System and method for dynamically changing island event allocation - Patents.com

Abstract

The present disclosure advantageously provides systems and methods that may apply machine learning to detect and attribute network disruptions to specific components or nodes within the network. In one aspect, the present disclosure provides a computer-implemented method that includes mapping a network comprising a plurality of islands, the plurality of islands comprising a plurality of individual components, which may change dynamically through the splitting and / or merging of one or more islands, and detecting and localizing one or more local events at the individual component level and at the island level using a discrete model.

Description

[Background technology]

[0001] (CROSS REFERENCE TO RELATED APPLICATIONS) This application claims priority to U.S. Provisional Patent Application No. 62 / 822,300, filed March 22, 2019, which is incorporated by reference in its entirety.

[0002] (background) A network is a general system of interconnected devices or components. The operation and control of a networked system can be very complex even under ideal circumstances, as network functions may require the performance of certain tasks synchronously, sequentially, or in more complex patterns. An example of a complex networked system is an electric power distribution network. The control and maintenance of the electric grid is a particularly complex problem given constantly changing supply and demand levels. The current proliferation of renewable energy sources and large-scale battery systems imposes unique opportunities and challenges for the control of electric power distribution networks, as they may bring additional resilience to the network, but they also tend to have intermittencies in their power output. When an unplanned event, such as a power line failure, occurs, the electric network control system is challenged to respond appropriately and minimize the propagation of the failure to other parts of the grid.

[0003] There is a need for predictive maintenance applications for networks such as electric grids. Currently, most predictive maintenance applications are only able to assign disruption events to a particular area of ​​the grid, rather than the specific component that caused the disruption. Some disruption events, such as the disconnection of an area of ​​the grid from the rest of the grid, can be predicted using relatively simple logic. However, as the complexity of the network increases, the predictive maintenance logic becomes much more difficult. Thus, there is a need for improved predictive maintenance applications for complex networks that enhance the ability of control systems and system operators to respond to unexpected changes in the network configuration. Summary of the Invention [Problem to be solved by the invention]

[0004] There is a need for predictive maintenance applications for networked systems that are capable of handling the complex logic that describes interruption and reconnection events. Specifically, there is a need for network maintenance applications that can detect and attribute network interruptions to specific components or nodes within the network. There is also a need for predictive maintenance applications that can learn from individual network interruption events and use the obtained data to predict and respond to future interruption events. [Means for solving the problem]

[0005] The present disclosure addresses the need for enhanced predictive maintenance applications of networked systems by utilizing at least machine learning latent variable models. In some aspects, the present disclosure describes discrete models that may be applied to predictive maintenance of networked systems, such as electric grids or telecommunications networks. Discrete models may provide increased sophistication of the network description, and disruptive events can be traced to specific nodes or components rather than broader regions. The described latent variable models may be capable of handling more complex logical problems such as islanding of grid sections and re-merging of islands back into the network. Predictive maintenance applications may be able to use probabilistic modeling to assign disruptive events to specific components and use the converged set of probabilities to predict subsequent network behavior.

[0006] In some aspects, the present disclosure may describe a computer-implemented event allocation method that includes mapping a network comprising multiple islands, which may be dynamically changed by splitting and / or merging one or more islands such that the islands comprise multiple individual components, and predicting and / or detecting and localizing one or more local events at the individual component level and at the island level, which may utilize discrete models.

[0007] In some aspects, the present disclosure may describe a computer-implemented event allocation method having a number of individual components, including nodes and branches.

[0008] In some aspects, this disclosure may describe computer-implemented event allocation methods in which a discrete model is configured to detect one or more local events at an individual component level instead of at the island level.

[0009] In some aspects, the present disclosure may describe a computer-implemented event allocation method that utilizes a discrete model that is configured to localize or isolate one or more local events at an individual component level.

[0010] In some aspects, the present disclosure may describe a computer-implemented event allocation method that utilizes a discrete model that is configured to detect one or more local events at the individual component level when one or more of the islands are dynamically changing.

[0011] In some aspects, this disclosure may describe computer-implemented event allocation methods in which dynamic changes to islands include multiple instances of splitting, merging, and / or re-merging of one or more islands.

[0012] In some aspects, this disclosure may describe a computer-implemented event allocation method in which dynamic changes to an island include joining and / or splitting two or more of a plurality of nodes and branches.

[0013] In some aspects, this disclosure may describe computer-implemented event allocation methods in which dynamic changes to an island include splitting and / or merging subsets of individual components.

[0014] In some aspects, this disclosure may describe computer-implemented event allocation methods in which subsets of individual components are the same size.

[0015] In some aspects, this disclosure may describe a computer-implemented event allocation method in which subsets of individual components are of variable size.

[0016] In some aspects, the present disclosure may describe a computer-implemented event allocation method that utilizes a discrete model that is a machine learning latent variable model.

[0017] In some aspects, the present disclosure may describe a computer-implemented event allocation method that utilizes machine learning latent variable models implemented at the level of individual components.

[0018] In some aspects, this disclosure may describe a computer-implemented event allocation method in which a machine learning latent variable model comprises a plurality of latent variables corresponding to one or more local events in each of the individual components.

[0019] In some aspects, the present disclosure may describe a computer-implemented event allocation method in which one or more local events occurring at an individual component level are associated with one or more island events occurring at an island level.

