Optimizing sensor placement in pipeline network

Abstract

Tool (1) for determining an optimized sensor placement (33) of a plurality of sensors (2) in a pipeline network (9) for detecting and localizing a leakage (7); the pipeline network (9) comprising a network topology (4) comprising network nodes (21) and edges (22) connecting the network nodes (21); the tool (1) comprising a proposition functionality (3), and an evaluation functionality (5); wherein the proposition functionality (3) is configured for providing propositions of sensor placements (31, 32, 33, 34) for a given network topology; wherein the evaluation functionality (5) is configured for predicting for a proposed sensor placement (31, 32, 33, 34) a respective placement quality (51, 52, 53, 54); wherein the placement quality (51, 52, 53, 54) takes into account a suitability of the respective sensor placement (31, 32, 33, 34) to localize a leakage in at least a part of the pipeline network; wherein the evaluation functionality (5) is configured for providing a placement evaluation (61, 62, 63, 64) for the proposed sensor placement (31, 32, 33, 34), taking into account the respective predicted placement quality (51, 52, 53, 54); wherein the evaluation functionality (5) is configured to take into account the presence of doppelgangers (7c, 7d) for the prediction of the placement quality (51, 52, 53, 54).

Description

[0001] Description

[0002] Optimizing Sensor Placement in Pipeline Network

[0003] The invention relates to the technical field of leak detection in a pipeline system.

[0004] Pipelines of various types at critical industrial facilities, such as power plants or thermal energy complexes, have a vital role in maintaining urban infrastructure and providing basic survival services. When a leakage occurs in a large pipeline network, in order to be able to repair the leakage, it needs to be localized. Pipeline networks can therefore be provided with a sensor system for localizing the leakage. The sensor system can comprise a plurality of sensors that detect signals that are emitted from the leak, for example acoustic signals or pressure signals. Acquirement, installation, and maintenance comes with cost. Therefore, it is desirable to optimally place the sensors.

[0005] Many methods for optimizing a sensor placement exist, e.g. the following disclosures: lean graph method (example with water distribution network) : https : / / www .mdpi . com / 2073-4441 / 12 / 12 / 3439; geostatistical tools and genetic algorithm (example with water distribution network) : https : / / iwaponline . com / jh / article / 21 / 6 / 1030 / 70034 / Optimal- se 1 ection-of -number- and- loca t ion-o f ; greedy algorithm (example with oil pipes) : https : / / www , sciencedirect . com / science / article / pii / S1877050915 009588? via %3 Dihub; information theory method (example with water distribution network) : https : / / www, mdpi . com / 1424-8220 / 22 / 2 / 443; some multi-objective some optimization algorithms (example with gas pipelines) : https : / / ieeexplore .ieee.org / document / 8396219; a method of minimizing the number of non-isolable leaks according to the isolability criteria (example with water distribution network) : https : / / ieeexplore .ieee.org / document / 6693876 .

[0006] Sensor placement is one of the factors affecting the detectability of pipeline leaks localization accuracy. Decreasing the localization accuracy can lead to financial implications, as well as to a reputation damage. Understanding which points are best for placing sensors is crucial for providing an optimized network coverage. Higher localization accuracy allows to save material and human resources. Acquirement, installation, and maintenance of the sensors comes with a cost.

[0007] Difficulties arise also when detecting bursts or other leakages: Some events cannot be localized. Some events are "fake" since they are reflections of real events. In order to avoid such difficulties, it is desirable to optimize sensor placement considering the ability to localize events. Though many sensor placement methods like the ones mentioned above already exist, a suitable algorithm with respect to event localization has not been developed.

[0008] It is an objective of the present invention to optimize the placement of leak detection sensors in a pipeline network.

[0009] This object is achieved by the solutions presented in the independent claims.

