Device and method for operating a photovoltaic system

Abstract

The invention relates to a device and to a method for operating a photovoltaic system (1000). The method comprises at least the following steps: detecting (S10) a current value in the photovoltaic system (1000); calculating (S20) an estimated value for an electrical power of a possible arc (1) in the photovoltaic system (1000) using a model, wherein the detected current value is used as an input variable of the model; adapting (S30) a maximum detection time duration (Tmax, Tmax,2) on the basis of the calculated estimated value for the electrical power; detecting (S40) a warning state (W) which indicates a possible occurrence of an arc (1) in the photovoltaic system (1000); continuously measuring (S50) a time duration (td) of the detected warning state (W); triggering (S70) a safety procedure when the continuously measured time duration (td) of the detected warning state (W) reaches the maximum detection time duration (Tmax, Tmax,2); and resetting (S80) the measuring (S50) of the time duration (td) of the detected warning state (W) when the detected warning state (W) comes to an end.

Description

Field of the Invention The invention relates to a device and a method for operating a photovoltaic system, particularly in the event of a possible (i.e., apparent or actual) occurrence of an electric arc. Technical Background Arcing can occur repeatedly in photovoltaic systems. There are extensive regulatory standards that specify how quickly an arc must be extinguished and how frequently such an event may occur before further measures become necessary, such as a long-term shutdown of the photovoltaic system, maintenance by a qualified technician, and the like. While compliance with the applicable regulations is essential, there is also a desire to keep photovoltaic systems in continuous operation with as few restrictions as possible. To this end, for example, work is continuously being done on more precise methods for detecting arcing that result in as few false negatives / false detection (i.e., failure to detect an arc that actually occurred) as possible, while minimizing false positives (i.e., erroneously detecting an arc that did not actually occur). CN 117 421 651 A discloses a method for identifying arc faults, particularly in photovoltaic modules. The method involves capturing a power spectrum of the photovoltaic module and generating a target power spectrum. The acquired data is represented in matrix form (feature matrix) to obtain features through deep convolution and channel shuffling. These are then processed through a plurality of matrix concatenations and channel rearrangements to reduce the dimensionality of the matrix and obtain global features. On this basis, an arc fault is detected. CN 115 021 209 A describes a protection mechanism against short circuits in photovoltaic systems, particularly when the short circuit is caused by a line-to-line fault. To this end, I-V and P-V curves are analyzed immediately after a short circuit, and the number of faulty cells in the photovoltaic system is determined. Based on this, a protection mechanism is activated. In principle, this is intended, among other things, to prevent a fire triggered by an electric arc. Since it is not possible to unambiguously measure an electric arc in real photovoltaic systems in operation, and indirect methods must be used to detect electric arcs, a certain percentage of false decisions (both negative and positive) may occur. Failure to detect an actual electric arc may result in a fire. A false detection, where no electric arc is present, leads to reduced performance and poor customer satisfaction. Therefore, it makes sense to take full advantage of the permitted limits and time window for detection. Various models are known for the general theoretical description of an electric arc, such as those found in the scientific publications by Hertha Marks Ayrton, *The Electric Arc*, New York, D. Van Nostrand Company, 1902; by J. Paukert, “The arc voltage and arc resistance of IV fault arcs,” in *Proceedings of the 7th International Symposium on Switching Arc Phenomena*, 1993, pp. 49-51; or by A. D. Stokes and W. T. Oppenlander, “Electric Arcs in Open Air,” in “Journal of Physics D: Applied Physics,” 1991, pp. 26-35. According to some standards, a safety procedure must be triggered when the electric arc reaches a certain energy level. While this can be precisely measured in a test setup, it is very difficult to do so in photovoltaic systems in operation. Summary of the Invention It is an objective of the present invention to provide an improved method and an improved device for operating a photovoltaic system, in particular for handling potential electric arcs, i.e., apparent or actual electric arcs. This objective is achieved by the subject matter of the independent patent claims. Accordingly, in a first aspect, a method for operating a photovoltaic system is provided, comprising the steps of: detecting a current value in the photovoltaic system; calculating an estimated value for the electrical power of a potential electric arc in the photovoltaic system using a model, wherein the detected current value is used as an input variable for the model; adjusting a maximum detection time duration as a function of the calculated estimated electrical power; Detecting a warning state that indicates a possible occurrence of an electric arc in the photovoltaic system (i.e., determining that an electric arc may have occurred); continuously measuring the time duration of the detected warning state (i.e., measuring how long the detected warning state has persisted since its detection); triggering a safety procedure (in particular for the purpose of extinguishing the possible electric arc) if the continuously measured time duration of the detected warning state reaches the maximum detection time duration; and resetting the measurement of the time duration of the detected warning state when the detected warning state ends (i.e., if another warning state occurs, its time duration is measured again starting from zero). A fundamental concept of the present invention is thus that, in a warning state, i.e., when there are indications of an electric arc, a safety procedure is triggered (preferably only at that precise moment) when a maximum detection time duration has elapsed since the onset of the warning state; this maximum detection time duration, according to the invention, is not constant but is adjusted. The maximum detection time duration is adjusted as a function of a calculated estimate of the electrical power of the potential electric arc, whereby the maximum detection time duration advantageously decreases monotonically or strictly monotonically with the calculated estimate of the electrical power of the potential electric arc. The maximum detection time duration can also be referred to as “triggering time duration,” as it indicates when the safety procedure is to be triggered. This makes it possible, for example, to meet requirements regarding the maximum energy that the electric arc may emit before the safety procedure must be triggered. However, according to the invention, a particularly long period of time is provided before this occurs. This additional time, compared to the prior art, can be used, in particular, to perform further measurements, for example, to determine with greater certainty whether an electric arc is actually present or not, to allow the problem to resolve itself, and / or to better evaluate short-term events in the photovoltaic system within their context. This additional