METHOD AND ASSOCIATED DEVICE FOR ASSIGNING EACH COUNTER OF A MULTINUMBER OF COUNTERS TO A THREE-PHASE OUTPUT OF A TRANSFORMER.

Description

TECHNICAL FIELD OF THE INVENTION

[0001] The technical field of the invention is that of electrical network mapping. The present invention relates to a method for associating each meter of a plurality of meters with a three-phase feeder of a transformer, and an associated device. TECHNOLOGICAL BACKGROUND OF THE INVENTION

[0002] It often happens that the relationship between meters and transformers is only partially known or still to be determined. This step is particularly important in order to estimate the load on each transformer, on each of the three-phase feeders of each transformer and on each phase of that feeder, or to know which meters need to have their phase changed to rebalance the network.

[0003] To perform such an association, an initial method based on the conservation of energy was proposed. More specifically, this method uses the principle of conservation of energy to determine which transformers supply which meters: the sum of the energy consumed by the meters must be found at each transformer. However, this method has drawbacks. First, it only works if almost all the power consumed is measured and these measurements are recorded to be taken into account by the method. It is therefore not very robust to measurement errors, synchronization errors, "lost" data, and losses (technical – due to cable heating – and non-technical – due to power theft). Furthermore, since the calculation is combinatorial, the number of meters to be sorted must be limited (generally a few hundred).Furthermore, it requires measurements to be taken at each meter, as well as at each three-phase feeder of each transformer, which may necessitate the installation of a temporary system to enable this measurement. Finally, it does not allow for the determination of the network topology, that is, how the meters in the network are connected to one another. See also document CA 2 915 674 A1.

[0004] There is therefore a need for a method to map the electrical network, that is to say to associate each meter with a phase, a three-phase feeder and a transformer without requiring a measurement feedback from all or most of the meters and without requiring measurements at the level of each transformer, while allowing the processing of a network with a large number of meters. SUMMARY OF THE INVENTION

[0005] The invention offers a solution to the problems mentioned above, by using the connection between each phase of a three-phase feeder to determine the network topology without having to perform measurements at the transformer level, but only at the meters to be characterized. Furthermore, the principles of this measurement make it relatively insensitive to measurement anomalies such as the omission of one or more meters or the existence of unauthorized sampling. A first aspect of the invention relates to an association method according to claim 1.

[0006] Thanks to the invention, it is possible to associate each meter with a group and therefore with a three-phase feeder. Since this three-phase feeder is itself associated with a transformer, the method, according to a first aspect of the invention, makes it possible to associate each meter in a plurality of meters with a three-phase feeder of a transformer. The inventors discovered that it was possible to utilize the relationship that exists between the different phases of a three-phase feeder. This relationship is reflected, for example, by the fact that a change in consumption on one phase will impact the voltage of the other two phases of the three-phase feeder. The measurement of this relationship is carried out in the invention using voltage measurements taken by meters.Indeed, generally speaking, if two voltages measured by two different meters show a negative similarity, for example if they are negatively correlated, then these meters are in principle connected to two different phases of the same three-phase feeder.

[0007] Similarity can be measured, for example, using Pearson correlation or by calculating a regression coefficient (this aspect will be detailed later). Similarly, the voltages measured by two meters connected to two phases of the same three-phase circuit do not necessarily exhibit negative similarity. However, in the case of a significant imbalance, this similarity becomes negative. In this situation, the voltages measured by three meters, each connected to one phase of a three-phase circuit, will show negative similarity in pairs.

[0008] In addition to the characteristics mentioned in the preceding paragraph, the process according to a first aspect of the invention may have one or more complementary characteristics from among the following, considered individually or according to all technically possible combinations.

[0009] Advantageously, the method according to a first aspect of the invention further includes, before the step of determining the similarity between each voltage measurement of each meter of the plurality of meters, a pre-processing step of the voltage measurement carried out by each meter over the measurement period.

[0010] Thus, the operation of the process is improved by ensuring that the data is purged of all or part of the measurements that could impair the relevance of the results obtained, for example by removing the component due to voltage variations at the transformer level.

[0011] Advantageously, the step of grouping pairs of meters is only carried out on pairs of meters with a similarity of less than -0.2.

[0012] Thus, when the similarity measured between two counters is too high, the process does not take into account the couple formed by these two counters, which improves the results obtained by the process according to a first aspect of the invention.

[0013] Advantageously, the grouping step includes, for each pair of meters: when the two meters of said pair of meters do not belong to any group, a sub-step of creating a group and associating the two meters with said group; when one of the two meters of said pair of meters is already associated with a group, a sub-step of associating the meter not yet associated with a group with said group; when the two meters of said pair of meters are already associated with two different groups, a sub-step of merging said two groups.

[0014] Thus, the grouping allows each group to be associated with a three-phase start of the network and therefore with a transformer of the network.

[0015] Advantageously, the grouping step includes a first sub-step of ranking the pairs of similarity counters from the most negative to the least negative and, for each pair of counters and in the order of their ranking: when the two meters of said pair of meters do not belong to any group, a substep of creating a group and associating the two meters with said group; when one of the two meters of said pair of meters is already associated with a group, a substep of associating the meter not yet associated with a group with said group; when the two meters of said pair of meters are already associated with two different groups, a substep of merging said two groups if there is at least one other pair of meters exhibiting a negative similarity and whose two meters are associated with the same two groups.

