Fault detection in an electric system

Abstract

A method for detecting a fault in an electric system (400) comprises determining (1010, 1012) a distribution of signal values, which are based on at least one electric signal of the electric system (400), over a plurality of signal segments of the at least one electric signal. Each of the plurality of signal segments comprises at least one phase interval. Each signal value is assigned to one of the at least one phase interval. The method also comprises determining (1014, 1016) whether there is a fault, based on the distribution of the signal values in the at least one phase interval.

Description

[0001]Fault detection in an electrical system FIELD OF THE INVENTION The present invention relates to the field of fault detection in an electrical system, in particular to a method and a device for detecting a fault in an electrical system, for example to improve personnel protection. BACKGROUND OF THE INVENTION Test systems are used during the commissioning and maintenance of operating equipment in electrical systems, for example power generation systems or transformer substations. For example, high-voltage insulation measurements, capacitance measurements, and dissipation factor measurements are established methods for testing the condition of high-voltage equipment, such as transformer insulation or bushing insulation. High test voltages, in particular high AC voltages, of 1000 V or more are used. A measuring current can be several 100 mA.For measurements on, for example, current transformers, lines, cables, earthing systems, or circuit breakers, high test currents of several hundred A or more can be used. Corresponding test systems are typically designed to protect the user from the high test voltage or current, for example, through suitable insulation of the test system and the test leads used. However, there may be points or areas on the device under test itself, for example, on a transformer or other component of the electrical system, that may be subjected to high voltage or high current during the test.To protect a person from the risk of electric shock during testing, the test area can be fenced off, for example, and warning signals can be issued, such as visual warning signals with a red light or a flashing light, or acoustic warning signals, such as a loud warning tone. Personal protection is therefore based on the assumption that all persons involved in the test and all persons in the test area are familiar with and comply with the relevant safety regulations. Despite these safety regulations, accidents do occur again and again, for example when safety regulations are not observed or circumvented. Due to the high voltages and / or currents used during testing, there is a significant risk to life if a person comes into contact with live parts.If a person's life is in danger due to contact with live parts, rapid detection and shutdown is of paramount importance in order to minimize the extent of injury and improve the chances of survival. At the same time, unnecessary interruptions to testing due to spurious shutdowns must be avoided. Modern fault protection relays, for example, are capable of a response time of approximately 7 ms. For a 50 Hz signal, this is less than half a signal period, but still too slow to prevent fatal accidents with currents above 500 mA. IEC standard 60479-1 specifies various time / current zones AC-1 to AC-4 for alternating currents. These zones define the effects of alternating currents (in the range 15 Hz to 100 Hz) on persons for a current path corresponding to a path from the left hand to the feet. Zone AC-1 applies to a body current of up to 0.5 mA for any exposure duration.In zone AC-1, perception is possible, but a "startle" reaction usually does not occur. Zone AC-2 covers a range in which the maximum body current depends on the exposure duration. For an exposure duration of, for example, 10 ms, the maximum body current is 200 mA; for an exposure duration of, for example, 10 seconds, the maximum body current is only 5 mA. In zone AC-2, perceptions and involuntary muscle contractions are likely, but normally no harmful physiological effects occur. Zone AC-3 covers a range in which the maximum body current also depends on the exposure duration. For a short exposure duration of, for example, 10 ms, the maximum body current is 500 mA; for an exposure duration of, for example, 1 second, the maximum body current is only 50 mA.In zone AC-3, strong involuntary muscle contractions, difficulty breathing, reversible disturbances in cardiac function, and immobilization can occur. As the current intensity increases, the effects increase, but organic damage is normally not expected. Zone AC-4 follows zone AC-3 in the direction of higher currents. Pathophysiological effects such as cardiac arrest, respiratory arrest, burns, or other cellular damage can occur. The probability of ventricular fibrillation increases with current intensity and exposure duration, up to a probability of over 50%. With an exposure duration of less than 200 ms, ventricular fibrillation is only triggered within the vulnerable period if certain thresholds are exceeded. Regarding ventricular fibrillation, the effects of the current depend on its path through the body.If the exposure duration is very short, for example less than 10 ms, alternating current no longer poses a hazard. In this case, IEC 60479-2 on the effects of short current pulses can be applied. For example, IEC 60479-2 defines risk probabilities for ventricular fibrillation depending on a current flow from the left hand to the feet and the duration of the current pulse. The three curves c1, c2, and c3 shown in Fig. 1 illustrate these probabilities. Below curve c1, the probability of ventricular fibrillation is almost zero; between curves c1 and c2, the probability is up to 5%; between curves c2 and c3, the probability is up to 50%; and above curve c3, the probability is over 50%. The goal is therefore to stay below curve c1.Reducing the exposure time, i.e., the reaction time from the beginning of contact with the person until the current is switched off, can have a significant impact. For example, reducing the reaction time from 2 ms to 1 ms at a current of 1000 mA can be crucial to preventing ventricular fibrillation. The level of the effective current through the body depends on several