time thus reduces, in particular, the probability that a false-positive detection of an electric arc will result in an unnecessary, disruptive intervention in the operation of the photovoltaic system. In the description of the present invention, the term “possible electric arc” is used repeatedly. This stems from the fact that, in the context of the present invention, initially it will only be determined whether a warning state exists, i.e., a state in which an electric arc may have occurred. There are various methods in the prior art for this purpose (e.g., spectral analyses of the photovoltaic system’s power lines), all of which can be used here. During the warning state, calculations and similar procedures must therefore assume that an electric arc is present, even if this may not actually be the case. Therefore, reference will be made, for example, to the power of a possible electric arc, i.e., the power that this electric arc would have if the warning state described were actually caused by an electric arc. In other words, the “power of a possible electric arc” is not that of just any electric arc, but rather the power of the specific electric arc that is precautionarily assumed to be present during the warning state. The term “photovoltaic system,” as used here, may include, in particular, all elements arranged between the incident solar radiation on the one hand and the alternating current output of an inverter on the other hand, specifically photovoltaic modules, module electronics, protective devices (circuit breakers and the like), and the inverter itself. According to some preferred embodiments, variants, or refinements of embodiments, the maximum detection time duration is continuously adjusted, at least when no warning state is detected. In other words, the method may be configured to continuously determine the current maximum detection time duration, i.e., roughly speaking, how long a warning state could currently persist before the safety procedure would need to be triggered. During a warning state, i.e., when a warning state is currently detected, the maximum detection time duration need not be adjusted. In this case, for example, the most recently adjusted maximum detection time duration can be used as a time threshold to determine whether or not the safety procedure should be triggered. In some embodiments, the adjustment of the maximum detection time duration may continue even during a warning state. In this case, the most recently adjusted maximum detection time duration may also be a maximum detection time duration adjusted during the warning state. Alternatively, upon the occurrence of a warning state, a previously measured and stored current value may be used to calculate the maximum detection time duration specifically for the current warning state. According to some preferred embodiments, variants, or refinements of embodiments, an estimated value for the electrical voltage of the potential electric arc is determined in the model using the detected current value as an input parameter. The estimated value for the electrical power can then be calculated using the detected current value and the estimated value for the electrical voltage, in particular by multiplying them. According to some preferred embodiments, variants, or refinements of embodiments, the detected current value is the only variable input parameter of the model. In this case, the model is relatively easy and quick to compute. In alternative embodiments, variants, or refinements of embodiments, additional input variables may be used in addition to the detected current value, such as a voltage change upon the occurrence of the potential electric arc, a peak in the power density spectrum, and / or the like. According to some preferred embodiments, variants, or refinements of embodiments, the model uses a plurality of candidate submodels to calculate the estimated value for the electrical power of the potential electric arc. In this process, each candidate submodel can perform its own estimation. The model’s estimate of the electrical power may be based on at least one of the estimates from the candidate submodels. For example, the largest estimate (i.e., the maximum) generated by one of the candidate submodels can be used as the model’s estimate. However, it is also possible to calculate the model’s estimate from the estimates of the candidate submodels, for example as their moving average or median, or as a specific quantile x (i.e., a value below which 100*x% of the estimates from the candidate submodels lie, where the median is the special case of the 0.5 quantile). According to some preferred embodiments, variants, or refinements of embodiments, the maximum detection time duration is adjusted such that a product of the estimated electrical power of the potential electric arc and the maximum detection time duration does not exceed a predetermined electrical energy threshold. In other words, the maximum detection time duration can be set such that the specified electrical energy threshold is just reached if the potential electric arc emits electrical power in accordance with the calculated estimated value for the electrical power for the entire maximum detection time duration. In this way, for example, regulatory requirements regarding the maximum energy that the potential electric arc is permitted to discharge before triggering a safety procedure can be implemented in such a way that as much time as possible is available. According to some preferred embodiments, variants, or refinements of embodiments, the current value is detected at an input of an inverter and / or module electronics and / or a protective device of the photovoltaic system. The current value may thus be a measured value that is preferably detected (i.e., measured) regularly or continuously. According to some preferred embodiments, variants, or refinements of embodiments, the detected current value is a current value generated by the photovoltaic system at a point in time prior to the occurrence of the detected warning state, e.g., the last current value measured before the occurrence of the detected warning state (i.e., before the start of the detection of the warning state), the highest current value within a predefined time window prior to the occurrence of the detected warning state, or the like. According to some preferred embodiments, variants, or refinements of embodiments, the current value is measured periodically or continuously. Accordingly, the estimated value for the electrical power can also be calculated at regular intervals or continuously, and thus the detection time can also be advantageously adjusted at regular intervals or continuously, always based on the most recent estimated value for the electrical power, which in turn is always based on the most recently recorded (e.g., measured) current value. According to some preferred embodiments, variants, or refinements of embodiments, the method further comprises verifying the detected warning state during a period in which the duration of the detected warning state is between zero and the maximum detection time duration. Preferably, the warning state is terminated if the verifying step reveals that no electric arc is actually present. Accordingly, the warning state advantageously remains active if the verifying step does not yield a sufficiently clear result or confirms the presence of the warning state. Alternatively, confirming the warning state with sufficiently high probability during verifying may also trigger the immediate execution of a safety procedure (e.g., the next, the mildest, the most