, and calculates compensation signals of the current beat voltage and current;
the harmonic prediction neural network module receives the voltage and current real-time acquisition values and the stored values from the previous beat to the previous n beats output by the harmonic voltage detection link and the harmonic current detection link, calculates a hysteresis error compensation signal of the output harmonic voltage and current, and outputs the hysteresis error compensation signal to the UPQC compensation calculation module; adding the hysteresis error compensation signal and the compensation signal of the current beat voltage and current calculated by the UPQC compensation calculation module to finish the acquisition of the harmonic voltage signal and the harmonic current signal;
the UPQC compensation calculation module respectively sends the compensation signals of the voltage and the current into a voltage and current driving circuit, and controls a voltage inverter to output harmonic compensation voltage and controls a current inverter to output harmonic compensation current through the driving circuit; after the output of the two inverters is respectively filtered by the filtering module, harmonic compensation voltage is added with the power grid voltage by the series transformer, and harmonic compensation current is added with the power grid current by the parallel transformer.
The voltage and current compensation signal output by the UPQC compensation calculation module consists of two parts, wherein one part is a harmonic compensation value of the current beat obtained by adopting a traditional ip-iq algorithm, and the other part is a harmonic voltage and harmonic current lag error compensation signal calculated by a harmonic prediction neural network module;
the obtained traditional ip-iq algorithm of the harmonic compensation value of the current beat is that the acquired distorted load current is firstly subjected to Clark conversion, then the converted signal is subjected to Park conversion, then high-frequency harmonic waves are filtered through a low-pass filter, a direct-current component is obtained, then the direct-current component is subjected to coordinate inverse conversion to obtain fundamental wave current, and the difference value between the fundamental wave current and the acquired distorted load current is the current beat harmonic current compensation signal; the acquired distorted load voltage is subjected to Clark conversion firstly, then the converted signal is subjected to Park conversion, then high-frequency harmonic waves are filtered through a low-pass filter, a direct-current component is obtained, then the direct-current component is subjected to coordinate inverse conversion to obtain fundamental wave voltage, and the difference value between the fundamental wave voltage and the distorted load voltage is the current beat harmonic wave voltage compensation signal; the Clark transformation matrix is existing, and the Park transformation matrix is existing;
the harmonic prediction neural network module adopts a three-layer radial basis function neural network (RBF), the input layer is 3 x 2 x n signals, the input layer is respectively provided with three-phase voltage values of 1 to n beats and three-phase current values of 1 to n beats, the output layer is provided with six signals, and the output layer is respectively provided with three-phase harmonic voltage lag error compensation signals and three-phase harmonic current lag error compensation signals;
the harmonic prediction neural network training and checking method specifically comprises the following steps: firstly, applying given low-frequency harmonic operation to a power grid, taking N groups of three-phase voltage harmonic signals of 1 to N beats and three-phase current harmonic signals of 1 to N beats as input data of training of a radial base neural network, and taking the difference value between the corresponding harmonic signal of each beat and the harmonic signal of the previous beat as output data of training of the neural network. The training N sets of data are divided into two parts, the first part is trained and the second part is verified.
The neural network training has two stages, wherein the first stage adopts a K-means clustering algorithm: selecting the voltage value and the current value of the first three beats, respectively calculating Euclidean distance input to each clustering center, distributing the Euclidean distance to each clustering center according to the principle of nearest distance, calculating the average value of data in each clustering center, taking the average value as a new clustering center, redistributing input data, and repeating the processes until the position of the clustering center is not changed; the second stage uses a least squares method to determine the output weights.
The hidden layer of the harmonic prediction neural network module adopts a radial base vector, and the radial base vector is H= [ H ] 1 ,h 2 ,…,h m ] T Wherein h is m Is a Gaussian basis function:
the expression of the expression is European norms. The center vector of RBF network neuron is C= [ C ] 1 ,c 2 ,L,c m ] T . The base width vector of the RBF network is B= [ B ] 1 ,b 2 ,L,b m ] T 。
The weight vector of the RBF network is as follows: w= [ W ] 1 ,w 2 ,L,w m ] T . The output of the network at time k is: f (x) k )=WH=w 1 h 1 +w 2 h 2 +L+w m h m 。
In summary, the UPQC system based on time lag compensation has the advantages of high detection precision, no influence of distorted voltage on detection effect, capability of compensating time lag of signal extraction, and improvement of signal compensation instantaneity.
