CN113625054A - Power grid impedance detection device, inverter and power grid impedance detection method - Google Patents
Power grid impedance detection device, inverter and power grid impedance detection method Download PDFInfo
- Publication number
- CN113625054A CN113625054A CN202110958436.1A CN202110958436A CN113625054A CN 113625054 A CN113625054 A CN 113625054A CN 202110958436 A CN202110958436 A CN 202110958436A CN 113625054 A CN113625054 A CN 113625054A
- Authority
- CN
- China
- Prior art keywords
- power grid
- neural network
- grid impedance
- network model
- inverter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 96
- 238000003062 neural network model Methods 0.000 claims abstract description 77
- 238000005070 sampling Methods 0.000 claims abstract description 28
- 230000008878 coupling Effects 0.000 claims abstract description 24
- 238000010168 coupling process Methods 0.000 claims abstract description 24
- 238000005859 coupling reaction Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims description 63
- 238000012360 testing method Methods 0.000 claims description 26
- 238000000034 method Methods 0.000 claims description 20
- 238000012795 verification Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 10
- 230000003044 adaptive effect Effects 0.000 claims description 9
- 238000002474 experimental method Methods 0.000 claims description 9
- 238000013135 deep learning Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 230000001960 triggered effect Effects 0.000 claims description 3
- 238000013480 data collection Methods 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 description 12
- 238000011176 pooling Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 8
- 238000013527 convolutional neural network Methods 0.000 description 6
- 238000013473 artificial intelligence Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 3
- 230000009471 action Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 229910052799 carbon Inorganic materials 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000004146 energy storage Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000035699 permeability Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006386 neutralization reaction Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R27/00—Arrangements for measuring resistance, reactance, impedance, or electric characteristics derived therefrom
- G01R27/02—Measuring real or complex resistance, reactance, impedance, or other two-pole characteristics derived therefrom, e.g. time constant
- G01R27/16—Measuring impedance of element or network through which a current is passing from another source, e.g. cable, power line
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Inverter Devices (AREA)
Abstract
The invention discloses a power grid impedance detection device, an inverter and a power grid impedance detection method, which are used for detecting the power grid impedance at a public coupling point of the inverter connected to a power grid, wherein the detection device comprises the following modules: the sampling module is used for sampling three-phase voltage signals and three-phase current signals at the common coupling point; and the intelligent chip is used for receiving the three-phase voltage signals and the three-phase current signals sampled by the sampling module and inputting the three-phase voltage signals and the three-phase current signals into an internal power grid impedance detection neural network model, and the power grid impedance detection neural network model is pre-established and trained so as to output a predicted value of power grid impedance according to the input three-phase voltage signals and three-phase current signals. The power grid impedance detection device disclosed by the invention can be used for accurately predicting the power grid impedance in real time through the deep neural network model without injecting harmonic current into a public coupling point, so that the self-adaptive control of an inverter is realized.
Description
Technical Field
The invention relates to the field of power electronics, in particular to a power grid impedance detection device, an inverter and a power grid impedance detection method.
Background
Under the targets of 'carbon neutralization' and 'carbon peak-to-peak' in all countries in the world, the installed increment of new energy including photovoltaic, wind power and energy storage is greatly increased year by year, but the continuous improvement of the permeability of the new energy reduces the carbon emission and simultaneously causes the reduction of the strength of a power grid, and the power grid with lower strength is generally called as a weak power grid.
The weak power grid can cause instability of the new energy inverter, and great challenges are brought to safe and stable operation of the power system. In order to overcome the problems caused by a weak power grid, a feasible method is to detect the power grid impedance value at a Point of Common Coupling (PCC) where the inverter is connected to the power grid in real time, and adjust the control parameters of the inverter according to a power grid impedance adaptive algorithm, so that the stability of the inverter can be greatly improved.
Currently, commonly used power grid impedance detection algorithms are mainly divided into two categories, namely active detection algorithms and passive detection algorithms. The active detection algorithm utilizes the inverter to inject harmonic current into the PCC, and calculates the impedance value of the power grid according to the collected harmonic voltage and current of the PCC; the passive detection algorithm estimates the impedance value of the power grid in real time according to the self voltage and current variables of the PCC, does not need to inject harmonic current, and has high popularization value. However, at present, passive detection algorithms generally have at least the following defects:
firstly, voltage and current signals contained in the PCC are extremely complex and change along with the change of states of an inverter, a power supply, a load and the like, so that a calculation model of the detection method is complex and the detection precision is not high;
secondly, the calculated amount of the passive detection algorithm is usually large, the requirement on the controller is high, the passive detection algorithm cannot be embedded into a CPU of the existing inverter, and the passive detection algorithm is not beneficial to practical engineering application;
thirdly, compared with the traditional passive algorithm, the existing artificial neural network method can improve the accuracy of power grid impedance detection, the patent CN110247427A proposes that BP artificial neural network is adopted to identify the power grid impedance, but the artificial neural network with four nodes at each layer of two hidden layers is adopted in the patent, the identification precision is low, and the robustness of identified power grid parameters is poor by using instantaneous values as input quantities.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a power grid impedance detection device, an inverter and a power grid impedance detection method.