[0016] The ranking substep at the beginning of the grouping step improves the reliability of the grouping. Furthermore, only merging two groups if there are two links between them further enhances the reliability of the grouping.

[0017] Advantageously, the distance between two meters is equal to the absolute value of the similarity between the two meters when said similarity is negative and to zero when said similarity is positive, and, for each group, the process includes a step of creating three subgroups, each subgroup being constructed so that the average distance between two meters of different subgroups is as large as possible and the average distance between two meters belonging to the same subgroup is as small as possible, preferably zero, each subgroup corresponding to a phase of the three-phase starter associated with the group considered.

[0018] Thus, it is possible to associate each meter with a phase of a three-phase feeder.

[0019] Advantageously, alternatively, the distance between two meters is equal to the absolute value of the similarity between the two meters when said similarity is negative and to zero when said similarity is positive, and wherein, for each group, the method includes a step of implementing a clustering algorithm so as to obtain three clusters of meters, each cluster corresponding to a phase of the three-phase starter associated with the group considered.

[0020] Thus, as before, it is possible to associate each meter with a phase of a three-phase feeder.

[0021] Advantageously, and alternatively, for each group, the process includes: for each counter in the group, a step of determining the two other counters in the group with which the counter in question presents the most negative similarities, the three counters thus determined forming a triplet associated with a score so as to obtain a plurality of triplets, the counters in a triplet always presenting a negative similarity two by two; a step of determining the triplet presenting the highest score, called the reference triplet; a step of associating a subgroup with each counter in the reference triplet so as to obtain three subgroups; Then, for each counter in the group not belonging to the reference triplet, the process includes: when the counter exhibits a negative similarity with two and only two of the three counters in the reference triplet, a step of associating said counter with the subgroup associated with the counter in the reference triplet with which the similarity is positive or zero; otherwise, when the counter has a negative similarity with at least two other counters associated with two of the three subgroups: ▪ for each subgroup among the three subgroups, a substep of determining the counter associated with said subgroup with which the counter in question has the weakest similarity, called intermediate counters, so as to obtain one intermediate counter per subgroup; ▪ if the counter in question exhibits a negative similarity with each intermediate counter, then a substep of classifying said counter as being difficult to classify, the latter then being not associated with any subgroup;▪ otherwise, a sub-step of associating the meter in question with the subgroup associated with the intermediate meter with which the meter in question has a positive or zero similarity; otherwise, a step of classifying said meter as being difficult to classify, the latter then not being associated with any subgroup. ;

[0022] Thus, as before, it is possible to associate each meter with a subgroup and therefore with a phase of a three-phase feeder, each phase being itself associated with a subgroup.

[0023] A second, unclaimed aspect of the invention relates to a method for determining the topology of a plurality of meters on a three-phase feeder, said three-phase feeder comprising three phases, each meter in the plurality of meters being connected to one and only one phase of the three-phase feeder. The method according to a second aspect of the invention comprises: For each counter in the plurality of counters, a step of determining the two other counters in the plurality of counters with which the considered counter has the most negative similarities, the three counters thus determined forming a triplet associated with a score so as to obtain a plurality of triplets, the counters in a triplet always having a negative similarity in pairs; a step of determining a distance between each of the triplets in the plurality of triplets, said distance between two triplets being a function of the similarities between the counters in each of the two triplets; a step of determining the triplet with the best score, called the reference triplet; a step of connecting the triplet closest to the reference triplet to said reference triplet so as to form two connected triplets;then, a step of determining, among the triplets not yet connected, the triplet closest to a connected triplet and a step of connecting said triplet not yet connected to the connected triplet to which it is closest, these two steps being repeated until all the triplets are connected.

[0024] Thus, it is possible to determine the topology of a given three-phase circuit, that is, how the meters are connected to each other and to the phases of said three-phase circuit. A connected triplet refers to a circuit where each phase of the first connected triplet is associated with a phase of the second connected triplet.

[0025] A third, unclaimed aspect of the invention relates to a method for determining the topology of a network comprising a plurality of meters and at least one transformer, each transformer comprising at least one three-phase feeder, each three-phase feeder comprising three phases, and each meter being connected to one and only one phase. The method according to a third aspect of the invention comprises: For each counter in the plurality of counters, a step of determining the two other counters in the plurality of counters with which the considered counter has the most negative similarities, the three counters thus determined forming a triplet associated with a score so as to obtain a plurality of triplets, the counters in a triplet always having a negative similarity pairwise; a step of determining a distance separating the triplets pairwise, the distance between two triplets being a function of the similarities between the counters of each of the two triplets; a step of connecting the triplets comprising: ▪ a sub-step of determining the two triplets closest to each other and not yet connected to each other; ▪ a sub-step of connecting the two triplets thus determined a set of connected triplets forming a group, the substeps of the triplet connection step being repeated until a number of groups equals a predetermined value is obtained.

[0026] Thus, a method according to the invention makes it possible, using the same principles as the method according to a first aspect of the invention, to achieve a complete network topology from the implementation of triplets of counters.

[0027] Advantageously, the method according to a third aspect of the invention includes, after the step of determining the two other counters from the plurality of counters with which the counter in question has the most negative similarities, a step of sorting the triplets obtained so that triplets with a score below a predetermined threshold value are not taken into account.

[0028] Thus, this sorting prevents triplets with a low score from leading to the merging of two groups that should not have been merged.

[0029] A fourth aspect of the invention relates to an electronic device according to claim 9.

