CN113094983B - Online simulation method for multi-dimensional time-varying characteristics of direct-current fault electric arc of photovoltaic system - Google Patents
Online simulation method for multi-dimensional time-varying characteristics of direct-current fault electric arc of photovoltaic system Download PDFInfo
- Publication number
- CN113094983B CN113094983B CN202110333953.XA CN202110333953A CN113094983B CN 113094983 B CN113094983 B CN 113094983B CN 202110333953 A CN202110333953 A CN 202110333953A CN 113094983 B CN113094983 B CN 113094983B
- Authority
- CN
- China
- Prior art keywords
- fault arc
- model
- arc
- direct
- current
- 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 abstract description 53
- 238000004088 simulation Methods 0.000 title claims abstract description 48
- 238000010891 electric arc Methods 0.000 title claims description 4
- 230000008569 process Effects 0.000 claims abstract description 18
- 230000003068 static effect Effects 0.000 claims abstract description 11
- 238000001228 spectrum Methods 0.000 claims description 35
- 238000007637 random forest analysis Methods 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 14
- 238000010183 spectrum analysis Methods 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 5
- 238000000926 separation method Methods 0.000 claims description 5
- 238000012546 transfer Methods 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 2
- 238000012937 correction Methods 0.000 claims description 2
- 238000007670 refining Methods 0.000 claims description 2
- 238000002485 combustion reaction Methods 0.000 abstract description 2
- 238000013178 mathematical model Methods 0.000 abstract description 2
- 230000008859 change Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 6
- 238000001514 detection method Methods 0.000 description 6
- 238000011161 development Methods 0.000 description 6
- 230000018109 developmental process Effects 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/04—Power grid distribution networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/10—Noise analysis or noise optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses an online simulation method of multi-dimensional time-varying characteristics of a photovoltaic system direct-current fault arc, which comprises the steps of simulating the time domain characteristics of the direct-current fault arc through a U-I model, simulating the frequency domain characteristics of the direct-current fault arc through a pink noise model, constructing a filter to correct the pink noise model, and superposing the two models to obtain a static photovoltaic system direct-current fault arc model for synchronously simulating the characteristics of the multiple fault arcs; time factor variables are introduced into the U-I model and the pink noise model, and then dynamic association models which influence the key parameters of the characteristics of the fault arc and the time-frequency characteristics of the fault arc, such as arc generation gaps, loop currents and the like in the combustion evolution process of different arcs are respectively established. The method establishes a dynamic and static photovoltaic system direct-current fault arc mathematical model with a multidimensional time-varying fault arc characteristic evolution overall-process expression function, and realizes effective simulation of fault arc characteristic changes under different application scenes and system structure conditions.
Description
Technical Field
The invention belongs to the technical field of photovoltaic electrical fault detection, and particularly relates to an online simulation method for multi-dimensional time-varying characteristics of a photovoltaic system direct-current fault arc.
Background
With the large number of photovoltaic, energy storage and electric vehicles connected to the power distribution network, the number of power electronic devices is greatly increased, and the direct current of the power distribution and utilization system becomes an increasingly important development direction. Compared with an alternating current system, the direct current comprehensive energy system can conveniently realize the efficient access and flexible management of renewable energy, direct current and variable frequency loads, and provide efficient and safe electricity utilization service.
The inherent composition of a photovoltaic system makes it prone to problems such as loose wiring, aging of components, and insulation damage, i.e., fault arcing may occur at multiple locations in the photovoltaic system, resulting in higher frequency of photovoltaic arcing faults than other types of dc arcing faults. The normal operation of the photovoltaic system and the power system is endangered by the power utilization safety problem caused by fault electric arcs in the long-term operation process of the photovoltaic system. And as the voltage grade of the photovoltaic system is continuously improved and the wiring form is increasingly complicated, the probability of occurrence of fault arcs is increased. Therefore, the method has very important significance for timely detecting fault electric arc and extinguishing arc. However, compared with widely-developed ac fault arc research, there are many problems in the related research of dc fault arc, for example, there is a lack of a dynamic photovoltaic system dc fault arc model capable of realizing a full-process expression function of multidimensional fault arc characteristic evolution, which limits the development of a fault arc detection algorithm, and is not beneficial to further research of a fault arc characteristic change mechanism under different application scenarios and system structure conditions, and the development of a high-performance photovoltaic system dc fault arc detection apparatus.