[0010] According to an aspect, a tool for determining an optimized sensor placement of a plurality of sensors in a pipeline network is disclosed. The sensors are for detecting and localizing a leakage in the pipeline network. The pipeline network comprises a network topology comprising network nodes and edges . The network edges connect the network nodes . The tool comprises a proposition functionality, and an evaluation functionality . The proposition functionality is configured for providing propositions of sensor placements for a given network topology . The evaluation functionality is configured for predicting for a proposed sensor placement a respective placement quality . The placement quality takes into account a suitability of the respective sensor placement to locali ze a leakage in at least a part of the pipeline network . The evaluation functionality is configured for providing a placement evaluation for the proposed sensor placement . Therefore , the evaluation functionality is configured for taking into account the respective predicted placement quality . The evaluation functionality is configured to take into account the presence of doppelgangers for the prediction of the placement quality .

[0011] According to another aspect , a computer-implemented method for determining an optimi zed sensor placement of a plurality of sensors in a pipeline network is disclosed . Propositions of sensor placements for a given network topology are automatically provided . For a proposed sensor placement , a respective placement quality is predicted . Said placement quality takes into account a suitability of the respective sensor placement for locali zing a leakage in at least a part of the pipeline network . A placement evaluation for the proposed sensor placements is provided, taking into account the respective predicted placement quality .

[0012] According to yet another aspect , a computer program product is disclosed . The computer program product comprises instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method .

[0013] According to yet another aspect , a data processing system is disclosed . The data processing system comprises means for carrying out the method . More speci fic embodiments are presented in an exemplary manner in the independent claims , in the figures , and in the description .

[0014] According to an embodiment the tool is a software component . According to another embodiment , the tool is a data processing system comprising a software component .

[0015] According to an embodiment the tool is configured for providing a selection of evaluated sensor placements , wherein the selection comprises one or more recommended sensor placements based on the placement evaluation .

[0016] According to an embodiment a sensor placement comprises a plurality of sensors , and for each sensor of the plurality of sensors a location in the pipeline network .

[0017] According to an embodiment the suitability of the respective sensor placement to locali ze a leakage in at least a part of the pipeline network is based on detecting at least one time di f ference in the receipt of a signal coming from a site of the leakage in two or more of the sensors of the sensor placement . Said two or more sensors are placed at di f ferent locations in the pipeline network .

[0018] According to an embodiment a doppelganger is represented by two or more edges in the network topology for which a given sensor placement is not suited to unambiguously determine on which one of the two or more edges the leakage occurred . However, according to a more speci fic embodiment , for a doppelganger, the given sensor placement can be suited to determine where on a first edge of the doppelganger the leakage would have occurred in case it really has occurred on said first edge of the doppelganger, and not a second one of the edges of the doppelganger .

[0019] According to an embodiment an external edge is represented by two or more edges in the network topology for which a given sensor placement is not suited to determine on what location of said edge (22a) the leakage occurred. However, according to a more specific embodiment, for an external edge, the given sensor placement can be suited to determine that a leakage has occurred on said given edge.

[0020] According to an embodiment, a sensor placement comprises locations of a given number of sensors on nodes and / or edges of the network topology.

[0021] According to an embodiment, placement quality can comprise a percentage of the network in which a leakage can be localized by the sensor placement.

[0022] According to an embodiment, the leakage is a burst. Bursts are often particularly serious events that happen rapidly. Bursts are preferably monitored in real time. For the monitoring high-pressure sensors with a relatively low frequency (e.g. 200hz) are used.

[0023] According to an embodiment, the leakage is a non-burstleakage. According to a more specified embodiment, the pipeline network is monitored for non-burst-leakages based on schedule analytics via acoustic sensors (e.g. ~4khz) , for example lOsec every night.

