Method for generating input data for a state estimation of a power grid, state estimation and control of a power grid

Abstract

The invention relates to a method for generating input data for a state estimation of a power grid (1), the power grid (1) having at least two medium-voltage nodes (MV2,..., MV4), and the input data comprising at least covariances and expected values of electrical power at the medium-voltage nodes (MV2,...,MV4), wherein a low-voltage network (LV2,..., LV4) comprising Ν households is connected to one of the medium-voltage nodes (MV2,..., MV4), and a low-voltage network (LV2,..., LV4) comprising Ν households is connected to the other of the medium-voltage nodes (MV2,..., MV4). The method is characterised in that at least one of the low-voltage networks (LV4) has a smart meter penetration rate lower than a defined threshold, the input data comprising at least one covariance between electrical powers aggregated at the medium voltage nodes (MV2,...,MV4), the covariance being determined by a provided covariance value Cov(Ν,Μ) depending on the number of households Ν,Μ in the associated low-voltage networks. The invention also relates to: a method for state estimation of a power grid (1); a method for controlling a power grid (1) in an open-loop or closed-loop manner; and a computer program product.

Description

[0001] 202307903 1 Description Method for generating input data for a state estimation of a power grid, state estimation and control of a power grid The invention relates to a method according to the preamble of patent claim 1, a state estimation according to the preamble of patent claim 12, a control of a power grid according to the preamble of patent claim 14 and a computer program product according to the preamble of patent claim 15. Due to the progressive installation of additional consumers (e-mobility) and volatile decentralized generation units (photovoltaics and wind) in medium and low voltage grids, critical operating states can increasingly occur due to the higher capacity utilization of the power grid in the two lower voltage levels (medium and low voltage).In addition to expensive grid expansion, attractive operating approaches are those that specifically control these consumers in critical grid situations to avoid overloads, such as violations of the voltage range or the triggering of fuses due to exceeding maximum power. However, these approaches can only be technically implemented if the state of the power grid or its subgrids is known. Therefore, state estimation in the lower voltage levels, especially for medium-voltage and low-voltage grids, is becoming increasingly relevant. Since very few real-time measurements are installed in medium-voltage and low-voltage grids, additional data sources and / or prior assumptions about the grid state are required. Such data can be derived, for example, from historical or predicted data.However, when using them as input data for state estimation, their uncertainty, for example their variance, must be determined and taken into account. Using probabilistic state estimators, it is possible to process these uncertainties in the input data and calculate output data with probabilistic information, i.e., a probability distribution for the state. A technical challenge here is to combine measurement data and prior assumptions in a suitable way to obtain a reliable state estimation, i.e., an estimate of the grid state of the power grid. It is particularly important to note that loads typically exhibit correlations, i.e., they are not stochastically independent. According to the state of the art, such correlations or covariances are typically not taken into account.Methods for estimating the state of distribution systems based on the WLS (Weighted Least Square) approach, which is state-of-the-art for higher voltage levels, convert prior knowledge into so-called pseudomeasurements. These are passed to the state estimator as input data equivalent to the real-time measurements, but with high standard deviations for measurement accuracy (and thus low weight). Bayesian methods are probabilistic methods that consider the prior knowledge in the form of a prior distribution for the possible states. Using Bayes' theorem, a posterior distribution of the expected grid state is determined using (real-time) measurement data.For medium-voltage grids (MV grids), there is high potential for useful prior information in the distribution of load data (power data) from the underlying low-voltage grids, which can be estimated, for example, using power profiles from intelligent measuring systems (smart meters), annual energy demands from analog and digital meter data, or from standard load profiles. Exogenous sources, such as solar radiation for estimating photovoltaic generation, can also be used as a source of information. The aforementioned prior information has in common that its distribution, unlike typical measurement errors, is not stochastically independent. For example, if the load of a household is significantly higher than the annual average, the probability increases that the load of a neighboring household is also higher than the annual average.Advanced WLS methods therefore consider the correlation between pseudo-measurements. However, these methods place very high demands on the available information sources for load distributions, as they calculate correlations between individual