parameters, such as the voltage V and a power limit P. max the amplifier of the test system , a source impedance ZQ, the impedance ZB of the body, the period of the cardiac cycle, the signal phase ϕ fat the time of occurrence of the fault or contact, an impedance of the device under test Zdut and a pulse duration ti. In a test system, the exposure duration is essentially the only quantity that can be influenced by a safety device of the test system. Properties of the test signal are generally predetermined and are limited only by the output power of the test system. Fig. 2 shows an equivalent circuit for an exemplary test setup that corresponds to a person's contact with an electrical system during a test. In this test setup, ZQ≈ 100 mΩ is the source impedance of, for example, a test system, ZB= 2 kΩ is the impedance of the person's body, and Z dut = C is a capacitance representing the device under test. During the high voltage output of V = 2 kV, Z B"switched on" to simulate body contact. Fig. 3 shows a waveform of the current I measured at the output of the test system. At the high-voltage contact of Z B a current peak of Î = −2.35^ occurs. According to IEC 60479-2, for sinusoidal pulses, the effective total current = 1.66^. If the physical contact is after, for example, ^ ^ = 1 ^^ is detected (range 302) and the test current is then switched off, the capacitance C is then measured for further ^ ^^^^ ^^^^ = 3.5 ^^ discharged (range 304). The total pulse time is ^ ^^^^ = ^ ^ + ^ ^^^^ ^^^^ = 4.5 ^^. Here ZQ is small, so that ZQ << ZB and thus the discharge time is given by the current I >> I B RMS and the significant pulse time here is ti = tf = 1 ms, in which only half of I AC RMS totflows. According to Fig. 1, regardless of how I is divided into IB and IC, an amount of IAC RMS = 830 mA and ti = 1 ms is below the limit value c1, ie no ventricular fibrillation occurs. Rapid detection of body contact is therefore desirable. SUMMARY OF THE INVENTION The object of the present invention is therefore to provide rapid and reliable fault detection in an electrical system which can be implemented simply and inexpensively and avoids false shutdowns. According to the present invention, a method and a device for detecting a fault in an electrical system are provided, as defined in the independent claims. The dependent claims define embodiments of the invention. A method for detecting a fault in an electrical system comprises determining a distribution of signal values.The signal values ​​are based on at least one electrical signal of the electrical installation. The distribution of the signal values ​​is determined across a plurality of signal sections of the at least one electrical signal. Each of these signal sections comprises one or more phase intervals. Each signal value is assigned to one of these phase intervals. The method further comprises determining whether a fault exists depending on the distribution of the signal values ​​in the phase intervals. The fault may, for example, be a fault current flowing through physical contact between a person and a conductor of the electrical installation. In other examples, the fault may be a fault voltage or fault power occurring at a component of the electrical installation, in particular due to contact between a person and the electrical installation.The electrical system can, for example, be a high-voltage power supply system to which a test system is connected, which generates a test signal, for example a test current or a test voltage, to test the high-voltage power supply system or components thereof as part of a commissioning or maintenance measure. The signal values, the distribution of which is determined across the plurality of signal sections, can, for example, each comprise a voltage value of the electrical signal, a current value of the electrical signal, or a combination of the current value and the voltage value.In other examples, the signal values ​​can be any combination, for example a mathematical combination, of several signals from different sources, for example several signals which are sampled at different points in the electrical system or whose signal waveform is generated or at least influenced by the test system and is therefore known. In particular in high-voltage systems or power supply systems, alternating voltages and / or alternating currents are often used as test signals for testing the systems, so that currents, voltages and powers occurring in the systems which are generated by the test signals are also alternating signals. In the case of alternating signals, a signal section can, for example, be a period of the alternating signal. The period of the alternating signal can be divided into a specific number of phase intervals.For example, a respective signal section or period can be divided into a predetermined number of non-overlapping phase intervals. The total length of the phase intervals into which the respective signal section or period is divided can correspond to the length of the respective signal section or period. The predetermined number of phase intervals in a signal section or period can be determined or set depending on a sampling rate and a fundamental frequency of the electrical signal. The predetermined number can, for example, be in a range from 4 to 100. For example, a period can be divided into 16 phase intervals. The phase intervals can each have the same temporal length or be of different lengths.For example, the signal values ​​can be based on an electrical signal, such as a voltage signal, which is carried by a conductor of the electrical system. The signal values ​​can also be based on several electrical signals that are sampled at several points on the electrical system. A respective signal value can, for example, be formed from a combination of several electrical signals. The signal values ​​can also be based on simulated electrical signals or on values ​​derived from one or more electrical signals. The electrical signal can be sampled at a sampling frequency that is significantly higher than a fundamental frequency of the electrical signal. For example, the electrical signal can have a fundamental frequency of 50 Hz and be sampled at a sampling frequency of, for example, 3000 Hz.In this example, the period of the electrical signal is 20 ms and is divided into, for example, 20 phase intervals of 1 ms