[0024] Various features will become apparent to those skilled in the art from the following detailed description of the disclosed non-limiting embodiments. The drawings that accompany the detailed description can be briefly described as follows:

[0025] FIG 1 schematically depicts a block diagram of a data processing system according to an embodiment of the invention;

[0026] FIG 2A and 2B schematically depicts two block diagrams of a network topology with a sensor placement for illustrating the difference between internal and external edges; FIG 3 schematically depicts a block diagram of a network topology with a sensor placement for illustrating the presence of doppelgangers ;

[0027] FIG 4 schematically depicts a block diagram of a network topology illustrating damping ef fects of signals ;

[0028] FIG 5A schematically depicts a block diagram of a network topology with a sensor placement for illustrating the di f ference between internal and external edges ;

[0029] FIG 5B schematically depicts the block diagram of FIG 4A, illustrating the simultaneous presence of doppelgangers and external edges on several edges ;

[0030] FIG 6 schematically depicts the block diagram illustrating the order in which nodes get expanded according to an embodiment of the invention;

[0031] FIG 7 schematically depicts a diagram illustrating an example of optimal sensor placement by BFS algorithm with a given radius of sensitivity without adding sensors to endpoints ;

[0032] FIG 8 schematically depicts a block diagram illustrating an example of optimal sensor placement by BFS algorithm with a given radius of sensitivity with adding sensors to endpoints ;

[0033] FIG 9 schematically depicts a block diagram illustrating an example of optimal sensor placement of 30 sensors by greedy algorithm;

[0034] FIG 10 schematically depict a view of an application interface according to an embodiment of the invention;

[0035] FIG 11 schematically depicts a block diagram of a network topology with a sensor placement illustrating edges exceeding a sensitivity radius . FIG 1 shows a data processing system 1 according to an embodiment of the invention. The data processing system 1 is implemented as a tool 1 comprising an interface 16, a data processor 17, and an accessible memory 18 for carrying out a method according to an embodiment of the invention. An embodiment of the invention does not require that the tool data processing system is identical with the tool. In other embodiments, an embodiment of a tool according to the invention can be a software tool.

[0036] Referring to the embodiment of FIG 1, the tool 1 is a combination of hardware and software and configured for determining an optimized sensor placement 33 of a plurality of sensors 2 in a pipeline network 9 for detecting and localizing a leakage 7. Therefore, a model of a network topology 4 of a pipeline network 9 is loaded into the tool 1. The network topology 4 comprises network nodes 21. The edges 22 are connected via network nodes 21 edges 22.

[0037] The sensors are acoustic sensors and / or pressure sensors (preferably high pressure sensors) , and part of a sensor system having a sensor control system (not shown) . The sensor system is configured to determine a location of the occurrence of a leakage in the pipeline network 9.

[0038] The tool 1 comprises a proposition functionality 3, and an evaluation functionality 5. The proposition functionality 3 is configured for providing propositions of sensor placements 31, 32, 33, 34 for a given network topology.

[0039] The evaluation functionality 5 is configured for predicting for a proposed sensor placement 31, 32, 33, 34 a respective placement quality 51, 52, 53, 54. Each of the placement qualities 51, 52, 53, 54 takes into account a suitability of the respective sensor placement 31, 32, 33, 34 to localize a leakage in at least a part of the pipeline network 9. The evaluation functionality 5 is further configured for providing a placement evaluation 61 , 62 , 63 , 64 for each of the proposed sensor placements 31 , 32 , 33 , 34 , taking into account the respective predicted placement quality 51 , 52 , 53 , 54 .

[0040] The evaluation functionality 5 is configured to take into account the presence of doppelgangers for the prediction of the placement quality 51 , 52 , 53 , 54 .

[0041] The evaluation functionality is configured to take into account the presence of external edges 22a for the prediction of the placement quality 51 , 52 , 53 , 54 . For the prediction of the placement quality of a sensor placement , the evaluation functionality is configured to weight a presence of a doppelganger di f ferently than a presence of an external edge , preferably the presence of the doppelganger is weighted lower than the presence of the external edge .

[0042] Additionally, the proposition functionality 3 comprises an algorithm that is configured for reducing the presence of doppelgangers and / or reducing the presence of external edges at least in critical parts of the pipeline network 9 .

