A method for detecting dc series arc faults in a photovoltaic array

Abstract

This invention discloses a method for detecting DC series arc faults in photovoltaic arrays, relating to the field of DC safety detection technology for photovoltaic systems. The method involves injecting AC disturbance signals of a first frequency sequence and a second frequency sequence into the DC bus when the photovoltaic string is in a healthy state, obtaining reference impedance amplitude sets A and B, and storing them as personalized references. During operation, the same disturbance signal is periodically injected to obtain the current impedance amplitude sets X and Y, calculating the first differential value D1 and the second differential value D2, and constructing a degradation-specific feature parameter P = D1 / (D2 + e). P is input into a connector degradation mapping model to obtain a degradation degree value d. When d reaches a preset threshold, a DC series arc fault warning signal is issued. This invention, through dual-band disturbance injection and differential ratio feature construction, achieves quantitative perception and graded warning of arc risk during the connector degradation stage, shifting the timing of safety protection intervention from post-detection to pre-warning.

Description

Technical Field

[0001] This invention relates to the field of DC safety detection technology for photovoltaic systems, and more specifically, to a method for detecting DC series arc faults in photovoltaic arrays. Background Technology

[0002] A photovoltaic array consists of multiple photovoltaic modules connected in series via connectors. During long-term outdoor operation, the connectors are subject to oxidation, loosening, corrosion, and other factors, causing their contact resistance to gradually increase and their contact capacitance to shift. When the contact resistance increases to a critical value, a DC series arc will be generated under the normal operating voltage of the photovoltaic string. DC arcs do not have a natural zero-crossing point; once formed, they are difficult to extinguish on their own, and continued burning may cause a fire.

[0003] For DC series arc fault detection, existing technologies mainly employ the following approaches: methods based on current and voltage time-domain or frequency-domain feature analysis, identifying the arc by extracting current disturbances or voltage noise generated during arc combustion; artificial intelligence-based classification methods, using neural networks and other models to learn and discriminate arc signal characteristics; electromagnetic radiation detection methods, receiving high-frequency electromagnetic waves released by arc discharge through antenna arrays; and acoustic or ultrasonic detection methods, acquiring sound wave signals generated by the arc through sensors. All of these existing technologies assume that the arc has already been generated and is burning, identifying the fault by analyzing the physical signals associated with the arc. However, there is typically a degradation process of several hours to several days between the start of connector contact degradation and the actual occurrence of the arc, a stage that existing methods cannot detect. Even if some methods achieve a relatively fast detection response, their intervention time is still after the arc has occurred, at which point the fire risk already exists, making safety protection passive. Therefore, a method for detecting DC series arc faults in photovoltaic arrays is proposed to address the above problems. Summary of the Invention

[0004] To overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a method for detecting DC series arc faults in photovoltaic arrays, which aims to solve the problem that existing DC series arc fault detection methods can only identify the fault after the arc has occurred and cannot detect the arc risk during the connector degradation stage.

[0005] To achieve the above objectives, the present invention provides a method for detecting DC series arc faults in a photovoltaic array. When the photovoltaic string is in a healthy state and no DC series arc fault has occurred, an AC disturbance signal of a first frequency sequence and an AC disturbance signal of a second frequency sequence are injected into the DC bus of the photovoltaic string to obtain the set of reference impedance amplitudes corresponding to each frequency point of the first frequency sequence. and the set of reference impedance amplitudes corresponding to each frequency point in the second frequency sequence. ,Will and The storage serves as a personalized benchmark for this photovoltaic string.

[0006] The frequency band of the first frequency sequence covers the region of resonant frequencies determined by the contact capacitance of each connector in the photovoltaic string and the parasitic inductance of the DC bus. Within this frequency band, even a small change in the contact state of the connector can cause a detectable shift in the impedance amplitude.