each. Thus, 60 sampled values ​​are acquired during one period and assigned to the 20 phase intervals. The first three sampled values ​​are assigned to the first phase interval, the next three sampled values ​​are assigned to the second phase interval, and so on. This process is repeated for a plurality of periods, so that in this example, the number of sampled values ​​assigned to each phase interval increases by three with each period. For each phase interval, a corresponding distribution of the signal values ​​assigned to the phase interval, for example, the assigned sampled values, can be determined. The distribution can, for example, be a statistical distribution.In particular, for example, a statistical mean of the signal values ​​assigned to the respective phase interval across the plurality of signal sections and a statistical dispersion of the signal values ​​assigned to the respective phase interval across the plurality of signal sections can be determined for each phase interval. The statistical dispersion can, for example, be determined as the standard deviation from the mean. The statistical values, for example mean and dispersion or standard deviation, can be updated with each newly added signal value (sampled value). In principle, the method is also applicable when using direct voltages or direct currents. In this case, the length of a signal section can, for example, be chosen arbitrarily or depending on test patterns used when testing the system. Each signal section can, in turn, comprise one or more phase intervals.The plurality of signal sections can comprise a plurality of preferably equally long and directly consecutive signal sections. However, the signal sections can also be of different lengths, and time gaps can be provided between the signal sections. In one example, the signal sections can be of equal length and each comprise a phase interval into which all signal values ​​can fall. The distribution of the signal values ​​is then, for example, a statistic about the nominally constant value of the electrical signal. The distribution can, for example, comprise a statistical mean and a statistical variance. Outliers, which can be identified based on the statistical mean and the statistical variance, can indicate an error.In one embodiment, to determine whether a fault is present, for at least one of the at least one phase intervals, a signal value assigned to the phase interval is compared with the statistical values ​​determined for the phase interval. In particular, a signal value last assigned to the phase interval can be compared with the static values ​​previously determined for the phase interval. Such a comparison can be performed for each phase interval. If the signal value last assigned to the phase interval deviates from the signal values ​​assigned up to that point by more than a predetermined amount, this can be used to detect a fault, for example, a fault current due to a short-circuit between the elements.A deviation beyond a specified level can be detected, for example, by determining an outlier measure for the respective phase interval as a function of the signal value in question (e.g., the last assigned or recorded signal value) as well as the mean value and the scatter or standard deviation of the phase interval. The outlier measure determined in this way can be compared to a specified threshold value. If the outlier measure exceeds the specified threshold value, this can indicate the detection of the fault, e.g., a fault current due to body contact. The threshold value can be determined, for example, as a function of a desired maximum probability for an erroneous determination that a fault exists. This avoids unnecessary aborting of the test due to false shutdowns.To determine the at least one statistical value, the signal values ​​assigned to the respective phase interval can be filtered, for example with an infinite impulse response filter (IIR filter). In some embodiments, the method further comprises generating a test signal for the electrical system and feeding the test signal into the electrical system such that the electrical system generates the electrical signal as a function of the test signal. If it is determined that a fault exists, the test signal can be switched off. Since the fault can in principle be detected with each newly measured signal value, the test signal can be switched off very quickly. For example, with a sampling frequency of 3 kHz, in less than 1 ms. This can ensure personnel protection even at higher currents.The test signal can preferably have a period that is at least temporarily constant. Furthermore, the test signal can be optionally modulated, for example, to generate predetermined load profiles during the test. When recording the distribution of the signal values, for example, only periods of the electrical signal during which the test signal is not modulated can be taken into account. Since signal values ​​that could be identified as outliers are likely to occur during the modulation of the test signal, these signal values ​​are ignored in order to avoid, for example, false shutdowns and to prevent adversely changing statistical values ​​such as the mean value or standard deviation. This allows faults to be reliably detected and false shutdowns to be avoided, at least outside the times when the test signal is modulated.A device for detecting a fault in an electrical system comprises a device for determining a distribution of signal values ​​based on at least one electrical signal of the electrical system across a plurality of signal sections of the at least one electrical signal. Each of the plurality of signal sections comprises at least one phase interval. Each signal value is assigned to one of the at least one phase interval. The device further comprises a device for determining whether a fault exists depending on the distribution of the signal values ​​in the at least one phase interval. The device can be configured to carry out the method described above and therefore also comprises the properties and advantages described above.The features of the previously described embodiments and aspects of the invention can be combined with one another as desired, unless expressly stated otherwise. In particular, the features can be used not only in the described combinations, but also in other combinations or on their