[0043] The evaluation functionality 5 is further configured for predicting for the proposed sensor placements a respective placement ef ficiency 81 , 82 , 83 , 84 , said placement ef ficiency taking into account a suitability of the respective sensor placement to ef ficiently install , operate , and / or maintain the respective sensor placement . The placement ef ficiency is sometimes also addressed a cost function . But the does not necessarily need to be a monetary quantity, it can also be or correlate with the number of sensors or take into account a relatively low likelihood of a leakage or low expected damage in a part of the network where the respective sensor placement has one or several sensors placed . Or the sensor cost can also be or correlate to opportunity cost . The tool 1 is configured for determining a minimum number 68 of required sensors for the pipeline network 9 for the sensor placement 31, 32, 33, 34 to achieve one or more goals. According to a more specified embodiment, the goal is to achieve full coverage of the pipeline network. According to another more specified embodiment, goal is to achieve full coverage of critical parts of the pipeline network. According to another embodiment, goal is to optimize placement quality versus placement cost, in other words to find a compromise between optimizing placement quality and placement cost.

[0044] The tool 1 is further configured for determining an optimized sensor placement 31, 32, 33, 34 based on the placement evaluation 61, 62, 63, 64 of a plurality of sensor placements, wherein the sensor placements of said plurality each have a same number of sensors.

[0045] The tool 1 is configured for determining a respective optimized sensor placement 31, 32, 33, 34 based on the placement evaluation 61, 62, 63, 64 of several pluralities of sensor placements, the sensor placements of different pluralities of sensor placements having a different number of sensors. In a further embodiment, a minimum number or an optimum number can in a further embodiment be found by iteratively incrementing number of sensors.

[0046] The proposition functionality is configured to provide network placements ensuring that critical parts of the pipeline network are fully covered.

[0047] The tool 1 is configured for providing a selection of the evaluated sensor placements, wherein the selection comprises one or more recommended sensor placements based on the placement evaluations 61, 62, 63, 64.

[0048] A selection can be based on a rank order, wherein the rank order is based on the placement evaluations. The rank order can be configured to provide a placement quality rank order and / or a placement cost rank order, or a rank order that takes into account both, placement quality and placement cost, wherein the combined rank order can be based on weighting placement quality and placement cost differently.

[0049] According to a further embodiment, a computer-implemented method for determining an optimized sensor placement 33 of a plurality of sensors 2 in a pipeline network is proposed. The method can be carried out on the data processing system 1.

[0050] The method comprising the steps:

[0051] - automatically providing propositions of sensor placements for a given network topology;

[0052] - predicting for each of the proposed sensor placements a respective placement quality 51, 52, 53, 54, said placement quality 51, 52, 53, 54 taking into account a suitability of the respective sensor placement 31, 32, 33, 34 to localize a leakage in at least a part of the pipeline network; providing a placement evaluation 61, 62, 63, 64 for each of the proposed sensor placements 31, 32, 33, 34, thereby taking into account the respective predicted placement quality 51, 52, 53, 54.

[0053] FIGs 2A and 2B each show a logically equivalent network topology 104. The network topology 104 comprises an edge linear pipe for transporting a fluid. FIGS 1A and IB show a sensor placement comprising 2 sensors 2 that are placed at different locations on the edge 22.

[0054] FIG 2A shows a leakage 7a occurring between the two sensors 2. The part of the pipe 23 between the two sensors 2 is an internal edge 22b, while the other parts of the pipe 23 are called external edges 22a. When a leakage occurs on the internal edge 22a, the leakage 7a can be localized based on detecting the time difference in the receipt of a signal 71a or a signature of the signal 71a coming from a site of the leakage 7a to the two sensors 2. FIG 2B shows a leakage 7b occurring left to each one of the two sensors 2. A signal 71b is transmitted to the left and the right. Only the signal transmitted to the right passes the two sensors 2. However, irrespective of the location where the leakage 7b occurred left of the two sensors 7a, 7b, the time difference measured for receiving the signal 71b sensors is always the same. The only conclusion that can be made by the time difference is that the leakage occurred at the left sensor 2 or left to it.