[0030] A fifth unclaimed aspect of the invention relates to a computer program comprising instructions which lead the device according to a fourth aspect of the invention to perform the steps of the process according to a first, second or third aspect of the invention.

[0031] A sixth aspect of the unclaimed invention relates to a computer-readable medium on which the computer program is recorded according to a fifth aspect of the invention.

[0032] The invention and its various applications will be better understood by reading the following description and examining the accompanying figures. BRIEF DESCRIPTION OF THE FIGURES

[0033] The figures are presented for illustrative purposes only and are in no way limiting to the invention. There figure 1 shows a schematic representation of an electrical network to which a process can be applied according to a first, second, or third aspect of the invention. figure 2 shows a flowchart of one embodiment of a process according to a first aspect of the invention. figure 3 shows a schematic representation of the result obtained with a process according to a first aspect of the invention. figure 4 shows a flowchart of an embodiment of a process according to a second, unclaimed aspect of the invention. figure 5 shows a schematic representation of part of the steps of a process according to a second, unclaimed aspect of the invention. figure 6 shows a flowchart of an embodiment of a process according to a third, unclaimed aspect of the invention. figure 7shows a schematic representation of part of the steps of a process according to a third, unclaimed aspect of the invention. figure 8 shows a schematic representation of an embodiment of a device according to a fourth aspect of the invention. DETAILED DESCRIPTION OF AT LEAST ONE EMBODIMENT OF THE INVENTION

[0034] Unless otherwise specified, the same element appearing on different figures has a unique reference.

[0035] A first embodiment of a first aspect of the invention illustrated in figures 1 to 3 relates to a method 100 of associating each CPT meter of a plurality of CPT meters with a three-phase DTP feeder of a TRS transformer from among at least one TRS transformer, preferably from among a plurality of TRS transformers, each TRS transformer comprising at least one three-phase DTP feeder, each three-phase DTP feeder comprising three DDP phases, and each CPT meter being connected to one and only one DDP phase.

[0036] The process 100 includes a step 1ED of determining the similarity between each voltage measurement from each CPT meter in the plurality of CPT meters, thus forming a plurality of pairs of CPT meters. Each voltage measurement is taken over a period called the measurement period, and each pair of CPT meters is associated with a similarity. The similarity can, for example, be a correlation such as a Pearson correlation.

[0037] In one embodiment, similarity is determined by calculating regression coefficients. For this, for each CPT counter N, a step is implemented to determine the voltage VN of the CPT counter N from a linear combination of the voltages and / or powers of the other CPT counters. The coefficient associated with each counter then represents the similarity between the CPT counter N and the counter associated with that coefficient. Thus, consider a CPTi counter associated with the voltage V iwe determine the coefficients K j associated with the counter j ≠ i, such as: V i = ∑ j K j × V j

[0038] The similarity between CPT counter i and CPT counter j is given by the value of K j .

[0039] In one embodiment, the similarity is achieved in two steps: First, a correlation is calculated, for example a Pearson correlation between each counter in the plurality of counters; second, the counters are grouped by positive correlation into a set, this grouping being carried out by a clustering method, for example a so-called "affinity propagation"; Thirdly, each CPT i counter being associated with the voltage V i we determine the coefficients K j associated with the counter j ≠ i such as: V i = ∑ j K j × V j K jbeing imposed as equal to zero during regression when the CPT i and CPT j counters belong to the same set.

[0040] The similarity between CPT counter i and CPT counter j is given by the value of K j In other words, the similarity is zero when two meters belong to the same set and has a value determined by linear regression otherwise. The advantage of this definition of similarity is that it allows us to obtain, at the end of step 1ER of grouping pairs of CPT meters exhibiting negative similarity based on their similarity, not only a grouping by three-phase output as with the other definitions of similarity presented above, but also a grouping by phase output.

[0041] In one embodiment, the similarity between two counters is obtained by calculating the partial correlation matrix, which can be obtained, for example, by inverting the covariance matrix. This method of calculating similarity has the advantage of retaining only the significant links between counters: each counter is linked only to those that describe its variations. When two counters are unrelated (a link equal to zero in the matrix), their similarity is zero. When two counters are related, their similarity is equal to that link. In one example, this calculation is performed by forcing the matrix to be sparse (i.e., by increasing the number of zeros during matrix estimation), thereby decreasing the number of links between counters (for example, using a GraphicalLasso estimator), and thus obtaining more similarities equal to zero.

[0042] More generally, a similarity can be defined as the degree of resemblance between two signals, which can be calculated solely from those two signals (e.g., correlation or covariance), or by also considering other signals (e.g., regression coefficient or partial correlation). Following this step, each possible pair of CPT counters within the plurality of CPT counters is associated with a similarity. This similarity is calculated between the voltages measured at each CPT counter within that pair. A matrix of all the similarities between the counters, paired two at a time, is thus obtained. In other words, each CPT counter measures the voltage at regular intervals and records the time evolution of this voltage, and this measurement is used to calculate the similarity.In the following, the voltage change over the measurement period at a CPT counter will be referred to as the CPT counter's voltage measurement. Thus, a voltage measurement comprises multiple points, each point corresponding to a voltage value at a given instant. Similarly, the similarity between two CPT counters should be understood as the calculated similarity between the voltage measured at said CPT counters over a given period, called the measurement period. For this reason, in the following, the similarity between the voltage measurement taken by a first CPT counter and the voltage measurement taken by a second CPT counter will also be referred to as the similarity between the first and second CPT counters for the sake of brevity.

[0043] Process 100 also includes a 1ER step for grouping pairs of CPT meters with negative similarity to form a plurality of groups, each group being associated with a three-phase DTP feeder. In other words, only pairs with a similarity less than or equal to zero, preferably less than -0.2, are considered for the 1ER grouping step. This avoids considering pairs that are not very discriminating for determining the topology. Indeed, it is "easy" to find positive similarities, even on different feeders (because consumption patterns can be similar), whereas it is rarer to find negative similarities between two feeders. As detailed below, this grouping step can be carried out in several ways.

[0044] In one embodiment, when the number of three-phase feeders is known, grouping can be performed using a clustering method, for example, a k-means method. When the number of three-phase feeders is unknown, grouping can be performed using a method that does not require knowing the number of groups in advance, such as a so-called " affinity propagationThis last step allows us to determine the optimal number of groups. In other words, the grouping step does not necessarily require prior knowledge of the number of three-phase feeders on the network. It should be noted that, generally, when trying to group meters with negative similarities, a clustering algorithm can be used where the distance is the modified similarity value, retaining only the absolute values ​​of the negative values ​​and discarding the positive values. Similarly, when trying to group meters with positive similarities, only the positive values ​​can be retained.

[0045] In one embodiment, step 1 of the grouping process includes, for each pair of meters: when the two meters of said pair of meters do not belong to any group, a sub-step of creating a group and associating the two meters with said group; when one of the two meters of said pair of meters is already associated with a group, a sub-step of associating the meter not yet associated with a group with said group; when the two meters of said pair of meters are already associated with two different groups, a sub-step of merging said two groups.

[0046] Thus, it is possible to associate each CPT meter in a group with a DTP three-phase feeder, and consequently with the TRS transformer containing said DTP three-phase feeder. However, although simple to implement, this grouping method can in some cases lead to assigning a CPT meter to the wrong group and therefore to the wrong DTP three-phase feeder.

[0047] The results obtained with the method described above can be improved by ranking the CPT counters. More specifically, in an alternative embodiment, the grouping step 1ER includes a first substep of ranking the pairs of CPT counters from the most negative similarity to the least negative. It also includes, for each pair of CPT counters and in the order of their ranking: when the two CPT counters of said pair do not belong to any group, a sub-step of creating a group and a sub-step of associating the two CPT counters with said group; when one of the two CPT counters of said pair is already associated with a group, a sub-step of associating the CPT counter not yet associated with a group with said group;when the two CPT meters of said pair are already associated with two different groups, a merging step of said two groups if there exists at least one other pair of CPT meters exhibiting a negative similarity and for which the two CPT meters of said pair are associated with the same two groups (i.e., if the first CPT meter of the pair considered belongs to a first group and the second CPT meter of the pair considered belongs to a second group, the first and second groups will only be merged if there exists a second pair of meters exhibiting a negative similarity and one meter of which belongs to the first group and another meter belongs to the second group).

[0048] Thus, it is possible to associate each CPT meter of a group with a DTP three-phase feeder, and consequently with the TRS transformer including said DTP three-phase feeder.

[0049] In one embodiment, the method 100 according to a first aspect of the invention includes, before step 1 ED of determining the similarity between each voltage measurement of each CPT counter of the plurality of CPT counters, a step 1PT of preprocessing the voltage measurement carried out by each CPT counter over the measurement period.

[0050] In one embodiment, the preprocessing step 1PT includes a frequency filtering substep for the voltage measurements. For example, the filter might be a bandpass filter that retains only a given frequency range. In another example, the filter might be a high-pass filter that retains only the high frequencies, thus eliminating daily variations. A Butterworth filter could be used, for instance.

[0051] In one embodiment, the preprocessing step 1PT includes a substep of suppressing voltage measurements when they are zero during a given time interval, for example over a day.

[0052] In one embodiment, the preprocessing step 1PT includes a substep for removing points in measurements that deviate abnormally from other points in the same measurement, called outliers (or outliers (in English), for example, points that represent a value that deviates from the mean by more than 3 standard deviations. When several preprocessing substeps are performed, the outlier removal substep is preferably performed first.

[0053] When data is deleted during preprocessing, the similarity calculation can be done on the remaining data (and therefore on a modified measurement period corresponding to said remaining data) or the deleted data can be substituted with new data obtained by interpolation as will be described later.

[0054] In one embodiment, the preprocessing step 1PT includes a substep for synchronizing measurements, for example using a cross-correlation method. Such a method is known to those skilled in the art and will therefore not be described here.

[0055] In one embodiment, the preprocessing step 1PT includes an interpolation substep such that the number of points per measurement is identical for all measurements of the reference electrical quantity, for example, by linear interpolation. This interpolation also allows for the completion of any missing data.

[0056] In one embodiment, the preprocessing step 1PT includes a substep for determining a sub-period, referred to as the enhanced sub-period. This enhanced sub-period comprises one or more parts of the measurement period, and the similarity calculation between the different meters is then performed on this enhanced sub-period. This enhanced sub-period can, for example, correspond to a period during which the network is more unbalanced. An unbalanced network is defined as a situation where each phase of a three-phase feeder consumes different amounts of current and a significant neutral current is generated. An unbalanced period can be chosen as the period during which electricity consumption is higher than average (and therefore more likely to result in a significant imbalance), for example, during the evening peak or at midday when solar panels are present.It is also possible to determine this period of imbalance by measuring all the tensions at the same time and choosing the moments when the standard deviation is the greatest.

[0057] In one embodiment, the 1PT preprocessing step includes a substep for differentiating the signal measured by each counter; the derivative can be first-order or higher-order. This allows, in particular, the measurement of similarities in voltage variations rather than the voltage itself.

[0058] In one embodiment, the pretreatment step 1PT includes a substep for removing the component due to variations at the level of the TRS transformer(s). For example, the substep for removing the component due to variations at the level of the TRS transformer(s) can be performed by subtracting the voltage measured at the transformer. This component can be measured by a concentrator located at each TRS transformer. Note that for a given TRS transformer, the voltage is the same for all three-phase DTP feeders of the TRS transformer in question, but may be different for each of the DDP phases (i.e., the first DDP phases of the three-phase DTP feeders will have the same voltage, the second DDP phases of the three-phase DTP feeders will have the same voltage, and the third DDP phases of the three-phase DTP feeders will have the same voltage).The voltage of a TRS transformer can, for example, be approximated as being equal to the average voltage across the three phases of each three-phase DTP feeder (or of a reference three-phase DTP feeder), then subtracted from the voltage measured on each CPT meter connected to the TRS transformer in question. However, this assumes that each CPT meter has already been associated with a TRS transformer (which is the case when considering only one TRS transformer), which is not always possible.

[0059] In another example, when the voltage cannot be measured at each TRS transformer, the substep of removing the component due to variations at the TRS transformer(s) can be performed by averaging all the voltages measured at the CPT meters to find their common base, and then subtracting the component parallel to this common base from each voltage. In other words, the voltage at a TRS transformer is approximated by the average of the voltages measured at the CPT meters connected to that TRS transformer, and then this voltage is subtracted from the voltage measured at each of the CPT meters connected to that TRS transformer. In another example, the common base subtraction can also be performed for each voltage by subtracting the projection of that voltage onto that common base (using, for example, the dot product).In yet another example, the common base is directly subtracted from each tension.

[0060] In another example, the substep of removing the component due to variations in the TRS transformer(s) can be achieved by performing a principal component analysis or an independent component analysis and removing the first vector(s). In other words, a principal component analysis is performed on all voltages of a TRS transformer to obtain the principal axes, with the first axes being the most important. If the influence of the TRS transformer is significant, and since it impacts all the voltages measured at the CPT meters, then this voltage should be found in the first axis.

[0061] Once each CPT meter is associated with a DTP three-phase feeder, it may be useful to determine the association between each CPT meter and each DDP phase, this determination being done by DTP three-phase feeder.

[0062] In one embodiment, the distance between two CPT meters is defined as equal to the absolute value of the similarity between the two CPT meters when said similarity is negative, and to zero when said similarity is positive. Furthermore, for each group (i.e., the CPT meters associated with a given three-phase DTP feeder), the method 100 includes a step of creating three subgroups. These subgroups are constructed such that the average distance between two CPT meters from different subgroups is as large as possible, and the average distance between two CPT meters belonging to the same subgroup is as small as possible. Each subgroup corresponds to a phase of the three-phase DTP feeder associated with the group under consideration. Thus, depending on a meter's membership in a given subgroup, it is possible to associate said meter with a phase of a three-phase DTP feeder of a TRS transformer.Alternatively, the grouping of the counters into three subgroups can be done using a clustering method.

[0063] In an alternative embodiment, the distance between two CPT meters is defined as equal to the absolute value of the similarity between the two CPT meters considered when said similarity is negative, and to zero when said similarity is positive. Furthermore, method 100 includes, for each group (i.e., the CPT meters associated with a given three-phase DTP feeder), a step of implementing a clustering algorithm to obtain three clusters of meters, each cluster corresponding to a phase DDP of the three-phase DTP feeder associated with the group considered. The clustering algorithm can, for example, be the K-means algorithm or a self-organizing map type (e.g., Kohonen maps).

[0064] In an alternative embodiment, for each group (i.e., the CPT meters associated with a given three-phase DTP feeder), process 100 includes, for each CPT meter in the group, a step of determining the two other CPT meters in the group with which the CPT meter in question shares the most negative similarities. The three meters thus determined form a triplet associated with a score, as the meters in a triplet always share a negative similarity in pairs. Thus, a plurality of triplets is obtained, each triplet being associated with a score. It is important to note that a meter can belong to several triplets. The score can, for example, be equal to the absolute value of the average similarity between each CPT meter in the considered triplet. Process 100 then includes a step of determining the triplet with the highest score, called the reference triplet.

[0065] Process 100 also includes a step of associating a subgroup with each CPT meter in the reference triplet. This association will allow the other CPT meters in the group (those not belonging to the reference triplet) to be linked to each subgroup and therefore to a phase of the three-phase DTP feeder. This association will be based on the similarities between the CPT meters in the reference triplet and the CPT meters to be linked. To this end, two situations must be distinguished. When the CPT meter to be linked has a negative similarity with two and only two of the three CPT meters in the reference triplet, then said CPT meter is linked to the CPT-associated subgroup in the reference triplet with which it has the highest similarity.

[0066] If this first condition is not met, when the counter has a negative correlation with at least two other CPT counters associated with two of the three subgroups, the process includes, for each of the three subgroups, a substep to determine the CPT counter associated with that subgroup with which the considered CPT counter has the weakest similarity. This is called the intermediate CPT counter, so as to obtain one intermediate CPT counter per subgroup. Indeed, due to the previous iterations, other CPT counters besides those in the reference triplet are already associated with one of the three subgroups. Furthermore, it is possible to determine, among these CPT counters, the one that exhibits the weakest similarity with the CPT counter that one seeks to associate; the CPT counter thus determined becomes the intermediate CPT counter for the subgroup to which it is associated.At the end of this substep, three intermediate CPT meters are available, each associated with one of the three subgroups. The similarity between each of these intermediate CPT meters and the CPT meter to be associated is also known. The method then includes, if the meter in question exhibits a negative similarity with each intermediate CPT meter, a substep classifying said meter as difficult to classify, in which case it is not associated with any subgroup. Indeed, if the meter in question exhibits a negative similarity with the three intermediate meters, it is preferable to set it aside, for example, by using another method to assign it a phase with a three-phase distribution. Such an assignment will be facilitated by the fact that, thanks to the method according to the present invention, the set-aside meter is associated with a group and therefore with a three-phase feeder.If the previous condition is not met, i.e. that one and only one of the three reference CPT counters has a positive or zero similarity with the counter under consideration, the process includes a substep of associating the CPT counter under consideration with the subgroup associated with the intermediate CPT counter with which the CPT counter under consideration has a positive or zero similarity.

[0067] Finally, if none of the conditions stated above are met, the process includes a step of classifying the CPT counter as difficult to classify, meaning it is not associated with any subgroup. The unassociated CPT counter may be associated later using another technique.

[0068] As illustrated in the figure 3The method according to a first aspect of the invention allows each CPT meter to be associated with a three-phase DTP feeder (or even with a DDP phase of a three-phase DTP feeder). Once such an association is known, it may be useful to know the topology of the CPT meters at the level of each three-phase DPT feeder, in other words, how the different CPT meters are connected to each other.

[0069] To this end, an embodiment of a second, unclaimed aspect of the invention is illustrated in the figure 4 and to the figure 5This concerns a method 200 for determining the topology of a plurality of CPT meters on a three-phase DTP feeder. It is worth noting that if several three-phase DTP feeders are present, this method can be implemented for each of them to obtain the topology of a portion of the network or the entire network. As previously detailed, the three-phase DPT feeder comprises three phases, with each CPT meter in the plurality of CPT meters connected to one and only one phase of the three-phase CPT meter. The method 200 first comprises, for each CPT meter in the plurality of meters, a step 2ED1 for determining the two other CPT meters in the plurality of CPT meters with which the CPT meter in question shares the most negative similarities. The three CPT meters thus determined form a triplet, each associated with a score.As already mentioned, the score can, for example, be equal to the absolute value of the average similarity between each CPT meter in the considered triplet. As a reminder, for three CPT meters to form a triplet, they must have a negative similarity between pairs. At the end of this step 2ED1, a plurality of triplets is obtained, each associated with a score, and all that remains is to connect them to determine the topology at the three-phase DTP feeder.

[0070] To this end, process 200 includes a step 2ED2 of determining the distance between each of the previously obtained triples, this distance between two triples being a function of the similarities between the counters of each of the two triples. For example, to find the distance, we begin by pairing the CPT counters of the two triples two by two according to their DDP phase. For example, by finding the three pairs of CPT counters that are positively correlated and that maximize the quadratic sum of the three correlations. Once the three pairs are found, the distance can be obtained from these three correlations. For example, the distance is determined by calculating the absolute value of the mean of the three correlations corresponding to the three pairs formed minus 1 (in other words: abs ( average ( correlations between counters of each pair) - 1))

[0071] For example, in the case of a triplet (a, b, c) and a triplet (d, e, f), the CPT counters are first matched by finding the three pairs of CPT counters that are positively correlated and maximize the squared sum of the three correlations. In this example, the three pairs thus obtained are (a, e), (b, d), and (c, f). Once the three pairs are found, the distance can be obtained from these three correlations. For example, the distance between these two triplets is then given by the calculation detailed previously. It is important to note that when the same CPT counter belongs to both triplets, the correlation calculated between that CPT counter and itself is equal to 1.In the examples shown above, the similarity between two CPT counters is obtained using a correlation, but other methods for calculating the distance between two triplets can be considered depending on the method used to determine the similarity between two CPT counters.

[0072] This step is followed by a 2ED3 step to determine the triplet with the best score, called the reference triplet ( figure 5A ). This reference triplet will serve as the starting point for setting up the topology. The process 200 then includes a step 2EC1 of connecting the closest triplet (for example, according to the previously defined distance) to the reference triplet to said reference triplet so as to form two connected triplets ( figure 5B). Then, for each triplet not yet connected, the process includes a step 2ED4 of determining the triplet closest to a connected triplet and a step 2EC2 of connecting said triplet not yet connected to the connected triplet of which it is closest ( figure 5CThese two steps, 2ED4 and 2EC2, are repeated until all triplets are connected. At the end of the process, the topology of the CPT meters at the three-phase DTP feeder is obtained. A connected triplet means that each phase of the first connected triplet is associated with a phase of the second connected triplet. In other words, when a first triplet is connected to a second triplet, each CPT meter of the first triplet is connected to a CPT meter of the second triplet (they are therefore connected to the same phase). As shown above, each CPT meter of the first triplet is paired with a CPT meter of the second triplet when determining the distance between the first and second triplets. Furthermore, as already mentioned, a meter can belong to several triplets. In this case, as illustrated in figure 5Don which each counter is represented by a node, when connecting two triplets having a common CPT counter, the two nodes corresponding to this CPT counter are merged to form only one node, the connections between this node and the node of the previous or next triplet being adapted accordingly.

[0073] Advantageously, step 2ED1 of determining the two other CPT counters of the plurality of CPT counters with which the CPT counter in question has the most negative similarities is preceded by a pretreatment step as described in the process according to a first aspect of the invention.

[0074] In the method 200 for determining the topology of a plurality of CPT meters on a three-phase DTP feeder described above, in order to obtain the network topology, it is necessary first to associate each CPT meter with a three-phase DTP feeder (for example, by implementing a method 100 according to a first aspect of the invention) and then to determine the topology, three-phase DTP feeder by three-phase DTP feeder, using a method 200 according to a second aspect of the invention. Although such a solution is entirely feasible and even sometimes desirable (for example, when only the topology of one or more particular three-phase DTP feeders is important), it may be advantageous to have a method that allows each CPT meter to be associated with a three-phase DTP feeder while also allowing the topology to be determined at the level of each three-phase DTP feeder, even in the case of a plurality of three-phase DTP feeders.

[0075] To this end, an embodiment of a third, unclaimed aspect of the invention is illustrated in the figure 6 and to the figure 7 relates to a method 300 for determining the topology of a network. The network comprises a plurality of CPT meters and at least one TRS transformer, preferably a plurality of TRS transformers, each TRS transformer comprising at least one three-phase DTP feeder, each three-phase DTP feeder comprising three DDP phases, and each CPT meter being connected to one and only one DDP phase.

[0076] The process 300 includes, for each CPT counter in the plurality of CPT counters, a step 3ED1 of determining the two other CPT counters in the plurality of CPT counters with which the considered CPT counter has the most negative similarity. The three CPT counters thus determined form a triplet associated with a score. As already mentioned, the score can, for example, be equal to the absolute value of the average similarity between each CPT counter in the considered triplet. As previously stated, for three CPT counters to form a triplet, said CPT counters must have a pairwise negative similarity.

[0077] In one embodiment, process 300 then includes a 3ET step for sorting the resulting triplets so that triplets with a score below a predetermined threshold value (for example, a threshold equal to the mean value of the scores minus the standard deviation of said scores) are not included. This sorting step, although optional, prevents triplets with low scores from leading to the merging of two groups that should not have been merged.

[0078] The process then includes a 3ED2 step to determine the distance separating the triplets two by two. Finally, as illustrated in figures 7A, 7B and 7CThe 300 process includes a step 3EC for connecting triplets, comprising a substep for determining the two closest triplets not yet connected, and a substep for connecting these two triplets. Thus, a set of connected triplets forms a group, and the substeps of the triplet connection step are repeated until a predetermined number of groups is obtained. This predetermined value is, for example, the number of three-phase DTP feeders in the network, or the number of TRS transformers in the network if the feeders of each transformer are to be grouped together. Therefore, at the end of the process, a plurality of triplet groups are obtained, each group corresponding to a three-phase DTP feeder or a TRS transformer.Furthermore, the connections of the triplets to each other within a given group make it possible to know the topology at the level of the three-phase DTP starter or the TRS transformer corresponding to said group.

[0079] The method of connecting the triplets to each other is identical to the connection method used in the implementation of a method 200 according to a second aspect of the invention. In other words, when a first triplet is connected to a second triplet, each CPT counter of the first triplet is connected to a CPT counter of the second triplet (they are therefore connected to the same phase). Moreover, as already mentioned, a counter can belong to several triplets. In this case, as illustrated in figure 5Don which each counter is represented by a node, when connecting two triplets having a common CPT counter, the two nodes corresponding to this CPT counter are merged to form only one node, the connections between this node and the node of the previous or next triplet being adapted accordingly.

[0080] Advantageously, step 3ED1 of determining the two other CPT counters of the plurality of CPT counters with which the CPT counter in question has the most negative similarity is preceded by a pretreatment step as described in the framework of process 100 according to a first aspect of the invention.

[0081] A fourth aspect of the invention illustrated in figure 8relates to a DI device enabling the implementation of a process according to a first aspect, a second (unclaimed) aspect, or a third (unclaimed) aspect of the invention. As described previously, in an electrical distribution network, each TRS transformer comprises at least one three-phase DTP feeder, and each three-phase DTP feeder comprises three phase DDP feeders. In one embodiment, the device includes an MV processing means, said MV processing means being configured such that the device is capable of implementing a process according to a first, second, or third aspect of the invention. The processing means may, for example, take the form of a processor associated with memory, an FPGA ( Field-Programmable Gate Array (in English or programmable logic circuit) or an ASIC (Application-Specific Integrated Circuit(in English or an integrated circuit developed for a customer). In one embodiment, each CPT counter is capable of measuring the voltage at said CPT counter. In one embodiment, the device includes means for receiving the measurements taken at each associated CPT counter. This can be a wired connection, for example an Ethernet connection, or a wireless connection, for example Wi-Fi or Bluetooth. In an alternative or complementary embodiment, all the measurements taken at each CPT counter are centralized at one or more servers, and the DI device then includes means for receiving the data corresponding to said measurements sent by said servers.

Claims

1. Method (100) for associating each meter (CPT) in a plurality of meters (CPT) with a three-phase output (DTP) of a transformer (TRS) among at least one transformer (TRS), each transformer (TRS) comprising at least one three-phase output (DTP), each three-phase output (STP) comprising three phases (DDP), and each meter (CPT) being connected to one and only one phase (DDP); the method (100) comprising: - a step (1ED) of determining a similarity between each voltage measurement of each meter (CPT) in the plurality of meters (CPT) so as to form a plurality of pairs of meters (CPT), each voltage measurement being made over a period referred to as measurement period, each pair of meters (CPT) being associated with a similarity; the similarity between two signals being determined: - by means of a function taking: ∘ a positive value when the two signals, over the period taken into account, have an overall similar change; ∘ a negative value when the two signals, over the period taken into account, have an overall opposite change; ∘ a zero value, when the two signals, over the period taken into account, have an overall independent change; - by a calculation of regression coefficients; or - by a matrix calculation of partial correlations; the method being characterised in that it furthermore comprises: - a step (1ER) of grouping together pairs of meters (CPT) having a negative similarity according to their similarity so as to form a plurality of groups, each group being associated with a three-phase output (DTP).

2. Method according to the preceding claim, wherein the function taking a positive value when the two signals, over the period taken into account, have an overall similar change; a negative value when the two signals, over the period taken into account, have an overall opposite change; and a zero value, when the two signals, over the period taken into account, have an overall independent change, is a Pearson correlation.

3. Method (100) according to one of the preceding claims, furthermore comprising, before the step (1ED) of determining the similarity between each voltage measurement of each meter (CPT) in the plurality of meters (CPT), a step (1PT) of preprocessing the measurement of the voltage made by each meter (CPT) over the measurement period.

4. Association method (100) according to one of claims 1 to 3, wherein the grouping-together step (1ER) comprises, for each pair of meters (CPT): - when the two meters (CPT) in said pair of meters (CPT) do not belong to any group, a sub-step of creating a group and of associating the two meters (CPT) with said group; - when one of the two meters (CPT) in said pair of meters (CPT) is already associated with a group, a sub-step of associating the meter (CPT) not yet associated with a group with said group; - when the two meters (CPT) in said pair of meters are already associated with two different groups, a sub-step of merging said two groups.

5. Association method (100) according to one of claims 1 to 3, wherein the grouping-together step (1ER) comprises a first sub-step of classifying the pairs of meters (CPT) from the most negative similarity to the least negative and, for each pair of meters (CPT) and in the order of classification thereof: - when the two meters (CPT) in said pair of meters (CPT) do not belong to any group, a sub-step of creating a group and a sub-step of associating the two meters (CPT) with said group; - when one of the two meters (CPT) in said pair of meters (CPT) is already associated with a group, a sub-step of associating the meter (CPT) not yet associated with a group with said group; - when the two meters (CPT) in said pair of meters (CPT) are already associated with two different groups, a sub-step of merging said two groups if there exists at least one other pair of meters (CPT) having a negative similarity the two meters (CPT) of which are associated with the same two groups.

6. Method (100) according to one of the preceding claims, wherein the distance between two meters (CPT) is equal to the absolute value of the similarity between the two meters (CPT) when said similarity is negative and the value zero when said similarity is positive, and wherein, for each group, the method comprises a step of creating three subgroups, each subgroup being constructed so that the mean distance between two meters (CPT) of different subgroups is as great as possible and so that the mean distance between two meters (CPT) belonging to the same subgroup is as small as possible, each subgroup corresponding to a phase (DDP) of the three-phase output (DTP) associated with the group in question.

7. Method (100) according to one of claims 1 to 5, wherein the distance between two meters (CPT) is equal to the absolute value of the similarity between the two meters (CPT) when said similarity is negative and the value zero when said similarity is positive, and wherein, for each group, the method comprises a step of using a clustering algorithm so as to obtain three clusters of meters (CPT), each cluster corresponding to a phase (DDP) of the three-phase output (DTP) associated with the group in question.

8. Method (100) according to one of claims 1 to 5, comprising, for each group: - for each meter (CPT) in the group, a step of determining the other two meters (CPT) in the group with which the meter (CPT) in question has the most negative similarities, the three meters (CPT) thus determined forming a triplet associated with a score so as to obtain a plurality of triplets, the meters (CPT) in a triplet always having a negative similarity two by two; - a step of determining the triplet having the highest score, referred to as reference triplet; - a step of associating a subgroup with each meter (CPT) in the reference triplet so as to obtain three subgroups; then, for each meter (CPT) in the group not belonging to the reference triplet, the method (100) comprises: - when the meter (CPT) has a negative similarity to two and only two of the three meters (CPT) in the reference triplet, a step of associating said meter (CPT) with the subgroup associated with the meter (CPT) in the reference triplet to which the similarity is positive or zero; - otherwise, when the meter has a negative correlation with at least two other meters associated with two of the three subgroups: ▪ for each subgroup among the three subgroups, a sub-step of determining the meter (CPT) associated with said subgroup to which the meter (CPT) in question has the lowest similarity, referred to as intermediate meter (CPT), so as to obtain one intermediate meter (CPT) per subgroup; ▪ if the meter (CPT) in question has a negative similarity to each intermediate meter (CPT), then a sub-step of classifying said meter (CPT) as being difficult to classify, the latter then not being associated with any subgroup; ▪ otherwise a sub-step of associating the meter (CPT) in question with the subgroup associated with the intermediate meter (CPT) to which the meter (CPT) in question has a positive or zero similarity; - otherwise, a step of classifying said meter (CPT) as being difficult to classify, the latter then not being associated with any subgroup.

9. Electronic device (TI) comprising means configured to implement a method according to one of the preceding claims.