[0020] In some aspects, this disclosure may describe a computer-implemented event assignment method in which a machine learning latent variable model is configured to receive an observation of a selected island event occurring on a corresponding selected island, and to process the observation based on a hypothesis that one of a plurality of latent variables caused the selected island event to occur.

[0021] In some aspects, the present disclosure may describe a computer-implemented event assignment method in which a selected island event is triggered by an individual component that is on or associated with the corresponding selected island.

[0022] In some aspects, this disclosure may describe a computer-implemented event allocation method in which a machine learning latent variable model comprises a prior probability that each individual component causes each of one or more local events.

[0023] In some aspects, this disclosure may describe a computer-implemented event allocation method in which a machine learning latent variable model is configured to perform rigorous posterior inference based on prior probabilities to assign one or more local events to individual components.

[0024] In some aspects, the present disclosure may describe a computer-implemented event allocation method in which the allocation of one or more local events to individual components includes determining a probability estimate that each local event will occur in each of the corresponding components.

[0025] In some aspects, this disclosure may describe a computer-implemented event assignment method in which a machine learning latent variable model is configured to use probability estimates and iteratively update prior or prior probabilities until convergence.

[0026] In some aspects, this disclosure may describe computer-implemented event allocation methods in which the prior probabilities comprise initially known, actual, or estimated probabilities based on prior event data.

[0027] In some aspects, this disclosure may describe a computer-implemented event allocation method in which prior or prior probabilities are iteratively updated using an expectation-maximization (EM) algorithm.

[0028] In some aspects, the present disclosure may describe a computer-implemented event allocation method that further includes performing predictive maintenance at an individual component level based on one or more detected local events.

[0029] In some aspects, this disclosure may describe a computer-implemented event allocation method in which the network comprises an electrical distribution network.

[0030] In some aspects, the present disclosure may describe a computer-implemented event allocation method in which an electric power distribution network comprises a plurality of electric power distribution feeders.

[0031] In some aspects, the disclosure may describe a computer-implemented event allocation method in which a plurality of nodes and branches are associated with a plurality of electrical distribution feeders and connected nodes and branches within each feeder.

[0032] In some aspects, a plurality of individual components in a network are arranged in a geometric configuration selected from the group consisting of a one-dimensional configuration, a two-dimensional configuration, and an irregular configuration.

[0033] In some aspects, multiple individual components in a network are arranged in a two-dimensional configuration, the two-dimensional configuration being a rectangular configuration, a radial configuration, or a spoke and hub configuration.

[0034] In some aspects, the one or more local events comprise a non-technical loss associated with one or more of the plurality of individual components.

[0035] In some aspects, mapping the network includes identifying unexpected voltage levels in individual components of the network.

[0036] In some aspects, the disclosure may describe a computer-implemented event allocation method in which one or more events include opening or closing one or more breaker switches in an electrical distribution network, the one or more breaker switches being associated with one or more of the nodes and / or branches.

[0037] In some aspects, the disclosure may describe a system for event allocation comprising: a server in communication with a network comprising a plurality of islands; and a memory storing instructions that, when executed by the server, cause the server to perform operations including: mapping the network such that the plurality of islands can change dynamically by splitting and / or merging one or more islands, and the plurality of islands comprises a plurality of individual components; and detecting and localizing one or more local events at the individual component level and at the island level using a discrete model.