own. BRIEF DESCRIPTION OF THE FIGURES The present invention will be described in detail below with reference to the attached figures. Fig. 1 schematically shows the probability of the risk of ventricular fibrillation as a function of the current intensity and duration of a current pulse. Fig. 2 schematically shows an equivalent circuit diagram for an exemplary test setup with a source impedance of a test system, an impedance of a person's body and a capacitance which a device under test has. Fig. 3 schematically shows a temporal profile of a current in the test setup of Fig.2 upon body contact and test signal shutdown. Fig. 4 schematically shows an electrical system according to an embodiment of the present invention. Fig. 5 schematically shows an assignment of signal values ​​to phase intervals of a signal section according to an embodiment of the present invention. Fig. 6 schematically shows box diagrams of distributions of signal values ​​in phase intervals of a signal section according to an embodiment of the present invention. Fig. 7 schematically shows the stabilization of outputs from filters for determining statistical parameters of the phase intervals over a number of considered signal values. Fig. 8 schematically shows a normal distribution of signal values ​​in a phase interval. Fig. 9 schematically shows a distribution density diagram of signal values ​​in phase intervals with an amplitude and phase jump. Fig.10 schematically shows method steps of a method according to an embodiment of the present invention. DETAILED DESCRIPTION OF EMBODIMENTS The above-described properties, features, and advantages of this invention, as well as the manner in which they are achieved, will become clearer and more readily understood in connection with the following description of embodiments, which are explained in more detail in conjunction with the drawings. The present invention will now be explained in more detail using embodiments with reference to the drawings. In the figures, like reference numerals designate like or similar elements. The figures are schematic representations of various embodiments of the invention. Elements shown in the figures are not necessarily drawn to scale.The various elements illustrated in the figures are depicted in such a way that their function and general purpose are understandable to those skilled in the art. Connections and couplings between functional units and elements illustrated in the figures can also be implemented as indirect connections or couplings. Functional units can be implemented as hardware, software, or a combination of hardware and software. Fig. 4 shows an electrical system 400. The electrical system 400 includes, for example, a test system 410 connected to a component 402 of a high-voltage system or power supply system. The component 402 can include, for example, a transformer, a grounding system, a current transformer, a circuit breaker, a line, or a cable. The component 402 is to be tested using the test system 410. The component 402 is therefore also referred to as the device under test.The test system 410 comprises a test signal generating device 414 and a feed and sampling device 412. To test the component 402, electrical test signals, for example test voltages and / or test currents, can be generated by the test signal generating device 414 and fed into the component 402 by means of the feed and sampling device 412 via test lines, which are shown in simplified form in Fig. 4 as a connection between the test system 410 and the component 402. A reaction of the component 402 to the fed-in test signals can be detected via the test lines and the feed and sampling device, for example by detecting voltages and / or currents at elements of the component 402, for example at lines, connections, or a housing of the component 402. Fig. 4 only shows a connection between the test system 410 and the component 402.However, this one connection can comprise a plurality of electrical lines to supply and detect the aforementioned voltages and currents. The detection of voltages and currents can be achieved by sampling with appropriate converters at a predetermined time interval. For example, voltages can be sampled with an analog / digital converter with a sampling frequency of, for example, 3000 Hz. When testing component 402, voltages in the range of several thousand volts, for example, 2 kV to 12 kV, and / or currents in the range of up to several hundred amps can occur. Corresponding voltages and currents can be present, for example, on measuring lines between test system 410 and component 402 or on elements of component 402, for example, lines, connections, or housing parts.A person 450 located in the vicinity of the electrical system 400 may therefore, for example, accidentally come into contact with these current- or voltage-carrying elements. In this case, it is desirable to deactivate the test currents and test voltages as quickly as possible to avoid injury to the person 450. To achieve a rapid deactivation of the test currents and test voltages, it can be exploited that the reaction of the component 402 to the supplied test currents and test voltages changes when the person 450 touches the electrical system 400. If, for example, current flows through the person 450 due to contact between the person 450 and the electrical system 400, the current distribution in the electrical system 400 changes, causing, for example, a voltage change on a line monitored by the test system 410, which can be interpreted as a fault.As a result of the detection of this fault, the test currents and test voltages can be switched off. Therefore, the test system 410 comprises a device 416 for detecting a fault in the electrical system 400. The device 416 comprises a device 418 for determining a distribution of signal values ​​based on at least one electrical signal of the electrical system 400. The distribution is determined across a plurality of signal portions of the at least one electrical signal. The device 416 further comprises a device 420 for determining whether a fault exists. The device 420 makes this determination depending on the distribution of the signal values ​​in the at least one phase interval. The test system 410 can, for example, comprise one or more control devices, such as an electronic controller, which implements at least portions of the components 412 