Apparatus, system, and method for determining an erroneous signal window in a measurement signal

Abstract

The invention relates to a device for determining an error signal window, which represents a signal error in a measurement signal of an electrical device for energy transmission. A measurement signal detected by means of a sensor in energy transmission is first analyzed by means of a model in an initialization phase in order to determine a plurality of initial coefficient tuples, which form the coefficients of the model respectively. A noise tuple is determined by means of the device as an expected value. In a checking phase, further signal windows of the measurement signal are analyzed by means of the device by means of the model. The resulting coefficients produce a measurement tuple for each signal window. The distance of the respective measurement tuple from the noise tuple is used by the device in order to determine whether the analyzed signal window represents a signal error. If this is the case, the respective signal window is determined as an error signal window. The invention relates to a system for energy transmission, which comprises the device, and furthermore to a method for operating the device.

Description

Background Technology

[0001] Electrical equipment can be configured to transmit electrical energy. Electrical equipment can be, for example, an electrical cable with multiple lines used for transmitting electrical energy. Other electrical equipment can also be used to transmit electrical energy, such as transformers, or to quantify current and voltage, such as electrical converters or overhead power lines. Preferably, the electrical equipment is a cable for transmitting high-voltage direct current or high-voltage alternating current. If high voltage is used to transmit electrical energy, partial discharge (TE) may occur. A partial discharge is a location-restricted electrical discharge that partially bridges the insulation between conductors, particularly the aforementioned cables. Partial discharge may occur, for example, due to defects in the conductors and / or due to non-uniformity in the cable insulation. If partial discharge occurs, this generates a signal with a corresponding signal pattern, representing the partial discharge, which is also transmitted by the electrical equipment. Transmission in this case occurs from the location of the partial discharge to the connection point of the electrical equipment. The signal caused by the partial discharge must therefore travel a certain transmission path before it can be detected at the connection point of the electrical equipment. Inductive, capacitive, and / or resistive components in electrical equipment can cause signal attenuation and / or distortion due to partial discharge. Therefore, at connection points, attenuated and / or distorted signals caused by partial discharge are often only detectable by sensing. Signals caused by partial discharge can also be referred to as erroneous signal components of the measured signal, or simply erroneous signals. Other erroneous signals can also be caused by other errors in the electrical equipment.

[0002] To determine whether an error has occurred in the power transmission of electrical equipment, in practice, detectable measurement signals at the connection points of the electrical equipment are often recorded and then manually checked, for example, using an oscilloscope, to confirm whether an error has occurred in the power transmission of the electrical equipment. However, this method is not robust to the estimation of possible errors and is also very time-consuming. Summary of the Invention

[0003] The purpose of this invention is to provide an apparatus, system, and method that allows for the effective and error-robust identification of errors in the transmission of electrical power by means of electrical equipment.

[0004] According to a first aspect of the invention, the objective is achieved by an apparatus according to the invention. Therefore, an apparatus is provided having an input signal interface for receiving a digital measurement signal representing an analog signal detected at a connection point of an electrical device. The digital measurement signal is a sequence of scan values ​​of the detected signal. Furthermore, the apparatus has a processor unit configured to configure a signal window for each scan value of the measurement signal, each signal window being formed by a portion of the scan value sequence of the measurement signal including the corresponding scan value and a predetermined number of time-preceding scan values ​​of the measurement signal. The processor unit is configured to, during an initialization phase, determine M signal windows as initialization windows and K noise windows from the M initialization windows, analyze each noise window using a predetermined X-order model and obtain the corresponding coefficients, which form an initial coefficient tuple configured for the corresponding noise window, and obtain an expected value as a noise tuple from the initial coefficient tuple of the noise window, where X is an even number between 1 and 5. Furthermore, the processor unit is configured to, during the inspection phase, determine multiple signal windows as measurement windows, analyze each measurement window using the predetermined X-order model and calculate its associated coefficients, which form measurement tuples assigned to the corresponding measurement windows. For each measurement window, the distance between its associated measurement tuple and the noise tuple is calculated. The multiple measurement windows determine measurement windows as error signal windows, where the measurement tuples of each measurement window have a distance greater than a predetermined first boundary value from the noise tuple. Thus, each error signal window represents a signal error in the measurement signal. The signal error here can be an erroneous signal component of the measurement signal.

[0005] If electrical energy is transmitted via electrical equipment, signals can be detected at the connection point, indicating the voltage, power, current, and / or other values ​​representing the energy transmission. The detected signals preferably represent a time curve of the energy transmitted via the electrical system, the associated current, voltage, and / or power. Digital measurement signals provided to the input signal interface of the device represent the signals detected at the connection point of the electrical equipment. The digital measurement signals are formed from a sequence of scanned values ​​of the detected signals. The digital measurement signals are simply referred to as measurement signals.

[0006] However, frequently measured signals not only represent the electrical energy transmitted by electrical equipment (or the associated current, voltage, and / or power), but also noise and erroneous signals caused by the environment, which may arise from errors in the energy transmission process using electrical equipment.

[0007] The processor unit of the device is configured based on the fundamental concept that noise differs from the error signal components of the measurement signal, which are caused by errors in energy transmission via electrical equipment. Therefore, during the initialization phase, an initial set of coefficients should first be obtained using a mathematical model, which is characteristic of the noise occurring during energy transmission. This model is used to model the transmission function or transmission path of the electrical equipment in energy transmission. Based on coefficient matching, the model can be matched to the corresponding transmission characteristics of the electrical equipment. The processor unit is preferably configured to match the model coefficients based on a signal window, such that the model approximates the actual transmission function of the electrical equipment as accurately as possible.

[0008] The processor unit is further configured to determine M signal windows as initialization windows during the initialization phase. These M signal windows are preferably configured for a corresponding number of M scan values. The M scan values ​​are preferably M consecutive scan values ​​of a measurement signal. The measurement signal may have useful components, error signal components, and noise components. The useful components represent power transmission, the voltage, current, and / or power during power transmission. The error signal components may be caused by possible errors in power transmission, and the noise components may represent noise occurring in the measurement technique detection of power transmission and / or the measurement signal. The processor unit is therefore preferably configured to determine K noise windows from the M initialization windows. For this purpose, the processor unit may, for example, determine the frequency spectrum of each of the M initialization windows, the signal energy of the portion of the measurement signal represented by the corresponding initialization window, the number of zero crossings in the portion of the measurement signal represented by the corresponding initialization window, and / or other analysis parameters, thereby determining, for example, by the processor unit, which of the M initialization windows to evaluate as a noise window based on the frequency spectrum, signal energy, number of zero crossings, and / or analysis parameters. In this way, the processor unit can, for example, determine the K noise windows. In each of these noise windows, it is assumed that the corresponding noise component is particularly high or dominant.

[0009] The processor unit is configured to analyze each noise window using a predetermined X-order model and thereby determine its associated coefficients. These coefficients are therefore related to the coefficients of the predetermined model. These coefficients can also be referred to as model coefficients. The model preferably relates to a mathematical model. The model can be configured to represent a portion of the signal formed by the signal window by matching its associated coefficients. The model is matched in this way, through analysis performed by the processor unit for each noise window and by matching its associated coefficients, so as to represent the portion of the signal represented by the corresponding noise window as accurately as possible. Preferably, X is an even number, 2. It is particularly efficient and fast to analyze each noise window or each portion of the signal represented by the noise window using a predetermined model to determine the model's associated coefficients. Coefficients are thus assigned to each noise window, which are collectively referred to as an initial coefficient tuple for each noise window. Therefore, exactly one dedicated initial coefficient tuple is assigned to each noise window.