severe, or another predetermined safety procedure). According to a second aspect, the invention also provides a device for operating a photovoltaic system, comprising: a current-measuring apparatus configured to detect a current value; a power estimation module configured to calculate an estimated value for the electrical power of a possible (or: potential) electric arc using a model, wherein the detected current value is used as an input to the model; a maximum detection time duration adjustment module configured to adjust a maximum detection time duration based on the calculated estimated value for the electrical power; an arc detection module configured to detect a warning state indicating a possible occurrence of an electric arc in the photovoltaic system; a time-measuring apparatus configured to measure the elapsed time duration of the detected warning state; and a safety apparatus configured to: - execute a safety procedure (in particular for the purpose of extinguishing the potential electric arc), when the continuously measured time duration of the detected warning state reaches the maximum detection time duration, and - reset the measurement of the time duration of the detected warning state by the time-measuring apparatus when the detected warning state ends (for example, is no longer detected). The device according to any embodiment of the second aspect may be adapted in accordance with all embodiments, variants, or refinements of embodiments described in relation to the method according to the first aspect, and vice versa. When the terms “modules” or “apparatuses” are used herein, it is understood that this does not necessarily mean that such modules or apparatuses are configured as separate units distinct from one another. All modules and / or apparatuses may be implemented in hardware and / or software. In cases where modules or apparatuses are implemented as software, the modules or apparatuses may be realized as sections of program code or components of program code, which may be distinguishable from one another but may also be interwoven. Similarly, in cases where one or more modules or apparatuses are implemented as hardware, the functions of one or more modules or apparatuses may be implemented by one and the same hardware component. Alternatively or additionally, various functions of a single module or a single apparatus, or even various functions of different modules or different apparatuses, may be implemented on one or more separate hardware components, which need not necessarily have a one-to-one relationship with the modules or apparatuses. In this sense, any device, system, method, etc., that exhibits all the characteristics and functions attributed to a specific module or a specific apparatus can be understood as having, representing, or implementing such a module or apparatus. In particular, it may be possible for all modules and / or apparatuses to be implemented as program code executed by a computing device, e.g., a server or a cloud computing platform. According to some preferred embodiments, variants, or refinements of embodiments, the current-measuring apparatus is configured to measure the current value periodically or continuously; the power estimation module is configured to calculate the estimated value for the electrical power of the potential (or: possible) electric arc periodically or continuously using the detected current value; and / or the maximum detection time duration adjustment module is configured to adjust the maximum detection time duration periodically or continuously based on the calculated estimated value for the electrical power. The functions of the current-measuring apparatus, the power estimation module, and the maximum detection time duration adjustment module are advantageously synchronized with one another and, most advantageously, all operate continuously or at the same regular interval, for example every D milliseconds, where D may advantageously lie within a range between 1 millisecond and 5 milliseconds (including the boundary values). According to a third aspect, the present invention provides an inverter comprising the device according to one of the embodiments of the second aspect of the present invention. According to a fourth aspect, the present invention provides a photovoltaic system comprising the device according to an embodiment of the second aspect of the present invention and / or the inverter according to an embodiment of the third aspect of the present invention. According to a fifth aspect, the present invention provides a computer program product comprising executable program code which, when executed by a computing device, performs the method according to an embodiment of the first aspect of the present invention. Such a computing device may be or become any device capable of performing computations, and in particular of executing software, an app, or an algorithm. The computing device may, for example, include at least one processor unit, such as a central processing unit (CPU) and / or a graphics processing unit (GPU) and / or a field-programmable gate array (FPGA) and / or an application-specific integrated circuit (ASIC) and / or a combination thereof. The computing device may also include a main memory that is operatively coupled to the at least one processing unit, as well as a non-volatile memory that is operatively coupled to the at least one processing unit and the main memory. The computing device may be implemented entirely or in whole within a local device and / or entirely or in whole within a remote system, such as a remotely located server and / or a cloud computing platform. According to a sixth aspect, the present invention provides a non-volatile, computer-readable data storage medium comprising executable program code which, when executed by a computing device, performs the method according to an embodiment of the first aspect of the present invention. The data storage medium may, for example, be configured as a semiconductor memory, such as an SSD memory chip, or include such a memory chip. The data storage medium may also include or comprise a CD, DVD, Blu-ray, or a magnetic storage device. According to a seventh aspect, the present invention provides a data stream that comprises executable program code or is configured to generate executable program code, which, when executed by a computing device, performs the method according to an embodiment of the first aspect of the present invention. According to an eighth aspect, the present invention provides a computing device configured to perform the method according to an embodiment of the first aspect of the present invention. Further preferred embodiments, variants, and refinements of embodiments are set forth in the dependent patent claims and in the description with reference to the figures. Brief Description of the Figures The invention is explained in more detail below with reference to exemplary embodiments shown in the figures of the drawings. The partially schematic illustrations show the following: Fig. 1 shows a schematic block diagram illustrating a device according to an embodiment of the present invention, as well as an inverter and a photovoltaic system, each according to a further embodiment of the present invention; Fig. 2 shows an illustrative graph explaining the operation of the device shown in Fig. 1; Fig. 3  shows a further exemplary graph illustrating the operation of the device shown in Fig. 1 Fig. 4  shows power density spectra, on the one hand, under normal states and, on the other hand, under the warning state of the device shown in Fig. 1; Fig. 5  shows