loads empirically from historical data and therefore require fully recorded power time series (from real-time measurements or smart meters). In the medium-voltage grids in Europe and the USA, only a few real-time measurements are generally available. Furthermore, no country has 100 percent smart meter coverage. Furthermore, collected meter data and smart meter data may not be used for grid operation due to data protection reasons.Therefore, for the background distribution or pseudo-measurements of the state estimation, additional information from energy consumption measurements and standard load profiles or even from other external sources (solar radiation) must typically be considered. However, known methods cannot derive any information on correlations from this. Therefore, known methods assume that the data is stochastically independent. The present invention is based on the object of generating input data for a state estimation of a power grid that better considers a correlation or covariance between the medium-voltage nodes of the power grid.The problem is solved by a method having the features of independent patent claim 1, by a state estimation having the features of independent patent claim 12, by a method for controlling a power grid having the features of independent patent claim 14, and by a computer program product having the features of independent patent claim 15. Advantageous embodiments and developments of the invention are specified in the dependent patent claims.The method according to the invention for generating input data for a state estimation of a power grid, wherein the power grid has at least two medium-voltage nodes, and the input data comprise at least covariances and expected values ​​of electrical power at the medium-voltage nodes, wherein a low-voltage grid with ^^ households is connected to one of the medium-voltage nodes and a low-voltage grid with ^^ households is connected to the other of the medium-voltage nodes, is characterized in that at least one of the low-voltage grids has a smart meter penetration rate less than a specified threshold value, wherein the input data comprise at least one covariance between electrical power aggregated at the medium-voltage nodes, wherein the covariance is determined by a provided covariance value Cov^ ^^, ^^^ depending on the number of households ^^, ^^ in the respective low-voltage grids.The method according to the invention and / or one or more functions, features, and / or steps of the method according to the invention and / or one of its embodiments can be computer-aided. In particular, a control unit provided for controlling or regulating the power grid comprises one or more computing units that are designed and configured to carry out the method, for example, by means of commands. The smart meter penetration rate is the relative number of smart meters installed in the low-voltage grid. For example, 40 percent of households in a low-voltage grid have a smart meter. In this case, the smart meter penetration rate for this low-voltage grid then has a value of 0.4 or 40 percent. It should be noted that the data from the respective smart meter must also be available.A smart meter according to the present invention can thus be a general measuring unit that discretely or continuously records at least the temporal profile of the electrical active power and / or reactive power at a low-voltage node (household) and makes it available externally with respect to the household. A household within the meaning of the present invention can be an electricity consumer and / or an electricity generator. Typically, a household forms a low-voltage node of the respective low-voltage grid. Furthermore, a household could be formed by a large-scale consumer, i.e., by a consumer with an annual energy consumption above a specified threshold, for example, above 0.1 MWh. Such large-scale consumers can also be referred to as large-scale customers or commercial customers.These RLM consumers typically have recorded power measurements (RLM), which provide a time-resolved load profile. This power profile, like smart meter data or smart meter power profiles, typically has a temporal resolution of 15 minutes. For the purposes of the present invention, correlation and covariance are equivalent. If the covariance is known, the correlation is obtained in a known manner by normalizing to the respective standard deviations (root of the variances). The covariance between medium-voltage nodes ^^, ^^ can typically be represented as a matrix. Each medium-voltage node ^^ has a temporal progression of its electrical power ^^ ^,௧ , for example, its active power and / or reactive power. Therefore, the correlation ^^ ௫^,௫^ ൌ ^^ ௫^,௫^ / ^ ^^௫^ ^^ ௫^ , wobei ^^ ௫^ the variances of the time-dependent electrical power ^^ ^,௧ respectively ^^ ^,௧ For existing discrete time series, the covariance of the input data can thus in principle be calculated according to ^^ ௫^,௫^ ൌ ∑ ୀ^ ^ ^^ ^,௧ െ 1^ can be determined. Here, ^^௫^ ൌ∑ୀ^ ^^^,௧ / ^^ are the expected values ​​of the input data. However, for medium-voltage nodes in which at least one of the medium-voltage nodes has at least one low-voltage grid with a smart meter penetration rate less than the specified threshold, the covariance formed over the time series is not used, but rather the provided covariance value Cov^ ^^, ^^^. A basic idea of ​​the present invention is that