The invention provides a power grid impedance detection algorithm based on artificial intelligence, which adopts an artificial neural network method, can realize a complex artificial neural network algorithm such as a deep neural network due to the addition of a special intelligent chip, and adopts a deep learning algorithm to improve the accuracy and the robustness of detection. Compared with the traditional machine learning algorithm, the deep neural network algorithm does not need to artificially design a feature extractor, is obtained by network self-learning, is particularly suitable for data with complex changes, and has excellent generalization capability and robustness. Due to the excellent characteristics, the power grid impedance can be accurately and stably detected (predicted) by adopting the deep neural network, and the specific technical scheme is as follows:
in one aspect, the present invention provides a power grid impedance detection apparatus, configured to detect a power grid impedance at a point of a public coupling where an inverter is connected to a power grid, without injecting a harmonic current to the point of the public coupling, where the detection apparatus includes the following modules:
the sampling module is used for sampling the three-phase voltage signals and the three-phase current signals at the common coupling point;
and the intelligent chip is used for receiving the three-phase voltage signals and the three-phase current signals sampled by the sampling module and inputting the three-phase voltage signals and the three-phase current signals into an internal power grid impedance detection neural network model, and the power grid impedance detection neural network model is pre-established and trained so as to output a predicted value of power grid impedance according to the input three-phase voltage signals and three-phase current signals.
Further, a training sample for training the grid impedance detection neural network model and a test sample for verifying the grid impedance detection neural network model are obtained through the following steps:
s1, constructing an inverter experiment platform, and simulating the impedance of a power grid by using a resistor and a reactor;
s2, collecting data: adjusting at least one of the resistor and the reactor to obtain different impedance combinations, correspondingly sampling three-phase voltage and three-phase current data at a common coupling point to obtain a plurality of samples, and taking the resistance value of the resistor and the inductance value of the reactor corresponding to the same moment as labels corresponding to the samples, or taking the total impedance value of the impedance combinations corresponding to the same moment as labels corresponding to the samples;
and S3, respectively obtaining a training sample and a test sample from the samples.
Further, step S3 includes:
s31, collecting the sample data collected in the step S2 together, firstly carrying out normalization processing on the collected data according to the rated voltage and the rated current of the inverter, and carrying out digital filtering processing on the voltage signal and the current signal after the normalization processing;
and S32, obtaining training samples and test samples from the filtered data according to a preset rule.
Further, step S1 includes:
s11, selecting an inverter which is the same as the inverter in the actual detection working condition to build an experiment platform;
s12, determining the impedance value range of the simulated power grid impedance according to the rated power of the inverter;
s13, selecting a resistor in an adaptive resistance value range and an electric reactor in an adaptive inductance value range according to the determined impedance value range of the simulated power grid impedance;
and S14, setting different combinations of the resistance values of the resistors and the inductance values of the reactors according to a time sequence, and configuring different working condition parameters under the same or different simulated power grid impedance values.
Further, after the power grid impedance detection neural network model is obtained through training, the power grid impedance detection neural network model is downloaded into the intelligent chip; the power grid impedance detection neural network model is obtained through the following steps:
e1, building a neural network model by using a deep learning platform;
e2, training the neural network model by using the training sample, wherein the input quantity of the neural network model is three-phase voltage and three-phase current data in the training sample, and the output quantity of the neural network model is three-phase equivalent resistance and inductance of the power grid;
e3, verifying the neural network model by using the test sample, if the verification result does not meet the preset requirement, performing iterative training on the neural network model by using the training sample until the trained neural network model passes the verification, and taking the current neural network model as the power grid impedance detection neural network model.
The intelligent power supply system further comprises a CPU module and a PWM module, wherein the CPU module is electrically connected with the intelligent chip and the PWM module, and the PWM module is used for being electrically connected with the inverter;
the sampling module is used for acquiring and sending the three-phase voltage signals and the three-phase current signals to the intelligent chip in real time, the intelligent chip outputs an estimated power grid impedance value to the CPU module in real time, and the CPU module is used for generating a control instruction according to the current power grid impedance value and sending the control instruction to the PWM module;
and the PWM module is used for outputting an adaptive pulse signal according to the control instruction so as to control the running state of the inverter in real time.
Further, the inverter is provided with an anti-islanding device, and if the inverter determines that the islanding effect occurs according to the predicted value of the grid impedance, the anti-islanding device is triggered.