[0055] In other words: The edge is "external" if the localization within it isn't possible. Yet in other words, V two events on this edge and V pair of sensors the shifts between signals on sensors are equal. Conversely, the "internal" edge is the edge localization within which is possible. These findings can be used find an optimized sensor placement.

[0056] FIG 3 shows a network topology 204 together with a sensor placement having two sensors 202, one left, one right, wherein there are two paths with different edges 222a, 222b connecting the two sensors 202a, 202b. The two edges are "doppelgangers", if the leakage or other event occurred on the one of them but is localized on both. In other words: 3 d eventl 7c and event2 7d on the two edges 222a, 222b respectively, such that V pair of sensors 202a, 202b the shifts between signals on sensors are equal.

[0057] FIG 4 schematically depicts network topology illustrating damping effects of signals. Damping occurs on edges which length significantly exceeds the radius of sensor sensitivity or with weak events when the transient wave fades away before it reaches the sensor.

[0058] FIG 5A shows a network topology with a sensor placement. The sensor placement has three sensors 2 placed at nodes of the network topology. The sensor placement was evaluated by identifying internal and external edges. The internal edges being represented by unbroken lines , while the external edges are represented by dashed lines .

[0059] FIG 5B schematically the network topology with the sensor placement of FIG 5A, illustrating the simultaneous presence of doppelgangers and external edges on several edges . In addition to evaluation the sensor placement by identi fying internal and external edges , the presence or absence of doppelgangers were also evaluated . The sensor placement happens to only comprise internal edges and doppelganger edges that are additionally external , but no internal edges with doppelgangers and no external edges without doppelgangers .

[0060] In the embodiment , the proposed algorithms for optimal sensor placement are based on two well-known algorithms which can be implemented in the proposition functionality of a tool 1 according to an embodiment of the invention . These algorithms are the breadth- first search algorithm and the greedy algorithm . According to other embodiments , the two algorithms can be replaced by di f ferent optimi zation algorithms ( e . g . known algorithms disclosed in the introduction of this disclosure ) . According to an embodiment , the tool is providing network analysis for sensor placement ( calculation of inter- nal / external / doppelgangers ) with a calculation of a cost function and this information can be used by many other algorithms in order to estimate optimal sensor placement (minimization of the cost function) . The algorithms can thus be implemented in a method, and / or in a computer program product , and / or a data processing system according to further embodiments of the invention .

[0061] According to an embodiment with the two algorithms being the breadth- first search algorithm and the greedy algorithm, oth approaches are modi fied with network coverage evaluation . Network coverage refers to the ratio of internal edges ( those pipe areas where events can be locali zed) weight to the weight of the entire network . Below are detailed descriptions and examples of results of breadth-first search and greedy algorithms for estimation of how many sensors are required to cover a given network and how to optimally place them.

[0062] Breadth-first search algorithm:

[0063] FIG 6 shows schematically a breadth first search algorithm. The breadth-first search (BFS) starts by searching a start node Nl, followed by its adjacent nodes N2, N3, N4, then all nodes N5 - N8 that can be reached by a path from the start node containing two edges, three edges N9 - N12, and so on. Formally, the BFS algorithm visits all vertices in a graph G that are k edges away from the source vertex s before visiting any vertex k+1 edges away. This is done until no more vertices are reachable from s.

[0064] According to an embodiment, in order to provide a sensor placement,

[0065] Problem setup: given sensor sensitivity radius, optimize number of sensors and their placement.

[0066] According to an embodiment, in order to provide a sensor placement, an algorithm as follows is applied:

[0067] 1. Start with a random node of the network.

[0068] 2. Through the breadth-first search from the selected origin on the network graph, iterate over network nodes.

[0069] 3. If an encountered (through BFS) node is at distance at least R (radius of sensitivity) , place a new sensor.

[0070] 4. Iterate several times over original node and choose optimal configuration with respect to the cost function.