for medium-voltage nodes whose low-voltage grids have a low smart meter penetration, no technically usable calculation of correlations or covariances with other medium-voltage grids can be carried out. This is because for such medium-voltage nodes, neither direct real-time measurements nor sufficient measurement data, for example from smart meters, are available. However, it is disadvantageous, as typically assumed in the prior art,no correlation can be assumed. Even the use of standard load profiles may not solve this technical problem, as this would lead to an unrealistically perfect correlation. The present invention solves this problem by determining the correlation or covariance using the provided covariance value Cov^ ^^, ^^^ as a function of the number of households in the respective low-voltage grids. The covariance value Cov^ ^^, ^^^ can be determined using low-voltage grids that have a sufficient database. Thus, according to the invention, typical correlation values ​​or covariance values ​​between low-voltage grids comparable in terms of the number of households are used. The provided covariance value Cov^ ^^,^^^ between a medium-voltage node whose low-voltage grid has ^^ households and a medium-voltage node whose low-voltage grid has ^^ households can be provided in the form of a table, a graph, a matrix, and / or an analytical formula. An advantage of the invention is that correlation information can be used even if incompletely recorded power time series (from real-time measurements or smart meters) are available. Thus, other input variables such as the energy consumption of analog / digital meters, SLP, and exogenous sources such as solar radiation can be utilized without having to make the unrealistic assumption that these are stochastically independent. This leads to a significant improvement in the accuracy of a state estimate.which uses the input data generated according to the invention. The method according to the invention for estimating the state of a power grid with several medium-voltage nodes, wherein input data comprising expected values ​​and covariances of electrical power at the medium-voltage nodes are used, is characterized in that the input data are generated by a method according to the present invention and / or one of its embodiments. 202307903 8 Similar, equivalent, and equivalent advantages and / or embodiments of the state estimation according to the invention result from the method according to the invention. The method according to the invention for controlling a power grid by means of a control unit, wherein the power grid comprises several medium-voltage nodes, and the control or regulation is based on a state estimate, is characterized in thatthat the state estimation is carried out by a method for state estimation according to the present invention and / or one of its embodiments. Similar, equivalent, and equivalent advantages and / or embodiments of the method according to the invention for controlling a power grid result in the state estimation according to the invention. The computer program product according to the invention is characterized in that it comprises instructions which, when the program is executed by a computing unit, in particular a computer, cause the computer to execute a method and / or steps of the method according to the present invention and / or one of its embodiments. Particularly preferably, the control unit comprises the computing unit. Similar,equivalent and equivalent advantages and / or embodiments of the computer program product according to the invention. According to a preferred embodiment of the invention, the power grid has more than two medium-voltage nodes, wherein a low-voltage grid is connected to one or more of the medium-voltage nodes, and the medium-voltage nodes are divided into four categories, wherein - the first category comprises medium-voltage nodes whose aggregated electrical power has been measured; wherein 202307903 9 - the second category comprises medium-voltage nodes whose low-voltage grids have a full smart meter penetration rate; wherein - the third category comprises medium-voltage nodes whose low-voltage grids do not have a full smart meter penetration rate and have a smart meter penetration rate greater than or equal to the specified threshold; and - the fourth category comprises medium-voltage nodes,whose low-voltage networks have a smart meter penetration rate lower than the specified threshold; where - the covariance between a medium-voltage node of any category and at least one medium-voltage node of the fourth category is determined by the provided covariance value Cov^ ^^, ^^^ depending on the number of households ^^, ^^ in the respective low-voltage networks. In other words, the medium-voltage nodes are divided into four categories, with the categories differing essentially in whether immediate real-time measurements are available (first category) or not, and the respective smart meter penetration rate (second to fourth categories). The first category includes medium-voltage nodes,for which a direct measurement of their electrical power is available (real-time measurements). Therefore, the measurement data from subordinate low-voltage networks are not mandatory in this case. In other words, for the electrical power of a medium-voltage node of the first category, and ^^ ^^,௧ ൌ ^^ ^^^^^౩,௧ , where ^^ ^^^^^౩,௧ the measured active power and ^^ ^^^^^౩,௧ the measured reactive power. The first category can therefore be abbreviated as MV ୫^ୟ^and the recorded measurements at the medium-voltage nodes (MV nodes) can be used directly as time series input. For example, MV can include large customers (RLM consumers), the underlying transformer station, and the local network station. 