[0007] The frequency band of the second frequency sequence is to interval and to At least one interval in the range, which is insensitive to connector degradation but whose response to changes in ambient temperature and DC operating point is substantially consistent with the first frequency sequence, is used as a reference in subsequent differential processing to distinguish connector degradation signals from environmental interference signals.

[0008] During the DC power generation operation of the photovoltaic string, AC disturbance signals of the same first and second frequency sequences as those used when establishing the reference are injected into the DC bus of the photovoltaic string at a preset time period to obtain the impedance amplitude set corresponding to each frequency point of the first frequency sequence in the current state. and the set of impedance amplitudes corresponding to each frequency point in the second frequency sequence. By periodically applying the same perturbation conditions as when the reference was established, the difference between the current impedance spectrum and the reference impedance spectrum is made solely from connector degradation and changes in environmental conditions.

[0009] according to and Calculate the first difference value First difference It reflects the overall change in impedance amplitude within the resonant sensitive frequency band relative to the reference state. This change includes both impedance shift caused by connector degradation and common-mode drift caused by environmental factors.

[0010] according to and Calculate the second difference value Second difference It reflects the overall change in impedance amplitude within the reference frequency band relative to the reference state. This change is mainly caused by environmental factors and has a weak response to connector degradation.

[0011] Depend on and Constructing Degeneracy-Specific Feature Parameters , , For values ​​less than The constant. Common-mode interference components caused by environmental factors are suppressed through ratio calculations, thus making the characteristic parameters... It mainly reflects the impedance change component caused by connector degradation. When the connector is not degraded... Approaching a stable value, as connector degradation intensifies. Monotonically increasing.

[0012] Will By inputting the connector degradation mapping model, the degradation level of each connector in the current photovoltaic string is obtained. ,when When the preset degradation threshold is reached, a DC series arc fault warning signal is issued.

[0013] The connector degradation mapping model establishes degradation-specific feature parameters. The quantitative relationship between the connector contact resistance change and the online detection process allows for real-time calculation of the data. This is converted into an intuitive degradation level value, enabling quantitative assessment and graded early warning of arc risk.

[0014] Furthermore, the frequency band of the first frequency sequence is the resonant frequency measured by the photovoltaic string under healthy conditions. Expanding outwards from the center to both sides The range constituted by this is achieved by setting a frequency band centered on the measured resonant frequency, so that the first frequency sequence accurately covers the resonant region of the photovoltaic string, ensuring that different strings can be adapted to individual differences in component parameters and cable lengths, thereby guaranteeing the sensitivity and consistency of degradation detection.

[0015] Furthermore, the first difference value The calculation method is as follows ,in The number of frequency points in the first frequency sequence. For set The Middle The impedance amplitude corresponding to each frequency point For set The Middle The reference impedance amplitude corresponding to each frequency point.

[0016] Second difference value The calculation method is as follows ,in The number of frequency points in the second frequency sequence. For set The Middle The impedance amplitude corresponding to each frequency point For set The Middle The reference impedance amplitude corresponding to each frequency point. The differential value is calculated using an arithmetic mean method, assigning equal weight to each frequency point. The calculation method is simple and stable, and is suitable for application scenarios with small operating condition fluctuations or limited computing resources.

[0017] Furthermore, in calculating the first difference value At that time, weighting coefficients can be assigned to each frequency point within the first frequency sequence. Using a weighted formula calculate By using a weighted average method, frequency points that are more sensitive to connector degradation response receive a larger contribution to the first difference value, while frequency points that are less sensitive to degradation contribute less. This helps to improve the ability of degradation-specific characteristic parameters to distinguish different stages of connector degradation.

[0018] Furthermore, the weighting coefficients The determination method is as follows: When establishing a personalized benchmark, a simulated degradation resistor with a known resistance value is connected in series to both ends of any connector in the photovoltaic string. The relative change in impedance amplitude at each frequency point in the first frequency sequence under the condition that the simulated degradation resistor is connected is recorded relative to the condition that the simulated degradation resistor is not connected. ,Will Weighting coefficients for corresponding frequency points By simulating degradation experiments to pre-calibrate the sensitivity of each frequency point to changes in connector contact resistance, the weighting coefficients can objectively reflect the degradation detection contribution of each frequency point without relying on empirical settings or complex modeling.