The commonly used pink noise generation methods mainly include: paul Kellet's weighted sum filter method, robert Bristow-Johnson's zero-pole filter method, voss algorithm. The Paul Kellet weighting and filtering method is the best method in performance at present, but the calculation amount is large, and the calculation process is quite complex. The pole-zero filter method of Robert Bristow-Johnson has the advantages of large calculation amount, complex calculation process and poor fitting effect. Although the Voss algorithm is small in calculation amount and simple in process, the implementation performance is extremely poor, and the Voss algorithm has a great deviation from ideal pink noise. Meanwhile, pink noise, when used for simulating the frequency spectrum distribution of the fault arc in a certain period, can not clearly depict the fluctuation of the frequency spectrum along with the change of time.
Disclosure of Invention
The invention aims to provide an online simulation method for the multi-dimensional time-varying characteristics of a direct-current fault arc of a photovoltaic system.
In order to achieve the purpose, the invention adopts the following technical scheme:
the fault arc simulation method comprises the following steps:
1) Establishing a photovoltaic system simulation model, wherein the simulation model comprises a controlled current source module positioned at the output end of a photovoltaic array of a corresponding photovoltaic system, and the controlled current source module is selected from any one of the following direct current fault arc models:
a first direct-current fault arc model: simulating the static time domain characteristic of the direct-current fault arc by adopting a U-I model;
and D, direct-current fault arc model II: simulating the static frequency domain characteristic of the direct-current fault arc by adopting a pink noise model;
a third direct-current fault arc model: synchronously and statically simulating the multi-dimensional fault arc characteristics of the direct-current fault arc of the photovoltaic system by superposing the first model and the second model;
d, a direct-current fault arc model IV: introducing time factor variables into each superposition part (model I and model II) of the model III, and synchronously and dynamically simulating the multi-dimensional fault arc time-varying characteristic of the direct-current fault arc of the photovoltaic system;
2) Performing parameter fitting on the U-I model by using the DC fault arc experimental data under different application scenes and/or system structure conditions, and performing parameter setting on the pink noise model;
3) And operating the simulation model, and carrying out the whole process simulation of the characteristic evolution of the fault arc.
Preferably, the U-I model is represented as:
where A, B, C, D are the arc constants (obtainable by fitting) depending on the experimental conditions, L is the arc gap (in mm), V arc Is the fault arc voltage (in V), I arc Is the fault arc current (in units of a).
Preferably, A, B, C and D are respectively 29.4-36.25, 0.19-0.25, 11.77-13.811 and 0.019-0.030.
Preferably, a, B, C, and D are obtained by fitting fault arc current and fault arc voltage data measured experimentally at a certain arcing gap.
Preferably, the construction method of the pink noise model includes the following steps: white noise is converted to pink noise by a digital filter.
Preferably, the transfer function of the digital filter is expressed as:
the 7-order transfer function adopted by the invention has simple calculation process and higher fitting degree with ideal pink noise.
Preferably, the method for constructing the pink noise model further includes the following steps: and analyzing the frequency spectrum characteristics of the direct-current fault arc current in the experimental data, and correcting pink noise according to the analysis result.
Preferably, in the pink noise model, the pink noise is corrected by using a low-pass filter, a high-pass filter or a low-pass filter and a high-pass filter which are cascaded, so that the corrected pink noise forms a fault arc noise distribution form which is matched with the frequency spectrum characteristic of the direct-current fault arc current.
Preferably, the low-pass filter and the high-pass filter are selected from butterworth filters.