another illustrative graph explaining the operation of the device shown in Fig. 1; Fig. 6  shows yet another illustrative graph explaining the operation of the device shown in Fig. 1; Fig. 7  shows another illustrative graph explaining the operation of the device shown in Fig. 1; Fig. 8 shows a schematic flowchart illustrating a method according to another embodiment of the present invention; Fig. 9 shows a schematic block diagram illustrating a computer program product according to yet another embodiment of the present invention; and Fig. 10 shows a schematic block diagram illustrating a data storage medium according to yet another embodiment of the present invention. In all figures, identical or functionally equivalent elements and devices (unless otherwise specified) have been designated with the same reference numerals. The numbering of method steps is primarily intended to facilitate their distinction and does not necessarily imply a chronological order, although a chronological order corresponding to the sequence of the numbering is a possible variant. Various method steps may also be carried out simultaneously, either partially or completely. It is also possible to carry out method steps multiple times or iteratively. Detailed Description of the Figures Fig. 1 shows a schematic block diagram illustrating a device according to an embodiment of the present invention, i.e., a device 100 for operating a photovoltaic system 1000. Thus, Fig. 1 also shows a photovoltaic system 1000 according to an embodiment of another aspect of the present invention, as well as an inverter 300 according to an embodiment of a further aspect of the present invention. The photovoltaic system 1000 shown as an example in Fig. 1 comprises at least one photovoltaic module 200 and is depicted here with two photovoltaic modules 200, although the photovoltaic system 1000 may also comprise significantly more photovoltaic modules 200. The photovoltaic system 1000 also comprises an inverter 300, which is configured to convert the direct current (DC) received from the photovoltaic modules 200 (or from at least one photovoltaic module 200) into alternating current (AC). The photovoltaic system 1000 may also include additional inverters 300 as well as other components, such as module electronics on the photovoltaic modules 200, protective devices, and the like. Furthermore, the photovoltaic system 1000 comprises a device 100 for operating the photovoltaic system 1000, the operation of which is explained in more detail below. In Fig. 1, the device 100 is shown as being arranged on or in the (or in one of the) inverters 300. However, the device 100 may also be located on another device or distributed across multiple locations. The device 100 comprises at least one current-measuring apparatus 110, which is configured to measure a current value. The current value can be measured, for example, at an input of the inverter 300, at module electronics, and / or at a protective device of the photovoltaic system 1000. The current value may be measured periodically or continuously, as already explained above, specifically (and optionally also only) when no warning state is currently present that indicates a possible electric arc 1. It is understood that the at least one current-measuring apparatus 110 may also be located outside the inverter 300, depending on which current value is to be measured and where. Multiple current-measuring apparatuses 110 may be provided, and the detected current value may be based on individual current measurements recorded by the multiple currentmeasuring apparatuses 110, e.g., a supremum, a moving average, or a quantile (e.g., a median) of the individual current measurements. The current value may, for example, be the current currently generated by each of the photovoltaic modules 200, as measured at the maximum power point (MPP). Fig. 2 shows an example graph depicting a current value I in A (amperes) as a function of time t. Initially, the photovoltaic system 1000 is operated at the MPP, so that the recorded current value IMPP corresponds to the electrical current at the MPP. Starting at the point in time t0, an electric arc 1 with arc current Iarc is present. As can be seen from Fig. 2, the preceding current value IMPP can be used as an upper limit (or a supremum) for the magnitude of the subsequent arc current Iarc. Accordingly, for example, the last recorded current value before the occurrence of a detected warning state (or, in other words, before the warning state is detected) can be recorded, thereby providing a reasonable estimate of the possible arc current Iarc. For example, a predetermined number of the most recently detected current values (e.g., current values IMPP at the maximum power point, MPP) can be stored in a buffer, which are then used to determine the current value. The device 100 further comprises a power estimation module 120, which is configured to calculate an estimated value for the electrical power of a potential (or: possible) electric arc 1 using a model, wherein the recorded current value is used as an input to the model. Fig. 1 schematically shows two types or possible locations of electric arcs 1: in the series connection between two photovoltaic modules 200, and between a photovoltaic module 200 and an inverter 300. The detected current value may be the sole input parameter of the model, or one of several. Additional input parameters may include, for example, a voltage change upon the occurrence of the potential electric arc 1, a peak in the power density spectrum, and / or similar parameters. In the embodiment described herein, the power estimation module 120 is configured to first calculate an estimated value for the electrical voltage of the potential electric arc 1, i.e., a voltage that an electric arc 1 would exhibit if such an electric arc 1 were currently present. For example, the following formula (1) can be found in the scientific publication by J. Paukert mentioned at the beginning, according to which Uarc = (20 + 0.534%gap) / 0rC2 ,                (1) where Uarc represents the estimated value of the electrical voltage of the potential electric arc 1, and Iarc represents the measured current value, which is assumed here (and hereinafter) to be the electrical current of the potential electric arc 1. Here, xgap is the electrode spacing in millimeters, and is typically unknown for a spontaneous electric arc 1 (occurring outside a test bench) and can therefore be replaced by a fixed estimated value, for example, an estimated value of 0. It can be seen from Equation (1) that, for typical electrode distances in typical photovoltaic systems, the resulting uncertainty is in the single-digit percentage range. Accordingly, the estimated value Uarc for the electrical voltage of the possible electric arc 1 can be calculated, for example, using the following formula (2): Uarc = 20 10^2,          (2) In other words, the estimated value Uarc for the electrical voltage of the potential electric arc 1 may comprise or consist of a power of the measured current value multiplied by a constant. Alternatively, the following formula (3) can be derived from a model in the aforementioned scientific publication by H.M. Ayrton: Uarc = A + BXgap + ^^gap,        (3) arc where A is the voltage drop across the electrode, B is the voltage gradient, and C and D are constants that describe the nonlinear impedance response of electric arc 1, which are typically determined empirically. The estimated value Uarc for the electrical voltage of the potential electric arc 1 can thus be calculated (e.g., according to Formula (3)) as