[0019] Furthermore, in calculating the first difference value Second difference Previously, it was also possible to set and set Anomaly detection is performed on the impedance amplitude corresponding to each frequency point within the range. calculate the median of and interquartile range Remove or Frequency points; calculate the median of and interquartile range Remove or The frequency point.

[0020] Outlier detection methods based on median and interquartile range are robust to occasional electromagnetic interference or measurement noise. By removing outlier frequency points, individual distorted data can be avoided from adversely affecting the calculation of differential values, which helps to reduce the false alarm rate.

[0021] Furthermore, the connector degradation mapping model is a function ,in , , These are the model parameters. The model parameters are determined as follows: Accelerated aging tests were conducted on photovoltaic connectors of the same model as those used in the photovoltaic string, and the contact resistance values ​​of the connectors were measured at different aging stages. Simultaneously, the degradation-specific characteristic parameters calculated through the aforementioned steps under the corresponding aging state are acquired. ,by Compared to the pre-determined critical arc contact resistance value The ratio is used as a calibrator for the degree of degradation. For multiple groups and The data was fitted to obtain , , By establishing a quantitative mapping relationship between degradation-specific characteristic parameters and changes in contact resistance through accelerated aging experiments, online detection only requires calculation of the characteristic parameters. This allows for the direct acquisition of the current degradation level of the connector, providing a quantitative basis for arc risk early warning.

[0022] Furthermore, the degradation degree value For the range of values to Continuous quantity, This indicates that all connectors in the photovoltaic string are in good condition. This indicates that at least one connector in the photovoltaic string has reached the critical state of a DC series arc fault. The degradation level is represented by a normalized continuous quantity, allowing maintenance personnel to intuitively grasp the connector's current state relative to its healthy and critical states, facilitating the development of differentiated inspection and maintenance strategies.

[0023] Furthermore, the method also includes processing a sequence of degradation values ​​obtained from multiple consecutive detections. Perform a difference operation to obtain the degradation rate value. , , This represents the time interval between two consecutive detections. Degradation rate value. Reflecting the trend of degradation over time, when A sustained increase indicates that the connector degradation process is accelerating, providing an auxiliary basis for raising the warning level.

[0024] Furthermore, when When the first preset degradation threshold of 0.8 is exceeded, a first-level DC series arc fault warning signal is issued, indicating that the connector has undergone significant degradation. when When the second preset degradation threshold of 0.95 is exceeded, or when When the increment exceeds the preset rate increment threshold of 0.05 days for two consecutive time periods, a second-level DC series arc fault warning signal is issued, indicating that the connector degradation is approaching a critical state or the degradation process is accelerating.

[0025] By setting two threshold levels and combining them with degradation rate increment judgment, we can achieve graded early warning for different risk levels, which helps to optimize the allocation of operation and maintenance resources and improve the pertinence of security response.

[0026] The technical effects and advantages of this invention are as follows: (1) By designing a dual-band perturbation sequence and constructing a differential ratio feature, the enhancement of connector degradation signals and the suppression of environmental interference are achieved. AC perturbation signals of the first and second frequency sequences are injected into the DC bus of the photovoltaic string. The first frequency sequence covers the resonant frequency region determined by the connector contact capacitance and parasitic inductance. This frequency band is sensitive to changes in contact resistance and contact capacitance caused by connector degradation. The second frequency sequence is located in a reference frequency band far from the resonant frequency. This frequency band is not sensitive to connector degradation, but its response to changes in ambient temperature and DC operating point is basically consistent with that of the first frequency sequence. The first differential value D1 and the second differential value D2 of the impedance amplitude of the two frequency bands relative to the reference state are calculated, and the degradation-specific feature parameter P = D1 / (D2 + e) ​​is constructed. The common-mode interference caused by environmental factors is suppressed by the ratio calculation, so that the feature parameter mainly reflects the impedance change caused by connector degradation, which helps to achieve reliable detection in the early stage of degradation.