[0071] Cost function = kl * S - k2 * L, {1} where S - the ratio of the number of chosen nodes to the number of nodes of full graph, L - normalized network coverage (the ratio of internal edges weight to the weight of entire network) . Each edge is weighted by its length and the number of paths passing through this edge, kl - the weight coeffi- cient of internal network, k2 - the weight coefficient of the number of nodes. By default, kl = k2 = 1, but you can set custom weights to give a higher value to one of the components .

[0072] According to an embodiment, in order to provide a sensor placement, the algorithm can add sensors on endpoints - vertices that have only one neighbor. The distance to the endpoint can be significantly different from the sensitivity radius. Adding all the endpoints will increase the number of sensors and affect the quality of the sensor placement.

[0073] FIG 7 shows an example of optimized sensor placement by a BFS algorithm proposing sensor placements with a given radius of a sensitivity of 500 Meters without adding endpoints.

[0074] FIG 8 shows an example of optimized sensor placement by a BFS algorithm proposing sensor placements with a given radius of a sensitivity of 500 Meters with adding endpoints.

[0075] Greedy algorithm

[0076] Greedy algorithms take all of the data in a particular problem, and then set a rule for which elements to add to the solution at each step of the algorithm. The algorithm makes the optimal choice at each step as it attempts to find the overall optimal way to solve the entire problem.

[0077] Problem setup: given sensor sensitivity radius, optimize number of sensors and their placement.

[0078] Algorithm:

[0079] Sensors are placed iteratively while cost function (see BFS- based algorithm) are decreasing:

[0080] 1. If no sensors are placed, choose (first) sensor location as maximizing number of shortest paths passing through this node; Remark: this is a measure of im- portance of this node with respect to the graph topology .

[0081] 2 . I f N- l sensors are placed, place the N-th sensor at a distance at least R ( sensitivity) from all other sensors and maximi zing the number of shortest paths through this node .

[0082] Optionally, the number of sensors can be fixed . In this case greedy algorithm will ignore cost function .

[0083] FIG 9 An example of optimal placement of 30 sensors by greedy algorithm

[0084] Application

[0085] The methods can be assembled into an interactive application which allows to load task, and estimate obtained results .

[0086] FIG 10 shows a possible application interface .

[0087] The parameters for sensor placement :

[0088] • Set general parameters to both methods : the sensitivity radius , the custom weights for cost function { 1 } and the logical flag "Highlight edges exceeding sensitivity" .

[0089] • Set the "Number of sensors" parameter to estimate the optimal sensor placement of the limited number of sensors .

[0090] • Set the " Find minimum number of sensors" and the "Add endpoints" parameters to estimate the minimum number of sensors to cover an uploaded network .

[0091] For real topologies it is quite possible that the distance between two technical points exceeds the sensitivity radius of the sensors . Such edges can be highlighted on the interactive graph . FIG 11 shows a network topology with a sensor placement in which there are edges exceeding a sensitivity radius .

[0092] According to embodiments , the methods the minimum number of sensors required to cover a given topology and for optimal sensor placement is estimated .

[0093] In contrast to basic approaches used in known solutions ( such as lean graph or geostatistical tools ) , some embodiments :

[0094] • Use additional priori information about network topology ( technical points , sensor sensitivity)

[0095] • Evaluate resulting placement in terms of event localization speci fic to leak detection algorithm ( such as internal and external edges , doppelgangers )

[0096] Using the described analytical tool :

[0097] • Reduce the power consumption of sensor network .

[0098] • Improves the quality of network coverage .

[0099] According to an embodiment , the tool comprises at least two analytical components : topology analysis ( in terms of event locali zation) and two placement algorithms , and an interactive application for mapping results .

[0100] According to an embodiment , the method is configured to achieve two main goals of sensor placement - finding the minimum number of sensors to cover the network and the optimal placement of a given number of sensors . The input data for the methods is a network topology comprising points that are technically suitable for sensor installation ( or a topology in which such points are explicitly marked) .

[0101] According to an embodiment , a sensitivity radius of the sensors is speci fied, at which events can be detected and localized .