Drawings
FIG. 1 is a schematic diagram of a UPQC control system based on time lag compensation;
FIG. 2 is a schematic diagram of harmonic voltage detection of a UPQC control system based on time lag compensation;
FIG. 3 is a schematic diagram of harmonic current detection of a UPQC control system based on time lag compensation;
fig. 4 is a schematic diagram of harmonic voltage and harmonic current lag error prediction.
Detailed Description
The UPQC system based on time lag compensation comprises a power grid (1), and is characterized in that: the system further comprises a series transformer (2), a load (3), a first filtering module (4), a second filtering module (5), a voltage inverter (6), a current inverter (7), a harmonic voltage detection module (8), a UPQC compensation calculation module (9), a harmonic current detection module (10), a harmonic prediction neural network module (11), a three-phase voltage current beat (12), a three-phase voltage previous beat (13), a three-phase voltage previous n beat (14), a three-phase current beat (15), a three-phase current previous beat (16), a three-phase current previous n beat (17), a voltage driving circuit module (18), a current driving circuit module (19), a voltage sensor (20), a current sensor (21) and a parallel transformer (22);
the power grid (1) is connected to the series transformer (2), and the series transformer (2) is connected with the load (3) again to finish the power supply of the power grid (1) to the load (3);
the load (3) is connected with the harmonic current detection module (10) through the current sensor (21), and the harmonic current detection module (10) is connected to the harmonic prediction neural network module (11) and the UPQC compensation calculation module (9);
the harmonic prediction neural network module (11) is connected with the UPQC compensation calculation module (9), and the UPQC compensation calculation module (9) is connected with the first voltage driving circuit (18) and the second voltage driving circuit (19);
the voltage driving circuit (18) is connected with the voltage inverter (6), and the voltage inverter (6) is connected with the series transformer (2) through the first filtering module (4);
the current driving circuit module (19) is connected with the current inverter (7), the current inverter (7) is connected with the parallel transformer (22) through the second filtering module (5), and the parallel transformer (22) is connected with the load (3);
the power grid (1) is also connected with a voltage sensor (20), the voltage sensor (20) is connected with a harmonic voltage detection module (8), and the harmonic voltage detection module (8) is connected with a harmonic prediction neural network module (11) and a UPQC compensation calculation module (9).
The voltage sensor (20) receives a voltage signal of a power grid end and outputs the voltage signal to the harmonic voltage detection module (8);
the current sensor (21) receives a current signal at the end of the load (3) and outputs the current signal to the harmonic current detection module (10);
the harmonic prediction neural network module (11) extracts a current beat (12) of the three-phase voltage from the harmonic voltage detection module (8) to a lag error compensation signal of the harmonic voltage calculated by n beats (14) before the three-phase voltage, calculates a harmonic voltage signal through the UPQC compensation calculation module (9) and transmits the harmonic voltage signal to the voltage driving circuit (18);
the harmonic prediction neural network module (11) extracts a three-phase current beat (15) from the harmonic current detection module (10) to a three-phase current front n beat (17), and a corresponding hysteresis error compensation signal of the calculated harmonic current calculates a harmonic current signal through the UPQC compensation calculation module (9) and sends the harmonic current signal to the current drive circuit (19);
the UPQC compensation calculation module (9) receives the current beat (12) voltage detection signal output by the harmonic voltage detection module (8) and the current beat current detection signal output by the harmonic current detection module (10), calculates corresponding current beat voltage compensation signals and current beat current compensation signals, and sends the current beat voltage compensation signals and the current beat current compensation signals to the voltage driving circuit module (18) and the current driving circuit module (19) respectively; (the current beat voltage compensation signal is sent to the voltage driving circuit module (18), the current beat current compensation signal is sent to the current driving circuit module (19))
The voltage driving circuit module (18) receives the current beat voltage compensation signal and the harmonic voltage signal to generate a driving voltage signal, the driving voltage signal is sent to the voltage inverter (6), filtered by the first filtering module (4), sent to the series transformer (2) and then output to the load (3);
the second driving circuit module (19) receives the current beat current compensation signal and the harmonic current signal to generate a driving current signal, the driving current signal is sent to the current inverter (7), filtered by the second filtering module (5), and then sent to the parallel transformer (22) to be output to the load (3).