[0027] (2) The characteristic parameters are transformed into continuous degradation degree values ​​through the connector degradation mapping model, realizing the quantitative assessment and graded early warning of arc risk. The degradation-specific characteristic parameter P is input into the mapping model established by accelerated aging experiments and curve fitting to obtain the degradation degree value d ranging from 0 to 1. d=0 corresponds to the healthy state, and d=1 corresponds to the critical state of arc. The degradation rate v is calculated based on the d value sequence obtained from previous detections. When d exceeds the first threshold, the first-level early warning is issued. When d exceeds the second threshold or v continuously exceeds the threshold, the second-level early warning is issued. This method shifts the detection object from arc events to the connector degradation process, and can issue early warning signals several hours to several days before the actual occurrence of an arc, which helps to improve the initiative and timeliness of DC-side safety protection of photovoltaic systems. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating the method execution of the present invention.

[0029] Figure 2 The flowchart shows the parameter determination process for the connector degradation mapping model of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] Example 1 As attached Figure 1 and Figure 2 The method for detecting DC series arc faults in photovoltaic arrays is applicable to photovoltaic strings composed of multiple photovoltaic modules connected in series via connectors. This method quantifies the degree of degradation as the connectors gradually degrade from a healthy state to the point of DC series arc fault occurrence, before the actual arc occurs, and issues graded early warning signals based on the degree and trend of degradation.

[0032] In one implementation, the photovoltaic string is first confirmed to be in a healthy state and free from DC series arcing faults before being put into operation. This is confirmed by measuring the contact resistance at both ends of each connector in the photovoltaic string. A string is considered healthy when the contact resistance of all connectors is less than 1.5 times its factory nominal value.

[0033] Subsequently, an AC disturbance signal is injected into the DC bus of the photovoltaic string. The current amplitude of the AC disturbance signal is stabilized at one percent of the current operating current of the photovoltaic string through feedback control, so as not to affect the normal DC power generation and maximum power point tracking of the photovoltaic string. The frequency of the AC disturbance signal is output sequentially according to the frequency points specified by the first frequency sequence and the second frequency sequence.

[0034] The first frequency sequence covers the region of resonant frequencies determined by the contact capacitance of each connector in the photovoltaic string and the parasitic inductance of the DC bus. Before establishing a reference, this resonant frequency must first be determined. The method of determination is as follows: Frequency sweep measurements were performed on the photovoltaic strings in open-circuit or standby mode, with the sweep starting frequency being [frequency value missing]. The termination frequency is At each frequency point, the AC voltage signal across the photovoltaic string and the AC current signal flowing through the photovoltaic string are collected. The collected voltage and current signals are then input into an orthogonal demodulation-type digital phase-locked amplifier algorithm for processing.

[0035] The algorithm performs orthogonal demodulation on the acquired time-domain voltage and current sampling sequences to obtain the voltage amplitude, current amplitude, and phase difference between voltage and current at that frequency point. The ratio of voltage amplitude to current amplitude is used as the impedance amplitude at that frequency point, and the impedance amplitude and impedance phase angle corresponding to all frequency points are recorded.

[0036] After completing the frequency sweep measurement and recording the impedance amplitude and impedance phase angle at all frequency points, the frequency position near the zero-crossing point where the impedance phase angle crosses from a positive value to a negative value, or where the impedance amplitude exhibits a local maximum value, is determined as the resonant frequency of the photovoltaic string. .