[0102] According to an embodiment a topology analyses comprises a resulting quality of sensor placement that is evaluated by the coverage of the graph ( length of internal edges ) where events can be locali zed, by having edges with reflected events or doppelgangers and by edges which length signi ficantly exceeds the radius of sensor sensitivity .

[0103] According to embodiments :

[0104] • Topology analysis is performed with identi fying graph edges of di f ferent types : at which events can be localized, which have fake reflected events and which length signi ficantly exceeds the sensor' s sensitivity radius .

[0105] • An estimation of minimum number of sensors is performed with respect to topology analysis and sensitivity radius

[0106] • of sensors .

[0107] • An estimation of optimal sensor placement is performed with respect to weighted network graph .

[0108] • Customi zation of weights of a cost function is carried out to give preference for minimi zing the number of sensors or optimi zing the network coverage .

[0109] • An ability to compare various placement options is provided .

[0110] According to embodiments of the present invention the following technical problems of sensor placement are addressed :

[0111] - How to evaluate network coverage in terms of event locali zation?

[0112] - Which network nodes are critical ?

[0113] - What is the bare minimum number of sensors needed to cover the network? Finding the minimum number of sensors to cover a given network will reduce the power consumption of sensor network, preserving the network coverage .

[0114] - What is the best way to place a limited number of sensors ?

[0115] It should be understood that the foregoing relates only to certain embodiments of the invention and that numerous changes may be made therein without departing the scope of the invention as defined by the following claims . It should also be understood that the invention is not restricted to the illus- trated embodiments and that various modi fications can be made within the scope of the following claims .

Claims

Patent claims1. Tool (1) for determining an optimized sensor placement (33) of a plurality of sensors (2) in a pipeline network (9) for detecting and localizing a leakage (7) ; the pipeline network (9) comprising a network topology (4) comprising network nodes (21) and edges (22) connecting the network nodes (21) ; the tool (1) comprising a proposition functionality (3) , and an evaluation functionality (5) ; wherein the proposition functionality (3) is configured for providing propositions of sensor placements (31, 32, 33, 34) for a given network topology; wherein the evaluation functionality (5) is configured for predicting for a proposed sensor placement (31, 32, 33, 34) a respective placement quality (51, 52, 53, 54) ; wherein the placement quality (51, 52, 53, 54) takes into account a suitability of the respective sensor placement (31, 32, 33, 34) to localize a leakage in at least a part of the pipeline network; wherein the evaluation functionality (5) is configured for providing a placement evaluation (61, 62, 63, 64) for the proposed sensor placement (31, 32, 33, 34) , taking into account the respective predicted placement quality (51, 52, 53, 54) ; wherein the evaluation functionality (5) is configured to take into account the presence of doppelgangers (7c, 7d) for the prediction of the placement quality (51, 52, 53, 54) .

2. The tool (1) of claim 1, wherein the evaluation functionality is configured to take into account the presence of external edges (22a) for the prediction of the placement quality (51, 52, 53, 54.

3. The tool of claims 1 and 2, wherein for the prediction of the placement quality of a sensor placement, the evaluation functionality is configured to weight a presence of a doppelganger differently than a presence of an external edge, preferably the presence of the doppelganger is weighted lower than the presence of the external edge.

4. The tool (1) of any one of the preceding claims, wherein the proposition functionality (3) comprises an algorithm that is configured for reducing the presence of doppelgangers at least in critical parts of the pipeline network.

5. The tool (1) of any one of the preceding claims, wherein the proposition functionality (3) comprises an algorithm that is configured for reducing the presence of external edges at least in critical parts of the pipeline network (9) .

6. The tool (1) of any one of the preceding claims, wherein the evaluation functionality is configured for predicting for the proposed sensor placements a respective placement efficiency (cost) (81, 82, 83, 84) , said placement efficiency taking into account a suitability of the respective sensor placement to efficiently install, operate, and / or maintain the respective sensor placement.