[0038] In some aspects, the disclosure may describe a non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform an event allocation method including: mapping a network comprising a plurality of islands, which may dynamically change by splitting and / or merging one or more islands, such that the plurality of islands comprises a plurality of individual components; and using a discrete model to detect and localize one or more local events at the individual component level and at the island level. The present specification also provides, for example, the following: (Item 1) 1. A computer-implemented method for event allocation, comprising: Mapping a network comprising a plurality of islands, the plurality of islands comprising a plurality of individual components, the plurality of islands being capable of dynamically changing by splitting and / or merging one or more islands; Using the discrete model to predict and / or detect and localize one or more local events at the individual component level and at the island level; A method comprising: (Item 2) 2. The method of claim 1, wherein the plurality of individual components comprises a plurality of nodes and branches. (Item 3) 2. The method of claim 1, wherein the discrete model is configured to detect the one or more local events at the individual component level instead of at an island level. (Item 4) 2. The method of claim 1, wherein the discrete model is configured to localize or isolate the one or more local events at the individual component level. (Item 5) 2. The method of claim 1, wherein the discrete model is configured to detect the one or more local events at the individual component level when one or more of the islands are dynamically changing. (Item 6) 2. The method of claim 1, wherein the dynamic changes to the islands include multiple instances of splitting, merging, and / or re-merging of the one or more islands. (Item 7) 3. The method of claim 2, wherein the dynamic change to the island includes joining and / or splitting two or more of the plurality of nodes and branches. (Item 8) 2. The method of claim 1, wherein the dynamic alteration of the island comprises splitting and / or merging a subset of the individual components. (Item 9) 9. The method of claim 8, wherein the subsets of individual components are the same size. (Item 10) 9. The method of claim 8, wherein the subset of individual components is of variable size. (Item 11) 2. The method of claim 1, wherein the discrete model is a machine learning latent variable model. (Item 12) 12. The method of claim 11, wherein the machine learning latent variable model is implemented at the level of the individual components. (Item 13) 12. The method of claim 11, wherein the machine learning latent variable model comprises a plurality of latent variables corresponding to the one or more local events in each of the individual components. (Item 14) 14. The method of claim 13, wherein the one or more local events occurring at the individual component level are associated with one or more island events occurring at the island level. (Item 15) Item 15. The method of item 14, wherein the machine learning latent variable model is configured to receive an observation of a selected island event occurring on a corresponding selected island and process the observation based on a hypothesis that one of the plurality of latent variables caused the selected island event to occur. (Item 16) Item 16. The method of item 15, wherein the selected island event is caused by one of the individual components located on or associated with the corresponding selected island. (Item 17) 16. The method of claim 15, wherein the machine learning latent variable model comprises a prior probability that each individual component causes each of the one or more local events. (Item 18) 20. The method of claim 17, wherein the machine learning latent variable model is configured to perform rigorous posterior inference based on the prior probabilities to assign the one or more local events to the individual components. (Item 19) 20. The method of claim 18, wherein the allocation of the one or more local events to the individual components comprises determining a probability estimate that each local event will occur in each of the corresponding components. (Item 20) 20. The method of claim 19, wherein the machine learning latent variable model is configured to use the probability estimates and iteratively update prior or prior probabilities until convergence. (Item 21) 20. The method of claim 17, wherein the prior probability comprises an initially known, actual, or estimated probability based on previous event data. (Item 22) 21. The method of claim 20, wherein the prior or prior probability is iteratively updated using an expectation-maximization (EM) algorithm. (Item 23) 2. The method of claim 1, further comprising: performing predictive maintenance at the individual component level based on the one or more detected local events. (Item 24) 3. The method of claim 2, wherein the network comprises an electrical distribution network. (Item 25) 25. The method of claim 24, wherein the power distribution network comprises a plurality of power distribution feeders. (Item 26) 26. The method of claim 25, wherein the plurality of nodes and branches are associated with the plurality of distribution feeders and connected nodes and branches within each feeder. (Item 27) 25. The method of claim 24, wherein the one or more events comprise opening or closing of one or more breaker switches in the power distribution network, the one or more breaker switches being associated with one or more of the nodes and / or branches. (Item 28) 2. The method of claim 1, wherein the plurality of individual components in the network are arranged in a geometric configuration selected from the group consisting of a one-dimensional configuration, a two-dimensional configuration, and an irregular configuration. (Item 29) 2. The method of claim 1, wherein the plurality of individual components in the network are arranged in a two-dimensional configuration, the two-dimensional configuration being a rectangular configuration, a radial configuration, or a spoke and hub configuration. (Item 30) 25. The method of claim 24, wherein the one or more local events comprise a non-technical loss associated with one or more of the plurality of individual components. (Item 31) 25. The method of claim 24, wherein mapping the network includes identifying unexpected voltage levels in individual components of the network. (Item 32) 1. A system for event allocation, comprising: a server in communication with a network comprising a plurality of islands; a memory storing instructions that, when executed by the server, cause the server to: Mapping the network, wherein the plurality of islands can change dynamically by splitting and / or merging one or more islands, the plurality of islands comprising a plurality of individual components; Using the discrete model to predict and / or detect and localize one or more local events at the individual component level and at the island level; A memory and A system comprising: (Item 33) A non-transitory computer readable medium having instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform an event allocation method, the method comprising: Mapping a network comprising a plurality of islands, the plurality of islands comprising a plurality of individual components, the plurality of islands being capable of dynamically changing by splitting and / or merging one or more islands; Using the discrete model to predict and / or detect and localize one or more local events at the individual component level and at the island level; 16. A non-transitory computer readable medium comprising:

[0039] (Incorporated by reference) All publications, patents, and patent applications mentioned in this specification are incorporated herein by reference to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference. To the extent that the publications and patents or patent applications incorporated by reference conflict with a disclosure contained in this specification, the specification is intended to supersede and / or precede any such conflicting material. [Brief description of the drawings]

[0040] [Figure 1] FIG. 1 is a diagram of an example predictive maintenance system.

[0041] [Diagram 2] FIG. 2 is a schematic diagram of an exemplary power distribution network.

[0042] [Diagram 3] FIG. 3 is a flow chart of an exemplary process for detecting and localizing events occurring in a distributed system using a discrete model.

[0043] [Figure 4] FIG. 4 is a flow chart of an exemplary process for training a latent variable model.

[0044] [Figure 5a] FIG. 5a depicts the splitting, merging, and remerging of nodes in a one-dimensional distributed system.

[0045] [Figure 5b] FIG. 5b is a chart of the hypothetical probability that an event will occur at each node in FIG. 5a.

[0046] [Figure 6]FIG. 6 is a schematic diagram of a computer system that is programmed or otherwise configured to implement the methods provided herein. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0047] The present disclosure describes systems and methods for event allocation during splitting and merging of networked systems. Networks such as electric grids can be highly complex dynamic systems that require equally complex control systems to maintain proper functioning. Control of the system can be relatively simple during normal operation, but when unexpected changes occur, such as the removal of a component or the failure of a connection, the logic controlling the network can become more complex. The present disclosure provides systems and methods for improving the control of dynamically changing networks that can be applied to the control of electric grids, telecommunications networks, supply chain distribution networks, computer networks, social networks, oil and gas production networks, and the like.