The harmonic prediction neural network module (11) includes a first harmonic prediction neural network module (88) and a second harmonic prediction neural network module (116)
The first harmonic prediction neural network module (88) receives the current beats of the three-phase a, b and c power grid voltages, the previous beats of the three-phase a, b and c power grid voltages until the previous n beats of the three-phase a, b and c power grid voltages, and the first harmonic prediction neural network module (88) calculates a, b and c three-phase harmonic voltage hysteresis error compensation signals;
the second harmonic prediction neural network module (116) receives the current beats of the a, b and c three-phase load currents, the previous beats of the a, b and c three-phase load currents until the previous n beats of the a, b and c three-phase load currents, and the harmonic prediction neural network module (116) calculates a, b and c three-phase harmonic current lag error compensation signals.
The control method implemented by the UPQC system based on time lag compensation comprises a UPQC control system harmonic voltage method and a harmonic current detection method based on time lag compensation;
the harmonic voltage method comprises the following steps: the participation module comprises: a first phase-locked loop, PLL, (81), a first triangular transformation module (82), a first Clark transformation module (83), a first Park transformation module (84), a first low pass filter (85), a first Park inverse transformation module (86), and a first Clark inverse transformation module (87); (these are circuit block components of UPQC compensation calculation block (9))
The first phase-locked loop PLL (81) receives a-phase power grid voltage and extracts a rotation angle of the voltage for the next various coordinate transformations;
the first triangular transformation module (82) transforms the rotation angle extracted by the first phase-locked loop (81) into a trigonometric function value for the first Park transformation module (84) and the first Park inverse transformation module (86);
the harmonic current detection method comprises the following steps: the participation module comprises: comprises a second phase-locked loop PLL (119), a second triangular transformation module (110), a second Clark transformation module (111), a second Park transformation module (112), a second low-pass filter (113), a second Park inverse transformation module (114) and a second Clark inverse transformation module (115);
the second phase-locked loop PLL (119) receives the a-phase grid current and extracts the rotation angle of the current for the next various coordinate transformations;
the second triangular transformation module (110) transforms the rotation angle extracted by the second phase-locked loop (PLL) (119) into a trigonometric function value for the second Park transformation module (112) and the second Park inverse transformation module (114).
The harmonic voltage detection method of the UPQC control system based on time lag compensation comprises the following steps: distorted mains voltage u a 、u b 、u c Transformed by a first Clark transformation module (83) into u in an alpha beta coordinate system α 、u β The corresponding transformation matrix is present; u (u) α 、u β Then transformed to u under dq coordinate system by a first Park transformation module (84) p 、u q The corresponding transformation matrix is present; u (u) p 、u q Obtaining a direct current component by a first low pass filter (85)Then the fundamental wave active voltage u is obtained through a first Park inverse conversion module (86) af 、u bf 、u cf The method comprises the steps of carrying out a first treatment on the surface of the Three-phase network terminal voltage u a 、u b 、u c And fundamental wave active voltage u af 、u bf 、u cf The difference is made to obtain a command voltage signal u ah 、u bh 、u ch The method comprises the steps of carrying out a first treatment on the surface of the The first harmonic prediction neural network module (88) sends out a harmonic voltage hysteresis error compensation signal u aph 、u bph 、u cph Compensation signal u with the current beat voltage ah 、u bh 、u ch The sum is the harmonic voltage signal +.>
The harmonic current detection method of the UPQC control system based on time lag compensation is as follows: three-phase load side current i La 、i Lb 、i Lc Transformed into i in the alpha beta coordinate system by a second Clark transformation module (111) α 、i β The corresponding transformation matrix is present; i.e α 、i β Then the i is transformed to the dq coordinate system by a second Park transformation module (112) p 、i q The corresponding transformation matrix is present; i.e p 、i q Obtaining a DC component by a second low-pass filter (113)Then the fundamental wave active current i is obtained through a second Park inverse conversion module (114) af 、i bf 、i cf The method comprises the steps of carrying out a first treatment on the surface of the Distorted load current i La 、i Lb 、i Lc And then the fundamental wave active current i af 、i bf 、i cf The difference is made to obtain a compensation signal i of the current beat current ah 、i bh 、i ch The method comprises the steps of carrying out a first treatment on the surface of the The second harmonic prediction neural network module (116) sends out a harmonic current lag error compensation signal i aph 、i bph 、i cph Compensating signal i with current ah 、i bh 、i ch The sum is the harmonic current signal +.>