[0010] The processor unit is configured to obtain an expected value as a noise tuple from a corresponding number of K initial coefficient tuples. The noise tuple can here be the arithmetic mean of the initial coefficient tuples. If each initial coefficient tuple is formed by a vector of multiple coefficients, then the noise tuple can be obtained, for example, by adding the corresponding vectors and then dividing the result by the number of initial coefficient tuples. The processor unit can be configured for this purpose. The noise tuple can thus form the expected value for the initial coefficient tuples. The noise tuple can also be understood as the noise center for the initial coefficient tuples, because the initial coefficient tuples are distributed around the noise tuples in geometric space.

[0011] In the initialization phase, the processor unit determines M signal windows as initialization windows. A checking phase follows immediately after the initialization phase. The processor unit is therefore configured to determine multiple signal windows as measurement windows during the checking phase. Preferably, the measurement windows are different from the signal windows previously determined in the initialization phase. The multiple measurement windows are preferably significantly larger than the number of initialization windows. The basic concept of the invention is that, in the checking phase, multiple measurement windows are identified as error signal windows, each representing a signal error in the portion of the measured signal represented by the corresponding signal window. Therefore, it is meaningful that the number of measurement windows is significantly larger than the number of initialization windows.

[0012] The processor unit is configured to analyze each measurement window during the inspection phase using a predetermined X-order model to determine its associated coefficients. Each measurement window is thus analyzed using the same predetermined X-order model, which was also used to analyze noise windows during the initialization phase. The analysis of the measurement windows can therefore be performed by the processor unit in a similar manner to the analysis of the noise windows. Each measurement window is thus analyzed by the processor unit using the X-order model to determine the model's associated coefficients. The coefficients determined for each measurement window are called measurement tuples and are assigned to the corresponding measurement window. If the predetermined model is second-order, then two coefficients are also determined for each measurement window. However, if the model is of a higher order, then a number of coefficients equal to the order are determined for each measurement window. If one measurement tuple is determined for a measurement window, then the measurement tuple has a defined distance in geometric space from the noise tuple determined during the initialization phase. The dimension of the geometric space can be equal to the order of the model. If the predetermined model is of a higher order, particularly third, fourth, or fifth order, then the distance can be a distance defined in the corresponding space. The processor unit is configured to calculate the distance between the corresponding measurement element and the noise element for each measurement value. Therefore, the processor unit calculates a corresponding number of measurement elements and a corresponding number of distances based on the number of measurement windows. If the measurement element and the noise element have a small distance, the corresponding measurement window will very likely represent a portion of the measurement signal with a large noise component. However, if the distance between the measurement element and the noise element is larger, particularly greater than a predetermined first boundary value, the configured measurement window will very likely represent a portion of the measurement signal with a signal error. This is because the signal error arises in the analysis of the measurement windows by calculating other coefficients for a predetermined model, thereby correspondingly setting the measurement elements and noise elements at a larger distance. By determining the corresponding distances between the measurement elements and the noise elements, measurement windows representing signal errors in the measurement signal can be determined very effectively from the plurality of measurement windows.

[0013] Based on the preceding explanation, the processor unit is configured to calculate the distance between the corresponding measurement element and the noise element for each measurement window, and to determine such a measurement window as an error signal window, wherein the measurement element to which the measurement window belongs has a distance greater than a predetermined first boundary value G from the noise element. This ensures that each error signal window represents a portion of the measurement signal with a signal error. Therefore, each error signal window also represents a signal error in the measurement signal.

[0014] The processor unit's last-mentioned settings for determining the error signal window allow for the rapid and robust determination of the error signal window in which a signal error is expected. For further analysis, the error signal window can be further processed. In particular, the type and quality of the signal error and / or the type and / or quality of the error occurring during energy transfer via electrical equipment can be determined. The processor unit can be configured accordingly.

[0015] The processor unit configuration further offers the advantage of eliminating the need for manual analysis of the entire measurement signal. Moreover, the processor unit configuration allows for focused evaluation on specific segments, i.e., the error signal window of the measurement signal. The total cost for error identification and / or analysis in power transfer can thus be reduced to a small fraction.

[0016] As previously explained, the processor unit is configured to assign a signal window to each scan value of the measurement signal. This generates multiple signal windows. It is also mentioned that the process (Ablauf) that generates the signal windows moves co-currently from scan value to scan value. Each signal window may, for example, include at least 16, at least 32, or at least 64 scan values. Preferably, each signal window has, for example, 128 scan values. The initial phase can be very short. The processor unit is preferably configured to perform a check phase after each initialization phase. The number of measurement windows can be significantly greater than the number of initialization windows. Thus, the number of measurement windows can be at least ten times the number of initialization windows. The processor unit can also be configured to perform the initialization phase at least twice per second. The number of measurement windows can be determined based on the scan frequency. The number of repetitions of the initialization phase can also be increased. Thus, the processor unit can be configured to perform the initialization phase, for example, 2 to 10 times per second. For this purpose, a corresponding match can be made between the number of measurement windows. Furthermore, as previously explained, the noise tuple can be positioned in the center between the initial coefficient tuples. Therefore, the noise tuple can be located at the noise center of the "cloud" formed by the initial coefficient tuples of the noise windows. Preferably, the scanning frequency for the scan values ​​used to measure the signal is selected such that at least 10,000 signal windows are preferably determined as measurement windows during the inspection phase. Preferably, the signal windows can also be significantly more.

[0017] Furthermore, as previously explained, the processor unit is configured to calculate the distance between the corresponding measurement tuple and the noise tuple for each measurement window. The distance between the corresponding measurement tuple and the noise tuple is preferably the Mahalanobis distance between them. For calculating the Mahalanobis distance, refer to the following publication: Michaël Bensimhoun's N-Dimensional cumulative function, and other useful facts about gaussians and normal densities (Jerusalem, 06 / 2009).

[0018] The predetermined X-order model can be a linear model and / or an LPC model of the corresponding order. The coefficients belonging to the X-order model can be, for example, α coefficients, PARCOR coefficients, or prediction error coefficients and / or standardized errors.

[0019] Furthermore, as previously explained, the processor unit is configured such that if the distance between the associated measurement tuple and the noise tuple is greater than a predetermined first boundary value, then the measurement windows are determined by the plurality of measurement windows to be error signal windows. The boundary value can, for example, be one to five times the standard deviation of the expected value or the noise tuple. The predetermined boundary value can, for example, be predetermined based on expertise and previous testing. Other possible schemes for determining the boundary value are also possible.

[0020] An advantageous embodiment of the apparatus is characterized in that the processor unit is configured to calculate, for each initialization window, the number of zero crossings in the portion of the measurement signal represented by the corresponding initialization window. Furthermore, the processor unit is preferably configured such that the initialization windows determine a maximum of K initialization windows with the most zero crossings as noise windows. In practice, it has been confirmed that a large number of zero crossings in the portion of the measurement signal represented by the initialization window means that this portion of the measurement signal is highly susceptible to noise interference. Calculating the number of zero crossings in the initialization windows has proven to be an effective method for determining the noise windows with the most noise in the initialization windows. Preferably, zero crossings in the measurement signal occur if the value of the measurement signal preferably reaches or passes through zero or another predetermined signal value for a short period. The number of noise windows K is preferably less than, and particularly significantly less than, the number of initialization windows M. Thus, the number K can be, for example, up to two-thirds, up to half, or up to one-third of the number M. This allows for particularly robust determination of which initialization window is determined by the processor unit to be a noise window. The noise windows are used in the initialization phase to: calculate a corresponding number of K initial coefficient tuples and thereby calculate the noise tuples. As is well known, noise does not allow for meaningful correlation. Therefore, the application of zero-crossing and robust determination of the noise window are effective and robust possible schemes for determining noise tuples, so that the distance between the corresponding measurement tuple and the noise tuple is obtainable for each measurement window during the inspection phase. This, in turn, allows for the determination of the error signal window. Furthermore, the determination of the noise window is made possible by means of the zero-crossing method, i.e., the noise window is obtainable in real time or in time parallel during the initialization phase and / or before the start of the inspection phase.