to 420.The electronic control unit can, for example, comprise a processor, in particular a signal processor, as well as electrical and electronic components. The electronic control unit can be implemented in an analog or digital, discrete or integrated design. The electrical signal can, for example, be sampled on a line of the electrical system 400 and can be, for example, a voltage, a current, or a power. Based on this electrical signal, a signal value can be formed, for example using a mathematical method, taking into account other variables available in the test system 410, for example an impedance of the feed device 412 or a signal shape of the signal generated by the test signal generating device 414. Multiple electrical signals can also be sampled on the electrical system 400, and the signal value can be formed therefrom.For example, when AC voltage signals are fed in, a signal section can be a signal period of the fed-in AC voltage signal. If, for example, test voltages and / or test currents are fed in at a frequency of 50 Hz, a signal section can be a period with a period duration of 20 ms. Based on the one or more sampled electrical signals, a plurality of temporally successive signal values ​​are formed over a plurality of periods. Each period is divided into several phase intervals, and each signal value is assigned to one of the several phase intervals based on its temporal position in relation to the period duration. Within each phase interval, a distribution of the signal values ​​assigned to the respective phase interval is determined.Based on this distribution, the device 420 can, for example, determine for one or each phase interval whether the signal values ​​assigned to the phase interval exhibit expected distribution properties or not. For example, a statistical distribution of the signal values ​​assigned to the phase interval can be determined, and based on the statistical distribution, it can be determined whether there are any outliers in the phase interval, i.e., signal values ​​that deviate significantly from an expected statistical distribution. In the event of such a deviation, it can be determined that a fault is present. Due to the presence of a fault, the test system 410 can switch off the output of test voltages and test currents in order to de-energize the component 402 and thereby prevent injury to the person 450 from an electric shock. The operation of the device 416 for detecting the fault is described in detail below.This mode of operation is based on detecting outliers based on statistics over consecutive signal periods. For example, device 116 performs the following steps. The sampling of the one electrical signal or the multiple electrical signals takes place, for example, at a sampling rate fsmp. The sampling rate fsmp does not necessarily have to be constant, but it can be advantageous if it is constant at least temporarily. Constant sampling rates are typically used in test systems. The one electrical signal or the multiple electrical signals have, for example, a signal frequency fsig. The signal frequency fsig can be variable. The one electrical signal or the multiple electrical signals can comprise multiple frequencies, for example, a fundamental frequency and one or more harmonics. In this case, for example, the fundamental frequency can be used as the signal frequency fsig.The signal frequency fsig can determine the length of the above-mentioned signal sections. This means that a signal section, for example, has the length of one signal period, as determined by the signal frequency fsig. The signal period is divided into an integer number N of phase intervals. The phase intervals are non-overlapping and cover the entire length of the signal period. A phase interval j corresponds to a signal phase range ^^ of the signal period with a width of ^. ^ ∈ , (^+ 1) ^ ^, where N is the number of phase intervals. Fig. 5 shows a sinusoidal signal divided into 16 phase intervals. Each signal value xi is represented by a circle in Fig. 5 and is assigned to one of the 16 phase intervals via its acquisition time ti with respect to the signal period. A signal value xi can, for example, be a voltage value, a current value, or a combination thereof. The number of the phase interval can be calculated by incrementing the index i for each sample value. The time ^ ti of the measurement of the sample value with the index i is ^ ^ = ^ ^^^ . Therefore, the number ji of the phase interval for the sample with index i is given by ^ ^ with a fraction of For example, a mean μ and a variance var can be determined as statistical parameters for each phase interval. When a new signal value x is received, the mean μ and the variance var of the phase interval can be updated. For example, two first-order IIR filters can be used per phase interval: one for the mean μ and one for the variance var. The memory requirement is then an array of 2N floating-point values. First-order IIR filters can be defined recursively as follows: is the n-th iteration of the mean in the phase interval j and k is a filter constant with The standard deviation in the phase interval j is: ^ ^ = ^ ^^^ ^. Fig. 6 shows statistics for 16 phase intervals of a 50 Hz sine signal with 100% white noise in the form of box plots, also known as box plots or box-whisker plots. The box corresponds to the area containing the middle 50% of the data. The box is therefore bounded by the upper and lower quartiles, and the length of the box corresponds to the interquartile range. The median is shown as a horizontal line within the box. The antennas, also known as whiskers, have a maximum length of 1.5 times the interquartile range, although the respective antenna does not end exactly at this length, but at the value from the signal values ​​that still lies within this limit. Signal values ​​outside the antennas can be considered outliers.If the device 416 is connected to the test signal generation device 414 and thus has information about the generated test signals, it is known when an output amplitude, frequency, or phase is modulated. During these times, the output signal(s) is / are unstable. Consequently, measured signals may also be unstable, and statistics about these signals are generally undesirable. After the signal has stabilized, the filters require a certain amount of time to settle. For example, for a first-order IIR