Preferably, the method for introducing the time factor variable into the U-I model comprises the following steps: the arcing gap L is constructed as a function of time t, which is expressed as:
wherein v is the separation velocity (in mm/s) and t 0 Time required for separation to maximum arcing gap, L max For maximum arc generation clearance, the function is used as a correlation model of a model I (U-I model) and a time factor variable, and parameters (v, L) can be carried out under different application scenes and/or system structure conditions max ) And (6) setting.
Preferably, the method for introducing the time factor variable into the pink noise model comprises the following steps: respectively establishing random forest models for the frequency spectrum characteristic value output of the direct current fault arc frequency spectrum which changes along with time on N frequency bands, and training the random forest models established on the frequency bands to jointly form a whole frequency band distribution model of a fault arc current frequency spectrum prediction result under a certain time window. The time factor variable introduced into the pink noise model means that the trend of the full-band frequency spectrum characteristic amplitude changing along with the time is determined by a machine learning method, so that the frequency spectrum amplitude determined by the pink noise can accurately change along with the time, the fault arc time-frequency information is well reflected, and the clearer and more precise fault arc characteristics are obtained.
Preferably, the training of the random forest model comprises the following steps: refining (for example, averagely dividing) 0-100kHz into N small frequency bands, taking a current value and an arc generating gap in each time window as sample characteristic quantities (wherein the current value is a system loop current value, and the arc generating gap can influence the output of a subsequent machine learning frequency spectrum, so that the subsequent machine learning frequency spectrum is taken as a necessary characteristic quantity in the machine learning process), taking a frequency spectrum analysis result of the current in the time window as a sample label, constructing a training set and a test set aiming at a single small frequency band, and adjusting random forest model parameters in the training process to enable the average prediction accuracy of the training model on the full frequency band to be not lower than 80%. The random forest model is used as a correlation model of a model II (pink noise model) and a time factor variable, and parameter setting can also be performed under different application scenes and/or system structure conditions.
Preferably, N is 8 to 12 (e.g., 10), so that the characterization of the fault arc frequency spectrum is more precise and accurate; the corresponding time window is 1 to 4ms (e.g., 3 ms).
The invention has the beneficial effects that:
1) The method can effectively simulate the time domain characteristics of the fault arc: the arc voltage and the arc current vary with time. The established fault arc model is introduced into a photovoltaic system for simulation, so that the influence of key parameters such as arc generation gaps and loop current on the electrical parameters of the fault arc in the combustion evolution process of different arcs can be reflected, and a standardized simulation model of the time domain characteristic of the fault arc is established to reveal the state change rule of the key electrical parameters such as the voltage and the current of the fault arc in a direct current system.
2) The method can effectively simulate the frequency domain characteristics of the fault arc: spectral signature of the fault arc. The pink noise model is built, the built pink noise model is corrected by utilizing the low-pass filter and the high-pass filter, the frequency domain model of the fault arc is obtained, the frequency spectrum characteristics of the fault arc under different working conditions can be correctly simulated, and therefore the standardized simulation model of the frequency domain characteristics of the fault arc is built and used for revealing the change rule of the frequency domain characteristics of the fault arc in the direct current system.
3) The method has wide applicability, a dynamic photovoltaic system direct-current fault arc model for realizing the whole process of multidimensional fault arc characteristic evolution is constructed by introducing a time factor variable into a U-I model and introducing the time factor variable into a pink noise model based on a trained random forest model, the time-varying characteristics of fault arcs under the influence of different factors can be correctly simulated, and then an online simulation method of the fault arcs is provided.
Drawings
FIG. 1 is a structural block diagram of a photovoltaic system direct-current fault arc multi-dimensional time-varying characteristic online simulation system constructed in an embodiment; wherein, discrete represents Discrete time simulation (simulation time of one step is 1e-6s, namely simulation frequency is 1 MHz).
Fig. 2 is a block diagram of a fault arc model in the on-line simulation system shown in fig. 1.
Fig. 3 is a graph comparing simulation and experimental results of dc fault voltage and current of a photovoltaic system.
Fig. 4 is a spectrum analysis diagram of the pink noise model constructed.
FIG. 5a is a graph comparing simulation and experimental results of DC fault arc noise of a photovoltaic system (ohm Nick inverter: omniksol-20k-TL 2).