[0021] Another advantageous embodiment of the device is characterized in that the processor unit is configured to, for each initialization window, calculate the signal energy of the portion of the measurement signal represented by the corresponding initialization window, wherein the processor unit is further configured to determine, from the initialization windows, a number K initialization windows with the minimum signal energy as noise windows. This is because it has been confirmed in practice that a small signal energy in the portion of the measurement signal represented by the initialization window means that the portion of the measurement signal is highly susceptible to noise interference. To determine the noise window with the most noise in the case of initialization windows, the determination of the minimum signal energy of the initialization window has proven to be an equally effective method. Therefore, the noise window can also be determined by signal energy instead of by means of zero-crossing. This also allows for particularly robust determination of which initialization window among the initialization windows is determined by the processor unit to be a noise window. The noise window is used in the initialization phase to: calculate a corresponding number K initial coefficient tuples and thereby calculate the noise tuples. It is well known that noise does not allow meaningful correlation. Therefore, the application of minimum signal energy and robust determination of the noise window are effective and robust possible schemes for determining noise tuples, so that the distance between the corresponding measurement tuple and the noise tuple is achievable for each measurement window during the inspection phase. This, in turn, allows for the determination of erroneous signal windows. Furthermore, the possibility of determining the noise window is provided by means of the determination of minimum signal energy, i.e., the noise window is achievable in real time or in time parallel during the initialization phase and / or before the start of the inspection phase.

[0022] The so-called "spectral flatness" can also be applied instead of signal energy. The processor unit can be configured accordingly.

[0023] Another advantageous embodiment of the device is characterized in that the X-order model is configured as an X-order LPC model. LPC models are known in principle from the prior art. An LPC model can be described, for example, by means of the following recursive relation:

[0024] y(k) = e(k) + Σ1 N a i · y(k - i).

[0025] Here, k is a discrete time variable, i.e., a natural number greater than zero, y(k) is the value of the discrete measured signal at discrete time k, and N is the approximate order, where N corresponds to the value X of order X in the LPC model. i Preferably, it is a so-called linear variable of order N, and e(k) is the prediction error. Variable a iPreferably, the coefficients of the LPC model are formed. Since the values ​​of the discrete measurement signals are known, the variables or coefficients of the model can be determined by means of the processor unit in a way that minimizes the total squared error of the model's approximation with respect to the discrete measurement signals. Here, the portions of the measurement signals represented by the corresponding windows are considered respectively. The total squared error can be given by Q according to the following equation:

[0026] Q = Σ1 N e 2 (k) = Σ1 N (y(k) – Σ1 N a i · y(k - i)) 2 .

[0027] Therefore, we can distinguish variable a. i The total squared error of the coefficients is such that the corresponding result is equal to zero, and a system of equations consisting of N linear equations is solved. The processor unit described therein can be configured to obtain the coefficients of the LPC model. Further reference is made to the publication DE 10 2018 126 743 B3 regarding the LPC model and the determination of the coefficients.

[0028] While the X-order model is preferably configured as an X-order LPC model, it can also be specified that the X-order model is configured as another X-order mathematical model. In this case, the model can be formed, for example, using Fourier transform or wavelet transform.

[0029] Another advantageous embodiment of the device is characterized in that M is an integer of at least 100, preferably at least 10,000. Furthermore, it is preferably specified that K is an integer less than M. Preferably, K is less than 0.5 times M, and preferably K is at most 0.1 times M. This ensures robust and simple determination of the number of noise windows.

[0030] Another advantageous embodiment of the device is characterized in that each initialization phase lasts for a maximum of 0.1 seconds, preferably a maximum of 0.05 seconds. The check phase, for example, can last for several seconds or even several minutes or even longer. The short duration of the initialization phase allows for a particularly rapid transition to the check phase. If the device is used by a user, the user will generally not be aware of the initialization phase.

[0031] Another advantageous embodiment of the apparatus is characterized in that the processor unit is configured to analyze each error signal window by means of a predetermined N-order model and obtain the corresponding coefficients, which form error tuples configured for the respective error signal windows, wherein N is an integer of at least 6, preferably 8. The analysis can occur during the inspection phase. Preferably, the N-order model and the X-order model are of the same type. Both models, i.e., not only the N-order model but also the X-order model, can be formed by LPC models of corresponding orders. The X-order model is preferably a second-order or third-order model. The N-order model is preferably a sixth-order or higher-order model, such as an eighth-order model. The order can also be larger. Preferably, N is at least 10, 12, or 14. During the inspection phase, the processor unit determines, by means of the plurality of measurement windows, measurement windows as error signal windows, wherein the measurement tuples to which the measurement windows belong have a distance greater than a predetermined first boundary value from the noise tuples. It can thus be assumed that the portion of the measured signal represented by the corresponding error signal window includes signal errors. According to the advantageous embodiment mentioned last in the device, the processor unit is configured to analyze the error signal window by means of a predetermined Nth-order model, i.e., at least a sixth-order model. Through analysis, the corresponding coefficients are obtained. This analysis can then be performed to determine the coefficients of the at least sixth-order model. This allows for particularly accurate examination of the portion of the measured signal represented by the error signal window. However, this introduces a not insignificant computational cost in analyzing the error signal window by means of a predetermined Nth-order model. However, since only a limited number of error signal windows are determined, this is often short-term and possible during the examination phase. In other words, the error signal window is analyzed more thoroughly. Preferably, the processor unit can obtain the real and imaginary parts of the poles of the portion of the measured signal represented by the corresponding error signal window by means of the coefficients. The processor unit can be configured for this purpose. From the real and imaginary parts, it can be determined whether the portion of the measured signal represented by the corresponding error signal window is attributable to an error, particularly partial discharge. In practice, matching a predetermined first boundary value can often ensure that the error signal window represents only or mostly a portion of the measured signal, including signal components caused by errors and, in particular, partial discharge. However, in order to keep the computational cost of analyzing the error signal window at an acceptable level, it can be specified that the Nth order of the model is limited. Thus, N can be, for example, a maximum of 20 or 30.

[0032] Another advantageous embodiment of the device is characterized in that the processor unit is configured to divide the error elements into multiple error groups, such that error elements in the same error group are spaced apart from each other by a distance less than a predetermined second boundary value, so that all error signal windows assigned to the corresponding same error group by their respective error elements represent the same signal error. Errors in electrical equipment in an energy transfer system can lead to erroneous signals in the detected measurement signals. Multiple error signal windows can therefore represent multiple erroneous signals caused by the same error during energy transfer via electrical equipment. If, for example, an undesirable partial discharge occurs during energy transfer via electrical equipment, then multiple error signal windows can represent portions of the measurement signal significantly affected by the partial discharge. This often results in error elements obtained by the processor unit based on the error signal windows being affected by the same error source, i.e., partial discharge. Therefore, it is stipulated that the error elements are assigned to the same error group. It is determined here that such error elements have only a small distance between each other. The maximum distance between error elements can therefore be determined by a predetermined second boundary value. The processor unit is therefore preferably configured to divide error elements into multiple error groups, such that error elements in the same error group are spaced apart from each other by a distance less than a predetermined second boundary value. Furthermore, it can be specified that error elements in the same error group are spaced at least substantially the same distance from noise elements. The deviation of the distance between error elements and noise elements is preferably less than the predetermined second boundary value. This ensures that error elements in the same error group represent the same signal error. These error elements in the same error group can then be caused by the same error in power transmission via electrical equipment. An error group can also be referred to as a cluster. The number of different error groups preferably indicates the number of different errors in power transmission via electrical equipment. If, for example, three partial discharges occur in power transmission via electrical equipment, then predetermined first and second boundary values ​​can be predetermined and / or selected such that the processor unit generates three error groups, each with multiple error elements, each having the same partial discharge error as its cause.