[0037] The frequency band of the first frequency sequence is... Expanding from the center in both the direction of decreasing frequency and the direction of increasing frequency The interval formed by the frequency band. Select within this frequency band interval. The first frequency sequence is composed of several frequency points. The value of is chosen so that the frequency point covers the peak and the slope regions on both sides of the resonance peak, so as to ensure that the changes of the resonance peak are fully captured.

[0038] The frequency band of the second frequency sequence is to interval and to Intervals. Several frequency points are selected from the low-frequency interval and several frequency points are selected from the high-frequency interval to form a second frequency sequence. The total number of frequency points is denoted as . .

[0039] For each frequency point in the first frequency sequence, after injecting the corresponding frequency AC disturbance signal, the voltage and current signals are acquired using the same acquisition method as the frequency sweep, and then input into the quadrature demodulation type digital lock-in amplification algorithm to obtain the impedance amplitude at that frequency point.

[0040] All of the first frequency sequence The impedance amplitudes corresponding to each frequency point are denoted as follows: These elements constitute the set of reference impedance amplitudes. .

[0041] Perform the same operation on each frequency point in the second frequency sequence to obtain The impedance magnitudes are denoted as follows: These elements constitute the set of reference impedance amplitudes. .

[0042] set and set This data is stored as personalized baseline data for the photovoltaic string, and the storage location can be either the internal memory of the photovoltaic inverter or the database of the local monitoring system.

[0043] While the photovoltaic string is in DC power generation operation, the current state measurement is repeatedly performed at a preset time cycle. The moment of the first measurement is the first preset measurement moment after the baseline is established. The preset time cycle can be set to once a day or once every half day according to the power plant operation and maintenance requirements.

[0044] During each execution, an AC disturbance signal with a first frequency sequence and an AC disturbance signal with a second frequency sequence, identical to those used when the reference was established, are injected into the DC bus of the photovoltaic string. The current amplitude and frequency sequence of the AC disturbance signal remain consistent with those used when the reference was established.

[0045] Using the same acquisition method and orthogonal demodulation digital lock-in amplification algorithm, the impedance amplitude corresponding to each frequency point of the first frequency sequence in the current state is obtained, and the... The impedance amplitude corresponding to each frequency point is denoted as . ,all The impedance amplitude corresponding to each frequency point constitutes a set. .

[0046] Simultaneously, the impedance amplitude corresponding to each frequency point of the second frequency sequence is obtained, and the first... The impedance amplitude corresponding to each frequency point is denoted as . ,all The impedance amplitude corresponding to each frequency point constitutes a set. .

[0047] Next, calculate the first difference value. Second difference The calculation formula is: , .

[0048] in, The number of frequency points in the first frequency sequence. The number of frequency points in the second frequency sequence. For set The Middle The current impedance amplitude corresponding to each frequency point For set The Middle The reference impedance amplitude corresponding to each frequency point For set The Middle The current impedance amplitude corresponding to each frequency point For set The Middle The reference impedance amplitude corresponding to each frequency point.

[0049] Then, based on the first difference value Second difference Constructing Degeneracy-Specific Feature Parameters The construction formula is as follows: , in, For values ​​less than The constant is used to prevent numerical overflow when the denominator is zero.

[0050] The calculated By inputting a pre-established connector degradation mapping model, the degradation level of each connector in the current photovoltaic string can be obtained. .

[0051] The connector degradation mapping model is expressed in functional form as follows: ,in , , These are model parameters, the values ​​of which are predetermined through accelerated aging experiments and curve fitting described later.

[0052] Degradation level value For the range of values to Continuous quantity, This indicates that the connector is in a healthy state. This indicates that the connector has reached a critical state where a DC series arcing fault may occur. When When the preset degradation threshold is reached, a DC series arc fault warning signal is issued.

[0053] As an optional implementation method, the first difference value The calculation can be performed using a weighted average method instead of the arithmetic average method described above. In this method, the weighting coefficients corresponding to each frequency point of the first frequency sequence are predetermined. .