FIG. 5b is a graph comparing simulation and experimental results of DC fault arc noise of a photovoltaic system (solid topology inverter: GW 8000-DT).
FIG. 6a is a graph of simulated versus actual frequency spectrum of a fault arc current waveform over time in the 10-20kHz band.
FIG. 6b is a graph showing the result of the prediction accuracy of the simulated spectrum and the actual spectrum of the fault arc current waveform in different frequency bands of 0-100 kHz.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. The examples are given solely for the purpose of illustration and are not intended to limit the scope of the invention.
According to the invention, multiple direct-current fault arc models are fused, multiple direct-current fault arc characteristics in the photovoltaic system are accurately simulated, the on-line simulation of the whole process of the evolution of the direct-current fault arc in the photovoltaic system is realized, and an effective sensitivity verification simulation technical means is provided for the development of a high-performance direct-current fault arc detection device for the photovoltaic system. Therefore, the problem of deep research on the change mechanism of the characteristics of the fault arc under different application scenes and system structure conditions is solved.
Referring to fig. 1, the online simulation system for the multidimensional time-varying characteristics of the direct-current fault arc of the photovoltaic system integrally comprises a photovoltaic array, a fault arc model, an MPPT controller, a BOOST converter, an inverter, a transformer and an alternating-current power grid. Wherein, the output of the photovoltaic array can be adjusted by adjusting the temperature and the illumination intensity.
Examples of online simulations: discrete time simulation is set in matlab, and the simulation time of one step is 1e-6s, namely the simulation frequency is 1MHz. The photovoltaic array is a current source, generates main loop current, and can control the magnitude of the current output by the photovoltaic array according to temperature and illumination intensity. The MPPT controller operates the photovoltaic array at the maximum power point for maximum efficiency. The BOOST converter BOOSTs the voltage output by the photovoltaic array, and then inverts direct current into alternating current through the inverter and then the alternating current is merged into an alternating current power grid. The fault arc model is connected in series into a line bus, and the change of the Current in the loop is Controlled in a Controlled Current Source mode.
The fault arc model comprises: constructing a U-I model to simulate the static time domain characteristic of the direct-current fault arc; more than one pink noise model corrected by a filter is constructed to simulate the static frequency domain characteristics of the direct-current fault arc; introducing a time factor variable into the U-I model, introducing the time factor variable into the pink noise model based on the trained random forest model, and simulating the multi-dimensional time-varying characteristic of the direct-current fault arc of the photovoltaic system; therefore, the online simulation of the whole process of the evolution of the direct-current fault arc in the photovoltaic system is realized.
Referring to fig. 2, the construction of the fault arc model for the online simulation of the multidimensional time-varying characteristics of the dc fault arc of the photovoltaic system specifically includes the following steps:
1) And building a U-I model, fitting experimental data such as arc current and arc voltage measured by experiments to obtain parameters of the U-I model, and simulating the static time domain characteristics of the direct-current fault arc (for example, the online simulation of the static time domain characteristics of the direct-current fault arc can be realized).
The U-I model is as follows:
wherein A, B, C, D are arc constants depending on experimental conditions, L is arc gap (unit is mm), and V is arc Is the fault arc voltage (in V), I arc Is the fault arc current (in units of a).
The method can utilize the fault arc current and fault arc voltage data measured by an arc generating gap under the condition of 0.8-6.4 mm to fit constants A, B, C and D, the fitting effect evaluation index Adjusted R-square is at least larger than 0.95, the fitting parameters in the model are considered to be effective, and A, B, C and D are respectively 32.83, 0.22, 12.79 and 0.024.
Examples of the experiments: the arc generating gap of the fault arc generating device is set to be 3.2mm, and the speed is set to be 2mm/s. The arc current is measured with a current sensor and the arc voltage is measured with a voltage sensor. When the system works normally, the fault arc generating device is started, a corresponding 0-3.2 mm fault arc is generated in the experiment loop, and the arc voltage and the arc current at the moment are recorded for data fitting. By adjusting the Arc generating gap and the current of the system in normal operation, data of different Arc currents, arc generating gaps and Arc voltages can be obtained (reference: time-Frequency Distribution characteristics and Model Simulation of Photovoltaic Series Arc Fault With Power Electronic Equipment).