202307903 10 In the second category, all households in the low-voltage grid of the respective medium-voltage node have a smart meter. In other words, the smart meter coverage or the smart meter penetration rate in this case for the subordinate low-voltage grid is 100 percent. This allows the time-dependent electrical power of the medium-voltage nodes of the second category to be determined by aggregating the smart meter data. In other words, in this case and where ^^^^^^^౩,^,௧, ^^^^^^^౩,^,௧ denotes the respective smart meter active power data and reactive power data, respectively. ^^ is the number of households in the subordinate low-voltage grid, since in the second category there is complete smart meter coverage, i.e., every household has a smart meter. In other words, in this case, the active and reactive power of the smart meters are summed for the corresponding MV node, creating an aggregated time series for the respective MV node. Power losses are neglected here. The second category can be abbreviated as SM ^^^%The third category includes medium-voltage nodes whose subordinate low-voltage grid does not have complete smart meter coverage or smart meter penetration rate, but still has a smart meter penetration rate greater than or equal to the specified threshold. Thus, a sufficiently large proportion of smart meter data can be used in terms of the threshold. In this case, it is assumed that the summed or aggregated smart meter time series and ^^ ୗ^,௧ of the measured smart meter households are representative of the load behavior of this low-voltage grid. To determine the correct energy demand at the MV node, the summed smart meter time series are scaled with a scaling factor ^^ ^ୡୟ୪^multiplied. The scaling factor corresponds to the ratio between the total energy demand of all low-voltage nodes (LV nodes) or households, as recorded by analog or digital meters, and the total energy demand of the loads, whose temporal variation is recorded by smart meters. Here, the reactive power can be assumed to be constant with a power factor cos ^^. However, a non-constant power factor can also be provided. In other words, ^^ ^^,௧ ^ ^^ ^^,௧ ∙ ^^ ^ୡୟ୪^ and ^^ ^^,௧ ^ ^^ ^^^,௧ ∙ ^^ ^ୡୟ୪^ with ^^ ^ୡୟ୪^ ൌ ^^ ^ై^ / ^^ ^^^ ,, where the total energy consumption of the low-voltage network and ^^ ^^^ The total energy consumption of households in the low-voltage network with smart meters. The third category can be abbreviated as SM ୫୭^^୪^The fourth category includes medium-voltage nodes whose subordinate low-voltage networks do not have sufficient smart meter coverage or smart meter penetration rate with respect to the specified threshold. The fourth category can therefore be abbreviated as SM ୪^^^ Using standard profiles and the available smart meter data, an aggregated load or aggregated electrical power can be determined for the respective medium-voltage node. In other words, in this case, all loads / households not recorded via smart meters are assigned a suitable scaled standard load profile (SLP). These load profiles are aggregated with the measured profiles of loads measured by smart meters to form an aggregated total profile at the medium-voltage node. In other words, ^^ ^^,௧ ^ ^^ ^^,௧ ^ ^^ ^^|ୗ^^,௧ and ^^ ^^,௧ ^ ^^ ^^^,௧ ^ ^^^^|ୗ^^,௧. Using these aggregated electrical powers, the expected values ​​for the input data can be determined. However, the electrical powers generated in this way are unsuitable for determining the correlation or covariance, since the proportion of standard profiles is too high. This would result in an unrealistically high correlation of almost 100 percent. Thus, according to the invention, the covariances in this case, i.e., in the case of the fourth category, are determined using the provided covariance values ​​Cov^ ^^, ^^^ depending on the respective number of low-voltage nodes (households).The following procedure can be used to determine the provided and, in this sense, standardized covariance values: a) Using a sufficiently large amount of measurement data from representative loads, random subsets of different sizes are formed, and the pairwise correlation of the aggregated time series is calculated depending on the size of the subsets. This calculation only needs to be performed once for a grid area. b) For a medium-voltage node pair ^ ^^, ^^^, for which the covariance ^^. ௫^,௫^is to be determined, a typical covariance value or correlation value is determined from the previously calculated values, depending on the number of loads in the respective subnetworks. For example, for a medium-voltage grid ^^ with 75 low-voltage nodes / loads / households in the underlying low-voltage grid and a medium-voltage grid ^^ with 15 low-voltage nodes / loads / households, a typical correlation coefficient of approximately 0.6 results. c) From a sufficiently large number of households / consumers in the low-voltage grid, slightly fluctuating correlation values ​​result (low value for the standard deviation of the correlation coefficients), which makes the values ​​more reliable.Thus, the correlation or