a sum, wherein the measured current value Iarc appears in the denominator of one of the summands of the sum or forms this denominator. It goes without saying that, depending on the planned location of use and the design of the photovoltaic system 1000, other known formulas for calculating the estimated value Uarc for the electrical voltage of the potential electric arc 1 may also be used, for example, from the aforementioned scientific publication by Stokes and Oppenlander. From the estimated value Uarc for the voltage of the potential electric arc 1 and the measured current value as the estimated value Iarc for the current of the potential electric arc 1, the power estimation module 120 can, in accordance with Parc= Uarc Iarc       (4) calculate the estimated value Parc for the electrical power of the potential (or: possible) electric arc 1. A model for calculating the estimated value Parc can thus consist of, or include, a formula (or a submodel) for calculating the estimated value Uarc for the electrical voltage of the potential electric arc 1, in combination with formula (4), or comprise such a formula or submodel. Alternatively, a model for calculating the estimated value Parc for the electrical power of the potential electric arc 1 may also directly be or include a formula or a submodel for a power calculation. The device 100 further comprises a maximum detection time duration adjustment module 130, which is configured to adjust a maximum detection time duration Tmax as a function of the calculated estimated value Parc for the electrical power. The function of the maximum detection time duration Tmax has already been touched upon above and will be described in greater detail below. The model for calculating the estimate Parc of the electrical power may use a plurality of candidate submodels, wherein each candidate submodel performs its own estimation, and the estimate of the model for the electrical power is based on at least one of the estimates from the candidate submodels. For example, the largest estimate (i.e., the maximum) generated by one of the candidate submodels can be used as the model’s estimate. However, it is also possible to calculate the model’s estimate from the estimates of the candidate submodels, for example, as their moving average or median, or as a specific quantile x (i.e., a value below which 100*x% of the estimates from the candidate submodels lie, where the median is the special case of the 0.5 quantile). Equivalently, a plurality of candidate submodels can be used to generate the estimate Uarc for the electrical voltage of the potential electric arc 1, whereby each candidate submodel generates estimates for Uarc, and the estimate for Uarc, which is then used to calculate the estimate for the power Parc, is calculated from the estimates of the candidate submodels for Uarc, for example as their moving average or median, or as a specific quantile x. The device 100 further comprises an arc detection module 140, which is configured to detect a warning state indicating a possible occurrence of an electric arc 1 in the photovoltaic array. In other words, a warning state exists, for example, when, for safety reasons, it must be assumed that an electric arc 1 is present, even if it is not known whether this is actually the case. The detection of the warning state by the arc detection module 140 may be performed in accordance with any method known in the art for detecting an electric arc 1 and may, for example, include a Fourier transform and a spectral analysis. Fig. 3 shows a graph in which the vertical axis represents frequencies (in kHz) of electrical signals in at least one conductor of the photovoltaic system 1000, the horizontal axis represents a time course (in ms), and the color of the points indicates the intensity (in dB) of the signals at a specific frequency at a specific point in time, with a darker color indicating higher intensity. The electrical signals can, for example, be tapped from a negative conductor of the photovoltaic system 1000 (such as a common negative conductor), although, depending on the topology, other and / or multiple measurement points are also possible. In Fig. 3, as an example and schematically, a frequency range between 0 kHz and 100 kHz is shown on the vertical axis. It is understood that a wider, narrower, and / or shifted frequency range (or several separate sub-frequency ranges) can also be used to detect a warning state W. The photovoltaic system 1000 according to the invention may advantageously include powerline communication for control or communication signals without impairing the functions of the device 100 in any way. Fig. 3 clearly shows how the overall frequency response in the photovoltaic system 1000 changes abruptly at the point in time t0, and in particular, how the overall intensities are significantly higher. Since electric arcs 1 typically generate pink noise (similar to the “white noise”) in the electromagnetic spectrum, this can be interpreted as an indication of the occurrence of an electric arc 1 at the point in time t0. Accordingly, the arc detection module 140 may be configured to detect the presence of a warning state W at the point in time t0. Fig. 4 shows, by way of example, two graphs, one of which represents a power density spectrum FN in the normal state N of the photovoltaic system 1000, and the other a power density spectrum FW in the warning state W of the photovoltaic system 1000. Accordingly, the graph with the frequency spectrum FN can be viewed, for example, as a vertical cross-section through Fig. 3 in the region of the normal state N, e.g., at 1200 ms, and the graph with the frequency spectrum FW can be viewed, for example, as a vertical cross-section in the region of the warning state W, e.g., at 1700 ms. Typically, standards stipulate that a safety procedure (such as shutting down the photovoltaic system 1000) must not be executed immediately upon the occurrence of a warning state W (i.e., upon detection of a possible electric arc). As already explained above, a maximum energy quantity Emax is often allowed instead; that is, the electrical energy Earc emitted by the (possible or presumed) electric arc 1 may reach a maximum value of Emax before the safety procedure must be executed. Fig. 5 shows two example graphs, one of which depicts the power Parc of the potential electric arc 1 (left vertical axis, in watts) and the other the total energy Earc emitted by the electric arc 1 (right vertical axis, in joules), each as a function of time. The total energy Earc emitted can be calculated by integrating the power Parc. According to the invention, the fact that, given a known estimated value for the power Parc of the potential electric arc 1 and a specified maximum energy Emax, the maximum detection time duration Tmax can be calculated by the maximum detection time duration adjustment module 130 is now utilized. This thus specifies how long a warning state W before the required safety procedure whereby the following may apply in a rp     _ ^arc lmax = ,,  , arc arc wherein Earc is a constant, Uarc is the may (or can) be tolerated must be triggered, simple case: (5), estimated value for the electrical voltage of the potential electric arc 1, and Iarc is the estimated value for the electrical current, given by the detected current value (or calculated based on the detected current value) of the potential electric arc 1. Fig. 6 shows a version of the graph in Fig. 3 in which, among other things, the maximum detection time duration Tmax (corresponding, in the present example, to an energy Earc of 200 J emitted by the electric