[0033] By dividing error elements into multiple error groups, it is particularly simple and fast to identify the quantity of errors in energy transfer, especially discharge errors. Furthermore, based on the error elements and their associated error signal windows, rapid and accurate analysis can be achieved to determine where and at what intensity partial discharges occur.

[0034] Another advantageous embodiment of the device is characterized in that the processor unit is configured to divide the error signal window into multiple error groups, such that the error tuples of the error signal windows in the same error group are spaced apart from each other by a distance less than a predetermined second boundary value, so that all error signal windows in the corresponding same error group represent the same signal error. This embodiment is similar to a previous embodiment of the device. Instead of grouping error tuples, error signal windows are grouped here. Therefore, similarly, reference is made to the previous preferred description, advantageous features, technical effects, and / or advantages for this embodiment of the device.

[0035] Another advantageous embodiment of the device is characterized in that the processor unit is configured to determine a plurality of different errors in the electrical equipment based on the plurality of error groups. If, by means of the processor unit, the error elements are divided into three error groups or the error signal window is divided into three error groups, then the processor unit can infer the corresponding number of different errors based on the number of error groups. The number of errors preferably corresponds to the number of error groups.

[0036] Another advantageous embodiment of the device is characterized in that the processor unit is configured to generate an image signal representing a measurement signal as a signal graph, and the processor unit is configured to visually and identically indicate portions of the signal graph based on scan values ​​of the measurement signal, the scan values ​​being assigned to error signal windows of the same error group. Even with reference to error signal windows of the same error group, similarly applicable can be made to error tuples of the same error group, since each error tuple in the error tuple is assigned to one of the error signal windows. The portions of the signal graph caused by the same error can, for example, be indicated identically. If, for example, a partial discharge error occurs in energy transfer by means of electrical equipment, then multiple error signal windows can represent the partial discharge error by corresponding signal components. As described above, these error signal windows can be assigned to the same error group. However, it is also possible that error tuples assigned to error signal windows are assigned to the same error group. In both cases, the error group includes error signal windows or error signal windows configured by error tuples, which can visually indicate partial discharge errors or other errors particularly concisely. The corresponding portions of the signal diagram—determined by the same group (or tuple) of the error signal window—are visually indicated identically by means of the processor unit. This can be achieved, for example, by using the same color (e.g., red, green, or blue). However, it is also possible that the corresponding portions of the signal diagram are indicated identically by the same dashed lines, line thickness, and / or other features. The device can be configured to provide an image signal. Thus, the device can have, for example, an output signal interface, which the processor unit is configured to control for transmitting, in particular, sending, the image signal. This allows the image signal to be transmitted to a remote unit and displayed there if necessary.

[0037] Another advantageous embodiment of the device is characterized by having a display unit. Furthermore, the processor unit can be configured to control the display unit such that it displays an image based on an image signal, thereby visually depicting a signal graph. The signal graph visually depicted by the image can here visually indicate portions based on scan values ​​of the measured signal, which are assigned to error signal windows of the same error group. Therefore, it can be determined particularly quickly on the display unit which portion of the signal graph was caused by the same error, particularly the same partial discharge error.

[0038] Another advantageous embodiment of the device is characterized in that the processor unit is configured to perform an initialization phase multiple times, thereby re-evaluating the noise tuple with each initialization phase. This allows for updates to the noise, which may vary.

[0039] Another advantageous embodiment of the device is characterized in that the processor unit is configured to perform at least one check phase after each initialization phase. Preferably, it is specified that exactly one check phase is performed by means of the processor unit after each initialization phase. However, it is also possible to specify that the processor unit is configured to perform multiple check phases after each initialization phase. In this case, the check phases can be performed sequentially by the processor unit.

[0040] According to a second aspect of the invention, the object mentioned at the beginning is solved by a system according to the invention. Therefore, a system for energy transmission is provided, wherein the system has electrical equipment for transmitting an electrical energy signal from an input interface of the electrical equipment to a supply interface of the electrical equipment. The system further includes a sensor unit and a device. The device is configured according to the first aspect of the invention and / or one of its advantageous embodiments. Similarly, reference is made to the advantageous description, preferred features, technical effects, and / or advantages of the system's device, as described in conjunction with the device according to one of the first aspect and / or its advantageous embodiments. The sensor unit of the system is disposed at the connection point of the electrical equipment between the input interface and the supply interface. The sensor unit is configured to detect the electrical energy signal and to generate a digital measurement signal representing the energy signal detected at the connection point. The sensor unit is coupled to the signal interface of the device to transmit the measurement signal to the signal interface. Reference is made to the advantages and technical effects of the system as described in conjunction with the device according to one of the first aspect and / or its advantageous embodiments. (Repetition omitted.)

[0041] An advantageous embodiment of the system is characterized in that the electrical equipment is configured as a high-voltage line, a transformer, a rotating motor, a gas-insulated line, or a gas-insulated power distribution device. Preferably, the electrical equipment has a high-voltage line extending from the feed point to the supply point of the electrical equipment. The high-voltage line can be configured to transmit electrical energy having a high DC voltage or a high AC voltage. Alternatively and / or additionally, the electrical equipment can have a transformer. If the electrical equipment has a transformer instead of a high-voltage line, the feed point can be configured on the primary side of the transformer, and the supply point can be configured on the secondary side of the transformer. Thus, electrical energy can be transmitted from the feed point, i.e., the primary side of the transformer, to the supply point, i.e., the secondary side of the transformer. The aforementioned technical effects and / or advantageous embodiments similarly apply to other possible embodiments of the electrical equipment.

[0042] According to a third aspect of the invention, the object mentioned at the beginning is achieved by a method according to the invention. A method for operating an apparatus is provided, the apparatus including an input signal interface for receiving digital measurement signals, the digital measurement signals comprising a sequence of scan values, wherein the measurement signals represent signals detected at connection points of electrical equipment. The method includes at least the following steps: a) configuring a signal window for each scan value of the measurement signal by means of a processor unit, the signal windows being formed by a portion of the scan value sequence of the measurement signal including the corresponding scan value and a predetermined number of scan values ​​of the measurement signal that are temporally earlier. The method is configured such that, in an initialization phase, the processor unit performs the following steps: b) determining a number of M signal windows as initialization windows; c) determining a number of K initialization windows as noise windows from the M initialization windows; d) analyzing each noise window by means of a predetermined X-order model and obtaining the corresponding coefficients, the coefficients forming an initial coefficient tuple configured for the corresponding noise window, wherein X is an even number between 1 and 5; and e) obtaining an expected value from the initial coefficient tuple of the noise window as a noise tuple. Furthermore, the method is configured such that, during the inspection phase, the processor unit performs the following steps: f) determining a plurality of signal windows as measurement windows; g) analyzing each measurement window using a predetermined X-order model and obtaining the coefficients to which it belongs, the coefficients forming measurement tuples configured for the corresponding measurement windows; h) obtaining the distance between the measurement tuple to which each measurement window belongs and the noise tuple for each measurement window; and i) determining measurement windows as error signal windows from the plurality of measurement windows, the measurement tuples to which the measurement windows belong have a distance greater than a predetermined first boundary value from the noise tuples, so that each error signal window represents a signal error of the measurement signal.

[0043] The method corresponds at least substantially to the apparatus according to the first aspect of the invention. Therefore, reference is made, at least similarly, to the advantageous descriptions, preferred features, advantages, and / or technical effects, as described with respect to the apparatus according to the first aspect of the invention and / or its advantageous embodiments. Repetitions are omitted.