The harmonic prediction neural network module is divided into 3 layers: an input layer (201), an hidden layer (202) and an output layer (203); wherein the input layer (201) contains 3 x 2 x n input signals, which are respectively the current beat, the previous beat and the previous n beats of a, b and c three-phase power grid voltages and a, b and c three-phase load currents; the output layer (203) contains 6 signals, respectively a, b, c three-phase harmonicsA voltage hysteresis error compensation signal and a harmonic current hysteresis error compensation signal; the hidden layer (202) of the harmonic prediction neural network module adopts a radial base vector, and the radial base vector is H= [ H ] 1 ,h 2 ,…,h m ] T Wherein h is m Is a Gaussian basis function:
the I & is an European norm; the central vector of the neural element of the harmonic prediction neural network module is C= [ C ] 1 ,c 2 ,L,c m ] T The method comprises the steps of carrying out a first treatment on the surface of the The base width vector of the harmonic prediction neural network module is B= [ B ] 1 ,b 2 ,L,b m ] T The method comprises the steps of carrying out a first treatment on the surface of the The weight vector of the harmonic prediction neural network module is as follows: w= [ W ] 1 ,w 2 ,L,w m ] T The method comprises the steps of carrying out a first treatment on the surface of the The output of the network at time k is: f (x) k )=WH=w 1 h 1 +w 2 h 2 +L+w m h m 。
Training of the first harmonic prediction neural network module (88) and the second harmonic prediction neural network module (116) is divided into two phases, wherein the first phase adopts a K-means clustering algorithm: selecting the voltage value and the current value of the first three beats, respectively calculating Euclidean distance input to each clustering center, distributing the Euclidean distance to each clustering center according to the principle of nearest distance, calculating the average value of data in each clustering center, taking the average value as a new clustering center, redistributing input data, and repeating the processes until the position of the clustering center is not changed; and in the second stage, the output weight is determined by adopting a least square method.
Claims (7)
1. The UPQC system based on time lag compensation comprises a power grid (1), and is characterized in that: the system further comprises a series transformer (2), a load (3), a first filtering module (4), a second filtering module (5), a voltage inverter (6), a current inverter (7), a harmonic voltage detection module (8), a UPQC compensation calculation module (9), a harmonic current detection module (10), a harmonic prediction neural network module (11), a voltage driving circuit module (18), a current driving circuit module (19), a voltage sensor (20), a current sensor (21) and a parallel transformer (22);
the power grid (1) is connected to the series transformer (2), and the series transformer (2) is connected with the load (3);
the load (3) is connected with the harmonic current detection module (10) through the current sensor (21), and the harmonic current detection module (10) is connected to the harmonic prediction neural network module (11) and the UPQC compensation calculation module (9);
the harmonic prediction neural network module (11) is connected with the UPQC compensation calculation module (9), and the UPQC compensation calculation module (9) is connected with the first voltage driving circuit (18) and the second voltage driving circuit (19);
the voltage driving circuit (18) is connected with the voltage inverter (6), and the voltage inverter (6) is connected with the series transformer (2) through the first filtering module (4);
the current driving circuit module (19) is connected with the current inverter (7), the current inverter (7) is connected with the parallel transformer (22) through the second filtering module (5), and the parallel transformer (22) is connected with the load (3);
the power grid (1) is also connected with a voltage sensor (20), the voltage sensor (20) is connected with a harmonic voltage detection module (8), and the harmonic voltage detection module (8) is connected with a harmonic prediction neural network module (11) and a UPQC compensation calculation module (9);
the voltage sensor (20) receives a voltage signal of a power grid end and outputs the voltage signal to the harmonic voltage detection module (8);
the current sensor (21) receives a current signal at the end of the load (3) and outputs the current signal to the harmonic current detection module (10);