[0054] Specifically, the set is completed during the benchmark establishment phase. and set After obtaining the data, select any connector in the photovoltaic string and connect a simulated degradation resistor of known resistance value in series across its two ends. The resistance value of the simulated degradation resistor can be selected based on several to tens of times the healthy contact resistance of the connector.

[0055] With the simulated degradation resistor connected, an AC disturbance signal of the same first frequency sequence as when the reference was established is applied again. Using the same acquisition method and orthogonal demodulation-type digital lock-in amplification algorithm, the impedance amplitude corresponding to each frequency point of the first frequency sequence under this simulated degradation state is measured and calculated, denoted as . ,in .

[0056] For each frequency point Calculate the relative change in impedance amplitude at this frequency point under simulated degradation conditions relative to the healthy reference conditions. ,Will As the weighting coefficient corresponding to this frequency point All weighting coefficients With sets ,gather Store them together.

[0057] The weighting coefficients are read from the storage location each time a measurement is performed. Calculated using a weighted formula : , in, Indicates all The sum of several weighting coefficients. Through weighted processing, the frequency points that are more sensitive to connector degradation response are... This will contribute more significantly to improving the sensitivity of degradation detection. Second difference value The calculation still uses the arithmetic mean formula.

[0058] As another optional implementation, when calculating the first difference value Second difference Previously, an outlier detection and removal process could be added to eliminate the impact of occasional electromagnetic interference or measurement noise on the calculation results.

[0059] For the set obtained in the current state The elements in Calculate the median and interquartile range The calculation method is as follows: Will Sort by value from smallest to largest. If the number is odd, take the value in the middle position as... ,like If the number is even, take the average of the two middle values ​​as the mean. ; It equals the upper quartile minus the lower quartile. Establish an outlier detection interval, with a lower bound of... The upper boundary is .

[0060] Check each one one by one ,like If the value is less than the lower bound or greater than the upper bound, the frequency point is marked as an outlier and removed from subsequent calculations.

[0061] For sets The elements in The median was calculated using the same steps. and interquartile range Eliminate those that meet the requirements or The frequency point.

[0062] After removing outlier frequency points, the number of valid frequency points retained in the first frequency sequence is denoted as . The number of valid frequency points retained in the second frequency sequence is denoted as . Calculation using only the impedance amplitude data corresponding to the frequency points that were not excluded and If removed Less than the preset minimum number of valid points or If the number of valid points is less than the preset minimum, the test data is considered invalid, the test process is terminated, and the measurement is re-executed in the next preset time period to ensure the reliability of subsequent calculations.

[0063] As another alternative implementation method, after obtaining the degradation degree value Further degradation trend analysis and graded early warning can then be conducted. During the operation of the photovoltaic string, the degradation level value obtained after each measurement will be recorded. Stored together with the detection timestamp.

[0064] The results obtained from each test The values ​​are arranged in chronological order of detection time to form a sequence. ,in This represents the number of tests that have been completed. Indicates the first The degradation level value obtained from the second test.

[0065] when At that time, calculate the degradation rate value. The calculation formula is: , in, The time interval between two consecutive measurements. The unit is per day.

[0066] When the degree of degradation value When the first preset degradation threshold of 0.8 is exceeded, a first-level DC series arc fault warning signal is issued, indicating to the operation and maintenance personnel that the photovoltaic string connector has shown significant degradation and it is recommended to include it in the key inspection plan.

[0067] When the degree of degradation value When the degradation rate exceeds the second preset degradation threshold of 0.95, or when the degradation rate value... When the increment for two consecutive time periods exceeds the preset rate increment threshold of 0.05 days, a second-level DC series arc fault warning signal is issued, indicating that the connector degradation is approaching a critical state or the degradation process is accelerating. It is recommended to arrange shutdown and maintenance as soon as possible.