[0044] An advantageous embodiment of the method is characterized in that the processor unit performs the initialization phase multiple times, preferably 2 to 10 times per second. By performing the initialization phase multiple times, the noise tuple can be updated periodically, thus allowing for the most accurate possible distinction between the noise component and the erroneous component of the measurement signal. Preferably, the method is characterized in that at least one checking phase is performed by means of the processor unit after each initialization phase.

[0045] Another advantageous embodiment of the method is characterized in that the X-order model is constructed by an X-order LPC model. Moreover, reference is made to the corresponding advantageous descriptions, preferred features, effects, and / or advantages, as have been set forth for a corresponding advantageous embodiment of the apparatus according to the first aspect of the invention.

[0046] Another advantageous embodiment of the method is characterized in that step c) comprises the following sub-steps c.1) and c.2): c.1) for each initialization window, the number of zero crossings of the portion of the measurement signal represented by the corresponding initialization window is determined; and c.2) from the M initialization windows, K initialization windows with the most zero crossings are determined as noise windows. Each of the two sub-steps c.1) and c.2) is executed by means of the processor unit. In step c.1), the number of zero crossings of the portion of the measurement signal represented by the corresponding initialization window is determined by means of the processor unit. And in step c.2), the number K noise windows are determined from the M initialization windows by means of the processor unit based on the number of zero crossings. Moreover, similarly, reference is made to the advantageous description, preferred features, technical effects and / or advantages of the method, as has been previously described for corresponding advantageous embodiments of the apparatus according to the first aspect of the invention. Repetition is also omitted here.

[0047] Another advantageous embodiment of the method is characterized in that step c) comprises the following sub-steps c.1) and c.2): c.1) for each initialization window, the signal energy of the portion of the measurement signal represented by the corresponding initialization window is obtained; and c.2) the number of initialization windows with minimum signal energy, K, are determined as noise windows from the M initialization windows. Each of the two sub-steps c.1) and c.2) is executed by means of a processor unit. In step c.1), the signal energy of the portion of the measurement signal represented by the corresponding initialization window is obtained by means of a processor unit. And in step c.2), the number of noise windows, K, is determined from the M initialization windows according to the minimum signal energy, by means of a processor unit. Moreover, similarly, reference is made to the advantageous description, preferred features, technical effects, and advantages of the method, as has been previously described for corresponding advantageous embodiments of the apparatus according to the first aspect of the invention. Repetition is also omitted here.

[0048] Another advantageous embodiment of the method is characterized in that the method is configured to perform the following steps by means of a processor unit during the inspection phase: j) analyzing each error signal window by means of a predetermined N-order model and obtaining the corresponding coefficients, the coefficients forming error tuples configured for the corresponding error signal windows, wherein N is an integer of at least 6, preferably 8.

[0049] Another advantageous embodiment of the method is characterized in that the method is configured to perform the following steps by means of a processor unit during the inspection phase: k) dividing the error tuple into multiple error groups, such that the error tuples of the same error group have a spacing between each other that is less than a predetermined second boundary value G, thereby assigning all error signal windows of the corresponding same error group to their respective error tuples to represent the same signal error.

[0050] The two last embodiments of the method are similarly described with reference to the advantageous descriptions, preferred features, effects and / or advantages, as have been previously described with respect to the two corresponding advantageous embodiments of the apparatus according to the first aspect of the invention. Attached Figure Description

[0051] Other features, advantages, and applications of the invention may arise from the following description of the embodiments and drawings. Herein, all described and / or illustrated features, in themselves and in any combination, form the technical solutions of the invention, independent of their composition or their reference relationships. In the drawings, the same reference numerals further represent the same or similar objects.

[0052] Figure 1 Advantageous embodiments of the apparatus and system are illustrated schematically.

[0053] Figure 2 An exemplary distribution of multiple coefficient tuples is shown;

[0054] Figure 3 An exemplary portion of the measured signal is shown;

[0055] Figure 4 An exemplary implementation of the method is illustrated in a schematic flowchart; and

[0056] Figure 5 Another exemplary implementation of the method is illustrated in a schematic flowchart. Detailed Implementation

[0057] exist Figure 1 An advantageous embodiment of device 2 is schematically shown. Device 2 has an input signal interface 4 and a processor unit 10. Furthermore, device 2 may have a display unit 22. Preferably, device 2 has an output signal interface 24.

[0058] exist Figure 1The electrical device 8 is schematically shown in the diagram. The electrical device 8 is used to transmit electrical energy. Thus, the electrical device 8 can be configured, for example, for high-voltage direct current transmission. However, it is also possible that the electrical device 8 is configured in other ways for transmitting electrical energy. Thus, the electrical device 8 can be configured, for example, for high-voltage alternating current transmission. The electrical device 8 preferably has an input interface 28 and a supply interface 30. At least one electrical line 38 can extend from the input interface 28 to the supply interface 30. A current can be generated through the electrical line 38. The electrical device 8... Figure 1 The preferred embodiments shown are, however, to be understood merely as exemplary. Thus, in an alternative embodiment, the electrical device 8 may be formed by a transformer, with the primary side connection forming a feed interface 28 and the secondary side connection forming a supply interface 30.

[0059] Electrical equipment 8 and device 2 may be part of system 26. System 26 may further be configured to include a sensor unit 32, which is preferably coupled to the input signal interface 4 of device 2 via signal line 36. Sensor unit 32 may be located at connection point 6 of electrical equipment 8 to detect electrical energy transmission. Thus, sensor unit 32 may be configured, for example, to detect voltage and / or current through electrical line 38. Sensor unit 32 is therefore configured to detect electrical energy signals from electrical equipment 8. Sensor unit 32 is further configured to generate an analog measurement signal representing the energy signal detected at connection point 6. The measurement signal can be transmitted from sensor unit 32 to the input signal interface 4 of device 2 via signal line 36. There, processor unit 10 digitizes the measurement signal. The digital measurement signal is a sequence of scan values ​​of the detected signal. The input signal interface 4 of device 2 is configured to receive the digital measurement signal, which includes the sequence to which the scan values ​​belong.

[0060] The preceding and subsequent descriptions of device 2 may preferably refer only to device 2. However, it is also stipulated that the same description of device 2 applies similarly to system 26, since system 26 may include device 2. Conversely, corresponding descriptions may apply.

[0061] exist Figure 3 An exemplary portion of the measurement signal with respect to time is recorded and schematically shown. In principle, at least a portion of the measurement signal represents the energy signal detected at connection point 6 of the electrical device 8. However, if an error occurs, such as a partial discharge in the electrical circuit 38, this error, particularly a partial discharge, causes a signal component in the measurement signal. The difficulty here lies in determining which portion of the measurement signal includes the signal component that concisely represents the corresponding error.

[0062] Device 2 is therefore also for the purpose of obtaining error signal windows of the measurement signal, wherein each error signal window represents a signal error of the measurement signal. Error signal windows may represent different signal errors. However, it is also possible that at least multiple error signal windows represent the same signal error of the measurement signal. This is particularly true when these signal errors are caused by the same error, such as partial discharge. Each error signal can therefore be a component of the measurement signal that occurs due to an error in energy transmission in the electrical device 8, particularly partial discharge.

[0063] To ensure the desired purpose of device 2, an advantageous arrangement of processor unit 10 is provided, which will be described below.