covariance is only calculated using the above-mentioned aggregated temporal powers if the correlation or covariance is determined with at least one medium-voltage node of the fourth category. If no medium-voltage node of the fourth category is involved, the correlation or covariance can be calculated in a known manner from the above-mentioned time series for the aggregated electrical powers. For example, the following applies to discrete time series. The expected values, however, can be calculated independently of the category according to ^^ ^,௧ / ^^. In each case ^^ ^,௧ ൌ ^^ ^^,௧ for a medium voltage node ^^, where depending on the category as mentioned above. In an advantageous development of the invention, for each medium-voltage node in the first category, its electrical power is determined using the recorded electrical power associated with the respective medium-voltage node. As already explained above, in this case the aggregated power at the respective medium-voltage node is determined directly by real-time measurements. According to an advantageous embodiment of the invention, for each medium-voltage node in the second category, its electrical power is determined by aggregating the smart meter measurement data associated with the respective medium-voltage node. As already explained above, in this case the aggregated power at the respective medium-voltage node is determined directly by smart meter data.This is the case because, for medium-voltage nodes in the second category, the respective associated low-voltage grid has complete, i.e., 100 percent, smart meter coverage or smart meter penetration rate. In an advantageous development of the invention, for each medium-voltage node in the third category, its 202307903 14 electrical power is determined using a scaling factor and an aggregation of the smart meter measurement data associated with the respective medium-voltage node. In other words, the smart meter data available in this case is upscaled to the entire respective low-voltage grid. According to the definition of the third category, there are still a sufficient number of smart meters available for such medium-voltage nodes within their associated low-voltage grid.It can therefore be assumed that the smart meter data is representative of the respective low-voltage grid and thus of the aggregated power at the respective medium-voltage node. The following applies to the power at a medium-voltage node of the third category: ^^. ^^,௧ ^ ^^ ^^,௧ ∙ ^^ ^ୡୟ୪^ and ^^ ^^,௧ ^ ^^ ^^^,௧ ∙ ^^ ^ୡୟ୪^ with ^^ ^ୡୟ୪^ where ^^ ^ై^ the total energy consumption of the low-voltage network and ^^ ^^^the total energy consumption of households in the low-voltage grid with smart meters. According to an advantageous embodiment of the invention, for each medium-voltage node in the fourth category, its electrical power is determined by aggregating the smart meter measurement data and standard load profiles associated with the respective medium-voltage node. In this case, according to the definition of the fourth category, too little smart meter data is available, so that, in contrast to the third category, it cannot be assumed that the available smart meter data is representative of the respective low-voltage grid and the associated medium-voltage node. Thus, the missing data in this sense is supplemented using standard load profiles. In other words, the following applies to the aggregated electrical power at one of the medium-voltage nodes in the fourth category and where Smart meter data and ^^ ^^|ୗ^^,௧ , ^^^^|ୗ^^,௧the standard load profiles. 202307903 15 The problem is that due to the high proportion of standard load profiles, no technically usable determination of the correlation or covariance of a medium-voltage node of any category with at least one medium-voltage node of the fourth category can be made. The present invention therefore provides for the determination of the correlation or covariance not via the aforementioned electrical power, but rather via essentially standardized correlation values ​​or covariance values, which depend on the number of households or the number of low-voltage nodes in the participating low-voltage networks. The standardized correlation values ​​or covariance values ​​can be determined once in advance, for example using medium-voltage networks / low-voltage networks that are technically comparable and have a correspondingly large database.In an advantageous development of the invention, the covariances and expected values ​​of the input data are thus determined using the determined electrical power of the medium-voltage nodes, wherein the covariance between a medium-voltage node of any category and at least one medium-voltage node of the fourth category is determined by the provided covariance value Cov^ ^^, ^^^ depending on the number of households in the respective low-voltage networks. This advantageously ensures that technically realistic correlations or covariances are used for the state estimation. According to an advantageous embodiment of the invention, the defined threshold value for the smart meter penetration rate has a value in the range of 0.5 to 0.8. Particularly preferably, the smart meter penetration rate (smart meter coverage) has a value of 0.6, i.e. of 60 202307903 16 percent.This ensures that the smart meter data of the third category is still sufficiently representative of the respective low-voltage