Referring to fig. 3, at 0-2.4s, the photovoltaic system is operating normally and no fault arc occurs; at 2.4 seconds, the fault arc is ignited, the arcing gap L is linearly increased from 0 to 3.2mm at the speed of 2mm/s to form a 3.2mm arcing gap, the arcing gap is kept at 3.2mm after 4s, and the arc is stably burnt. When a fault arc occurs, the current sharply decreases and the voltage sharply increases. With the increase of the arcing gap, the arc current is gradually reduced, and the arc voltage is steadily increased. When the arcing gap reaches a set maximum, the arc voltage and current remain relatively stable. And comparing the simulation result with the fault arc current and voltage of the test result, wherein the result shows that the variation of the simulated fault arc voltage and current is consistent with the test result.
2) Constructing a pink noise model for simulating the static frequency domain characteristics of the direct-current fault arc; the construction method of the pink noise model is specifically as follows.
Designing a digital filter with a transfer function of:
white noise is converted into pink noise through a designed digital filter.
Referring to fig. 4, the pink noise is distributed at 3dB down per octave, and it can be seen that the constructed pink noise spectrum analysis graph conforms to this variation characteristic.
And constructing a filter, adjusting filter parameters according to a frequency spectrum result (experimental frequency spectrum) obtained by analyzing the DC fault arc current obtained by an experiment, and correcting the pink noise. Wherein, the filter of structure does: a low-pass filter is connected with a high-pass filter in cascade; the low-frequency characteristic of pink noise is corrected by a low-pass filter, and the high-frequency characteristic of pink noise is corrected by a high-pass filter.
The high-pass filter and the low-pass filter are Butterworth filters, and filter parameters are adjusted according to a fault arc current spectrum analysis result obtained through experiments.
Referring to fig. 5a and 5b, the pink noise spectrum shown in fig. 4 is corrected by using the constructed low-pass filter and high-pass filter, and after the correction, the pink noise is corrected by the low-pass filter and the high-pass filter through spectrum analysis of the fault arc waveform occurring under different inverter conditions, so that the simulation spectrum and the experimental spectrum of the fault arc have high goodness of fit.
In addition, if the fault arc current spectrum analysis result under the action condition of the photovoltaic system is a double-peak spectrum form, the corrected pink noise model is cascaded after peak frequency and amplitude are adjusted (by adjusting parameters of a high-low filter), and the fault arc noise distribution form is further accurately simulated. For example, two high-low pass filter modified pink noises are cascaded, wherein the first set of high-pass filter parameters: 1-order, cut-off frequency 1-20 kHz (e.g., 8 kHz); first set of low pass filter parameters: 4 th order, cut-off frequency 60-120 kHz (for example, 105 kHz); second set of high pass filter parameters: 9 th order, cut-off frequency 100-140 kHz (e.g., 120 kHz); second set of low pass filter parameters: 5-order, cut-off frequency 100-140 kHz (120 kHz).
3) And introducing a time factor variable into the U-I model, introducing the time factor variable into the pink noise model based on the trained random forest model, and then overlapping the two models with the introduced time factor variable to form a dynamic mathematical model for simulating the multiple characteristics of the direct-current fault arc of the photovoltaic system, so that the overall process expression function of the characteristic evolution of the multi-dimensional fault arc is realized.
The method for introducing the time factor variable into the U-I model comprises the following steps: the arcing gap L is constructed as a function of time t, which is selected from the following according to the UL1699B standard established by Underwriters Laboratories (UL):
wherein v is the separation speed (in mm/s), t 0 Time required for separation to maximum arcing gap, L max Is the maximum arcing gap.