[0064] Processor unit 10 is configured to assign a signal window to each scan value of the measurement signal, each signal window being formed by a portion of a scan value sequence of the measurement signal including the corresponding scan value and a predetermined number of scan values ​​of the measurement signal that are temporally preceding. Each signal window may include the same number of scan values ​​of the measurement signal. Thus, for example, each signal window may include 32, 64, or 128 scan values ​​of the measurement signal. The scan values ​​of the signal window are a portion of the scan value sequence of the measurement signal. The scan values ​​of the corresponding signal window are therefore scan values ​​that are temporally directly successive. If the processor unit 10 generates a signal window, for example, for the scan values ​​of the measurement signal and assigns it to the corresponding scan values, then the signal window includes the corresponding scan value of the measurement signal and a predetermined number of scan values ​​that are temporally preceding the corresponding scan value of the measurement signal, for example, 127 temporally preceding scan values. In total, the signal window in this example includes 128 scan values. Since a signal window is assigned to each scan value of the measurement signal, it can also be said that the correspondingly generated signal window is a sequence of shifts in the signal window. If initialization phase 12 and checking phase 18 are described below, it should be considered that the signal window is configured for each scan value of the measurement signal in parallel in time.

[0065] Processor unit 10 is configured to perform multiple steps during initialization phase 12, which will be described below. Processor unit 10 is configured to determine a number M signal windows as initialization windows during initialization phase 12. This refers to a portion of the signal windows, which are continuously configured to scan values ​​of the measurement signal by means of processor unit 10. The number M is an integer between 100 and 50000. Preferably, the number M is an integer between 1000 and 10000. Therefore, for example, 10000 initialization windows can be determined by means of processor unit 10 during initialization phase 12. A number K noise windows are determined from these initialization windows. The number K is less than the number M. Thus, the number K can be an integer and preferably less than 0.5 or 0.1 times the number M. For example, it is assumed that the number K is 1000. Therefore, for example, 1000 noise windows can be determined by means of processor unit 10 from a large number of initialization windows. Which of the initialization windows are determined as noise windows can be determined by calculating the number of zero intersections or by calculating the signal energy of the portion of the measurement signal represented by the initialization windows. The processor unit can be configured to, for each initialization window, determine the number of zeros or the signal energy of the portion of the measured signal represented by the corresponding initialization window. Each initialization window is assigned a number of zeros or a specific signal energy. The number of zeros or the specific signal energy of the initialization windows are typically different. Initialization windows configured with the maximum number of zeros or the minimum signal energy represent the portion of the measured signal with the maximum noise component. The processor unit 10 is therefore configured to determine, from the M initialization windows, an initialization window with the most zeros or the minimum signal energy as a noise window. The noise window thus forms a subset of all initialization windows, where the portion of the measured signal represented by the noise window is affected by a large noise component.

[0066] The processor unit 10, and particularly its associated storage unit, can store an X-order model. The model preferably relates to a mathematical model configured to model energy transfer via the device 8. The model can have coefficients to match the transfer function or path represented by the model. By matching the model coefficients, a particularly good approximation of the actual transfer function in energy transfer via the device 8 can be achieved based on the selected coefficients. The model is also referred to as a predetermined model. The processor unit 10 is configured to analyze each noise window in the initialization phase 12 using the predetermined X-order model and thereby obtain the associated coefficients. Thus, a set of coefficients is obtained for each noise window. The coefficients can vary from noise window to noise window. For fast calculation, it has proven advantageous that X is an even number between one and five. Thus, X can be, for example, the number 2. In this case, the predetermined model is a second-order model. The number of coefficients preferably corresponds to the order of the model. In the previous example, therefore, two coefficients are provided for the second-order model. The coefficients obtained by the processor unit 10 for each noise window form an initial coefficient tuple 14 configured for the corresponding noise window. Noise can vary from noise window to noise window, and thus the initial coefficient tuple 14 can also vary from noise window to noise window. The processor unit 10 is therefore configured to obtain the expected value of the initial coefficient tuple 14 of the noise window as the noise tuple 16 during the initialization phase 12. The noise tuple 16 can therefore be the arithmetic mean of K initial coefficient tuples 14. If each initial coefficient tuple 14 consists of two coefficients, and for example 1000 noise windows are obtained and thus 1000 initial coefficient tuples 14 are obtained, then the noise tuple 16 can represent the expected value used for the initial coefficient tuple 14. The noise tuple 16 can be simply set at the "noise center". If another signal window is now used as the initialization window by means of a predetermined model analysis and the coefficients are obtained, then it can be determined based on the distance between the noise tuple 16 and the corresponding tuple composed of coefficients: whether the corresponding signal window can be evaluated as noise, or if the distance is large enough, then the signal window represents a portion of the measured signal with signal errors.

[0067] Processor unit 10 is configured to perform multiple steps in inspection phase 18, which are described below. Thus, processor unit 10 is configured to determine multiple signal windows as measurement windows in inspection phase 18. Preferably, the measurement windows are distinct from the initialization windows. For example, signal windows that immediately follow the final initialization window can be determined as measurement windows. The number of measurement windows is preferably significantly greater than the number of initialization windows. For example, at least 10,000 signal windows can be determined as measurement windows in inspection phase 18. However, it is preferably significantly more, such as 20,000, 30,000, or 100,000. Processor unit 10 is further configured to analyze each measurement window using a predetermined X-order model and thereby determine its associated coefficients. This involves the same X-order model that has also been applied in initialization phase 12. For each measurement window, its associated coefficients are determined, forming a measurement tuple 20 configured for the corresponding measurement window. For each measurement window, a measurement tuple 20 consisting of the corresponding coefficients is provided.

[0068] exist Figure 2 Multiple coefficient tuples 34 are schematically shown by corresponding points. Each coefficient tuple 34 consists of coefficients obtained in the analysis of the signal window using an X-order model. This analysis with respect to the noise window occurs in the initialization phase 12. This generates multiple initial coefficient tuples 14, which are then... Figure 2 The cloud is depicted as a collection of points based on its quantity. A noise tuple 16 is placed at the center of the cloud, which consists of the initial coefficient tuple 14. The noise tuple 16 thus forms the desired value for the initial coefficient tuple 14.

[0069] In inspection phase 18, the measurement window is analyzed using the same X-order model, thereby forming the aforementioned measurement tuples 20 with their associated coefficients. One of the measurement tuples 20 is in Figure 2 The measurement elements are marked with a cross. To determine whether the measurement element 20 indicates a portion of the measurement signal represented by its corresponding measurement window, or whether the measurement element 20 for the corresponding measurement window indicates a signal error in the portion of the measurement signal represented by the measurement window, a predetermined first boundary value G is determined. If the distance D between the corresponding measurement element 20 and the noise element 16 is greater than the first boundary value G, then the portion of the measurement signal represented by its corresponding measurement window is considered a signal error, which occurs due to an error in power transmission via the device 8. However, if the distance D is less than the first boundary value G, then the measurement element 20 indicates that the portion of the measurement signal represented by its corresponding measurement window does not represent a signal error, but is affected by noise.

[0070] The processor unit 10 is therefore configured to calculate the distance between the corresponding measurement element 20 and the noise element 16 for each measurement window in the inspection phase 18, and to determine, from the plurality of measurement windows, such measurement windows as error signal windows, wherein the corresponding measurement element 20 of each measurement window has a distance greater than a predetermined first boundary value G from the noise element 16. This achieves the following: each error signal window determined in this way represents a portion of the measurement signal including a signal error. This is because each of the portions of the measurement signal represented by the error signal window is at least partially affected by errors in energy transmission via the device 8. A signal error in the measurement signal is, for example, a component of the measurement signal caused by a corresponding error. Each error signal window therefore also represents a signal error in the measurement signal.

[0071] The error signal windows identified in inspection phase 18 represent separate, relevant portions of the measurement signal, which can be used to check for potential errors in the energy transmission via electrical equipment 8.

[0072] Preferably, device 2 is therefore provided with an output signal interface 24 configured to transmit an output signal. The output signal may directly or indirectly represent an error signal window. Thus, the output signal may, for example, include data for identifying error signal windows in a measurement signal. However, it may also be provided that the output signal represents each of the signal windows. The processor unit 10 of device 2 may be configured to control the output signal interface 24 in such a way as to transmit, in particular, send, the output signal. The output signal can provide other devices and / or units with information about the error signal windows obtained by device 2.