the harmonic prediction neural network module (11) extracts a current beat (12) of the three-phase voltage from the harmonic voltage detection module (8) to a lag error compensation signal of the harmonic voltage calculated by n beats (14) before the three-phase voltage, calculates a harmonic voltage signal through the UPQC compensation calculation module (9) and transmits the harmonic voltage signal to the voltage driving circuit (18);
the harmonic prediction neural network module (11) extracts a three-phase current beat (15) from the harmonic current detection module (10) to a three-phase current front n beat (17), and a corresponding hysteresis error compensation signal of the calculated harmonic current calculates a harmonic current signal through the UPQC compensation calculation module (9) and sends the harmonic current signal to the current drive circuit (19);
the UPQC compensation calculation module (9) receives the current beat (12) voltage detection signal output by the harmonic voltage detection module (8) and the current beat current detection signal output by the harmonic current detection module (10), calculates corresponding current beat voltage compensation signals and current beat current compensation signals, and sends the current beat voltage compensation signals and the current beat current compensation signals to the voltage driving circuit module (18) and the current driving circuit module (19) respectively;
the voltage driving circuit module (18) receives the current beat voltage compensation signal and the harmonic voltage signal to generate a driving voltage signal, the driving voltage signal is sent to the voltage inverter (6), filtered by the first filtering module (4), sent to the series transformer (2) and then output to the load (3);
the current driving circuit module (19) receives the current beat current compensation signal and the harmonic current signal to generate a driving current signal, the driving current signal is sent to the current inverter (7), filtered by the second filtering module (5), and then sent to the parallel transformer (22) to be output to the load (3).
2. A UPQC system as in claim 1 wherein:
the harmonic prediction neural network module (11) includes a first harmonic prediction neural network module (88) and a second harmonic prediction neural network module (116)
The first harmonic prediction neural network module (88) receives the current beats of the three-phase a, b and c power grid voltages, the previous beats of the three-phase a, b and c power grid voltages until the previous n beats of the three-phase a, b and c power grid voltages, and the first harmonic prediction neural network module (88) calculates a, b and c three-phase harmonic voltage hysteresis error compensation signals;
the second harmonic prediction neural network module (116) receives the current beats of the a, b and c three-phase load currents, the previous beats of the a, b and c three-phase load currents until the previous n beats of the a, b and c three-phase load currents, and the harmonic prediction neural network module (116) calculates a, b and c three-phase harmonic current lag error compensation signals.
3. A control method implemented by a UPQC system based on time lag compensation as in claim 1 wherein: the method comprises a UPQC control system harmonic voltage method and a harmonic current detection method based on time lag compensation;
the harmonic voltage method comprises the following steps: the participation module comprises: a first phase-locked loop, PLL, (81), a first triangular transformation module (82), a first Clark transformation module (83), a first Park transformation module (84), a first low pass filter (85), a first Park inverse transformation module (86), and a first Clark inverse transformation module (87);
the first phase-locked loop PLL (81) receives a-phase power grid voltage and extracts a rotation angle of the voltage;
the first triangular transformation module (82) transforms the rotation angle extracted by the first phase-locked loop (81) into a trigonometric function value for the first Park transformation module (84) and the first Park inverse transformation module (86);
the harmonic current detection method comprises the following steps: the participation module comprises: comprises a second phase-locked loop PLL (119), a second triangular transformation module (110), a second Clark transformation module (111), a second Park transformation module (112), a second low-pass filter (113), a second Park inverse transformation module (114) and a second Clark inverse transformation module (115);
the second phase-locked loop PLL (119) receives a-phase grid current and extracts a rotation angle of the current;
the second triangular transformation module (110) transforms the rotation angle extracted by the second phase-locked loop (PLL) (119) into a trigonometric function value for the second Park transformation module (112) and the second Park inverse transformation module (114).