[0048] Networks such as electric grids can be composed of many components in numerous configurations. For example, in an electric grid, the network may consist of power plants, transmission lines, transformers, substations, and local energy sources such as rooftop solar panels. A network may be highly connected, such as an electric grid in an urban area, or highly unilinear and less interconnected, such as an electric grid in a rural area. In general, the more isolated components in a networked system are more vulnerable due to a lack of fallback connections if another connection fails. Conversely, highly connected networks may be prone to cascading effects, e.g., power surges associated with blackouts, when components are intricately linked to each other. A current challenge and opportunity in the control of networked systems is to develop methods to both understand the past behavior of the network and apply that understanding to predict the future behavior of the network.

[0049] In complex networks, such as the electric grid, specific areas of the network with limited connectivity may become isolated if one or more connections fail. Such an event may be referred to as "islanding" due to the isolation of a subset of network components from the rest of the network. In highly complex networks, islands may frequently form and then re-merge with the network when connections are restored. As an extended example, a windstorm may cause multiple grid failures due to downed power lines, leading to the formation of multiple isolated islands of outage customers. When the power lines are restored, the islands will re-merge, shifting the way power must be distributed. In some cases, power line failures may be random, but in some cases the failures may be predictably tied to particularly vulnerable grid components. The ability to accurately predict likely failures, localize the failures at the island / component level, and proactively anticipate shifts in power demand as the grid is restored can help enhance reliability and reduce the cost of grid operation.

[0050] The present disclosure describes the use of one or more machine learning algorithms to analyze the behavior of a complex network during an islanding event and predict how the network will react when an island is formed and re-merges with the network. The machine learning algorithm can be trained with a data set over time. As the machine learning algorithm becomes better trained, it gains an improved ability to predict specific components within the islanding system that may be failing. Using the predicted results, the control of the rest of the network can be adjusted accordingly, taking into account the eventual re-merging of the island based on past performance. The present disclosure provides a faster and more efficient way for traditional network control systems to handle the control of complex networks.

[0051] FIG. 1 is a schematic diagram of an exemplary predictive maintenance system 100 configured to assign island-level events to specific components in a network and predict future component-level events. System 100 includes network 110. Network 110 includes nodes, e.g., nodes 1-10, and branches that interconnect the nodes, e.g., branches 21, 22, and 23, which connect node 1 to nodes 2, 6, and 7, respectively. For convenience, only a few branches in network 110 are identified. Also, the nodes and branches depicted in network 110 are merely illustrative. In general, a network can have any useful combination and any useful number of nodes and branches. For example, a network can have any of the properties, characteristics, or features described in the following paragraphs.

[0052] A network as described herein may comprise a system of multiple interconnected components. In some aspects, a network may comprise at least 2 components, at least 3 components, at least 4 components, at least 5 components, at least 10 components, at least 25 components, at least 50 components, at least 100 components, at least 150 components, at least 200 components, at least 250 components, at least 500 components, at least 1,000 components, at least 5,000 components, at least 10,000 components, at least 25,000 components, at least 50,000 components, at least 100,000 components, at least 500,000 components, at least 1,000,000 components, or more. In some aspects, each component may comprise a source, device, connector, junction, switch, or sink. In some aspects, a component in a networked system may comprise a dynamic component, e.g., a power plant, so that it actively affects the behavior of the network, or it may be a passive component, e.g., a light bulb, that reacts to the behavior of the network. A network may comprise interconnected components of similar size or behavior, e.g., a computer network, or it may comprise interconnected components of highly variable size, e.g., household appliances.

[0053] Networks as described herein may be directly or indirectly connected. In some aspects, a network may include a network connection that comprises a tangible link between components, e.g., power lines in an electric grid. In other aspects, a network may comprise a remote communication link between components, e.g., a wireless computer network. In some aspects, a network connection may be fixed, static, or permanent. In other aspects, a network connection may be variable, alterable, or reconfigurable. In some aspects, a network connection may have physical or other properties that affect the behavior of the network. In some aspects, a network connection may comprise a network component, e.g., a power line, a transformer, or a wireless router.

[0054] A network as referred to herein may also be described as a system of multiple interconnected nodes. In some aspects, a network may comprise at least 2 nodes, at least 3 nodes, at least 4 nodes, at least 5 nodes, at least 10 nodes, at least 25 nodes, at least 50 nodes, at least 100 nodes, at least 150 nodes, at least 200 nodes, at least 250 nodes, at least 500 nodes, at least 1,000 nodes, at least 5,000 nodes, at least 10,000 nodes, at least 25,000 nodes, at least 50,000 nodes, at least 100,000 nodes, at least 500,000 nodes, at least 1,000,000 nodes, or more. A node may comprise one or more components. In some aspects, a node may be a single device, connector, junction, switch, or sink. In some aspects, a node may comprise multiple components grouped together, e.g., a house with many appliances and other electrical sinks. In some aspects, a node in a network may have one or more connections to other nodes. In some aspects, a node may not have any connections to other nodes.