The method for introducing the time factor variable into the pink noise model based on the trained random forest model comprises the following steps:
s1) 0-100kHz is refined into N (for example, N is 10) small frequency bands, so that the fault arc frequency spectrum can be accurately depicted, the fault arc frequency spectrum can be better and finely simulated, and a corresponding time window is 3ms.
S2) taking each time window as a sample, taking the current value and the arc generation gap in each time window as sample characteristic quantities, and taking the frequency spectrum analysis result of the current in the time window as a sample label to construct a database sample. Selecting 50% of samples as a training set, inputting the characteristics and labels of the training set into the model for training, using the rest 50% of data as a test set, obtaining the prediction accuracy of the trained model according to the comparison effect of the prediction result and the actual result of the test set, and adjusting the parameters of the random forest model to ensure that the average prediction accuracy of the training model on the full frequency band is not lower than 80%.
And S3) respectively establishing random forest models for the frequency spectrum characteristic value output varying with time on each small frequency band according to the step S2, and finally synthesizing N small frequency band models to form a full-frequency-band distribution integral model corresponding to a 3ms fault arc current frequency spectrum prediction result, namely forming an integral prediction model group.
Referring to fig. 6a and 6b, by inputting current simulation data in a 3ms time window in real time, a spectrum prediction result of the time window can be output. It can be seen that the current frequency spectrum output by the random forest model has the same variation trend with the actual frequency spectrum in different frequency bands, and the numerical accuracy is higher and the frequency spectrum output is more accurate.
In a word, the invention relates to an online simulation method of a photovoltaic system direct-current fault arc multi-dimensional time-varying characteristic. The time domain characteristic of the direct-current fault arc is simulated by constructing a U-I model, the frequency domain characteristic of the direct-current fault arc is simulated by constructing a pink noise model corrected by a filter, a time factor variable is further introduced into the U-I model, and the time factor variable is introduced into the pink noise model based on a trained random forest model, so that the on-line simulation of the multi-dimensional time-varying characteristic of the direct-current fault arc of the photovoltaic system is realized. By changing the constructed model parameters, the fault arc electrical signals can be output on line, the effective expression of the whole process of the multi-dimensional characteristic evolution of the fault arc under different working conditions is realized, and the qualitative influence trend of different sources, charge types and system working voltage grades of the low-voltage direct-current power system on the fault arc electrical characteristics is favorably revealed. According to the invention, through the online simulation of the whole process of the evolution of the direct-current fault arc in the photovoltaic system, the deep research is facilitated, and the change mechanism of the fault arc characteristic under different application scenes and system structure conditions is revealed, so that a reference basis is provided for the development of a fault arc detection algorithm and the development of a high-performance photovoltaic system direct-current fault arc detection device.
Claims (4)
1. An online simulation method for multi-dimensional time-varying characteristics of a photovoltaic system direct-current fault arc is characterized by comprising the following steps: the method comprises the following steps:
1) Establishing a photovoltaic system simulation model, wherein the simulation model comprises controlled current source modules positioned in corresponding photovoltaic systems, the controlled current source modules are direct-current fault arc models, and the direct-current fault arc models synchronously and dynamically simulate the multi-dimensional fault arc time-varying characteristics of the direct-current fault arc of the photovoltaic systems by introducing time factor variables into each superposition part of the following three models:
a first direct-current fault arc model: simulating the static time domain characteristic of the direct-current fault arc by adopting a U-I model;
a second direct-current fault arc model: a pink noise model is adopted to simulate the static frequency domain characteristics of the direct-current fault arc;
a third direct-current fault arc model: synchronously and statically simulating the multi-dimensional fault arc characteristics of the direct-current fault arc of the photovoltaic system by superposing the first model and the second model;
2) Performing parameter fitting on the U-I model by using the DC fault arc experimental data under different application scenes and/or system structure conditions;
3) Operating the simulation model, and carrying out overall process simulation of the characteristic evolution of the fault electric arc;
the U-I model is represented as:
wherein A, B, C and D are arc constants,Lthe gap of the arc is generated,V arc in order to be able to detect a fault arc voltage,I arc is a fault arc current;
the construction method of the pink noise model comprises the following steps: converting white noise into pink noise through a digital filter;
the transfer function of the digital filter is expressed as:
the method for introducing the time factor variable into the U-I model comprises the following steps: will generate arc gapLConstructed as a function of time t, which is expressed as:
wherein,vas separation speed, t 0 The time required to separate to the maximum arcing gap,L max is the maximum arcing gap;
the method for introducing the time factor variable into the pink noise model comprises the following steps: respectively establishing random forest models for the frequency spectrum characteristic value output of the direct current fault arc frequency spectrum which changes along with time on N frequency bands, and training the random forest models established on the frequency bands to jointly form a full-frequency-band distribution integral model of a fault arc current frequency spectrum prediction result under a certain time window;
the training of the random forest model comprises the following steps: the method comprises the steps of refining 0-100kHz into N frequency bands, taking a current value and an arc generation gap in each time window as sample characteristic quantities, taking a frequency spectrum analysis result of the current in the time window as a sample label, constructing a training set and a test set aiming at a single frequency band, and adjusting random forest model parameters in the training process to enable the average prediction accuracy of a training model on the full frequency band to be not lower than 80%.