On the other hand, the invention provides an intelligent detection method for the power grid impedance, which does not need to inject harmonic current into a public coupling point of an inverter connected into a power grid, and comprises the following steps:
collecting three-phase voltage signals and three-phase current signals at a public coupling point of an inverter connected to a power grid;
inputting the data acquisition results of the three-phase voltage and the three-phase current to a pre-generated power grid impedance detection neural network model, and outputting corresponding prediction results of the power grid impedance by the power grid impedance detection neural network model; the power grid impedance detection neural network model is pre-generated through the following steps:
g1, data collection: building an inverter experiment platform, simulating the impedance of a power grid by using a resistor and a reactor, adjusting the resistance value of the resistor and/or the inductance value of the reactor, sampling three-phase voltage signals and three-phase current signals of a plurality of voltage periods as samples, and taking the resistance value of the resistor and the inductance value of the reactor corresponding to the same moment as labels corresponding to the samples;
g2, preprocessing data: collecting the sample data sampled in the step G1, and preprocessing the sample data;
g3, model training: building a neural network model, and learning by using the training sample obtained by preprocessing in the step G2 and the corresponding label thereof to obtain the trained neural network model;
g4, model verification: verifying the neural network model trained in the step G3 by using the test sample preprocessed in the step G2, and if the verification result meets the preset requirement, taking the current neural network model as the power grid impedance detection neural network model; and if the verification result does not meet the preset requirement, returning to execute the steps G3-G4.
Further, step G4 includes:
inputting a test sample into a neural network model to be verified to obtain a corresponding estimation value;
and if the root mean square value of the error between each estimated value and the corresponding label actual value of the corresponding test sample is less than 5%, judging that the verification result meets the preset requirement.
Further, in step G1, the resistance value of the resistor and/or the inductance of the reactor is adjusted, and the output power of the inverter and/or the grid voltage is adjusted to obtain sample data under different operating conditions.
And further, outputting a prediction result of the power grid impedance in response to the power grid impedance detection neural network model, and adjusting the operation parameters of the inverter by using the prediction result of the power grid impedance.
The technical scheme provided by the invention has the following beneficial effects:
a. the power grid impedance value of the inverter accessed to the power grid can be accurately predicted by running an artificial neural network algorithm based on the intelligent chip, harmonic current does not need to be injected into a public coupling point of the inverter accessed to the power grid, and the current quality is not reduced;
b. the grid impedance of the inverter connected to the PCC of the grid under the actual working condition can be detected in real time;
c. the method has the advantages that the power grid impedance is detected in real time in the actual working condition, the working stability of the inverter can be judged, and then the control parameters are improved through the self-adaptive algorithm, so that the stability of the inverter is improved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a schematic structural diagram of a power grid impedance detection apparatus provided in an embodiment of the present invention;
fig. 2 is a schematic flow chart of an intelligent detection method for the grid impedance according to an embodiment of the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, apparatus, article, or device that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or device.
With the improvement of the permeability of new energy, the power grid environment for the operation of the photovoltaic/energy storage inverter is increasingly complex, the intelligent requirement on the inverter is increased day by day, and the intelligent requirement on the inverter cannot be realized only by using the existing control chip. On this background, the present invention proposes an engineered, real-time, high-precision grid impedance detection technique for photovoltaic/energy-storage inverters.
In an embodiment of the present invention, a grid impedance detection apparatus is provided for detecting a grid impedance at a point of common coupling of an inverter into a grid, the detection apparatus does not need to inject a harmonic current into the point of common coupling, and referring to fig. 1, the grid impedance detection apparatus includes the following modules:
the sampling module is used for sampling the three-phase voltage signals and the three-phase current signals at the common coupling point;
the intelligent chip is used for receiving the three-phase voltage signals and the three-phase current signals sampled by the sampling module and inputting the three-phase voltage signals and the three-phase current signals into an internal power grid impedance detection neural network model, and the power grid impedance detection neural network model is pre-established and completes training so as to output a predicted value of power grid impedance according to the input three-phase voltage signals and three-phase current signals;
the CPU module is electrically connected with the intelligent chip and used for receiving the predicted value of the power grid impedance and generating a control instruction according to the current power grid impedance value;
and the input end of the PWM module is electrically connected with the CPU module, the output end of the PWM module is electrically connected with the inverter, and the PWM module is used for receiving the control instruction and outputting an adaptive pulse signal according to the control instruction so as to control the running state of the inverter in real time.
In addition, the detection device further comprises a communication module and a storage module, wherein the communication module is used for realizing communication between a device internal CPU and device external equipment (such as a controller or a mobile terminal), and the storage module is used for storing information data (such as processing information of the CPU) in the detection device.