[0055] Networks as described herein may be arranged in a number of geometric configurations. In some aspects, a network may comprise a one-dimensional connection of nodes, e.g., a node may be classified as a terminal, i.e., a terminal node that may only have a connection to one other node, a central node that may have two connections to other nodes, or a separate node that does not have any connections to other nodes. In some aspects, the connection between two nodes may be severed and the central node may become a terminal node or the terminal node may become a separate node.

[0056] In some aspects, a network may comprise a two-dimensional network. In some aspects, a node in the two-dimensional network may comprise one or more connections to other nodes. In other aspects, an isolated node in the two-dimensional network may not have any connections to other nodes. In some aspects, a connection in the two-dimensional network may comprise a bottleneck if no other connections exist that enable a path between two nodes at the ends of a selected connection.

[0057] In some aspects, a network may comprise a two-dimensional grid that may be represented spatially in a rectangular manner. In some aspects, the rectangular grid may comprise nodes that have only operational connections to their nearest neighbors. In some aspects, a network may comprise a two-dimensional grid that may be represented spatially in a radial manner. In some aspects, a radial grid may comprise a plurality of linear series of nodes radiating outward from one or more central nodes. In some aspects, a network may comprise a spoke and hub geometry where connections radiate outward from a central node to satellite nodes. In some aspects, satellite nodes of a spoke and hub network may have operational connections. In some aspects, a spoke and hub model may comprise several nested layers of satellite nodes whereby concentrically connected satellite rings may be formed around a hub node.

[0058] In some aspects, the network may comprise a hybrid geometry of linearly, rectangularly, or radially interconnected nodes. In some aspects, the network may comprise an irregular geometry with connectivity in a complex or asymmetric pattern. In some aspects, the network may comprise a network of more than two dimensions, whereby the additional dimensions may comprise different characteristics or physical properties that distinguish aspects of the network, e.g., a single-phase electrical grid connecting with a multi-phase electrical grid.

[0059] Networks as described herein may comprise many types of dynamic systems. The methods of the present disclosure may be applied to any system that meets the basic description of a network, such as that described above. The systems described below are examples to help illustrate the types of systems in which an event allocation algorithm may be beneficial.

[0060] A network may be defined by its physical connectivity, in some aspects the network may comprise an electrical grid system by which electricity may be distributed from various sources to various sinks.

[0061] A network as described herein may be defined by its data connectivity. In some aspects, the network may comprise a computer network. The computer network may comprise components such as a processor, a hard drive, an Ethernet card, an Ethernet cable, a wireless receiver, or a wireless router. In some aspects, the computer network may comprise a mainframe computer or a server system.

[0062] A network as described herein may be defined by spatial connectivity. In some aspects, the network may comprise an air traffic control system whereby individual airplanes are operatively associated with a particular control station and switch associations when they enter the control domain of a different station.

[0063] A network as described herein may be defined by multiple modes of connectivity. In some aspects, the network may comprise a manufacturing system. The manufacturing system may comprise electrical components, computer components, transportation devices, material delivery systems, part handling systems, robotic devices, or other interconnected devices.

[0064] A network as described herein may comprise one or more computers for control of the network. In some aspects, a computerized control system may automatically control the function of one or more components of the network. In some aspects, a computerized control system may require user input as part of its decision-making process. In some aspects, a computerized control system may comprise a reactive system that redistributes and balances network tasks or behaviors when changes occur. In some aspects, a computerized control system may comprise a predictive system that anticipates and proactively redistributes and balances tasks or behaviors before changes occur. In some aspects, a predictive control system may comprise machine learning algorithms.

[0065] A network as described herein may undergo an event that alters the behavior or characteristics of the network. An event may comprise the modification, removal, upgrade, failure, or other change of one or more components or nodes in the network. In some aspects, an event may comprise a failure in the network. The failure may comprise a broken link, a hardware failure, a data transfer disruption, or any other event that inhibits the performance of the network. An event may comprise a system upgrade, such as the addition of a new component or node, or a transient change in output from a component or node with variable operation. In some aspects, an event may create a bottleneck when the flow between two components or nodes becomes inhibited by a lack of connectivity. In some aspects, the disconnection of a bottleneck in a network may constitute islanding. An island may re-merge with the main network upon re-establishment of connectivity between the two entities.

[0066] Returning to FIG. 1 , network 110 also includes sensors 112, 114, and 116. Sensor 112 collects data at node 1, and sensors 114 and 116 collect data at branches 24 and 25, respectively. Sensors 112-116 may collect different data depending on the type of network of network 110. For example, if network 110 is an electrical distribution system, sensors 112-116 may be voltage sensors that collect voltage data. If network 110 is instead a supply chain network, sensors 112-116 may be vehicle sensors or traffic sensors, for example. Sensors 112-116 may communicate with event allocation server 120 via wired or wireless network 130.

[0067] System 100 also includes an event allocation server 120, which implements discrete model 125. Event allocation server 120 may be located at or near any node or branch of network 110, or it may be located remotely. Although referred to in this disclosure as a singular server, event allocation server 120 may actually be multiple servers in multiple locations.

[0068] The discrete model 125 can be implemented on the event allocation server 120 in software, hardware, or a combination of the two. The discrete model 125 is configured, i.e., trained, to use data from the sensors 112-116 to detect, isolate, and localize island-level events occurring within the network 110. That is, the discrete model 125 is configured to determine which nodes or branches within the network 110 are involved in, or are the source of, the event. For example, using data collected by the sensors 112-116, the trained discrete model 125 can determine that the source of the event 140 is node 3 of the network 110. The discrete model 125 can additionally or alternatively use data from external sources (e.g., weather data) to determine the source of the event. Events will be described in more detail later.