7. The tool (1) of any one of the preceding claims, wherein the tool (1) is configured for determining a minimum number (68) of required sensors for the pipeline network (9) for the sensor placement (31, 32, 33, 34) to achieve one or more goals .

8. The tool (1) of any one of the preceding claims, wherein the tool (1) is configured for determining an optimized sensor placement (31, 32, 33, 34) based on the placement evaluation (61, 62, 63, 64) of a plurality of sensor placements, the sensor placements of said plurality each having a same number of sensors.

9. The tool (1) of any one of the preceding claims, wherein the tool (1) is configured for determining a respective optimized sensor placement (31, 32, 33, 34) based on the placement evaluation (61, 62, 63, 64) of several pluralities of sensor placements, the sensor placements of different pluralities of sensor placements having a different number of sensors .

10. The tool (1) of any one of the preceding claims, wherein the proposition functionality is configured to provide network placements ensuring that critical parts of the pipeline network are fully covered.

11. The tool (1) of any one of the preceding claims, wherein the tool is configured for providing a selection of the evaluated sensor placements, wherein the selection comprises one or more recommended sensor placements based on the placement evaluations (61, 62, 63, 64) .

12. A computer-implemented method for determining an optimized sensor placement (33) of a plurality of sensors (2) in a pipeline network, the method comprising the steps:- automatically providing propositions of sensor placements for a given network topology;- predicting for each of the proposed sensor placements a respective placement quality (51, 52, 53, 54) , said placement quality (51, 52, 53, 54) taking into account a suitability of the respective sensor placement (31, 32, 33, 34) to localize a leakage in at least a part of the pipeline network; providing a placement evaluation (61, 62, 63, 64) for each of the proposed sensor placements (31, 32, 33, 34) , therebytaking into account the respective predicted placement quality (51, 52, 53, 54) .

13. The method of claim 12, wherein for predicting the respective placement quality (51, 52, 53, 54) , a presence of at least one doppelganger in the respective sensor placement (31, 32, 33, 34) is taken into account.

14. The method of any one of claims 12 or 13, wherein for predicting the respective placement quality (51, 52, 53, 54) a presence of at least one external edge in the respective sensor placement (31, 32, 33, 34) is taken into account.

15. The method of claims 13 and 14, wherein the placement quality (51, 52, 53, 54) of a sensor placement (31, 32, 33, 34) is predicted by weighting a negative impact on the placement quality (51, 52, 53, 54) of a presence of one or more doppelgangers lower than a presence of one or more external edges .

16. The method of any one of claims 12 to 15, comprising the method step of predicting for each of the proposed sensor placements (31, 32, 33, 34) a respective placement efficiency (81, 82, 83, 84) , said placement efficiency taking into account a suitablity of the respective sensor placement to efficiently install, operate, and / or maintain the respective sensor placement.

17. The method of any one of claims 12 to 16, comprising determining a minimum number (69) of required sensors for the pipeline network for the sensor placement (31, 32, 33, 34) to achieve one or more goals.

18. The method of any one of claims 12 to 17, comprising determining an optimized sensor placement (31, 32, 33, 34) based on the placement evaluation (61, 62, 63, 64) of a plu-rality of sensor placements, the sensor placements of said plurality each having a same number of sensors.

19. The method of any one of claims 12 to 19, comprising determining a respective optimized sensor placement (33) based on the placement evaluation (61, 62, 63, 64) of several pluralities of sensor placements, the sensor placements of different pluralities of sensor placements having a different number of sensors.

20. The method of any one of claims 12 to 20, comprising providing sensor placements (31, 32, 33, 34) ensuring that critical parts of the network topology (4) are fully covered.

21. The method of any one of claims 12 to 21, comprising providing a selection (67) of the evaluated sensor placements, wherein the selection comprises one or more recommended sensor placements based on the placement evaluations (61, 62, 63, 64) .

22. A computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method of claim 12.

23. A data processing system (1) comprising means for carrying out the method of claim 12.

Images