arc 1) has been plotted as a vertical line. It is understood that, according to the invention, the maximum detection time duration Tmax (at least when no warning state W is currently present, optionally always) is adjusted by the maximum detection time duration adjustment module 130 (in particular periodically or, more preferably, continuously) depending on the currently detected current value, which can accordingly also be detected (in particular, measured) periodically or, more preferably, continuously. Fig. 7 shows, by way of example, a graph illustrating how the maximum detection time duration Tmax is adjusted by the maximum detection time duration adjustment module 130 depending on the currently detected current value I1, I2, I3, with I1 < I2 < I3. A default value of the maximum detection time duration Tmax,0 can be provided, which is used for the maximum detection time duration Tmax when no usable measured current value is currently available. The maximum detection time duration adjustment module 130 may, for example, be configured to use the default value of the maximum detection time duration Tmax,0 if the recorded current value is not classified as plausible (i.e., does not meet predetermined plausibility rules), was recorded too long ago (i.e., is older than a predetermined time duration period), no recorded current value is available at all (e.g., in the event of a communication error), and / or similar circumstances. The device 100 also includes a time-measuring apparatus 150 that is configured to measure a continuous time duration td of the detected warning state W, or, in other words, to measure how long the detected warning state W has already lasted. This can be accomplished, for example, by starting a counter Z at the beginning of the detection of the warning state W, which counts fixed time intervals D (for example, with D in the range between 1 and 5 milliseconds, inclusive of the boundary values), i.e., approximately td = Z*D starting at t0. Alternatively, this can be achieved by recording the absolute point in time t0 at which the warning state W is detected, and subtracting this recorded absolute point in time from the current time t, i.e., td = t - t0. The device 100 also includes a safety apparatus 160 that is configured to execute a safety procedure for the purpose of extinguishing the potential electric arc 1 when the continuously measured time duration td of the detected warning state W reaches the maximum detection time duration Tmax, i.e., when td = Tmax. Preferably, therefore, no safety procedure is triggered (or only a different, in particular milder, safety procedure is triggered) as long as td ^ Tmax. In particular, preferably no shutdown of a photovoltaic module 200 or the photovoltaic system 1000 as a whole occurs as long as td ^ max. As Fig. 7 illustrates particularly well, the safety procedure is thus triggered relatively early if a relatively high current value I3 is detected, and relatively late if a relatively low current value I1 is detected, since the maximum detection time duration Tmax changes accordingly in each case. The safety procedure may comprise one or more measures (known from the prior art) intended to extinguish the potential electric arc 1, up to and including shutting down the photovoltaic system 100. The safety procedure may include, in each case, sending a message to a central control unit, for example to a virtual power plant operator, a maintenance technician, and / or the like. It may be provided that the safety apparatus 160 is configured to reset the measurement of the time duration td of the detected warning state W by the time-measuring apparatus 150 when the detected warning state W ends, i.e., is no longer detected, or, in other words, when the indications that pointed to the occurrence of an electric arc are no longer (to a sufficient extent) present. Advantageously, the period during which the time duration td of the detected warning state W increases from 0 to the maximum detection time duration Tmax is used by the arc detection module 140 to (further) verify the detected warning state W, i.e., to determine whether an electric arc 2 is actually present or not. To this end, for example, additional measurements (such as of the frequency spectrum) can be taken, or additional current values can be recorded; additional models can be applied and their results analyzed (e.g., a machine learning model such as an artificial intelligence entity, for example, an artificial neural network); more timeconsuming calculations than those required for the initial detection of the warning state W can be performed, and / or the like. If the step of verifying reveals that no electric arc 2 is actually present, the warning state W is preferably terminated immediately, so that, ideally, operation can continue in the normal state N. If, despite (or: in the course of) the verification, a warning state W is still present (i.e., in particular, if an electric arc 2 is considered possible or probable), the detected warning state W remains in effect, and the time duration time td continues to be incremented. Multiple maximum detection time durations Tmax may also be defined, in which case the safety apparatus 160 is configured to trigger, upon reaching each of the successive maximum detection time durations Tmax, a different and / or additional and / or renewed safety procedure, typically with increasing effectiveness in extinguishing the electric arc 1 but at the same time with greater intervention in the operation of the photovoltaic system 1000. For example, in Fig. 6, in addition to the first maximum detection time duration Tmax, which here corresponds to an (exemplary) maximum energy of Earc =200 J, a second maximum detection time duration Tmax,2 is also shown, which here corresponds to an (exemplary) maximum energy of Earc= 750 J. Thus, the safety apparatus 160 may be configured to first execute a first, milder safety procedure when the continuously measured time duration td of the warning state W reaches the first maximum detection time duration Tmax (i.e., when td = Tmax). In the event that this safety procedure was insufficient to resolve the warning state W (i.e., there is still a possibility that an electric arc 1 is present according to the assessment of the arc detection module 140), the timemeasuring apparatus 150 continues to measure (or: count) the time. If the continuously measured time duration td of the detected warning state W then reaches the second maximum detection time duration Tmax,2 (i.e., td = Tmax,2), a second, more stringent safety procedure is executed, such as shutting down the photovoltaic system 1000 (e.g., using one or more DC and / or AC disconnect switches), automatically summoning a maintenance technician, and / or the like. In contrast, the (relatively) milder safety procedure may, for example, involve a temporary reduction in DC power consumption via the lines associated with the detected current value (e.g., from the photovoltaic modules 200), for example, to a value below a threshold value between 200 mA and 300 mA (for example, below a threshold value of 250 mA). If multiple maximum detection time durations Tmax, Tmax,2, each associated with different safety procedures (i.e., differing in at least one safety procedure), are provided, all provided maximum detection time durations Tmax, Tmax,2, are preferably adjusted simultaneously. However, it may also be provided that at least one longer maximum detection time duration Tmax,2 is adjusted less frequently than at least one shorter maximum detection time duration Tmax. For example, it may be