[0069] Without the discrete model 125, the event allocation server 120 would not be able to localize an event to a particular node or branch in the network 110 because sensor data is not available at all nodes and branches. The discrete model 125 will be described in more detail with reference to Figures 3 and 4.

[0070] The event assignment server 125 also includes a training module 127 that assists in training the discrete model 125. The training module 127 may be a computer program that implements a training process for the discrete model 125. The training of the discrete model 125 will be described in more detail with reference to FIG.

[0071] FIG. 2 is a schematic diagram of an example power distribution network 200. The power distribution network 200 includes a power generating station 210, a power transmission substation 220, a high voltage power transmission line 230, switches 240a-c, a power receiving substation 250, a power distribution line 260, transformers 270a-d, and consumers 280a-d. The power generating station 210 can generate electricity. The power transmission substation 220 can convert the electricity generated by the power generating station 210 into high voltage electricity. The high voltage power transmission line 230 can transmit the high voltage electricity over long distances. The switches 240a-c can isolate portions of the power distribution network. The power receiving substation 250 can convert the high voltage electricity from the high voltage power line 230 into electricity that the power distribution line 260 can safely carry. And, the transformers 270a-d can convert the electricity from the power distribution line into electricity having a suitable voltage for the consumers. Although not depicted, the distribution network may include distribution feeders. A distribution feeder is a connection between an output terminal of a receiving substation 250 and an input terminal of a distribution line 260. The nodes and branches in the networks described herein may be associated with the distribution feeders and the connected nodes and branches in each distribution feeder. Any of the components described above may be equipped with or integrated with sensors that collect data about the component. Such data may be used by the algorithms described herein. Distribution networks are described in more detail in the following paragraphs.

[0072] A power distribution network as described herein may comprise any system that conveys electrical energy from a source to a sink. A power distribution network may comprise a grid system. In some aspects, a power distribution network may comprise an interconnected system of electrically connected components whereby electrical flow is used to drive or operate multiple components, for example, a factory assembly line. A power distribution network may comprise an interconnected system of power sources, power lines, junctions, switches, and power sinks. A power distribution network may comprise one or more computers for purposes such as control, monitoring, maintenance, optimization, or any other necessary network function.

[0073] An electrical distribution network as described herein may comprise one or more power sources. A power source may comprise any device, component, source, or equipment that delivers electrical energy to a distribution network. A power source may comprise a large-scale generating facility such as a combustion-driven power plant, a gasification power plant, a nuclear power plant, a hydroelectric power plant, a geothermal power plant, a solar thermal power plant, a photovoltaic power plant, a wind power plant, or a tidal power plant complex. A large-scale power source may be a base load, peak load, or load following source. Large-scale combustion and gasification systems may utilize a variety of fuels, including coal, petroleum, biomass, oil shale, and municipal solid waste. A power source may comprise a small-scale generating source such as a rooftop photovoltaic panel, a microturbine, a gas generator, a diesel generator, or a windmill. A power source may comprise an electrochemical device such as a hydrogen fuel cell, a solid fuel cell, or a battery. The battery system may include alkaline batteries, aluminum air batteries, lithium metal batteries, molten salt batteries, nickel-based primary batteries, solid state batteries, aluminum ion batteries, lead acid batteries, nickel metal hydride batteries, lithium cobalt batteries, lithium manganese batteries, lithium ion polymer batteries, lithium phosphate batteries, lithium sulfur batteries, magnesium ion batteries, and nickel cadmium batteries.

[0074] The power source may deliver electricity in various ways. The power source may comprise a DC source. The power device may comprise an AC source. In some aspects, the AC source may comprise a single-phase or three-phase source. The power source may comprise a specific voltage. The DC power source may comprise a value of 72V or more, a value of 48V or more, a value of 36V or more, a value of 24V or more, a value of 18V or more, a value of 12V or more, a value of 5V or more, a value of 1.5V or more, a value of 1V or more, a value of 0.1V or more, or a value of 0.01V or more. Alternatively, the DC power source may comprise up to 72V, up to 48V, up to 36V, up to 24V, up to 18V, up to 12V, up to 5V, up to 1.5V, up to 1V, up to 0.1V, or up to 0.01V. The AC power source may comprise a specific voltage. The AC power source may comprise a voltage of 765 kV or more, 500 kV or more, 230 kV or more, 115 kV or more, 69 kV or more, 240 V or more, 230 V or more, 220 V or more, 120 V or more, 110 V or more, 10 V or more, 1 V or more, or 0.01 V or more. In other cases, the AC power source may comprise up to 765 kV, up to 500 kV, up to 230 kV, up to 115 kV, up to 69 kV, up to 240 V, up to 230 V, up to 220 V, up to 120 V, up to 110 V, up to 10 V, up to 1 V, or up to 0.01 V. In some aspects, the AC power source may operate at 50 Hertz or 60 Hertz.