[0068] The method for determining whether the rate increment continuously exceeds the threshold is as follows: calculate the rate increment of the current period. The rate increment of the previous cycle ,like and If , then the condition is met. When At that time, based only on The comparison result between the value and the first preset degradation threshold is used for early warning judgment, without calculating... value.

[0069] Connector Degradation Mapping Model Medium parameters , , The process of determining is as follows.

[0070] Take several photovoltaic connector samples of the same model as those used in the target photovoltaic string, and build a photovoltaic string simulation test platform in a laboratory environment. The platform includes a photovoltaic module simulation power supply with the same open-circuit voltage and short-circuit current level as the real photovoltaic string, DC cables with the same wire diameter and length as the real photovoltaic string, and multiple connectors.

[0071] Accelerated aging tests were conducted on the connector samples. The accelerated aging tests included a combination of salt spray corrosion tests, temperature cycling tests, and mechanical vibration tests to simulate the various stresses that the connectors would experience during long-term outdoor operation.

[0072] At different aging stages, the aging experiment was paused and the connector sample was removed to measure the contact resistance values ​​at both ends of the connector. .

[0073] Then, the connector was connected to the photovoltaic string simulation test platform. Using the set of reference impedance amplitudes obtained when the connector was connected to the platform before the aging experiment began as a reference, the degradation-specific characteristic parameters were obtained as described above. In the same manner, the degradation-specific characteristic parameters of the connector under the current aging state are obtained. .

[0074] For each aging stage, based on the measured contact resistance value Calculate the degradation level calibration value ,in This is the critical arc contact resistance value determined in advance under the same test conditions.

[0075] The measurement method is as follows: In a photovoltaic string simulation test platform, an adjustable resistor is connected in series to the connector circuit. Under the normal operating voltage condition of the photovoltaic string, the resistance value is gradually increased, while the circuit current and voltage waveforms are monitored. When a continuous high-frequency oscillation is observed in the current waveform or an irregular drop occurs in the voltage waveform, the sum of the adjustable resistance value and the connector's own contact resistance is recorded. This critical resistance value indicates that the connector contact condition has deteriorated to a level sufficient to induce an arc at normal operating voltage.

[0076] The above measurements were performed at multiple different aging stages to obtain multiple sets of data pairs. Each set of data pairs was generated by... and constitute, , This represents the number of valid data sets. (In total) As input, all As the target output, the Levenberg-Marquardt algorithm is used for nonlinear curve fitting to solve for the sum of squared fitting errors. Minimum parameter , , After fitting, a connector degradation mapping model that can be used for online detection is obtained.

[0077] The weighted average calculation method, outlier detection method, and graded early warning method mentioned above can all be selected and used according to the actual application scenario, and can be arbitrarily combined to adapt to the needs of photovoltaic power plants of different sizes or different operation and maintenance strategies.

[0078] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting DC series arc faults in a photovoltaic array, characterized in that, Includes the following steps: S1: When the photovoltaic string is in a healthy state and no DC series arc fault has occurred, inject AC disturbance signals of a first frequency sequence and AC disturbance signals of a second frequency sequence into the DC bus of the photovoltaic string, and obtain the set of reference impedance amplitudes corresponding to each frequency point of the first frequency sequence. and the set of reference impedance amplitudes corresponding to each frequency point in the second frequency sequence. ,Will and The storage serves as a personalized reference for this photovoltaic string; wherein, the frequency band of the first frequency sequence covers the region of resonant frequencies determined by the contact capacitance of each connector in the photovoltaic string and the parasitic inductance of the DC bus, and the frequency band of the second frequency sequence is... to interval and to At least one interval in the interval; S2: During the DC power generation operation of the photovoltaic string, AC disturbance signals of the same first and second frequency sequences as in S1 are injected into the DC bus of the photovoltaic string at a preset time period to obtain the impedance amplitude set corresponding to each frequency point of the first frequency sequence in the current state. and the set of impedance amplitudes corresponding to each frequency point in the second frequency sequence. ; S3: According to and Calculate the first difference value ,according to and Calculate the second difference value ; S4: By and Constructing Degeneracy-Specific Feature Parameters , , For values ​​less than The constant; S5: Will By inputting the connector degradation mapping model, the degradation level of each connector in the current photovoltaic string is obtained. ,when When the preset degradation threshold is reached, a DC series arc fault warning signal is issued.