grid, so that a technically realistic correlation or covariance can be determined using the scaled electrical power. In an advantageous development of the invention, the provided covariance value Cov^ ^^, ^^^ is determined by determining low-voltage grids comparable in terms of the number of households, wherein measurement data of the electrical power of the households used for the low-voltage grids used here are available. This allows the covariance value to be advantageously determined in advance and standardized in this sense. According to an advantageous embodiment of the invention, electrical active power and / or reactive power are used as electrical power.This is advantageous because performance data is typically provided by the smart meters and / or the real-time measurements. In an advantageous development of the invention, the input data comprises correlations between the medium-voltage nodes, wherein the respective correlations are formed using the respective covariances. In the present case, correlations and / or covariances can thus be used equivalently as input data for the state estimation. The correlations result from the covariances by normalizing to the respective standard deviations. 202307903 17 According to an advantageous embodiment of the invention, the state estimation is carried out using a Bayesian state estimation or a WLS state estimation. The generated input data can advantageously be used for probabilistic methods and / or WLS methods.Further advantages, features, and details of the invention will become apparent from the exemplary embodiments described below and from the drawings. These show schematically: Figure 1 shows a power grid with several medium-voltage nodes and associated low-voltage networks; and Figure 2 shows a flow diagram of a method according to one embodiment of the present invention. Elements of the same type, value, or action may be provided with the same reference numerals in one of the figures or in the figures. Figure 1 shows a power grid 1 which, by way of example, comprises several medium-voltage nodes MV1,…,MV4. The medium-voltage node MV1 does not have a directly underlying low-voltage network, but rather forms the interface or connection to a higher-level high-voltage network 10. The further medium-voltage nodes MV2,…,MV4 each have a subordinate low-voltage network LV2,…,LV4.In the present exemplary embodiment, the medium-voltage nodes MV1,...,MV4 form a ring system, wherein the ring is interrupted at at least one point during normal operation, resulting in a radial grid topology. 202307903 18 According to one embodiment of the present invention, the medium-voltage nodes MV1,...,MV4 are divided into four categories K1,...,K4. The medium-voltage node MV1 lies in the first category K1. Real-time measurements of its electrical power, i.e., its electrical active power and / or reactive power, are therefore available for this medium-voltage node MV1. In other words, a real-time measurement is installed at the medium-voltage node MV1. The other medium-voltage nodes MV2,...,MV4 do not have any such real-time measurements installed.Thus, the measurements of the subordinate low-voltage networks, for example, from smart meters, must be used to determine the aggregated electrical power (total power) at the respective medium-voltage nodes MV2,…, MV4. The set of medium-voltage nodes in the first category K1 can be denoted by MV. ୫^ୟ^Furthermore, the medium-voltage nodes MV2,…,M4 are divided into the three further categories K2,…,K4 (a total of four categories with the medium-voltage node MV1). The medium-voltage nodes MV2,…,M4 are assigned according to the smart meter coverage or the smart meter penetration rate of their respective subordinate low-voltage grid LV2,…LV4. The second category K2 includes all medium-voltage nodes that have a complete, i.e. 100 percent, smart meter penetration rate. This means that all households or low-voltage nodes in the low-voltage grid associated with the respective medium-voltage node have a smart meter, i.e., a smart meter measurement. In this case, the second category K2 therefore only includes the second medium-voltage node MV2.In Figure 1, the installation of a smart meter is indicated by a black circle at the respective household. The set of medium-voltage nodes of the second category K2 can be identified with SM. ^^^% The third category K3 includes all medium-voltage nodes that have sufficiently complete smart meter coverage, i.e. a smart meter coverage above or equal to a specified threshold α ^୦୰^^ , but less than 100 percent. In other words, in this case, the inequality α applies to the smart meter coverage or the smart meter penetration rate α ^୦୰^^ ^ α ^ 1. In this case, the third category K3 thus includes the third medium-voltage node MV3. The set of medium-voltage nodes of the third category K3 can be denoted by SM வ^౪^౨^౩ or SM ୫୭^^୪^The fourth category, K4, includes all medium-voltage nodes that do not have sufficient smart meter coverage or a sufficient smart meter penetration rate in their associated low-voltage grid. This is the case if the smart meter penetration rate α is less than the specified threshold α ^୦୰^^ In other words, in this case the inequality 0 ^ α ^ α applies to the smart meter coverage or the smart meter penetration rate α ^୦୰^^. In the present case, the fourth category K4 thus includes the fourth medium-voltage node MV4. The set of medium-voltage nodes in the third category K4 can be designated SMழ^౪^౨^౩ or SM୪^^^. Accordingly, all medium-voltage nodes MV1,…,MV4 can be classified according to the embodiment of the present invention according to the presence of real-time measurements and / or according to their smart meter penetration rate of their subordinate low-voltage grid LV2,…,LV4. Figure 2 shows a flowchart of a method according to an embodiment of the present invention. 