provided that the at least one longer maximum detection time duration Tmax,2 is adjusted only when a warning state W has already persisted for a time duration td between 0 and the at least one shorter maximum detection time duration Tmax. Fig. 8 shows a schematic flowchart illustrating a method according to a further embodiment of the present invention, i.e., a method for operating a photovoltaic system 1000. The method can be carried out in particular with the device 100 of the invention and / or in the photovoltaic system 1000 of the invention, but also independently thereof. Accordingly, the method can be adapted in accordance with all options, variants, and refinements described in relation to the device 100 of the invention and the photovoltaic system 1000 of the invention, and vice versa. Accordingly, for the description of the method according to Fig. 8 reference will be made to the reference symbols from the preceding Figs. 1-7, without this being intended to be limiting. In a step S10, a current value is detected in the photovoltaic system 1000, for example as explained above with reference to the function of the current-measuring apparatus 110. The detected current value is preferably detected (in particular, measured) regularly or continuously, at least while no warning state W is present, and is particularly preferably a current value generated by the photovoltaic system 1000 at a point in time prior to the occurrence of the detected warning state W. In the case of multiple module strings in the photovoltaic system 1000, the present method can be carried out jointly for all module strings, or individually for each module string. Accordingly, the recorded current value may also be, for example, a current value at one of the module strings. In a step S20, an estimated value Parc for the electrical power is calculated using a model, wherein the detected current value is used as an input variable (either the sole input variable or one of several input variables) of the model, as described above with reference to the power estimation module 120. In a step S30, (at least) one maximum detection time duration Tmax, Tmax,2 is adjusted as a function of the calculated estimated value Parc for the electrical power, for example as explained above with reference to the maximum detection time duration adjustment module 130. In particular, this can be done periodically or continuously, at least when no warning state W is currently present (i.e., no warning state W is detected). In a step S40, a warning state W is detected, which indicates a possible occurrence of an electric arc 1 in the photovoltaic system 1000, as explained above with reference to the arc detection module 140. As previously explained, one of the methods known in the prior art can be used for this purpose, such as a method based on a spectral analysis of the current from the photovoltaic system 1000 (more specifically: from one or more photovoltaic modules 200) or the like. In a step S50, a time duration td of the detected warning state W, if present, is continuously measured, as explained above with reference to the time-measuring apparatus 150. In a step S60, the time period during which the time duration td of the detected warning state W rises from 0 to the maximum detection time duration Tmax, is used to verify the detected warning state W, i.e., to determine (or at least to make the best possible attempt to do so) with greater accuracy (or even or not. To this end, for example, additional measurements (such as of the frequency spectrum) can be taken, additional models can be applied and their results analyzed (e.g., a machine learning model such as an artificial intelligence entity, for example an artificial neural network), and more time-consuming calculations than those required for the initial detection S40 of the warning state W can be performed, and / or the like. The step of verifying S60 can, for example, also be performed by the arc detection module 140. If the step of verifying S60 determines that there is indeed no electric arc 2, the warning state W is preferably terminated immediately so that, ideally, operation can continue in the normal state N (see also step S80 below). The step of verifying S60 is preferably performed continuously until the maximum detection time duration Tmax is reached, so that every possible opportunity is utilized to avoid triggering an unnecessary safety procedure. During the step of verifying S60, data can thus be continuously aggregated to produce an increasingly reliable result on a regular (or iterative) basis. If, despite (or: in the course of) the verification S60, a warning state W persists (i.e., in particular, if an electric arc 2 is considered possible or probable), the detected warning state W remains in effect, and the time duration td continues to be incremented (in the continuously running step S50). In an optional variant, if the presence of the electric arc 2 is determined with certainty (or with a probability above a threshold value) during the verification S60, a safety procedure may also be triggered immediately. In a step S70, a safety procedure, specifically for the triggered if the continuously measured time duration td of the detected warning state W reaches the maximum detection time duration Tmax, as described above with reference to the safety apparatus 160. The safety procedure may, for example, involve reducing the current generated by one or more (or all) photovoltaic modules 200, shutting down one or more (or all) strings of the photovoltaic system 100, or shutting down the entire photovoltaic system 100, or the like. During the warning state, as long as the maximum detection time duration Tmax has not yet been reached, operation of the photovoltaic system 1000 can, for example, continue unchanged. In this way, if the warning state W turns out to be a false positive detection after all, it is possible to prevent the operation of the photovoltaic system 1000 from being unduly restricted. In a step S80, the step S50 of measuring the time duration td of the detected warning state W is reset when the detected warning state W ends, similar to what was described above with regard to the safety apparatus 160. Thus, the photovoltaic system 1000 is returned to its initial state (the normal state N) when a warning state W is no longer detected, and the process can continue uninterrupted. Preferably, the maximum detection time duration Tmax is continuously (or dynamically) adjusted during the normal state N, but during the warning state W, the most recently determined maximum detection time duration Tmax is retained and used. Fig. 9 shows a schematic block diagram of a computer program product 400 according to an embodiment of the present invention. The computer program product 400 comprises executable program code 450, which is configured to perform, when executed, the method according to an embodiment of the present invention, for example as shown in Fig. 8. Fig. 10 shows a schematic block diagram of a non-volatile computer-readable data storage medium 500 according to an embodiment of the present invention. The data storage medium 500 comprises executable program code 550, which is configured, when executed, to perform the method according to an embodiment of the present invention, for example as shown in Fig. 8. The non-volatile computer-readable data storage medium 300 may, for example, be configured as a semiconductor memory, such as an SSD memory chip, or include such a memory chip. The data storage medium 300 may also include or comprise a CD, DVD, Blu-ray, or a magnetic storage device.