The application scenario of the power grid impedance detection device is as shown in fig. 1: a public coupling Point (PCC) is arranged between the inverter and the power grid, and a sampling module of the detection device respectively samples three-phase voltage signals and three-phase current signals at the PCC as input parameters of a neural network model in the intelligent chip; because the sampling module can acquire and send the sampling data of the three-phase voltage signals and the three-phase current signals to the intelligent chip in real time, the intelligent chip can output the estimated power grid impedance value to the CPU module in real time, and finally, the real-time adaptive control of the inverter is realized, so that the stability of the inverter is improved, and the power supply quality of a power grid is improved.
An artificial intelligence chip is additionally arranged outside the existing inverter control chip (CPU module), can be a TPU, ARM, STM32 series chip, FPGA or other chips suitable for improving an intelligent algorithm, and is specially used for operating the artificial intelligence algorithm, so that the operating speed of the algorithm is improved, and the real-time performance of the algorithm is ensured. The intelligent chip is used for power grid impedance detection (prediction), but other artificial intelligence algorithms are not excluded from running in other scenes.
The invention utilizes the following working principle to detect the impedance of the power grid: the sampling module samples three-phase voltage v at PCC in real timeabcAnd three-phase current iabcThe sampling value is input into a CPU and an intelligent chip, and the intelligent chip estimates the power grid impedance in real time through an intelligent impedance detection algorithm, wherein the algorithm can be an artificial neural network algorithm or other intelligent algorithms. The intelligent chip inputs the estimated power grid impedance value into the CPU, the CPU uses the received power grid impedance value in a control program of the inverter and outputs a control signal to the PWM module, and the PWM module outputs a pulse signal to control the operation of the inverter.
According to the embodiment of the invention, three-phase current and three-phase voltage at the PCC are used as input parameters of the neural network model, and power grid impedance (or resistance value of a resistor and inductance value of an electric reactor) is used as output parameters of the neural network model. If the resistance value of the resistor and the inductance of the reactor are used as output parameters of the neural network model, the CPU can be used for converting the resistance value of the resistor and the inductance of the reactor into the total impedance value of the resistor and the reactor through a formula according to the resistance value of the resistor and the inductance of the reactor of the neural network model.
After determining the input parameter type and the output parameter type, establishing a neural network model by the following steps:
e1, building a convolutional neural network model by using a deep learning platform, configuring a network structure to comprise a convolutional layer, a pooling layer and a full-link layer, and carrying out network definition on the network structure.
Specifically, in this embodiment, a Convolutional Neural Network (CNN) may be built using a propeller deep learning platform, and a network structure is configured to include a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, and a full-link layer, where a convolutional kernel of the first convolutional layer is defined to be 10, the number of the convolutional layers is defined to be 5, and an activation function is relu; defining the size of a pooling core of the first pooling layer to be 2, the pooling type to be max and the pooling step length to be 2; defining the convolution kernel size of the second convolution layer to be 25, the number to be 5 and the activation function to be relu; defining the size of a pooling core of the second pooling layer to be 2, the pooling type to be max and the pooling step length to be 2; and defining the size of the full connection layer as 10 and the activation function as sigmoid.
Further, the network defining operation further comprises:
selecting the loss function to be any one of:
root mean square error loss function
Or
log-cosh loss function
Or
Mean square error loss function
Or
Mean absolute value error loss function
Wherein,
the loss value verified after each training, N is the number of test samples verified at each time,
is the predicted value of the i-th test sample, yiThe label value corresponding to the ith test sample;
defining an optimization method of the convolutional neural network model as Adam, AdaMax or Nadam;
the learning rate is defined in the range of 0.0001 to 0.002, preferably 0.001.
E2, configuring a training environment and a creating actuator (Executor) for the constructed convolutional neural network, and defining the dimensionality of the network output data according to the dimensionality of the training sample.
Specifically, after the CNN network is built, a training environment is configured and an execution is created, then the dimension of input data is defined to be equal to the dimension of a sample, and a training round is set.
And the actuator trains the neural network model by using training samples, wherein each training sample has a corresponding label, and the training is finished (completed) until the loss value and/or the accuracy of the trained neural network model meet requirements, namely the currently trained neural network model is used as the power grid impedance detection neural network model and deployed into an intelligent chip.
Referring to fig. 2, a training sample required for training a neural network model and a test sample for verifying the training sample are obtained through the following steps in one embodiment of the present invention:
and S1, constructing an inverter experiment platform, and simulating the impedance of the power grid by using a resistor and a reactor.