[0073] Alternatively or supplementing the output signal interface 24, the device 2 may have a display unit 22. Furthermore, the processor unit 10 may be configured to generate an image signal, which represents a measurement signal as a signal graph. The processor unit 10 may be configured to transmit the image signal to the display unit 22. The display unit 22 may be configured to generate an image on the display unit 22 based on the image signal.

[0074] Furthermore, it is advantageously specified that the processor unit 10 is configured to visually and identically indicate portions of a signal graph represented by an image signal, said portions being based on scan values ​​of a measurement signal, said scan values ​​being assigned to error signal windows of the same error group. Here, the processor unit 10 may be configured to divide the error signal windows into multiple error groups. Error signal windows of the same error group may respectively represent error signals caused by the same error in energy transmission via the electrical device 8.

[0075] exist Figure 3The diagram shows a portion of the measured signal, wherein the signal window 40 of the first error group is shown as a solid line, and the signal window 42 of the second error group is shown as a dotted line. Preferably, the processor unit 10 is configured to control the display unit 22 such that the display unit 22 displays a corresponding image based on the image signal, thereby visually depicting the signal graph and preferably depicting the portions of the signal graph with corresponding identical indications.

[0076] The initialization phase 12 and the checking phase 18 have already been described, namely, that they are analyzed using an X-order model. The X-order model preferably involves an X-order LPC model. Alternatively, another mathematical model may be applied. This can be formed, for example, using Fourier transform or wavelet transform.

[0077] Furthermore, it has proven advantageous to analyze each error signal window in the error signal window using a higher-order model. Here, the higher-order model can be of the same type as the X-order model. Preferably, the processor unit 10 is thus configured to analyze each error signal window using a predetermined N-order model and thereby obtain the corresponding coefficients, which form error tuples configured for the respective error signal windows, where N is an integer of at least six and preferably at least eight. The N-order model can also be stored by the processor unit 10 and, in particular, by its associated memory unit. Furthermore, it is preferably specified that the N-order model is an N-order LPC model. N can be at least six, at least eight, but also preferably at least ten, twelve, or fourteen, as described above. The higher order of the N-order model provides the advantage that the error signal windows can be analyzed more accurately, so that the corresponding errors can be checked more precisely. Furthermore, the error signal windows can be grouped based on error tuples. Thus, for example, it is advantageously specified that the processor unit 10 is configured to divide the error tuples into multiple error groups, such that the error tuples of the same error group have a spacing less than a predetermined second boundary value G between them. All error signal windows configured to the corresponding error group by their respective error tuples can thus represent the same or at least similar signal errors. Each of these signal errors can be caused by the same error in the energy transmission via the electrical device 8. Error tuples and / or error signal windows—which have the same error source, particularly caused by the same partial discharge—can be configured to the same error group. The number of groups can therefore correspond to the number of different errors in the energy transmission via the electrical device 8.

[0078] exist Figure 4An advantageous embodiment of the method is shown. The method includes method steps a) to i). According to method step a), by means of processor unit 10, a signal window is configured for each scan value of the measurement signal, the signal window being formed respectively by a portion of the scan value sequence of the measurement signal including the corresponding scan value and a predetermined number of time-preceding scan values ​​of the measurement signal. Step a) is performed continuously, that is, preferably also in parallel in time for initialization phase 12 and / or checking phase 18.

[0079] Method steps b) to e) are configured for initialization phase 12. The method is configured such that, in initialization phase 12, the following steps are performed by means of processor unit 10: b) determining a number of M signal windows as initialization windows; c) determining a number of K initialization windows as noise windows from the M initialization windows; d) analyzing each noise window by means of a predetermined X-order model and obtaining the coefficients thereto, the coefficients forming an initial coefficient tuple configured for the corresponding noise window, where X is an even number between 1 and 5; and e) obtaining the expected value from the initial coefficient tuple of the noise window as a noise tuple.

[0080] Furthermore, the method is configured such that, in the inspection phase 18, the processor unit 10 performs the following steps: f) determining a plurality of signal windows as measurement windows; g) analyzing each measurement window using a predetermined X-order model and obtaining the coefficients to which it belongs, the coefficients forming measurement tuples configured for the corresponding measurement windows; h) obtaining the distance between the measurement tuple to which each measurement window belongs and the noise tuple; and i) determining, from the plurality of measurement windows, measurement windows as error signal windows, wherein the measurement tuples to which the measurement windows belong have a distance greater than a predetermined first boundary value from the noise tuples, such that each error signal window represents a signal error of the measurement signal.

[0081] As by Figure 4 It is understood that the checking phase 18 immediately follows the initialization phase 12. With the aid of the processor unit 10, steps b) to e) are executed in the initialization phase 12. Subsequently, steps f) to i) are executed in the checking phase 18. Then the initialization phase 12 can begin again. In principle, however, it is also possible that the checking phase 18 is executed multiple times after the initialization phase 12 and before a new initialization phase 12 begins. The method is similarly described above with reference to the preferred options, advantageous features, technical effects, and / or advantages.

[0082] exist Figure 5 Another advantageous implementation of the method is shown, wherein reference is made to the description of the preceding steps a) to i). Figure 5The method in step c) includes the following sub-steps c.1) and c.2), which are executed by the processor unit 10: c.1) for each initialization window, the number of zero crossings and / or the signal energy of the portion of the measured signal represented by the corresponding initialization window are obtained; and c.2) the number of initialization windows with the most zero crossings and / or the minimum signal energy are determined as noise windows from the M initialization windows.

[0083] In addition, Figure 5 The method shown also includes step j) of inspection phase 18, which is executed by means of processor unit 10. According to step j), each error signal window is analyzed by means of a predetermined N-order model and the coefficients thereto are obtained, which form error tuples configured for the corresponding error signal windows, where N is an integer of at least 6, preferably 8.

[0084] It should be further noted that "having" does not exclude other elements or steps, and "a" or "an" does not exclude multiple. Furthermore, it should be noted that the features described in one of the above embodiments can also be applied in combination with other features of the other embodiments described above. Reference numerals should not be considered limiting.

[0085] List of reference numerals

[0086] a1 First coefficient

[0087] a2 second coefficient

[0088] D distance

[0089] G boundary value

[0090] 2 devices

[0091] 4 Input Signal Interface

[0092] 6 connection positions

[0093] 8 devices

[0094] 10 processor units

[0095] 12 Initialization Phase

[0096] 14 initial coefficient tuples

[0097] 16 noise tuples

[0098] 18 Inspection Phase

[0099] 20 measurement elements

[0100] 22 display units

[0101] 24 output signal interfaces

[0102] 26 system

[0103] 28 feed inlet

[0104] 30 supply interfaces

[0105] 32 sensor units

[0106] 34-coefficient tuple

[0107] 36 signal lines

[0108] 38 Electrical Circuits

[0109] 40 Signal window for the first error group

[0110] 42 Signal window for the second error group

Claims

1. A device (2) for determining an error signal window in a measurement signal, comprising: The input signal interface (4) is used to receive digital measurement signals, which include a sequence of scan values ​​and represent signals detected at the connection position (6) of the electrical equipment (8). as well as Processor unit (10) in, The processor unit (10) is configured to configure a signal window for each scan value of the measurement signal, the signal window being formed by a portion of a scan value sequence of the measurement signal including the corresponding scan value and a predetermined number of scan values ​​of the measurement signal that are earlier in time. The processor unit (10) is configured to determine a number of M signal windows as initialization windows in the initialization phase (12) and determine a number of K noise windows by the M initialization windows, analyze each noise window by means of a predetermined X-order model and obtain the initial coefficients thereto, the initial coefficients form an initial coefficient tuple (14) configured for the corresponding noise window, and obtain the expected value as a noise tuple (16) from the initial coefficient tuple (14) of the noise window, where X is an even number between 1 and 5; as well as The processor unit (10) is configured to determine multiple signal windows as measurement windows in the inspection phase (18), analyze each measurement window by means of a predetermined X-order model and obtain the measurement coefficients thereto, the measurement coefficients forming measurement tuples (20) configured for the corresponding measurement windows, obtain the distance between the measurement tuple (20) and the noise tuple (16) for each measurement window, and determine the following error measurement windows as error signal windows by the measurement windows, the measurement tuples (20) of the error measurement windows having a distance greater than a predetermined first boundary value from the noise tuple (16), so that each error signal window represents a signal error of the measurement signal.