Claims

1. A method for operating a photovoltaic system (1000), comprising the steps of:detecting (S10) a current value in the photovoltaic system (1000);calculating (S20) an estimated value for the electrical power of a potential electric arc (1) in the photovoltaic system (1000) using a model, wherein the detected current value isused as an input parameter of the model;adjusting (S30) a maximum detection time duration (Tmax, Tmax,2)based on the calculated estimate of the electrical power;detecting (S40) a warning state (W) that indicates a possible occurrence of an electric arc (1) in the photovoltaic system (1000);continuously measuring (S50) a time duration (td) of thedetected warning state (W);triggering (S70) a safety procedure if the continuously measured time duration (td) of the detected warning state (W)reaches the maximum detection time duration (Tmax, Tmax,2); and resetting (S80) the measuring (S50) of the time duration (td)of the detected warning state (W) when the detected warningstate (W) ends.

2. The method according to claim 1, wherein the maximum detection time duration (Tmax, Tmax,2) is continuously adjusted, at least when no warning state (W) is detected.

3. The method according to claim 1 or 2, wherein, in the model, an estimated value for the electrical voltage of the potential electric arc (1) is determined using the detected current value as an input variable, and theestimated value for the electrical power is calculated using the detected current value and the estimated value for the electrical voltage, in particular by multiplying them.

4. The method according to any one of claims 1 to 3,wherein the detected current value is the only variable inputparameter of the model.

5. The method according to any one of claims 1 to 4,wherein the model uses a plurality of candidate submodels tocalculate the estimated value for the electrical power, wherein each candidate submodel performs its own estimation,and the estimated value of the model for the electrical power is based on at least one of the estimates from the candidate submodels.

6. The method according to any one of claims 1 to 5,wherein the maximum detection time duration (Tmax, Tmax,2) is adjusted such that the product of the estimated electricalpower of the potential electric arc and the maximum detection time duration (Tmax, Tmax,2) does not exceed a predetermined electrical energy threshold.

7. The method according to any one of claims 1 to 6,wherein the current value is detected at an input of aninverter (300) and / or module electronics and / or a protectivedevice of the photovoltaic system (1000).

8. The method according to claim 7,wherein the detected current value is a current value generated by the photovoltaic system (1000) at a point in timeprior to the occurrence of the detected warning state (W).

9. The method according to claim 7 or 8,wherein the current value is detected periodically or continuously, the estimated value for the electrical power is calculated periodically or continuously, and the detection time duration is adjusted periodically or continuously.

10. The method according to any one of claims 1 through 9, further comprising verifying (S60) the detected warning state (W) during a period in which the time duration (td) of the detected warning state (W) is between zero and the maximum detection time duration (Tmax), wherein, preferably, the warning state (W) is terminated if the verification (S60) reveals that no electric arc (2) is actually present.

11. A device (100) for operating a photovoltaic system (1000), comprising:a current-measuring apparatus (110), which is configured to detect a current value;a power estimation module (120) configured to calculate an estimated value for the electrical power of a possible electric arc (1) using a model, wherein the detected current value is used as an input to the model;a maximum detection time duration adjustment module (130) configured to adjust a maximum detection time duration (Tmax, Tmax,2) based on the calculated estimated value of the electrical power;an arc detection module (140) configured to detect a warning state (W), which indicates the possible occurrence of an electric arc (1) in the photovoltaic system (1000);a time-measuring apparatus (150), which is configured to measure a continuous time duration (td) of the detected warning state (W); anda safety apparatus (160), configured to:- execute a safety procedure for the purpose of extinguishing the potential electric arc (1) when thecontinuously measured time duration (td) of the detected warning state (W) reaches the maximum detection time duration (Tmax,  Tmax,2)  and- reset the measurement of the time duration (td) of the detected warning state (W) by the time-measuring apparatus (160) when the detected warning state (W) ends.

12. The device (100) according to claim 11, whereinthe current-measuring apparatus (110) is configured to measure the current value periodically or continuously;the power estimation module (120) is configured to calculate the estimated value for the electrical power of the potential electric arc (1) periodically or continuously using the detected current value; and / orthe maximum detection time duration adjustment module (130) is configured to adjust the maximum detection time duration (Tmax, Tmax,2) periodically or continuously as afunction of the calculated estimated value for the electrical power.

13. An inverter (300) comprising the device according to any one of claims 11 or 12.

14. A photovoltaic system (1000) comprising the device (100) according to any one of claims 11 or 12 and / or the inverteraccording to claim 13.

15. A computer program product (400) comprising executable program code (450) which, when executed by a computing device,performs the method according to any one of claims 1 through 10.

Images