Selecting an inverter which is the same as the inverter in an actual detection working condition to build an experimental platform, on the other hand, taking an equivalent model of the power grid impedance as an example, wherein the equivalent model comprises resistance and inductance, simulating the power grid impedance by using a resistor and a reactor, the resolution of the resistor is 0.1 omega, the resolution of the reactor is 0.1mH, the selection value range of the resistor and the reactor is selected according to the rated power of the inverter, taking a three-phase inverter with the rated power of 200kW as an example, and the selection value range of the total impedance value of the resistor and the reactor is 0.1-3.1 omega so as to simulate the extreme power grid impedance condition of the power grid connected to the inverter with the rated power.
Selecting a value range according to the determined total impedance values of the resistor and the reactor, and selecting a resistor in a resistance value range and a reactor in an inductance value range; and according to the time sequence, setting different combinations (namely impedance combinations) of the resistance values of the resistors and the inductance values of the reactors, and optionally configuring different working condition parameters under the same or different simulated grid impedance values, such as adjusting the output power and/or the grid voltage of the inverter, so as to obtain sample data under different working conditions.
S2, collecting data: and adjusting at least one of the resistor and the reactor to obtain different impedance combinations, correspondingly sampling three-phase voltage and three-phase current data at the common coupling point to obtain a plurality of samples, and taking the resistance value of the resistor and the inductance value of the reactor corresponding to the same moment as labels corresponding to the samples, or taking the total impedance value of the impedance combinations corresponding to the same moment as labels corresponding to the samples.
Specifically, taking the sampling frequency of 16KHz as an example, each set of sampling data includes voltage and current sampling values within one voltage cycle (20ms), that is, includes data of 6 × 320, and the total impedance values (or the resistor resistance value and the reactor inductance value) of the resistor and the reactor used at the corresponding time in the experiment are used as tags; samples under different working conditions are acquired through values of the resistor and the reactor and/or modes (one or any combination) of changing active power, reactive power, direct-current bus voltage, power grid voltage and the like of the inverter, wherein the number of the acquired samples exceeds 104And setting the training round to 200 (obviously, the number of samples and the training round can be adjusted correspondingly).
S3, preprocessing data: the sample data collected in step S2 are collected together, normalization processing is performed on the collected data according to the rated voltage and the rated current of the inverter, and digital low-pass filtering processing is performed on the voltage signal and the current signal after normalization processing, so as to filter out high-frequency interference in the signals.
The preprocessed data may be used as training data in one part and test data in another part, for example, 80% of the preprocessed data may be used as training data and 20% of the preprocessed data may be used as test data. The sample proportion is only an example, and the division of the training sample and the test sample from the total sample is only an embodiment, and the application does not exclude a mode of obtaining the test sample by building a second identical inverter experiment platform, for example, all samples of the former platform are used as the training sample, and the test sample is randomly selected from the samples of the latter platform.
After obtaining the training sample set and the testing sample set, the neural network model is trained as follows:
inputting the training sample set with the label into a neural network model, wherein the neural network model learns according to each training sample in the training sample set until the training samples in the training sample set are trained once, completing one round of learning, and if the training round is set to be 200, starting a new round of learning again until 200 rounds of learning are completed, and completing one round of training.
After each training is finished, inputting a test sample set without a label into a neural network model, outputting corresponding predicted values according to the neural network model, comparing the labels, and calculating corresponding loss values Cost and accuracy rates acc, wherein the calculation accuracy rate can set a certain tolerance range, for example, the absolute value of the difference between the predicted values and the label values is considered to be accurate in a certain range, the absolute value is considered to be inaccurate when the absolute value exceeds the range, and the ratio of the number of the accurate predicted values to the total number of the predicted values is the accuracy rate.
The loss value is calculated according to a loss function, taking the root mean square error loss function as an example,
The smaller the loss value is, or the higher the accuracy is, the higher the precision of the currently trained neural network model is, and vice versa. Setting a training target, wherein the loss value is smaller than a loss threshold, or the accuracy is higher than an accuracy threshold, or both the loss value and the accuracy threshold are met; and when the verification loss value and/or the verification accuracy meet the training target, the currently trained model is considered to meet the requirements. In this embodiment, if the calculated loss value using the root mean square error loss function is less than 5%, the current model is considered to meet the requirement.
If the training target is reached, stopping training, and downloading the current neural network model into the intelligent chip;
if the training target is not reached, performing iterative training, namely inputting the training sample set into the currently trained neural network model for multiple times again until the next training is completed, and then verifying whether the loss value and/or the accuracy meet the training target or not, if not, performing iterative training and verification again until the training target is reached.
The power grid impedance detection scheme of the invention can be used for different specific models of the applied three-phase inverter.