4. A control method implemented by a UPQC system based on time lag compensation as claimed in claim 3 wherein:
the harmonic voltage detection method of the UPQC control system based on time lag compensation comprises the following steps: distorted mains voltage u a 、u b 、u c Transformed by a first Clark transformation module (83) into u in an alpha beta coordinate system α 、u β ;u α 、u β Then transformed to u under dq coordinate system by a first Park transformation module (84) p 、u q ;u p 、u q Through the firstA low pass filter (85) for obtaining the DC componentThen the fundamental wave active voltage u is obtained through a first Park inverse conversion module (86) af 、u bf 、u cf The method comprises the steps of carrying out a first treatment on the surface of the Three-phase network terminal voltage u a 、u b 、u c And fundamental wave active voltage u af 、u bf 、u cf The difference is made to obtain a command voltage signal u ah 、u bh 、u ch The method comprises the steps of carrying out a first treatment on the surface of the The first harmonic prediction neural network module (88) sends out a harmonic voltage hysteresis error compensation signal u aph 、u bph 、u cph Compensation signal u with the current beat voltage ah 、u bh 、u ch The sum is the harmonic voltage signal +.>
5. A control method implemented by a UPQC system based on time lag compensation as claimed in claim 3 wherein:
the harmonic current detection method of the UPQC control system based on time lag compensation is as follows: three-phase load side current i La 、i Lb 、i Lc Transformed into i in the alpha beta coordinate system by a second Clark transformation module (111) α 、i β ;i α 、i β Then the i is transformed to the dq coordinate system by a second Park transformation module (112) p 、i q ;i p 、i q Obtaining a DC component by a second low-pass filter (113)Then the fundamental wave active current i is obtained through a second Park inverse conversion module (114) af 、i bf 、i cf The method comprises the steps of carrying out a first treatment on the surface of the Distorted load current i La 、i Lb 、i Lc And then the fundamental wave active current i af 、i bf 、i cf The difference is made to obtain a compensation signal i of the current beat current ah 、i bh 、i ch The method comprises the steps of carrying out a first treatment on the surface of the The second harmonic prediction neural network module (116) sends out a harmonic current lag error compensation signal i aph 、i bph 、i cph Compensating signal i with current ah 、i bh 、i ch The sum is the harmonic current signal +.>
6. The control method according to claim 5, characterized in that: the harmonic prediction neural network module is divided into 3 layers: an input layer (201), an hidden layer (202) and an output layer (203); wherein the input layer (201) contains 3 x 2 x n input signals, which are respectively the current beat, the previous beat and the previous n beats of a, b and c three-phase power grid voltages and a, b and c three-phase load currents; the output layer (203) comprises 6 signals, namely an a, b and c three-phase harmonic voltage hysteresis error compensation signal and a harmonic current hysteresis error compensation signal; the hidden layer (202) of the harmonic prediction neural network module adopts a radial base vector, and the radial base vector is H= [ H ] 1 ,h 2 ,…,h m ] T Wherein h is m Is a Gaussian basis function:
the I & is an European norm; the central vector of the neural element of the harmonic prediction neural network module is C= [ C1, C 2 ,L,c m ] T The method comprises the steps of carrying out a first treatment on the surface of the The base width vector of the harmonic prediction neural network module is B= [ B ] 1 ,b 2 ,L,b m ] T The method comprises the steps of carrying out a first treatment on the surface of the The weight vector of the harmonic prediction neural network module is as follows: w= [ W ] 1 ,w 2 ,L,w m ] T The method comprises the steps of carrying out a first treatment on the surface of the The output of the network at time k is: f (x) k )=WH=w 1 h 1 +w 2 h 2 +L+w m h m 。
7. The control method according to claim 4, characterized in that: training of the first harmonic prediction neural network module (88) and the second harmonic prediction neural network module (116) is divided into two phases, wherein the first phase adopts a K-means clustering algorithm: selecting the voltage value and the current value of the first three beats, respectively calculating Euclidean distance input to each clustering center, distributing the Euclidean distance to each clustering center according to the principle of nearest distance, calculating the average value of data in each clustering center, taking the average value as a new clustering center, redistributing input data, and repeating the processes until the position of the clustering center is not changed; and in the second stage, the output weight is determined by adopting a least square method.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011056002 | 2020-09-30 | ||