202307903 20 The aim of the method presented is to generate input data for a state estimation of a power grid, which includes the covariances or correlations between the electrical power at the medium-voltage nodes as well as the expected values ​​of their electrical power.In other words, each medium-voltage node has a temporal profile of its active and / or reactive electrical power. These are essentially obtained by aggregating the respective subordinate low-voltage grids. Using these electrical powers, their covariance or correlation, as well as their temporal expected value, can be determined. The determined covariances / correlations and expected values ​​are included in the generated input data, which are then used for a state estimation, i.e., an estimate of the state of the power grid. The start of the process shown in Figure 2 is marked with the reference symbol S. The end of the process shown in Figure 2 is marked with the reference symbol E.In a first step S1 of the method, the classification of the medium-voltage nodes (MV nodes) already described in Figure 1 takes place. This means that the medium-voltage nodes of an existing power grid, whose status is to be estimated, are assigned to one of the four categories MV୫^ୟ^, SM^^^%, SM୫୭^^୪^ or SM୪^^^. For this purpose, additional recorded power measurements and / or energy consumption measurements, SLP classes of the low-voltage nodes and / or data on the grid topology can be used. As a result of the first step S1, each medium-voltage node is assigned to one of the four categories. In a second step S2 of the method, the time series, i.e. the time-dependent electrical power at each of the medium-voltage nodes, are determined. The determination of the electrical power differs depending on the category.For medium-voltage nodes in the first category, the available real-time measurements can be used directly. For medium-voltage nodes in the second category, the respective time series are created by aggregating the smart meter data, which in this case is fully available. For medium-voltage nodes in the third category, the respective time series are determined by scaling and aggregating the available smart meter data. It is therefore assumed that the smart meter data or the smart meter measurement data are representative, and the aggregated smart meter power is scaled up to the entire respective low-voltage grid. For medium-voltage nodes in the fourth category, the respective time series are generated using the available smart meter data and the use of standard load profiles. The result of the second step S2 is therefore the time series of the electrical power ^^. ^^,௧ , for all considered times and all considered medium-voltage nodes are determined. In the third step S3 of the procedure, the time series generated in this way can be used to determine ^^ ^,௧ the respective expected values, or the expected value if as a vector ^^ ௌ,^ୟ୫୮ summarized, by means of ^^ ^,௧ / ^^ can be determined directly. Regarding the determination of the covariance or covariances ^^ ௫^,௫^ A distinction is made as to whether the covariance contains a medium voltage node of the fourth category, i.e. category SM ழ^౪^౨^౩ or SM ୪^^^ is involved. 202307903 22 If no covariance is formed with a medium voltage node of the fourth category, the covariance is calculated over the generated time series according to ^^ ௫^,௫^ ൌ െ 1^. The totality of these covariances, which can be in the form of a matrix, is shown in the figure with ^^ ௌ^^,^ୟ୫୮If a medium-voltage node of the fourth category is involved, i.e., if the covariance of a medium-voltage node of any category with at least one medium-voltage node of the fourth category is to be determined, the covariance is not determined by the determined time series, but by a pre-provided covariance value Cov^ ^^, ^^^, which depends on the size, i.e., the number of households ^^, ^^ of the low-voltage networks involved. This covariance formed or generated in this way is shown in Figure 2 with ^^ ௌ^^,ୡ୭୫୮ Thus, after the third step S3 and its case differentiation regarding the categories, all covariances ^^ ௌ and expected values ​​^^ ௌThe covariances and expected values ​​determined in this way form the input data for a grid state estimate of the power grid under consideration. In other words, the input data, which includes the determined covariances and expected values, are used for a grid state estimate of the power grid under consideration. Since technically realistic covariances are determined through the classification and differentiation in step S3, the state estimate based thereon has significantly improved accuracy, i.e., significantly reduced uncertainty. Although the invention has been illustrated and described in detail by the preferred embodiments, the invention is not limited by the disclosed examples, and other variations can be derived therefrom by a person skilled in the art without departing from the scope of the invention.