After the neural network model reaching the training target (passing verification) is downloaded into an intelligent chip, the trained model is used, under the working condition of an inverter, the power grid impedance value is estimated according to the three-phase voltage signal and the three-phase current signal at the PCC (point of common coupling) obtained by actual sampling, and therefore the method for detecting the power grid impedance by the power grid impedance detection device is the method for detecting the power grid impedance; compared with the existing passive algorithm, the power grid impedance detection method provided by the invention adopts a deep learning algorithm, the accuracy and robustness of impedance detection under the condition of a complex power grid are greatly improved, and a special intelligent chip is adopted, so that the problem that the deep learning algorithm cannot be embedded into an inverter CPU is solved, and the real-time performance of power grid impedance detection is ensured;
after the grid impedance of the current inverter accessed to the grid PCC is detected, the detection result is output to the CPU, and the CPU modulates the pulse signal through the self-adaptive algorithm to control the running state of the inverter, so that the method for controlling the running state of the inverter by the grid impedance detection device provided by the embodiment of the invention effectively improves the robustness of a system where the inverter is located.
The grid impedance detection device of the embodiment of the invention can be integrated in the inverter, and can also be electrically connected with the inverter in the form of an independent device.
And the inverter is provided with an anti-islanding device, and if the inverter judges that the islanding effect occurs according to the predicted value of the grid impedance, the anti-islanding device is triggered to stop grid-connected operation of the inverter.
The invention provides an artificial intelligence-based power grid impedance detection algorithm, a special intelligent chip is added, a complex artificial neural network algorithm can be realized, the detection accuracy and robustness are improved, and the power grid impedance at the PCC can be accurately and stably predicted by adopting a deep neural network. The power grid impedance detection device disclosed by the invention can be used for accurately predicting the power grid impedance in real time through the deep neural network model without injecting harmonic current to the public coupling point of the inverter connected to the power grid, so that the self-adaptive control of the inverter is realized.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The foregoing is directed to embodiments of the present application and it is noted that numerous modifications and adaptations may be made by those skilled in the art without departing from the principles of the present application and are intended to be within the scope of the present application.
Claims (12)
1. A grid impedance detection device for detecting grid impedance at a point of common coupling of an inverter into a grid, wherein a harmonic current does not need to be injected into the point of common coupling, the detection device comprising the following modules:
the sampling module is used for sampling the three-phase voltage signals and the three-phase current signals at the common coupling point;
and the intelligent chip is used for receiving the three-phase voltage signals and the three-phase current signals sampled by the sampling module and inputting the three-phase voltage signals and the three-phase current signals into an internal power grid impedance detection neural network model, and the power grid impedance detection neural network model is pre-established and trained so as to output a predicted value of power grid impedance according to the input three-phase voltage signals and three-phase current signals.
2. The grid impedance detection device according to claim 1, wherein the training sample for training the grid impedance detection neural network model and the test sample for verifying the training sample are obtained by:
s1, constructing an inverter experiment platform, and simulating the impedance of a power grid by using a resistor and a reactor;
s2, collecting data: adjusting at least one of the resistor and the reactor to obtain different impedance combinations, correspondingly sampling three-phase voltage and three-phase current data at a common coupling point to obtain a plurality of samples, and taking the resistance value of the resistor and the inductance value of the reactor corresponding to the same moment as labels corresponding to the samples, or taking the total impedance value of the impedance combinations corresponding to the same moment as labels corresponding to the samples;
and S3, respectively obtaining a training sample and a test sample from the samples.
3. The grid impedance detection device according to claim 2, wherein step S3 includes:
s31, collecting the sample data collected in the step S2 together, firstly carrying out normalization processing on the collected data according to the rated voltage and the rated current of the inverter, and carrying out digital filtering processing on the voltage signal and the current signal after the normalization processing;
and S32, obtaining training samples and test samples from the filtered data according to a preset rule.
4. The grid impedance detection device according to claim 2, wherein step S1 includes:
s11, selecting an inverter which is the same as the inverter in the actual detection working condition to build an experiment platform;
s12, determining the impedance value range of the simulated power grid impedance according to the rated power of the inverter;
s13, selecting a resistor in an adaptive resistance value range and an electric reactor in an adaptive inductance value range according to the determined impedance value range of the simulated power grid impedance;
and S14, setting different combinations of the resistance values of the resistors and the inductance values of the reactors according to a time sequence, and configuring different working condition parameters under the same or different simulated power grid impedance values.
5. The grid impedance detection device according to claim 2, wherein after the grid impedance detection neural network model is trained, the grid impedance detection neural network model is downloaded into the smart chip; the power grid impedance detection neural network model is obtained through the following steps:
e1, building a neural network model by using a deep learning platform;
e2, training the neural network model by using the training sample, wherein the input quantity of the neural network model is three-phase voltage and three-phase current data in the training sample, and the output quantity of the neural network model is three-phase equivalent resistance and inductance of the power grid;
e3, verifying the neural network model by using the test sample, if the verification result does not meet the preset requirement, performing iterative training on the neural network model by using the training sample until the trained neural network model passes the verification, and taking the current neural network model as the power grid impedance detection neural network model.