CN2020110560024 | 2020-09-30 |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112398131A CN112398131A (en) | 2021-02-23 |
CN112398131B true CN112398131B (en) | 2023-08-04 |
Family
ID=74600960
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011285909.8A Active CN112398131B (en) | 2020-09-30 | 2020-11-17 | UPQC control system based on time lag compensation and control method thereof |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112398131B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104836232A (en) * | 2015-05-14 | 2015-08-12 | 电子科技大学 | Frequency wide-range self-adaptive repetition control method of active power filter |
CN106451466A (en) * | 2016-11-01 | 2017-02-22 | 辽宁工程技术大学 | Grid power quality control system and method based on unified power quality conditioner |
CN106849082A (en) * | 2017-03-10 | 2017-06-13 | 国网江苏省电力公司常州供电公司 | Research on Unified Power Quality Conditioner Harmonic Control Method containing photovoltaic power distribution network |
-
2020
- 2020-11-17 CN CN202011285909.8A patent/CN112398131B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104836232A (en) * | 2015-05-14 | 2015-08-12 | 电子科技大学 | Frequency wide-range self-adaptive repetition control method of active power filter |
CN106451466A (en) * | 2016-11-01 | 2017-02-22 | 辽宁工程技术大学 | Grid power quality control system and method based on unified power quality conditioner |
CN106849082A (en) * | 2017-03-10 | 2017-06-13 | 国网江苏省电力公司常州供电公司 | Research on Unified Power Quality Conditioner Harmonic Control Method containing photovoltaic power distribution network |
Non-Patent Citations (1)
Title |
---|
统一电能质量控制器的建模与仿真;任永峰 等;《电力***及其自动化学报》;第21卷(第04期);30-35 * |
Also Published As
Publication number | Publication date |
---|---|
CN112398131A (en) | 2021-02-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107294527B (en) | Synchronous rotating coordinate system phase-locked loop and testing method and device thereof | |
CN110460112B (en) | Micro-grid sag control method based on power deviation value consistency control | |
CN110838731B (en) | Three-phase four-bridge arm photovoltaic grid-connected inverter and multi-target cooperative fault control method | |
CN104578182B (en) | A kind of sagging multiple feedback loop method of low delay robust power | |
CN109347147B (en) | quasi-PR grid-connected inverter optimization control method for self-adaptive harmonic elimination | |
CN105743091B (en) | A kind of double close-loop decoupling control method of Active Power Filter-APF | |
CN112803461B (en) | Active disturbance rejection control method of battery energy storage converter based on extended state observer | |
CN110739678A (en) | Control method for series virtual impedance of grid-connected converters | |
CN108233394A (en) | A kind of capacitive coupling voltage balancing control method suitable for Y type chain types STATCOM | |
CN106410858A (en) | Software digital phase-locking method based on dual dq coordination conversion | |
CN110943632A (en) | Energy storage converter virtual inertia control method based on cascade generalized integrator | |
CN115954877A (en) | Control method and system for railway traction network and railway power supply and distribution network interconnection converter | |
CN110401195B (en) | Closed loop compensation method, system and medium based on DFT rotation transformation | |
CN112346439A (en) | Micro-grid sensor fault tolerance control method based on PD type learning observer | |
CN112398131B (en) | UPQC control system based on time lag compensation and control method thereof | |
CN113691155A (en) | Single-phase inverter control method and system based on PI and repetitive control and inverter | |
CN107123991B (en) | Dynamic voltage restorer control method and device based on fuzzy particle swarm algorithm | |
CN109510548B (en) | Double-fed motor flexible power control method and device | |
CN116979535A (en) | Dual harmonic current detection method for active power filter | |
Kunjumuhammed et al. | Comparison of single phase shunt active power filter algorithms | |
CN109962480A (en) | Static reactive generator no phase-locked loop control system, method and application | |
CN115986745A (en) | Harmonic voltage compensation method for intelligent power distribution network | |
CN110048466B (en) | Prediction control method for grid-side converter model of double-fed wind power generation system | |
CN115276439A (en) | LCL type grid-connected inverter resonance suppression method adapting to impedance change of weak grid | |
CN114744626A (en) | On-grid/off-grid seamless switching control method and device for renewable energy hydrogen production system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20240311 Address after: 230000 B-2704, wo Yuan Garden, 81 Ganquan Road, Shushan District, Hefei, Anhui. Patentee after: HEFEI LONGZHI ELECTROMECHANICAL TECHNOLOGY Co.,Ltd. Country or region after: China Address before: No.111 Shenyang West Liaoning Economic Development Zone Patentee before: SHENYANG University OF TECHNOLOGY Country or region before: China |