2. The on-line simulation method for the multi-dimensional time-varying characteristics of the direct-current fault arc of the photovoltaic system according to claim 1, characterized by comprising the following steps of: the A, B, C and D are respectively 29.4 to 36.25, 0.19 to 0.25, 11.77 to 13.811 and 0.019 to 0.030.
3. The method for on-line simulation of the multi-dimensional time-varying characteristics of the direct-current fault arc of the photovoltaic system according to claim 1, wherein: the construction method of the pink noise model further comprises the following steps: and analyzing the frequency spectrum characteristics of the direct-current fault arc current in the experimental data, and correcting pink noise according to the analysis result.
4. The on-line simulation method for the multi-dimensional time-varying characteristics of the direct-current fault arc of the photovoltaic system according to claim 1, characterized by comprising the following steps of: in the pink noise model, a low-pass filter and/or a high-pass filter are/is adopted for correction, so that the corrected pink noise forms a fault arc noise distribution form which is matched with the frequency spectrum characteristic of the direct-current fault arc current.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110333953.XA CN113094983B (en) | 2021-03-29 | 2021-03-29 | Online simulation method for multi-dimensional time-varying characteristics of direct-current fault electric arc of photovoltaic system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110333953.XA CN113094983B (en) | 2021-03-29 | 2021-03-29 | Online simulation method for multi-dimensional time-varying characteristics of direct-current fault electric arc of photovoltaic system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113094983A CN113094983A (en) | 2021-07-09 |
| CN113094983B true CN113094983B (en) | 2022-12-09 |
Family
ID=76670412
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110333953.XA Active CN113094983B (en) | 2021-03-29 | 2021-03-29 | Online simulation method for multi-dimensional time-varying characteristics of direct-current fault electric arc of photovoltaic system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113094983B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113552447A (en) * | 2021-07-27 | 2021-10-26 | 上海电机学院 | Series arc fault detection method based on random forest |
| CN113705147B (en) * | 2021-09-01 | 2024-03-19 | 上海大学 | DC micro-grid fault arc modeling and simulation method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103400019A (en) * | 2013-08-26 | 2013-11-20 | 上海交通大学 | Matlab/Simulink-based alternating current (AC) fault arc simulation method |
| CN107086855A (en) * | 2017-04-25 | 2017-08-22 | 西安交通大学 | The photovoltaic system fault arc detection method of many time-frequency characteristics is merged in a kind of machine learning |
| CN110417351A (en) * | 2019-08-01 | 2019-11-05 | 上海理工大学 | Photovoltaic system DC side arc fault detection systems and detection method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104092440B (en) * | 2014-07-21 | 2017-07-28 | 阳光电源股份有限公司 | Photovoltaic system direct current arc fault detection method, device, processor and its system |
-
2021
- 2021-03-29 CN CN202110333953.XA patent/CN113094983B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103400019A (en) * | 2013-08-26 | 2013-11-20 | 上海交通大学 | Matlab/Simulink-based alternating current (AC) fault arc simulation method |
| CN107086855A (en) * | 2017-04-25 | 2017-08-22 | 西安交通大学 | The photovoltaic system fault arc detection method of many time-frequency characteristics is merged in a kind of machine learning |
| CN110417351A (en) * | 2019-08-01 | 2019-11-05 | 上海理工大学 | Photovoltaic system DC side arc fault detection systems and detection method |
Non-Patent Citations (2)
| Title |
|---|