2. The method for detecting DC series arc faults in photovoltaic arrays according to claim 1, characterized in that, The frequency band of the first frequency sequence is the resonant frequency measured by the photovoltaic string under healthy conditions. Expanding outwards from the center to both sides The interval formed by these.

3. The method for detecting DC series arc faults in photovoltaic arrays according to claim 1, characterized in that, The first difference value in S3 The calculation method is as follows: ， The number of frequency points in the first frequency sequence. For set The Middle The impedance amplitude corresponding to each frequency point For set The Middle The reference impedance amplitude corresponding to each frequency point; Second difference value The calculation method is as follows: ， The number of frequency points in the second frequency sequence. For set The Middle The impedance amplitude corresponding to each frequency point For set The Middle The reference impedance amplitude corresponding to each frequency point.

4. The method for detecting DC series arc faults in photovoltaic arrays according to claim 1, characterized in that, Calculation in S3 At that time, a weighting coefficient is assigned to each frequency point within the first frequency sequence. The following weighted formula is used to calculate... : 。 5. The method for detecting DC series arc faults in photovoltaic arrays according to claim 4, characterized in that, The weighting coefficient The determination method is as follows: When establishing a personalized benchmark, a simulated degradation resistor with a known resistance value is connected in series to both ends of any connector in the photovoltaic string. The relative change in impedance amplitude at each frequency point in the first frequency sequence under the condition that the simulated degradation resistor is connected is recorded relative to the condition that the simulated degradation resistor is not connected. ,Will Weighting coefficients for corresponding frequency points .

6. The method for detecting DC series arc faults in a photovoltaic array according to claim 1 or 3, characterized in that, Calculated in S3 and Previously, for sets and set Anomaly detection is performed on the impedance amplitude corresponding to each frequency point within the range. calculate the median of and interquartile range Remove or Frequency points; calculate the median of and interquartile range Remove or The frequency point.

7. The method for detecting DC series arc faults in photovoltaic arrays according to claim 1, characterized in that, The connector degradation mapping model described in S5 is a function ,in , , The model parameters are determined by conducting accelerated aging tests on photovoltaic connectors of the same model as those used in the photovoltaic string, and measuring the contact resistance value of the connectors at different aging stages. Simultaneously, the degradation-specific characteristic parameters calculated through steps S1 to S4 under the corresponding aging state are acquired. ,by Compared to the pre-determined critical arc contact resistance value The ratio is used as a calibrator for the degree of degradation. For multiple groups and The data was fitted to obtain , , .

8. The method for detecting DC series arc faults in a photovoltaic array according to claim 1 or 7, characterized in that, The degradation level value described in S5 For the range of values to Continuous quantity, This indicates that all connectors in the photovoltaic string are in good condition. This indicates that at least one connector in the photovoltaic string has reached the critical state of a DC series arc fault.

9. The method for detecting DC series arc faults in photovoltaic arrays according to claim 8, characterized in that, Also includes: A sequence of degradation values ​​obtained by repeatedly executing S1 to S5. Perform a difference operation to obtain the degradation rate value. , , The time interval between two consecutive executions of steps S1 to S5.

10. The method for detecting DC series arc faults in a photovoltaic array according to claim 9, characterized in that, when When the degradation exceeds the first preset threshold of 0.8, a first-level DC series arc fault warning signal is issued; when When the second preset degradation threshold of 0.95 is exceeded, or when When the increment for two consecutive time periods exceeds the preset rate increment threshold of 0.05 days, a second-level DC series arc fault warning signal is issued.

Images