2. The apparatus (2) according to claim 1, characterized in that, The processor unit (10) is configured to, for each initialization window, determine the number of zero crossings and / or the signal energy of the portion of the measured signal represented by the corresponding initialization window, and the processor unit (10) is further configured to determine, by the initialization window, a number of K initialization windows having the most zero crossings and / or the least signal energy as noise windows.

3. The apparatus (2) according to claim 1 or 2, characterized in that, The X-order model is configured as an X-order LPC model.

4. The apparatus (2) according to claim 1 or 2, characterized in that, M is an integer of at least 100, and K is an integer less than M.

5. The apparatus (2) according to claim 4, characterized in that, M is an integer of at least 10,000.

6. The apparatus (2) according to claim 1 or 2, characterized in that, Each initialization phase (12) lasts for a maximum of 0.1 seconds.

7. The apparatus (2) according to claim 6, characterized in that, Each initialization phase (12) lasts for a maximum of 0.05 seconds.

8. The apparatus (2) according to claim 1 or 2, characterized in that, The processor unit (10) is configured to analyze each error signal window by means of a predetermined N-order model and obtain the corresponding error coefficient, the error coefficient forming an error tuple assigned to the corresponding error signal window, wherein N is an integer of at least 6.

9. The apparatus (2) according to claim 8, characterized in that, N is an integer of at least 8.

10. The apparatus (2) according to claim 8, characterized in that, The processor unit (10) is configured to divide the error element into multiple error groups, such that the error elements of the same error group have a spacing of less than a predetermined second boundary value between them, so that all error signal windows of the corresponding same error group are configured to represent the same signal error through their respective error elements.

11. The apparatus (2) according to claim 1 or 2, characterized in that, The processor unit (10) is configured to divide the error signal window into multiple error groups, such that the error tuples of the error signal windows in the same error group have a spacing of less than a predetermined second boundary value between them, so that all error signal windows in the same error group represent the same signal error.

12. The apparatus (2) according to claim 10, characterized in that, The processor unit (10) is configured to identify multiple different errors in the electrical equipment (8) based on the multiple error groups.

13. The apparatus (2) according to claim 10, characterized in that, The processor unit (10) is configured to generate an image signal representing a measurement signal as a signal graph, and the processor unit (10) is configured to visually indicate the following error portions of the signal graph, the error portions being based on the following scan values ​​of the measurement signal, the scan values ​​being configured to error signal windows of the same error group.

14. The apparatus (2) according to claim 13, characterized in that, The device (2) has a display unit (22), and the processor unit (10) is configured to control the display unit (22), so that the display unit (22) displays an image based on an image signal, thereby the image visually depicts a signal diagram.

15. The apparatus (2) according to claim 1 or 2, characterized in that, The processor unit (10) is configured to execute the initialization phase (12) multiple times, thereby re-evaluating the noise tuple (16) with each initialization phase (12).

16. The apparatus (2) according to claim 15, characterized in that, The processor unit (10) is configured to perform at least one check phase (18) after each initialization phase (12).

17. A system for energy transfer (26), wherein, The system (26) has: Electrical equipment (8) is used to transmit electrical energy signals from the feed interface (28) of the electrical equipment (8) to the supply interface (30) of the electrical equipment (8). Sensor unit (32); and The apparatus (2) according to any one of claims 1 to 16; The sensor unit (32) is located at the connection position (6) between the feed interface (28) and the supply interface (30) of the electrical equipment (8); The sensor unit (32) is configured to detect electrical energy signals and generate digital measurement signals representing the energy signals detected at the connection point (6); and The sensor unit (32) is coupled to the signal interface of the device (2) so as to transmit the measurement signal to the signal interface.

18. The system (26) according to claim 17, characterized in that, The electrical equipment (8) is configured as a high-voltage line, a transformer, a rotating motor, a gas-insulated line or a gas-insulated power distribution device.

19. A method for operating an apparatus (2) for determining an error signal window in a measurement signal, the apparatus comprising an input signal interface (4) for receiving a digital measurement signal comprising a sequence of scan values, the digital measurement signal representing a signal detected at a connection point (6) of an electrical device (8), in, The method includes at least the following steps: a) By means of the processor unit (10), each scan value of the measurement signal is configured with a signal window, the signal window being formed by a portion of the scan value sequence of the measurement signal including the corresponding scan value and a predetermined number of scan values ​​of the measurement signal that are earlier in time; The method is configured such that, during the initialization phase (12), the following steps are performed by means of the processor unit (10): b) Determine the number of M signal windows as initialization windows; c) Determine K initialization windows from the M initialization windows to serve as noise windows; d) Each noise window is analyzed using a predetermined X-order model, and its initial coefficients are determined, forming an initial coefficient tuple (14) assigned to the corresponding noise window, where X is an even number between 1 and 5; and e) Obtain the expected value from the initial coefficient tuple (14) of the noise window as the noise tuple (16). The method is configured such that, during the inspection phase (18), the following steps are performed by means of the processor unit (10): f) Determine multiple signal windows as measurement windows; g) Analyze each measurement window using a predetermined X-order model and obtain the corresponding measurement coefficients, which form measurement tuples (20) assigned to the corresponding measurement windows. h) For each measurement window, calculate the distance between the corresponding measurement element (20) and the noise element (16); and i) The measurement window determines the following error measurement window as the error signal window, wherein the measurement tuple (20) to which the error measurement window belongs has a distance greater than a predetermined first boundary value from the noise tuple (16), so that each error signal window represents a signal error of the measurement signal.

20. The method according to claim 19, characterized in that, The initialization phase (12) is executed multiple times by the processor unit (10).

21. The method according to claim 20, characterized in that, The initialization phase (12) is executed by the processor unit (10) 2 to 10 times per second.

22. The method according to any one of claims 19 to 21, characterized in that, The X-order model is an X-order LPC model.

23. The method according to any one of claims 19 to 21, characterized in that, Step c) includes the following sub-steps c.1) and c.2): c.1) For each initialization window, determine the number of zero crossings for the portion of the measured signal represented by the corresponding initialization window; and / or for each initialization window, determine the signal energy for the portion of the measured signal represented by the corresponding initialization window; and c.2) The K initialization windows with the most zero intersections are determined from the M initialization windows and used as noise windows; and / or The K initialization windows with the minimum signal energy are determined from the M initialization windows and used as noise windows.

24. The method according to any one of claims 19 to 21, characterized in that, The method is configured such that, during the inspection phase (18), the following steps are further performed by means of the processor unit (10): j) Analyze each error signal window using a predetermined N-order model and obtain the corresponding error coefficient, which forms an error tuple assigned to the corresponding error signal window, where N is an integer of at least 6.

25. The method according to claim 24, characterized in that, N is an integer of at least 8.

26. The method according to claim 24, characterized in that, The method is configured such that, during the inspection phase (18), the following steps are further performed by means of the processor unit (10): k) Divide the error tuple into multiple error groups, such that error tuples in the same error group have a spacing between each other that is less than a predetermined second boundary value, so that all error signal windows in the corresponding same error group are configured to represent the same signal error through their respective error tuples.

Images