[0075] An electrical distribution network as described herein may comprise a number of transmission elements that operatively connect components or nodes in the network. A transmission element may comprise a component or node that has properties that may affect the behavior of the network. A transmission element may comprise any object or device capable of transmitting an electric charge. A transmission element may comprise a wire, cable, conduit, or cord. A transmission element may comprise a bundle or collection of a number of individual wires or cables. A transmission element may comprise a metallic conductor, a superconductor, or a semiconductor. In some aspects, a transmission element may comprise a power line. In some aspects, a power line may comprise a high tension power line. In other aspects, a transmission element may comprise a sub-transmission power line, a feeder line, or a distribution line. In some aspects, a transmission element may be above ground. In other aspects, a transmission element may be buried or underwater. In some aspects, a transmission element may be permanently affixed. In other aspects, a transmission element may be removable or separable.

[0076] An electrical distribution network as described herein may comprise a plurality of junctions or switches. A junction may comprise a point of connection between two or more transmission elements. A junction may comprise one transmission element that splits into two or more transmission elements. A junction may comprise two or more transmission elements that merge to result in a fewer number of transmission elements. In some aspects, a junction may comprise a transformer. In some aspects, a junction may comprise a substation. In some aspects, a junction may comprise a terminal or a connection plate. An electrical distribution network may comprise a plurality of switches or opens. A switch may comprise any element that may interrupt or switch electrical flow from on a transmission element to a different transmission element. A switch may comprise a manually or mechanically controlled device. A switch may be electronically controlled. A switch may be controlled by the action of a computer. A switch may comprise a relay, such as a mechanical relay or a solid state relay. An electrical distribution system may comprise a plurality of opens. An open circuit may comprise any device or component that temporarily or permanently disrupts electrical flow. An open circuit may comprise a fuse or an open switch. An open circuit may be a passive or active element in an electrical distribution system.

[0077] Electrical energy may be consumed or dissipated in various sinks. Electrical sinks may comprise any device that converts electrical energy into mechanical, thermal, or electromagnetic energy. Electrical sinks may comprise common devices including, but not limited to, computers, lights, washers, dryers, refrigerators, pumps, compressors, microwave ovens, furnaces, stoves, fans, winches, motors, tools, and electric dissipation water heaters. In some aspects, a battery system may behave as a source during discharge and as a sink during charging. In some aspects, a sink may comprise a ground or neutral connection.

[0078] A power distribution network as described herein may include other components. The electrical components may be passive or active. The electrical components may include resistors, diodes, rectifiers, transistors, light emitting diodes, amplifiers, inductors, fuses, potentiometers, capacitors, or solenoids.

[0079] An electrical distribution network as described herein may comprise one or more computers for control of the network. In some aspects, a computerized control system may automatically control the functioning of the electrical distribution network. In some aspects, the computerized control system may require user input as part of its decision-making process. In some aspects, the computerized control system may comprise a reactive system that redistributes and balances electrical loads when changes occur. In some aspects, the computerized control system may comprise a predictive system that anticipates and proactively redistributes and balances electrical loads before changes occur. In some aspects, the predictive control system may comprise machine learning algorithms.

[0080] An electrical distribution network as described herein may comprise a conventional power grid system. An electrical grid may comprise an interconnection of power sources, transmission devices, and power sinks in a local, regional, national, or international grid. A grid may comprise one or more large-scale power sources that provide a continuous base load. In some aspects, a grid may comprise a network of small-scale devices, e.g., photovoltaic cells, that feed electrical energy into the grid in a regular or irregular pattern. A grid may comprise a synchronous grid, whereby alternating current occurs at the same phase and frequency throughout the grid. An electrical grid may comprise more than one voltage, with variations mediated by devices such as transformers.

[0081] An electrical distribution network as described herein may comprise a microgrid. A microgrid may comprise a small or isolated grid. A microgrid may comprise conventional energy sources such as gas generators. A microgrid may comprise renewable energy sources such as wind turbines. In some aspects, a microgrid may operate independently from a conventional power grid. In other aspects, a microgrid may be connected to a larger grid. A microgrid may comprise one or more computers that control the functioning of the microgrid. In cases where one or more connections between the microgrid and the conventional grid are disconnected, the microgrid may be capable of autonomous operation without supply or control from a primary grid system.

[0082] Electrical grids as described herein may be arranged in a number of geometries. The power distribution network may comprise a radial network. The radial network may comprise a tree-like network whereby individual radial branches are subdivided into smaller branches which may in turn also be subdivided into smaller branches. The power distribution network may comprise a mesh network. The mesh network may comprise a loop system or a tidal ring system. The power distribution network may comprise a more complex geometry that blends aspects of radial and mesh networks.

[0083] An electrical distribution network as described herein may undergo an event that alters the distribution and flow of electrical energy. The event may comprise an alteration, removal, upgrade, failure, or other change of one or more components or nodes in the electrical network. In some aspects, the event may comprise a fault in the network. The fault may comprise a broken transmission line, a power outage, a power plant emergency, or a cascading failure. The event may comprise a system upgrade, such as the addition of a new source of power, or a transient fluctuation in power output from a source such as a wind turbine or photovoltaic cell. In some aspects, an event may create a bottleneck when the flow between two components or nodes becomes inhibited by a lack of connection. In some aspects, the disconnection of a bottleneck in an electrical distribution network may constitute islanding. An electrical island may re-merge with the main power grid upon re-establishment of a connection between the two regions.