filter, after n iterations, the transfer function of a single step is: ^) ^ To achieve an accuracy of p stable percent to reach ^ ^ > ^^^^ ^ ^ ^^^^^^ ^^^^ ^ ^ Iterations per phase interval are required. If ^ e.g. ^^^^^^^= 0.9 is desired and ^ = is, the following applies: A corresponding transition function is shown, for example, in Fig. 7. If all phase intervals have an accuracy of p percent, the number of samples required for stabilization is therefore ^ The time required for stabilization in this case is When the signal is stable, the signal values ​​within a phase interval have a specific distribution. For example, with a normal distribution of the signal values, the probability distribution is as shown in Fig. 8. In many cases, it can be assumed that the measurements are normally distributed. However, it is clear that the techniques described herein are not limited to signal distributions of a specific type. An incoming (new) measured value ^ ^ can be used as ^ ^ = ^ ^ + ^ ^ ^ ^where ^i is a multiple of the standard deviation σj. The probability P that a value of xi lies in the range of ^ ^ ± ^^ ^ as a function of ^ is the integral of the probability density function (PDF) over the same interval. For a normal distribution, this can be solved analytically: ^ ( ^ ) = req with ction req ( ^ ) = ^^ the error fun ^^. The following table shows probability values ​​^ ( ^ )depending on ^: ^ P(^) 1 0.682689492137085897170465091264075844955825933453208781974789 2 0.954499736103641585599434725666933125056447552596643132032668 3 0.997300203936739810946696370464810045244341263683238701271556 4 0.999936657516333760157492458486555697403112333249039446982901 5 0.999999426696856241612176652495334250709292291153972776220854 6 0.999999998026824709924603718598271735203915962660417500041943 7 0.999999999997440374912228329991232752618438334003934311691603 8 0.999999999999999875580788514564317529680096548236231550225654 9 0.9999999999999999999774282318809231870452899584806250548403992 10 0.999999999999999999999984760293951678947868053313496801383273 11 0.9999999999999999999999999999617868085100264857769916873258442 12 0.9999999999999999999999999999999996447035775844642004607657996 13 0,99999999999999999999999999999999999987765671200900240635450 14 0,99999999999999999999999999999999999999999984412926361614399 15 0,9999999999999999999999999999999999999999999999999999992658067601 16 0.99999999999999999999999999999999999999999999999999998722 If a measurement x is outside the limits, ^ ^ + ^ ^ ^ ^ ] occurs, P is the probability that it is an outlier. Outliers can be filtered out by a threshold |^ ^ | > ^ ^^^ which is independent of μ and σ. ^^ ^ can be easily calculated as ^ = ^ ^ ^ ^. Another way to calculate consists in solving the equation = ^ ^ + ^ ^ ^ ^ into the above equations. Assuming that the values ​​of the phase intervals are stable at ^= 0, the following applies: Then and A deviating or abnormal measurement in which ^ ^ > ^ ^^^ with constant ^ ^^^can be identified as a fault, and in practice, the test current and / or the test voltage can be switched off in this case to avoid endangering a person in the vicinity of the electrical installation. For example, a person touching a live component of the electrical installation 400 can cause a sudden current increase, which leads to a deviating measurement, in which > ^ ^^^ is applied, after which the test voltage is switched off. Since a fault condition or an error can be detected after each scan, the error response time of the previously described method can be very short. In principle, there is a linear relationship between the error response time tf ^ and the sampling frequency fsmp: ^ ^ ~ ^ ^^^The fault response time tf can be further extended by a time delay in the acquisition of measured values, for example, by a time delay in an analog-to-digital converter that converts an analog measurement signal into a digital measured value. Furthermore, the response time tf can be further extended by a time delay required to switch off an amplifier of the feed device 412, for example, by processing in a control device of the test system 410. Depending on the time of occurrence of the fault, the fault may be detected only with low reliability or not at all in some phase intervals. Such phase intervals can be referred to as "blind" phase intervals. In the case of a sinusoidal signal whose amplitude and phase jump suddenly, this can occur, for example, in two ranges, as shown in connection with Fig. 9. Fig.Figure 9 shows a so-called "distribution density diagram," in which the bulges correspond to the number of measured values ​​in amplitude. An amplitude and phase jump occurred there during signal recording. Figure 9 shows the statistical distribution of the values ​​per phase interval. A temporal resolution is not shown in Figure 9. At two points, 902 and 904, no dumbbell-shaped bulge in the distribution around two mean values ​​is visible: these are the blind phase intervals, where no significant statistical change occurs. However, since the error is evident in the neighboring phase intervals, even such an error can be detected quickly. The reliability of the error detection method can be expressed using a probability for the absence of false triggering. In general, the probability for individual errors in a scenario with normally distributed measurements can be taken from the table above.The single error probability is:. For n iterations, ^ ^,^ = 1 − ^ ^,^ = 1 − ^ ^ ^ . For example, with a sampling rate of ^ ^^^ = 3 ^^^ and a test duration of, for example, 30 minutes, ^ should be selected as follows so that ^ ^ < 0.01 (1%) is: ^ > 6.00948 At a sampling rate of ^ ^^^ = 3 ^^^ is a reliable error detection with a threshold value ^ ^^^ ≥ 6 is ensured. The sensitivity of a phase interval can be defined, for example, as follows: where a relative error amplitude for an error value x is: A system is more sensitive when faults are detected with small values ​​of ^. For a single phase interval j, ^ ^ a signal property, namely the mean value of the signal amplitude in the phase interval j, and ^ ^depends on noise and signal oscillations, which are further signal properties, as well as a signal phase range, which depends on the choice of the number of phase intervals. Furthermore, ^ ^ also depend on numerical oscillations that may occur due to inherent computational properties, such