[0002] 202307903 24 List of reference symbols 1 Power grid 10 High-voltage grid MV1,…,MV4 Medium-voltage nodes LV2,…,LV4 Low-voltage grids K1 first category K2 second category K3 third category K4 fourth category S1 first step S2 second step S3 third step S Start E Ende

Claims

202307903 25 claims 1. Method for generating input data for a state estimation of a power grid (1), wherein the power grid (1) has at least two medium-voltage nodes (MV2,…,MV4), and the input data comprise at least covariances and expected values ​​of electrical power at the medium-voltage nodes (MV2,…,MV4), wherein a low-voltage grid (LV2,…,LV4) with ^^ households is connected to one of the medium-voltage nodes (MV2,…,MV4) and a low-voltage grid (LV2,…,LV4) with ^^ households is connected to the other of the medium-voltage nodes (MV2,…,MV4), characterized in that at least one of the low-voltage grids (LV4) has a smart meter penetration rate less than a specified threshold, wherein the input data comprise at least a covariance between electrical power aggregated at the medium voltage nodes (MV2,…,MV4), wherein the covariance is determined by a provided covariance value Cov^ ^^,^^^ is determined depending on the number of households ^^, ^^ in the respective low-voltage networks.

2. Method according to claim 1, characterized in that the power grid (1) has more than two medium-voltage nodes (MV1,...,MV4), wherein a low-voltage network (LV2,...,LV4) is connected to one or more of the medium-voltage nodes (MV2...,MV4), and the medium-voltage nodes (MV1,...,MV4) are divided into four categories, wherein - the first category (K1) comprises medium-voltage nodes (MV1) whose aggregated electrical power has been measured; wherein - the second category (K2) comprises medium-voltage nodes (MV2) whose low-voltage networks (LV2) have a full smart meter penetration rate; wherein - the third category (K3) comprises medium-voltage nodes (MV3),whose low-voltage networks (LV3) do not have full smart meter penetration and a smart meter penetration rate greater than or equal to the specified threshold; and, 202307903 26 - the fourth category (K4) comprises medium-voltage nodes (MV4) whose low-voltage networks (LV4) have a smart meter penetration rate lower than the specified threshold; wherein - the covariance between a medium-voltage node (MV1,…,MV4) of any category (K1,…,K4) and at least one medium-voltage node (MV4) of the fourth category is determined by the provided covariance value Cov^ ^^, ^^^ depending on the number of households ^^, ^^ in the respective low-voltage networks (LV2,…,LV4).

3. The method according to claim 2, characterized in that for each medium-voltage node (MV1) of the first category (K1), its electrical power is determined by the recorded electrical power associated with the respective medium-voltage node (MV1). 4.Method according to claim 2 or 3, characterized in that for each medium-voltage node (MV2) of the second category (K2), its electrical power is determined by aggregating the smart meter measurement data associated with the respective medium-voltage node (MV2).

5. Method according to one of claims 2 to 4, characterized in that for each medium-voltage node (MV3) of the third category (K3), its electrical power is determined by a scaling factor and aggregating the smart meter measurement data associated with the respective medium-voltage node (MV3).

6. Method according to one of claims 2 to 5, characterized in that for each medium-voltage node (MV4) of the fourth category (K4), its electrical power is determined by aggregating the smart meter measurement data and standard load profiles associated with the respective medium-voltage node (MV4). 202307903 27 7. The method according to one of claims 2 to 6, characterized in that the covariances and expected values ​​of the input data are determined using the determined electrical power of the medium-voltage nodes (MV1,...,MV4), wherein the covariance between a medium-voltage node (MV1,...,MV4) of any category and at least one medium-voltage node (MV4) of the fourth category (K4) is determined by the provided covariance value Cov^ ^^, ^^^ depending on the number of households in the respective low-voltage networks (LV2,...,LV4).

8. The method according to one of the preceding claims, characterized in that the specified threshold value for the smart meter penetration rate has a value in the range from 0.5 to 0.

8. 9.Method according to one of the preceding claims, characterized in that the provided covariance value Cov^ ^^, ^^^ is determined by determining low-voltage networks comparable in terms of the number of households, wherein measurement data of the electrical power of their households is available for the low-voltage networks used here.

10. Method according to one of the preceding claims, characterized in that electrical active power and / or reactive power are used as electrical power.

11. Method according to one of the preceding claims, characterized in that the input data comprise correlations between the medium-voltage nodes (MV1, ..., MV4), wherein the respective correlations are formed using the respective covariances. 12.Method for estimating the state of a power grid (1) with several medium-voltage nodes (MV1,…,MV4), wherein input data which are expected values ​​and covariances of. 202307903 28 electrical power at the medium-voltage nodes (MV1,...,MV4) are used, characterized in that the input data is generated by a method according to one of the preceding claims.

13. Method according to claim 12, characterized in that the state estimation is carried out by means of a Bayesian state estimation or a WLS state estimation.

14. Method for controlling a power grid (1) by means of a control unit, wherein the power grid (1) comprises a plurality of medium-voltage nodes (MV1,...,MV4), and the control or regulation is carried out based on a state estimation, characterized in that the state estimation is carried out by a method according to claim 12 or 13. 15.Computer program product, characterized in that it comprises instructions which, when the program is executed by a computing unit, in particular a computer, cause the latter to carry out a method and / or steps of the method according to one of claims 1 to 14.

Images