6. The grid impedance detection device according to any one of claims 1 to 5, further comprising a CPU module and a PWM module, wherein the CPU module is electrically connected with both the smart chip and the PWM module, and the PWM module is used for being electrically connected with the inverter;
the sampling module is used for acquiring and sending the three-phase voltage signals and the three-phase current signals to the intelligent chip in real time, the intelligent chip outputs an estimated power grid impedance value to the CPU module in real time, and the CPU module is used for generating a control instruction according to the current power grid impedance value and sending the control instruction to the PWM module;
and the PWM module is used for outputting an adaptive pulse signal according to the control instruction so as to control the running state of the inverter in real time.
7. An inverter, characterized by comprising a grid impedance detection device according to any one of claims 1-6.
8. The inverter according to claim 7, wherein the inverter is provided with an anti-islanding device, and the anti-islanding device is triggered if the inverter determines that the islanding effect occurs according to the predicted value of the grid impedance.
9. An intelligent detection method for power grid impedance is characterized in that harmonic current is not required to be injected into a public coupling point of an inverter connected into a power grid, and the detection method comprises the following steps:
collecting three-phase voltage signals and three-phase current signals at a public coupling point of an inverter connected to a power grid;
inputting the data acquisition results of the three-phase voltage and the three-phase current to a pre-generated power grid impedance detection neural network model, and outputting corresponding prediction results of the power grid impedance by the power grid impedance detection neural network model; the power grid impedance detection neural network model is pre-generated through the following steps:
g1, data collection: building an inverter experiment platform, simulating the impedance of a power grid by using a resistor and a reactor, adjusting the resistance value of the resistor and/or the inductance value of the reactor, sampling three-phase voltage signals and three-phase current signals of a plurality of voltage periods as samples, and taking the resistance value of the resistor and the inductance value of the reactor corresponding to the same moment as labels corresponding to the samples;
g2, preprocessing data: collecting the sample data sampled in the step G1, and preprocessing the sample data;
g3, model training: building a neural network model, and learning by using the training sample obtained by preprocessing in the step G2 and the corresponding label thereof to obtain the trained neural network model;
g4, model verification: verifying the neural network model trained in the step G3 by using the test sample preprocessed in the step G2, and if the verification result meets the preset requirement, taking the current neural network model as the power grid impedance detection neural network model; and if the verification result does not meet the preset requirement, returning to execute the steps G3-G4.
10. The intelligent grid impedance detection method according to claim 9, wherein step G4 comprises:
inputting a test sample into a neural network model to be verified to obtain a corresponding estimation value;
and if the root mean square value of the error between each estimated value and the corresponding label actual value of the corresponding test sample is less than 5%, judging that the verification result meets the preset requirement.
11. The method according to claim 9, wherein in step G1, the resistance of the resistor and/or the inductance of the reactor are adjusted, and the output power of the inverter and/or the grid voltage are adjusted to obtain sample data under different operating conditions.
12. The method of claim 9, wherein the prediction of the grid impedance is output by the grid impedance detection neural network model, and the operation parameters of the inverter are adjusted by using the prediction of the grid impedance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110958436.1A CN113625054B (en) | 2021-08-20 | 2021-08-20 | Power grid impedance detection device, inverter and power grid impedance detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110958436.1A CN113625054B (en) | 2021-08-20 | 2021-08-20 | Power grid impedance detection device, inverter and power grid impedance detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113625054A true CN113625054A (en) | 2021-11-09 |
| CN113625054B CN113625054B (en) | 2024-07-23 |
Family
ID=78386828
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110958436.1A Active CN113625054B (en) | 2021-08-20 | 2021-08-20 | Power grid impedance detection device, inverter and power grid impedance detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113625054B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118277924A (en) * | 2024-06-04 | 2024-07-02 | 深圳创芯技术股份有限公司 | Desktop power supply intelligent detection method based on data driving |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102565571A (en) * | 2010-12-03 | 2012-07-11 | 西门子公司 | Arrangement and method for testing an electric power generation system |