| PV Series Arc Fault Recognition under Different Working Conditions with Joint Detection Method;Silei Chen等;《IEEE Xplore》;20161212;全文 * |
| 基于支持向量机的光伏系统电弧检测方法;马健凯等;《山东建筑大学学报》;20200415(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113094983A (en) | 2021-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Beniwal et al. | A critical analysis of methodologies for detection and classification of power quality events in smart grid | |
| Yao et al. | Characteristic study and time-domain discrete-wavelet-transform based hybrid detection of series DC arc faults | |
| CN113094983B (en) | Online simulation method for multi-dimensional time-varying characteristics of direct-current fault electric arc of photovoltaic system | |
| CN110376497B (en) | Low-voltage distribution system series fault electric arc identification method based on full-phase deep learning | |
| CN103592528B (en) | A kind of photovoltaic DC-to-AC converter model parameter identification method based on dynamic trajectory sensitivity | |
| Yılmaz et al. | Automated classification of power quality disturbances in a SOFC&PV-based distributed generator using a hybrid machine learning method with high noise immunity | |
| CN110543921A (en) | cable early fault identification method based on waveform learning | |
| Wang et al. | A novel series arc fault detection method based on mel-frequency cepstral coefficients and fully connected neural network | |
| CN105514975B (en) | A kind of efficiency Forecasting Methodology of photovoltaic generating system | |
| Kaufhold et al. | Transient response of single-phase photovoltaic inverters to step changes in supply voltage distortion | |
| Saini et al. | Signal Processing Tool & Artificial Intelligence for Detection & Classification of Voltage Sag | |
| Chu et al. | TDV-LSTM: A new methodology for series arc fault detection in low power AC systems | |
| Samet et al. | A two-level neural network approach for flicker source location | |
| Nath et al. | Application of machine learning for online dynamic security assessment in presence of system variability and additive instrumentation errors | |
| CN110174602A (en) | Nonlinear-load series arc faults determination method and application | |
| CN110542835A (en) | Detection method, detection system and test system for vehicle arc fault | |
| Darmawardana et al. | Important Considerations in Development of PV Inverter Models for High Frequency Emission (Supraharmonic) Studies | |
| Joga et al. | Harmonic source identification in Microgrid using wavelet time frequency analysis | |
| Fahad et al. | A voltage flicker severity analysis module for multiple electric arc furnace operation | |
| CN114325107A (en) | Method for judging accuracy of harmonic impedance calculation result | |
| Zhi et al. | Intelligent Island detection method of DC microgrid based on Adaboost algorithm | |
| Andami et al. | Voltage notch detection using fuzzy expert system | |
| Jamil et al. | High Impedance Fault Detection in Electrical Power Feeder by Wavelet and GNN | |
| Shang et al. | Series arc fault identification based on complete ensemble empirical mode decomposition with adaptive noise and convolutional neural network | |
| Omran et al. | Photovoltaic system DC series arc fault: a case study |
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 |
Effective date of registration: 20240129 Address after: No.28, Xianning West Road, Beilin District, Xi'an City, Shaanxi Province Patentee after: XI'AN JIAOTONG University Country or region after: China Patentee after: Electric Power Research Institute of State Grid Shaanxi Electric Power Co.,Ltd. Address before: 710049 No. 28 West Xianning Road, Shaanxi, Xi'an Patentee before: XI'AN JIAOTONG University Country or region before: China Patentee before: STATE GRID SHAANXI ELECTRIC POWER Research Institute |
|
| TR01 | Transfer of patent right |