as numerical limitations. Finally, ^ is the error deviation. Thus, the sensitivity is high when ^ is small, which is the case when ^ ^ is small, for example by a suitable choice of N, and the threshold for the error deviation ^ ^^^ can also be kept small. If reliability should not be compromised, ^ ^^^can be reduced if only a certain number of consecutive faults lead to tripping. The number M of consecutive faults required for a fault to be detected and the test current and voltage to be switched off can be referred to as fault accumulation. For example, ^ = 2 can be chosen, meaning only consecutive double faults lead to tripping. The probability for a single double fault is ^ ^^,^ = (1 − ^ ^,^ )^ For ^ iterations there are ^ − 1 pairs of measurements. The probability of zero double faults after ^ iterations is: The probability of double faults after ^ iterations is therefore: For example, with a sampling rate of ^ ^^^ = 3 ^^^ and a test duration of, for example, 30 minutes is ^ ^^ to be determined as follows: and results depending on ^ to: ^ Q 4 ,0 0.0214332 4 ,1 0.00917507 4,2 0.00383976 4,3 0.00157405 4 ,4 0.000632585 4 ,5 0.000249323 With a desired error probability of less than 1% and a sampling rate of ^ ^^^ = 3 ^^^ is a reliable error detection with ^ ^^^ ≥ 4.1 is ensured. Since this value for ^ ^^^,^ ^ ^ = 4.1 is lower than the above value of ^ ^^^,^ ^ ^ = 6, the system becomes more sensitive, but at the expense of speed. The general formula for the probability ^ ^^,^ for ^ consecutive errors after ^ iterations at a sampling rate of, for example, ^ ^^^ = 3 ^^^ and a test duration of, for example, 30 minutes is: For the resulting error probability ^ ^^ for ^ consecutive errors in ^ iterations: and results depending on ^ for M=3 for example: ^ Q 3 ,0 0.100813 3 ,1 0.03838 3 ,2 0.013918 3 ,30,00486866 3,4 0,00165098 3 ,50.000543698 In summary, Fig. 10 shows a corresponding method 1000, which can be carried out, for example, by a control device of the test system 410. The control device can comprise an electronic controller, such as a processor, in particular a signal processor. However, the electronic controller can also be implemented in an analog design. The method 1000 shown in Fig. 10 comprises method steps 1002 to 1018. Although the method steps are shown in a specific order in Fig. 10, the method steps can be carried out in any other order or in parallel.In particular, method steps 1006 to 1018 example page can be executed in parallel in a pipeline structure, for example by further processing an electrical signal sampled in step 1006 in steps 1008 to 1018, while the electrical signal is already sampled at a next point in time in step 1006. At the beginning of a test of component 402, a test signal, for example a sinusoidal test signal, is generated in step 1002 by means of test signal generation device 414. In step 1004, the test signal is fed into component 402 by means of feed and sampling device 412, for example as a high-voltage signal or current signal corresponding to the test signal. For this purpose, feed and sampling device 412 can have corresponding amplifiers. In step 1006, an electrical signal is sampled by means of feed and sampling device 412.For example, the electrical signal can be sampled on a housing or a conductor of component 402. Alternatively, the input signal can also be sampled, since in the event of a fault, this can also exhibit anomalies that can be identified as faults. In step 1008, a signal value corresponding to the sampled electrical signal is determined. The sampling and determination of the signal value can be implemented, for example, using an analog / digital converter and, if appropriate, an upstream measuring transducer. In particular, the further processing in steps 1010 to 1016 can be performed by a digital control device, for example a digital signal processor or general-purpose processor. Alternatively, this processing can also be implemented in analog circuits or discretely constructed digital circuits.Steps 1006 to 1016 are executed multiple times in succession, for example, in time with the sampling, for example 3000 times per second. This results in a large number of consecutive signal values. Next, a distribution of these signal values ​​is determined. The distribution of the signal values ​​can, for example, be a statistical distribution that occurs over a large number of signal sections of the electrical signal. A signal section of the electrical signal can, for example, be a period of a fundamental oscillation of the electrical signal, for example 20 ms for an electrical signal with a fundamental frequency of 50 Hz. The signal section or the period is divided into several phase intervals, for example, into 16 phase intervals, as described above in connection with Figures 5 and 6.In step 1010, a current signal value is assigned to a respective phase interval depending on its sampling time in relation to the period duration. The respective phase interval now comprises another signal value, and the distribution of the signal values ​​in this phase interval is updated accordingly. As previously described, the statistical distribution can, for example, comprise a mean value of the signal values ​​assigned to the respective phase interval and a standard deviation. In step 1014, an outlier measure is determined at least for the phase interval to which the new signal value was assigned, for example by comparing the value of the newly added signal value with the mean value and the standard deviation of the phase interval.If the value of the newly added signal value is within prescribed limits (step 1016), it is presumably not an outlier, and the method continues in step 1006 with the next sampling of the electrical signal and the processing of the next sample. If, however, an outlier is identified in step 1016, the test signal is switched off in step 1018. For example, amplifiers in the feed and sampling device 412 can be switched off in order to de-energize the system 400 as quickly as possible in order to reduce the risk to the person 450. Of course, the features of the previously described embodiments and aspects of the invention can be combined with one another. In particular, the features can be used not only in the described combinations, but also in other combinations or on their own, without departing from the scope of the invention.