| CN105932678A (en) * | 2016-06-07 | 2016-09-07 | 湖南大学 | Virtual impedance comprehensive control method for power inductive filtering system |
| CN106487240A (en) * | 2015-08-28 | 2017-03-08 | 中国电力科学研究院 | A kind of grid simulator with accurate harmonic voltage and virtual impedance control |
| CN106771786A (en) * | 2017-02-28 | 2017-05-31 | 合肥工业大学 | The verification method and experimental provision of electric network impedance identification |
| WO2019184389A1 (en) * | 2018-03-29 | 2019-10-03 | 山东大学 | System and method for voltage-free sampling coordination control for multiple grid-connected inverters |
| CN110456161A (en) * | 2019-08-09 | 2019-11-15 | 合肥工业大学 | The impedance measurement method of forcing frequency and disturbance amplitude self adaptive control |
| CN112946363A (en) * | 2019-12-11 | 2021-06-11 | 国网湖北省电力有限公司电力科学研究院 | Neural network-based power grid impedance online identification method |
-
2021
- 2021-08-20 CN CN202110958436.1A patent/CN113625054B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102565571A (en) * | 2010-12-03 | 2012-07-11 | 西门子公司 | Arrangement and method for testing an electric power generation system |
| CN106487240A (en) * | 2015-08-28 | 2017-03-08 | 中国电力科学研究院 | A kind of grid simulator with accurate harmonic voltage and virtual impedance control |
| CN105932678A (en) * | 2016-06-07 | 2016-09-07 | 湖南大学 | Virtual impedance comprehensive control method for power inductive filtering system |
| CN106771786A (en) * | 2017-02-28 | 2017-05-31 | 合肥工业大学 | The verification method and experimental provision of electric network impedance identification |
| WO2019184389A1 (en) * | 2018-03-29 | 2019-10-03 | 山东大学 | System and method for voltage-free sampling coordination control for multiple grid-connected inverters |
| CN110456161A (en) * | 2019-08-09 | 2019-11-15 | 合肥工业大学 | The impedance measurement method of forcing frequency and disturbance amplitude self adaptive control |
| CN112946363A (en) * | 2019-12-11 | 2021-06-11 | 国网湖北省电力有限公司电力科学研究院 | Neural network-based power grid impedance online identification method |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118277924A (en) * | 2024-06-04 | 2024-07-02 | 深圳创芯技术股份有限公司 | Desktop power supply intelligent detection method based on data driving |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113625054B (en) | 2024-07-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105633956B (en) | A kind of power distribution network three-phase state method of estimation that modeling is measured based on Spiking neutral nets puppet | |
| CN108053128B (en) | Electric network transient stability rapid evaluation method based on ELM and TF | |
| CN110119570B (en) | Actually measured data driven wind farm model parameter checking method | |
| CN106646253A (en) | On-line identification method for intrinsic parameters of battery | |
| CN110838758B (en) | Transformer area low-voltage topological method based on signal injection mode | |
| CN105938578A (en) | Large-scale photovoltaic power station equivalent modeling method based on clustering analysis | |
| CN110943453A (en) | Power system fault sample generation and model construction method facing transfer learning | |
| CN110619321B (en) | Charging pile technical condition diagnosis method and system based on deep neural network | |
| CN109992911B (en) | Photovoltaic module rapid modeling method based on extreme learning machine and IV characteristics | |
| CN114091816A (en) | Power distribution network state estimation method based on gated graph neural network of data fusion | |
| CN110363334A (en) | Grid-connected grid line loss prediction technique based on Grey Neural Network Model | |
| CN116401532B (en) | Method and system for recognizing frequency instability of power system after disturbance | |
| CN113625054B (en) | Power grid impedance detection device, inverter and power grid impedance detection method | |
| CN115201563B (en) | Multi-harmonic source positioning method and system based on joint entropy | |
| CN115469184A (en) | New energy transmission line fault identification method based on convolutional network | |
| CN112686380A (en) | Neural network-based echelon power cell consistency evaluation method and system | |
| CN113746132B (en) | Photovoltaic power station based on cloud edge cooperation and control method thereof | |
| CN111177973B (en) | Photovoltaic array online modeling method based on reinforcement learning | |
| CN112085043B (en) | Intelligent monitoring method and system for network security of transformer substation | |
| CN117272184A (en) | ISSA-RF algorithm-based photovoltaic power generation system fault classification method | |
| CN112036010A (en) | Photovoltaic system dynamic process hybrid equivalent modeling method based on data driving | |
| CN115114964B (en) | Sensor intermittent fault diagnosis method based on data driving | |
| CN107069710B (en) | Power system state estimation method considering new energy space-time correlation | |
| CN114089181B (en) | XG-Boost-based MMC switching tube open-circuit fault detection method | |
| CN113761793B (en) | Inverter output impedance detection device and method and inverter operation control method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 215000 No. 90, Zijin Road, high tech Zone, Suzhou, Jiangsu Applicant after: Goodway Technology Co.,Ltd. Address before: 215163 No.90 Zijin Road, high tech Zone, Suzhou City, Jiangsu Province Applicant before: JIANGSU GOODWE POWER SUPPLY TECHNOLOGY Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |