ANOMAL IDENTIFICATION IN AN ELECTRICAL DISTRIBUTION NETWORK

Description

[0001] The invention relates to the field of management of an electrical distribution network.

[0002] It is the responsibility of an electricity distribution network manager to monitor the proper functioning of the network. Preventive measures, including maintenance, can limit the risk of malfunctions. Curative measures, including field operator interventions, can reduce the duration and extent of malfunctions when they occur. To enable rapid and effective interventions, it is important to detect the occurrence of an incident as early as possible and to locate the affected areas of the network as precisely as possible.

[0003] Electricity transmission and distribution networks represent the upstream part of the electrical infrastructure and carry electrical energy over long distances. Such networks are generally equipped with at least partially automated monitoring systems. Transmission and distribution networks include sensors capable of remotely transmitting data to monitor the state of the network. Distribution networks represent the downstream part of the infrastructure and carry electrical energy over short distances, but to each consumption node and from a few production nodes. Distribution networks are therefore denser and more numerous than transmission and distribution networks. Therefore, equipping the distribution network with monitoring systems and ensuring its maintenance would be expensive.Distribution networks generally lack such monitoring systems.

[0004] In practice, an incident on a distribution network is generally reported to the network manager by one or more users noticing the interruption in the supply of electrical energy. Such reports, generally made by telephone, are imprecise and unresponsive. For example, when an entire geographical area is affected, only a few users inform the network manager of an anomaly. It is then impossible to identify with certainty each of the network nodes experiencing an anomaly. Consequently, it is difficult to deduce the precise location of a source of the problem, such as a cable break. Research into the sources of anomalies is therefore costly and time-consuming.

[0005] Recently, distribution network nodes can be equipped with communicating devices. For example, communicating meters, known by the commercial reference " Linky » are being deployed in France. Such meters include a communication module capable of transmitting data remotely in an at least partially automated manner. Such modules are designed to transmit information relating to consumption and / or production, particularly for billing purposes. Such modules are not designed to transmit information in real time about outages in the upstream distribution network. The communication modules of the nodes therefore do not, as they stand, allow for the detection of anomalies on the distribution network.

[0006] Furthermore, the document Jan Van De Vyver et al, "The need for a distributed algorithm for control of the electrical power infrastructure", in Proceedings of the International Symposium on Computational Intelligence for Measurement Systems and Applications 2003, CIMSA'03, July 29-31, 2003, pages 211-215 discloses an anomaly detection module on an electrical distribution network including a plurality of nodes, the document FR 3 036 906 A1 relates to a method of obtaining a communication path between a node and a concentrator using power lines.

[0007] The invention set forth in the attached set of claims improves the situation.

[0008] The method and the location system of the invention make it possible to reduce the average time of power outage of nodes of an electrical distribution network. The method provides reliable identification of nodes temporarily isolated from the network, which has the effect of facilitating and accelerating the location of a likely origin of the anomaly. The monitoring of the network can be largely automated and carried out substantially continuously. In the event of a malfunction, the anomaly can be detected in real time and corrective measures adapted to the extent and location of the anomaly can be implemented as soon as possible. The inconvenience caused by a malfunction is short-lived.

[0009] Other characteristics, details and advantages of the invention will appear on reading the detailed description below, and on analyzing the attached drawings, in which: there figure 1schematically represents a set of nodes of a distribution network; the figure 2 represents a functional diagram of a method according to the invention; and the figures 3 to 8 show a portion of the network in various situations.

[0010] The drawings and the description below contain, for the most part, elements of a certain character. They may therefore not only serve to better understand the present invention, but also contribute to its definition, if necessary.

[0011] In an electricity distribution network 100 on which it is desired to apply the method described below, upstream nodes and downstream nodes are usually identified. For example, an upstream node corresponds to a transformer arranged between the medium voltage transmission network (or high voltage type A HTA) and the low voltage (LV) distribution network downstream. The low voltage (LV) network, chosen as an example, has a tree structure. The top of the tree structure corresponds to the upstream node, materialized by an HTA / LV transformer station. Each downstream node forms a connection point along one of the sections of the network. The downstream nodes correspond substantially to the “Delivery Points (PDL)” when they are consumer nodes. The downstream nodes also include producer nodes, or suppliers, from which energy is supplied to the network.For example, wind turbine or photovoltaic panel installations can be connected to the distribution network. A downstream node can be a consumer node and a producer node. Each downstream node forms an interface between the distribution network and a secondary, generally private, electricity network. In practice, the downstream node is materialized by a set of equipment, including a meter capable of recording the electricity consumption and / or production of the secondary network for billing purposes. By misuse of language, the downstream nodes and their equipment can be called meters.

[0012] Each of the nodes is equipped with at least one communication module and a memory capable of storing data. The communication module, or communication interface, is capable of exchanging data with a set of at least one neighboring node among the plurality of nodes and via the network 100. Here, the communication modules exchange data with each other by Power Line Communication (PLC). The communication modules transmit data to a particular communication module called a master module or common “concentrator”. Each node can transmit data to the concentrator directly or via other nodes acting as communication relays. The concentrator is, for example, located at an upstream node while the downstream nodes are equipped with non-concentrator communication modules. Each node is therefore communicating.In the following, transmitting data from one node to another means that the respective communication modules equipping the nodes implement the data transmissions.

[0013] For example, smart meters or AMM for “ Automated Meter Management » are able to exchange data with the concentrator. The concentrator collects data from several AMMs. In the example described here, the concentrator in turn transmits information to a remote processing unit common to several concentrators, for example via a mobile telephone network. Thus, the data processing unit comprises at least one processor operatively coupled with a memory to implement at least in part a method as defined herein. Alternatively, at least part of the data processing may be implemented by the concentrators and / or locally by equipment adjacent to the concentrators.

[0014] Among the AMMs developed, some include communication modules operating via PLC. Thus, data is exchanged via the electrical network itself without the need for a dedicated communication network between the concentrator and the other communication modules equipping the network nodes. The communication network physically uses the electrical cable network. The communication modules of the nodes, including the concentrator, can communicate with each other via PLC, here exclusively via PLC, while a concentrator also communicates with a processing unit by other means, such as mobile telephone networks.

[0015] Among the PLC communication protocols, some consist of a coordination terminal (here the concentrator) and several device-type terminals (here the communication modules of the downstream nodes). The set of nodes, or terminals, is sometimes called a "PLC cluster". The communication modules of the nodes are generally supplied with energy by the distribution network itself. Therefore, the nodes stop functioning (and transmitting data) when the energy supply is interrupted.

[0016] In figure 1 , nodes of an electricity distribution network are represented by small circles, with or without a numerical reference. The relative positions of the nodes on the figure 1 do not necessarily correspond to the actual geographical locations of the nodes. The relative positions of the nodes on the figure 1are representative of a metric for routing data on the electrical network by PLC. In other words, the greater the distance between two small circles on the figure 1 , the longer and / or more degraded the electrical connection between the two corresponding nodes on the ground, the more the probability of success of direct communication by PLC between the two nodes decreases. Thus, the large ovals referenced 10, 30 and 50 on the figure 1 represent the “neighborhood” of the respective nodes 1, 3 and 5. The node referenced 1 corresponds to the upstream node. Node 1 here includes a communication module forming a concentrator.

[0017] Here and in the following, the notion of neighborhood is understood in the sense of PLC communications and therefore of real wired distance. In this context, the notion of neighborhood should not be confused with that used in other fields such as communications networks operating in low current and under IP protocol, or in wireless communications such as mobile telephony. In a simplified manner, it can be considered that only nodes located on the network at a mutual wired distance less than a threshold wired distance are considered as neighboring nodes for each other. In practice, the ability of two nodes to communicate with each other by PLC depends not only on the wired distance but also on the background noise, the noise generated locally by interfering equipment, the network impedances, etc.

[0018] According to the example shown in figure 1 : node 2 belongs to neighborhood 10 of node 1 and does not belong to neighborhoods 30, 50 of the respective nodes 3 and 5; nodes 4 and 6 belong to neighborhood 50 of node 5 and do not belong to neighborhoods 10, 30 of the respective nodes 1 and 3; node 8 belongs to neighborhood 30 of node 3 and does not belong to neighborhoods 10, 50 of the respective nodes 1 and 5; node 12 belongs to both neighborhoods 10 and 30 of the respective nodes 1 and 3 and does not belong to neighborhood 50 of the respective node 5; and nodes 1, 3 and 5 do not belong to each other's neighborhoods.

[0019] In the examples described here, the nodes communicate via PLC in compliance with the standard known as "PLC G3". The standard is defined in particular in the document referenced "ITU-T G.9903, version approved on February 22, 2014, published by the International Telecommunication Union (ITU) and bearing the identifier 11.1002 / 1000 / 12088", which the reader is invited to consult. Thus, the implementation of the method is compatible with existing PLC protocols. At the time of writing, the latest versions of the meters "Linky» deployed in France by the applicant are compatible with the PLC G3 standard. Alternatively, the method can be implemented with nodes equipped with communication modules operating by PLC according to other standards, for example the older PLC G1 standard, and probably with future developments of the ITU-T G9903 standard. For example, older generation communication modules can be equipped with additional hardware and / or software to make them compatible with the method.

[0020] According to the standard, each node stores in memory and updates a set of data, including a "neighbor table." Therefore, there is one neighbor table per node. Although the term "neighbor" is used in the plural, at least some of the nodes may have a single neighbor.

[0021] The neighbor table of a node 1 takes the form of a data set containing information relating to the state of the neighborhood 10 of the node 1. The neighbor table includes at least one list of identifiers corresponding to each of the neighbor nodes 2, 12. The identifier here takes the form of a short address, called " Short_Address » according to the CPL G3 standard. The identifier is unique for each node in a network.

[0022] In each neighbor table, each of the identifiers is associated with at least one state data item. In embodiments, the state data item takes the form of a Boolean value, for example “OK” or “KO”. The state data item can therefore take the form of a parameter as such. Alternatively, it can be considered that the simple presence of a node identifier in the table is equivalent to a state data item, for example “OK”, while the absence of a node identifier in the table is equivalent to another state data item, for example “KO”.

[0023] In the examples described below, the status data includes a value representative of a communication interruption duration. Each status data takes the form of a parameter whose value is an integer between 0 and 255. The value can be interpreted as representative of a validity duration of the functional status (“OK”) of the neighboring node whose identifier is associated. This value is called “ Neighbor_Valid_Time» according to the CPL G3 standard. The parameter, or attribute, takes the form of a countdown initialized to a maximum value (here 255 according to the CPL G3 standard). Each time an activity of the neighbor node 2, 12 corresponding to the identifier is detected by the node 1 to which the neighbor table belongs, the state variable is reset, here to its value 255. Then, the value is decremented by one unit at regular time intervals, for example every minute, until reaching zero and as long as no activity from the corresponding neighbor node 2, 12 is detected. When the value zero is reached, it can be considered that the validity of the functional status (“OK”) of the associated neighbor node 2, 12 has expired and that the associated node 2, 12 is therefore non-functional (“KO”). For example, when the value zero is reached, the associated identifier is deleted from the table.

[0024] A neighbor table can be represented by the following tables I, II and III, corresponding respectively to neighborhoods 10, 30 and 50. Each row of the table corresponds here to one of the neighbor nodes. In the following tables, only a few rows are represented. Alternatively, additional columns containing additional information on each of the neighbor nodes are provided. For example, state data is further associated with date data. For example, a date can provide information on the date of the last activity observed by a node of a neighbor node. Table I (neighborhood 10 of node 1): Short_Address Neighbor_Valid_Time 0002 255 0012 255 ... ... Table II (neighborhood 30 of node 3): Short_Address Neighbor_Valid_Time 0008 255 0012 255 ... ... Table III (neighborhood 50 of node 5): Short_Address Neighbor_Valid_Time 0004 255 0006 200 ... ...

[0025] When the status data takes the form of a Boolean value, for example "OK" or "KO", rather than a validity period, each time activity from the neighbor node corresponding to the identifier is detected by the node to which the neighbor table corresponds, the status variable is reset, for example to the value "OK". Then, the value is changed to "KO", or the associated identifier is deleted from the table after a predefined period of time without detected activity has elapsed, for example after 255 minutes.

[0026] The detected activities of a neighbor node may include data transmissions from the neighbor node. For example, when node 6 belonging to neighborhood 50 sends data to upstream node 1, node 5 detects the sending of the data and interprets this sending as an activity of neighbor node 6. Such detection may result from the relaying of data from a third-party node to another third-party node by node 6.

[0027] The nodes may be arranged to initiate data transmission at regular intervals, preferably at intervals shorter than the validity period selected for the state variable. In the example described above with a validity period of 255 minutes, the nodes may be arranged, for example, to transmit data every hour. The transmitted data may include information, for example for billing purposes, or may consist of a brief good-operation signal such as "ping".

[0028] Alternatively, a message intended to be repeated by all nodes in the network may be sent across the network, for example from upstream node 1. Such a message may for example be transmitted by broadcasting (“ broadcasting"). Thus, any node in the network receiving such a message relays it, which constitutes an activity detectable by neighboring nodes. In this case, it is not necessary for the nodes to be arranged to initiate data transmission themselves at regular intervals or at intervals shorter than the validity period selected for the state variable. For example, the nodes may be configured to initiate data transmission only once every 24 hours or so, and to relay a broadcast frame each time said frame is received. The sending of a broadcast frame may be scheduled at regular intervals (less than the validity period of the state variable). Such sending may be scheduled, in addition to or instead of regular sending, prior to the collection of the neighbor tables.Sending a broadcast frame updates, or resets, the state variable of each operational neighbor node connected to the network in each neighborhood table of operational nodes connected to the network.

[0029] In the context of nodes connected to the power supply network, the associated equipment can easily be powered by said network (as opposed to battery operation). In these cases, data transmissions are inexpensive in terms of energy and can therefore be implemented regularly and / or at will. For example, a broadcast-type frame can be transmitted over the network, by human action or in an automated manner, whenever it is desired to check the proper functioning of the network nodes.

[0030] According to the protocol defined in the CPL G3 standard, message propagation by "broadcasting» is intended to create or repair routes in the PLC network. Such routes allow PLC messages to be transported from device to device between a source and a destination on the electrical distribution network. Broadcast frames can be triggered voluntarily from the upstream node (or hub), or can be triggered by themselves, for example to repair an existing route following a communication failure along it.

[0031] In a situation where all the nodes in the network are operational, and after sending a broadcast frame, all the neighborhood tables include all the respective neighboring nodes and all the state variables have a value corresponding to proper operation, for example "OK" or a validity period " Neighbor_Valid_Time » equal to 255.

[0032] Tables I, II and III above correspond to a situation in which each node of the electrical network is functional except for node 6 which has recently been powered down (or disconnected from the network). Node 6, with the identifier "0006", is located in the neighborhood 50 of node 5. Therefore, the state variable " Neighbor_Valid_Time » associated with the identifier “0006” of the neighbor table of node 5 (table III above) has not been recently initialized to 255: it has the value “200”. This means that the last activity of node 6 detected by node 5 dates back 55 minutes.

[0033] Indeed, since node 6 is powered off, it is no longer able to communicate in PLC, and therefore to participate in the repetition of broadcast type messages. The validity period associated with it, or generally the state variable associated with it, in the network neighborhood tables is therefore not refreshed. In the example described, only the neighborhood table corresponding to neighborhood 50 of node 5 counts node 6 among its neighboring nodes (Table III). The validity period « Neighbor_Valid Time » associated with node 6 (whose identifier is “0006”) in another neighborhood table, for example that of node 4, would also be less than 255 if it had been represented.

[0034] Reference is now made to the figure 2representing the progress of an embodiment of the method. In the example described here, the method comprises an optional preliminary operation 101. Operation 101 is, for example, implemented by the concentrator or equipment associated with it. An initialization signal is sent on the network by PLC. The signal is arranged so as to generate an activity of each of the nodes of the network, activity capable of being detected by neighboring nodes. Operation 101 comprises, for example, the sending of a broadcast type frame on the network 100. Each of the nodes is arranged to relay the frame from node to node by PLC.

[0035] The method comprises an operation 103. The operation 103 is implemented by each of the nodes or the equipments respectively associated with them. The operation 103 comprises the recording in the memory of each node of a set of state data. Each state data is associated with a neighboring node of the node to which the memory belongs, for example by means of an identifier of each neighboring node. The state data set takes, for example, the form of a neighborhood table as described above. The state data are a function of the detection or absence of detection by the node of an activity of the neighboring node. As described above, the state data can take the form of a neighbor table conforming to the CPL G3 standard.

[0036] The method further comprises an operation 105. The operation 105 is, for example, implemented by the concentrator or equipment associated with it. The operation 105 comprises the collection of sets of status data stored in a plurality of nodes. In other words, the status data stored separately by each node of the network are at least partly centralized. For example, the status data are transmitted to a concentrator common to all the nodes of the network. The concentrator may be located at an upstream node. The status data are transmitted, here by PLC, to the concentrator. The data are relayed to the concentrator by the nodes of the network, in particular when a node cannot communicate directly with the concentrator by PLC.For example, when the channel propagation conditions do not allow a PLC signal to be transmitted directly under good conditions, then the intermediate nodes act as relays. Such operation is, for example, provided for by the PLC G3 standard.

[0037] The method further comprises an operation 107. The operation 107 is, for example, implemented by the concentrator or equipment associated therewith, for example an anomaly detection module including at least one processor associated with the concentrator. The operation 107 comprises the cross-referencing (or cross-referencing) of the state data collected from the plurality of nodes.

[0038] State data associated with each node and obtained from several different nodes are gathered and / or compared with each other. In theory, data relating to a node should be identical regardless of the neighbor table from which it is taken. In practice, such cross-referencing makes it possible to detect errors or inconsistencies.

[0039] Data cross-referencing can include sorting through the information from the collected data. This improves the reliability of the status data. a posteriori.For example, when the status data associated with the same node and obtained from several different nodes are identical or similar, the status data may be considered reliable. This is applicable whether the data indicates that the node is operational or the data indicates that the node is non-operational. In these cases, reliable status data (after cross-referencing the data) is established. When discrepancies appear between the status data associated with the same node and obtained from several different nodes, then only some of this data may be retained. For example, it may be considered that a single data set indicating that a node is operational is sufficient to consider said node as operational even if other data contain contradictory information. Reliable status data is then established.

[0040] The method further comprises an operation 109. Operation 109 comprises the identification of the nodes for which the status data assigned to it, or the evolution of said status data, indicate an absence of activity. Operation 109 is, for example, implemented by the concentrator or equipment associated with it, for example an anomaly detection module including at least one processor associated with the concentrator. Depending on the status data collected, optionally made reliable, the operational nodes are distinguished from the non-operational nodes. The non-operational nodes are identified.

[0041] The collected state data, optionally made reliable, are in turn stored, for example in a memory associated with the concentrator. The collected state data take the form, for example, of a neighborhood matrix.

[0042] The method further comprises an operation 111. The operation 111 is, for example, implemented by the concentrator or equipment associated with it, for example an anomaly detection module including at least one processor associated with the concentrator. The operation 111 comprises the deduction of a likely location of an anomaly on the network 100 based on the nodes identified in the operation 109 and a map of the network. By associating the identified non-operational nodes with their respective position on the network and / or their geographical position, a probable area of ​​the source of an anomaly on the network can be defined.

[0043] The deduction can be implemented based on the identified neighboring nodes for which the absence of communication is of a duration greater than a threshold duration. For example, when the state data is representative of a duration, a comparison with a threshold duration can constitute a criterion for deducing the operational or non-operational nature of each node.

[0044] In an operation 113, an alert signal including location data of the anomaly can be generated. The operation 113 is, for example, implemented by the concentrator or equipment associated with it, for example an anomaly detection module including at least one processor, associated with the concentrator. Thus, upon receipt of the signal, it is possible not only to be informed of the occurrence of an anomaly but also of the area to be treated on the network. This contributes to the speed of repair of the network. This also makes it possible to define intervention priorities in the event of multiple anomalies. For example, it may be decided to treat as a priority the anomalies for which the number of impacted nodes (and therefore a priori the number of users) is the most important.

[0045] In embodiments, the operations are implemented in the order of their description above. In embodiments, the method comprises several iterations of the series of operations defined above. For example, the implementation of successive iterations of the operations allows for substantially continuous monitoring of the network. In embodiments, operations 101 to 113 are implemented in turn (once each operation) forming a series.

[0046] In variants, the same operation is implemented several times successively during the same series. For example, the operation 103 of recording the state data of each node is itself implemented recursively during an iteration of the operations 103 to 113. After several iterations of the operation 103 during the same series, the operation 105 of collection (or centralization) is implemented. This optional reiteration of the operation 103 is represented in figure 2 by the loop “T1 = T1max?”. In other words, each node maintains substantially continuous updates of the state data relating to neighboring nodes. At a lower frequency, the updated state data is collected (centralized) on the hub by operation 105.

[0047] Similarly, a series of operations 103-105-optionally 107 can be implemented recursively so as to keep the neighborhood matrix up to date. This optional reiteration of the succession of operations is represented in figure 2 by the loop “T2 = T2max?”. At a lower frequency, operations 109 and 111 are implemented.

[0048] When iteratively implementing the above operations, the first iteration can be seen as the creation of a neighborhood matrix while the following iterations can be seen as an update and / or an enrichment of the neighborhood matrix.

[0049] The operation 109 of identifying non-operational nodes may include the implementation of an algorithm, for example by the anomaly detection module, aimed at detecting the non-refreshing in the neighborhood matrix of data relating to certain nodes. Indeed, when the state data of a node are deleted from the neighbor tables because no activity is detected, it may be necessary to identify such nodes by the absence of updating of said data in the neighborhood matrix.

[0050] Operation 109 for identifying non-operational nodes can include comparisons with a history. This avoids false positives: a node that usually has communication difficulties may appear to be faulty even though the power supply at the node remains operational. Measuring a deviation from the history allows a deviation from normality to be determined. A small deviation from normality may indicate a situation of poor communication without triggering an alert being relevant.

[0051] The collection operation 105 may be implemented automatically by a computer routine, for example for continuous monitoring. The operation 105 may, as an alternative or in addition to automatic implementation, be triggered following a suspicion of an anomaly. Such triggering may itself be automated according to predefined parameters or be manual.

[0052] The implementation of such a process is compatible with at least some of the equipment currently being deployed by the applicant on French territory (“meters” Linky "). Thus, the implementation of the method does not require the installation of additional hardware. In most cases, a software update of the computer components equipping the nodes is sufficient to make them suitable for implementing the method. Such an update can be carried out remotely without requiring an operator to travel.

[0053] The applicant carried out tests to implement the process. The installation included the following elements: an upstream node forming a concentrator comprising a communication module complying with the CPL G3 standard and installed in a High Voltage-Low Voltage (HTA / LV) transformer station; and one hundred communication modules complying with the CPL G3 standard distributed over 20 panels of 5 devices.

[0054] The modules were distributed over two separate transformer feeders, each feeder including a line of approximately 600 meters. 5-volt white noise was injected at four points in the network to artificially limit the propagation of PLC signals and to limit the neighboring radius of each module so that routing was necessary (indirect communications). In practice, a maximum of 2 to 3 hops were observed among the routes.

[0055] For the tests, the same network structure was implemented. Several situations were tested during which different nodes were deliberately powered down. figures 3 to 8each represent a tested situation. In the figures, the empty circles represent groups of 60, 70 functional nodes while the solid circles represent groups of switched-off nodes. The size of the circles is representative of the number of nodes constituting each group, here between 1 and 30 nodes.

[0056] The test results are grouped in Table IV below. Case Figure Composition of extinct node groups Cut detection time Identification of switched off meters A 3 None 1 minute Complete and without false positives B 4 All (100 knots total) 2 minutes and 20 seconds Complete and without false positives C 5 1 group of 30 2 minutes and 30 seconds Complete and without false positives D 6 1 group of 20 2 minutes Complete and without false positives E 7 1 group of 20 and 1 group of 10 3 minutes and 20 seconds Complete and without false positives F 8 1 group of 20 and 1 group of 1 2 minutes and 10 seconds Complete and without false positives

[0057] Although it does not appear in Table IV, each of the switched-off nodes is identified. Thus, maps of the network 100 can be constructed, on which the non-functional nodes are distinguished from the functional nodes, like the figures 3 to 8The location details make it possible to adapt the corrective measures to be taken accordingly and therefore further reduce the duration of interruptions in the energy supply for the affected nodes.

[0058] In the above, the method is applied to all the nodes of an electrical energy distribution network. In variants, the method can be applied in parallel to several groups of nodes of the same network. The groups can be completely distinct or include common nodes. The composition of each set can be selected according to the positions of the nodes on the network so that an alert signal is associated with a predefined area of ​​the network. Thus, when an alert is generated for a group, the signal can be associated with an area of ​​the network.

[0059] In the above, the implementation of the process is described as a succession of operations, regardless of the element implementing the various operations. Depending on the actors involved in the network, different parts of the succession of operations are implemented. In other words, the succession of operations can also be seen as a combination of processes that complement each other.

[0060] The invention can also be seen as a detection module arranged to implement only some of the operations, for example those implemented by the upstream node, or hub. The invention can also be seen as a computer program comprising instructions for implementing the method when this program is executed by a processor, for example a processor equipping the anomaly detection module. The invention can also be seen as a non-transitory recording medium readable by a computer on which such a program is recorded.

[0061] The invention is not limited to the examples of methods and modules described above, solely by way of example, but it encompasses all the variants that a person skilled in the art may envisage within the framework of the protection sought.

[0062] It should be noted that the solutions proposed here are particularly well suited to existing physical installations. This eliminates the need for many costly hardware replacement operations. A PLC communication module compliant with the G3 standard is particularly well suited. In France, for example, a large proportion of the equipment already deployed in the field complies with the G3 standard. To implement the solutions proposed here, it is therefore unnecessary to deploy PLC modules whose properties would be superior to (better than) those of the G3 standard, particularly in terms of throughput, reliability and processing capacity.

[0063] Furthermore, the methods and variants defined herein can be implemented locally, on a neighborhood scale, typically at the level of a transformer arranged between the medium voltage (or high voltage type A HTA) transmission network and the low voltage (LV) distribution network downstream. The anomaly detection module mentioned herein can form part of the equipment of such a transformer. Indeed, modest computing resources are sufficient for the implementation of the method. It is therefore unnecessary to relocate the implementation of the method to remote and powerful computing equipment (use of the " cloud " for example). Centralizing the implementation of the process on the scale of a national electricity distribution network, such as the French network, is superfluous.

[0064] It is understood from the above context that one of the anomalies that we wish to locate, optionally the only anomaly, is the interruption of the power supply.

Claims

1. Method for locating a fault on an electrical distribution network (100) including a plurality of nodes (1, 2, 3, 4, 5, 6, 8, 12) connected directly or indirectly to a concentrator node, each of the nodes in the plurality of nodes being equipped: - with a powerline communication module able to exchange data with said concentrator node and with one or more adjacent nodes among the plurality of nodes and via the network (100), and - with a memory able to store data; the concentrator node comprising: - a powerline communication module configured to communicate with said nodes, and - a fault detection module, the concentrator node being configured to transmit information to a remote processing unit common to a plurality of concentrator nodes, for example via a mobile telephony network; the method comprising: a) recording, in the memory of the nodes, one or more sets of state data, for which each state data item is associated with a node adjacent to the node to which the memory belongs, the state data being dependent on the detection or absence of detection by said node of an activity of said adjacent node, b) collecting (105), in the concentrator node, sets of state data stored in a plurality of nodes, each of the sets being stored in a respective node, c) by comparison and cross-referencing of the state data collected by the concentrator node, identifying (109), at the concentrator node, nodes (6) for which the associated state data or the change in the associated state data indicates an absence of activity, deducing (111), in the concentrator node or in said remote processing unit common to a plurality of concentrator notes, a probable location of a fault on the network from the nodes identified and from a mapping of said network.

2. Method according to claim 1, wherein each of the state data items is furthermore associated with a date data item.

3. Method according to one of the preceding claims, wherein the state data comprise a value representing a duration of interruption of communication.

4. Method according to one of the preceding claims, wherein the deduction (111) is implemented according to the identified adjacent nodes for which the absence of communication is of a duration greater than a threshold duration.

5. Method according to one of the preceding claims, wherein only the nodes disposed on the network in a zone allowing direct communication by PLC are considered to be nodes adjacent to each other.

6. Method according to one of the preceding claims, wherein a succession of operations is implemented in a manner reiterated over time, so that the fault detection is implemented substantially in real time.

7. Method according to one of the preceding claims, furthermore comprising a prior step of sending (101) over the network a broadcast frame intended for all the nodes.

8. System for locating faults on an electrical distribution network (100) adapted to implement the method of any one of the preceding claims, the system including a plurality of nodes (1, 2, 3, 4, 5, 6, 8, 12) provided: - with storage means adapted to record, in a memory of the nodes, one or more sets of state data, for which each state data item is associated with a node adjacent to the node to which the memory belongs, the state data being dependent on the detection or absence of detection by said node of an activity of said adjacent node, and - with a powerline communication module connecting said nodes with each other and with a concentrator node to which said nodes are directly or indirectly connected, said concentrator node being: - equipped with a powerline communication module for communicating with said nodes, - configured to transmit information to a remote processing unit common to a plurality of concentrator nodes, for example via a mobile telephony network, and - provided with a fault detection module adapted to: a) collect (105), in the concentrator node, sets of state data stored in a plurality of nodes, b) by comparison and cross-referencing of the state data collected, identifying (109) nodes (6) for which the associated state data or the change in the associated state data indicates an absence of activity, deducing (111) a probable location of a fault on the network from the nodes identified and from a mapping of said network.

9. Concentrator node for a system for locating faults on an electrical distribution network (100) according to claim 8, the concentrator node comprising: - a powerline communication module configured to communicate with a plurality of nodes in the electrical distribution network, and - a fault detection module, the concentrator node being configured to transmit information to a remote processing unit common to a plurality of concentrator nodes, for example via a mobile telephony network; and the fault detection module including at least one processor arranged to: a) collect (105) sets of state data stored in the plurality of nodes, each of the sets being stored in a respective node, b) by cross-referencing and comparison of the state data collected, identifying (109) nodes (6) for which the associated state data or the change in the associated state data indicates an absence of activity, deducing (111), from the nodes identified and from a mapping of said network, a probable location of a fault on the network.

10. Computer program comprising instructions that lead the system according to claim 8 to implement the steps of the method according to any one of claims 1 to 7.

11. Computer-readable medium on which the computer program according to claim 10 is recorded.