Claims

CLAIMS 1. A method for detecting a fault in an electrical system (400), comprising: - determining (1010, 1012) a distribution of signal values ​​based on at least one electrical signal of the electrical system (400) across a plurality of signal sections of the at least one electrical signal, wherein each of the plurality of signal sections comprises at least one phase interval, wherein each signal value is assigned to one of the at least one phase interval, and - determining (1014, 1016) whether a fault is present depending on the distribution of the signal values ​​in the at least one phase interval.

2. The method according to claim 1, wherein determining (1010, 1012) the distribution of signal values ​​comprises: - determining (1012), for each of the at least one phase interval, at least one statistical value depending on the signal values ​​assigned to the respective phase interval. 3.The method of claim 2, wherein determining (1014, 1016) whether an error is present for at least one of the at least one phase interval comprises: - comparing (1014) a signal value associated with the phase interval with the at least one static value determined for the phase interval.

4. The method of claim 2 or claim 3, wherein determining (1014, 1016) whether an error is present for each of the at least one phase interval comprises: - comparing (1014) a signal value last assigned to the phase interval with the at least one static value determined for the phase interval.

5. The method according to any one of claims 2-4, wherein the at least one statistical value comprises a statistical mean of the signal values ​​assigned to the respective phase interval across the plurality of signal sections, and a statistical dispersion of the signal values ​​assigned to the respective phase interval across the plurality of signal sections.

6. The method according to claim 5, wherein determining (1014, 1016) whether an error is present for at least one of the at least one phase interval comprises: - determining (1014) an outlier measure as a function of a signal value assigned to the phase interval and the statistical mean and the statistical dispersion of the phase interval, and - comparing (1016) the outlier measure with a predetermined threshold. 7.The method of claim 6, wherein the outlier measure ^. ^ for the signal value assigned to the phase interval ^ is defined as: where the statistical mean and ^ ^ is the standard deviation of the statistical dispersion of the phase interval ^.

8. The method according to claim 6 or claim 7, wherein the threshold value is determined as a function of a desired maximum probability for an erroneous determination that a fault is present.

9. The method according to any one of claims 2-8, wherein determining the at least one statistical value comprises filtering the signal values ​​associated with the respective phase interval.

10. The method according to any one of the preceding claims, further comprising: - generating (1002) a test signal for the electrical system (400), - feeding (1004) the test signal into the electrical system (400) such that the electrical system (400) generates the electrical signal as a function of the test signal.

11. The method according to claim 10, wherein the test signal has an at least temporarily constant period. 12.The method of claim 11, wherein the test signal is selectively modulated, wherein, when detecting the distribution of the signal values, only periods of the electrical signal during which the test signal is not modulated are taken into account.

13. The method of any of claims 10-12, further comprising: - switching off (1018) the test signal when it is determined that a fault is present.

14. The method according to any one of the preceding claims, wherein the signal values ​​each comprise a voltage value of the electrical signal, a current value of the electrical signal, or a combination of the current value and the voltage value.

15. The method according to claim 14, wherein a respective signal section of the plurality of signal sections comprises a predetermined number of non-overlapping phase intervals whose total length corresponds to the length of the respective signal section.

16. The method according to claim 15, wherein the predetermined number is determined as a function of a sampling rate and a fundamental frequency of the electrical signal.

17. The method according to claim 15 or claim 16, wherein the predetermined number is in a range from 4 to 100, preferably in a range from 8 to 30, and more preferably is 16. 18.Device for detecting a fault in an electrical installation (400), comprising: - a device (420) for determining a distribution of signal values ​​based on at least one electrical signal of the electrical installation over a plurality of signal sections of the at least one electrical signal, wherein each of the plurality of signal sections comprises at least one phase interval, wherein each signal value is assigned to one of the at least one phase interval, and. - means (420) for determining whether an error is present, depending on the distribution of the signal values ​​in the at least one phase interval.

19. The device according to claim 18, wherein the device (416) is configured to carry out the method according to any one of claims 2-17.

Images