[0084] An event in an electric distribution network may comprise a loss event. A loss event may comprise an expected or unexpected drop in an electric load in one or more regions of an electric grid. A loss event may comprise a technical loss. A technical loss may comprise a source of constant loss, such as resistive losses in an electric transmission line or transformer losses. In some aspects, an electric grid control system may consider technical losses in its algorithms. A loss event may comprise a non-technical loss (NTL). An NTL may comprise any unexpected source of loss. In some aspects, an NTL may comprise theft of electric service, tampering with an electric meter, a meter malfunction, a control system error, or sabotage of an electric device. In some aspects, an NTL may not be considered or may be insufficiently considered in an electric grid control system. In other aspects, an NTL may be considered by a grid control system applying machine learning algorithms to detect unexpected losses.

[0085] In some cases, the components described herein may include a power sensing meter. The power sensing meter may be attached to an electric main unit in a building. In some cases, the power sensing meter may be located at a substation, on a utility pole serving the building, or in a meter box adjacent to or within the building. The building may be a residential building or a commercial building. The building may be a home or any type of structure that may produce and / or consume electricity.

[0086] A power sensing meter as described herein may comprise an electric smart meter and / or a solar energy monitor. A distribution network as described herein may comprise a power monitoring module.

[0087] The smart meter may comprise one or more sensors configured to measure and record the building's electricity consumption in real time or near real time. Appliances and fixtures in the building may consume electricity, and their electricity consumption may be measured and recorded by the smart meter. For example, the smart meter may be configured to record the electrical energy consumption at time intervals and provide the data to the power monitoring module at a predetermined frequency. The predetermined frequency may range from every minute, every 5 minutes, every 10 minutes, every 15 minutes, every 30 minutes, every hour, every 12 hours, or every 24 hours. Any range of frequencies may be envisioned. In some cases, the predetermined frequency may range from every hour, every 30 minutes, every minute, every 45 seconds, every 30 seconds, or less than every 15 seconds. Thus, the electrical consumption data may comprise measurements of electrical consumption as a function of time. For example, the measured electrical consumption may be time stamped. In some cases, the electrical consumption data may be recorded and stored in a remote database or server.

[0088] In some cases, smart meters may be configured to collect electrical consumption data for remote reporting to a utility provider, for monitoring and / or billing by the utility provider. Smart meters may also be configured to send outage notifications to a utility provider and monitor the quality of the power delivered to the building (e.g., transmission rate, power continuity, voltage peaks, etc.). Smart meters may be connected to an electrical grid and used to determine the building's electrical load on the grid.

[0089] Any description of a smart meter may also apply to any type of device used to measure electricity consumption. For example, this may include a built-in electric meter in a building that transmits electricity consumption data to a utility. Any description of a smart meter may also apply to a meter that includes "smart" functionality, such as bidirectional real-time or near real-time signal communication between the meter and the utility.

[0090] A solar energy monitor as described herein may comprise one or more sensors configured to measure and record solar-based electricity generation in real-time, near real-time, or intermittent cycles. Electricity may be generated using a solar power system located on a building. The solar power system may comprise one or more solar panels configured to convert sunlight into electricity. In some embodiments, the solar power system may be a photovoltaic power system (also known as a solar PV power system or PV system). A PV system is a power system designed to provide usable solar power using photovoltaic power. A PV system may comprise an arrangement of several parts, including solar panels to absorb sunlight and convert it to electricity, solar inverters to change the current from DC to AC, and a stand, cabling, and other electrical accessories to set up the operating system. PV systems may be generally categorized into three distinct market segments: residential rooftop, commercial rooftop, and ground-mounted utility-scale systems. Their capacity may range from a few kilowatts to hundreds of megawatts. A typical residential system may be around 10 kilowatts and mounted on a sloping roof, while commercial systems can reach the megawatt scale and are generally disposed on gently sloping or even flat roofs.

[0091] The solar energy monitor described herein may be configured to record electrical energy generation at time intervals and provide the data to the power monitoring module at a predetermined frequency. The predetermined frequency may range from every minute, every 5 minutes, every 10 minutes, every 15 minutes, every 30 minutes, every hour, every 12 hours, or every 24 hours. Any range of frequencies may be envisioned. In some cases, the predetermined frequency may range from less than every hour, every 30 minutes, every minute, every 45 seconds, every 30 seconds, or every 15 seconds. Thus, the electrical generation data may comprise a measurement of electrical generation as a function of time.

[0092] In some instances, the solar energy monitor may be configured to collect solar power production data. The solar energy monitor may include any device or system capable of monitoring solar production data. The solar power production data may be obtained through one or more inverters in a solar power system located in a building. The inverters may convert a variable direct current (DC) output of a photovoltaic (PV) solar panel in the solar power system to utility frequency alternating current (AC) for powering conventional AC-powered equipment. Measurements of the converted current may indicate the solar power production data. The solar power production data may be transmitted to a solar monitoring system. The solar monitoring system may be configured to store, update, and monitor the solar production data.

[0093] The predictive maintenance systems described herein can be used to predict component-level events in networks other than electrical distribution networks, for example, the predictive maintenance systems can be used to predict events in telecommunications networks, supply chain networks, computer networks, oil and gas production networks, or social networks.

[0094] A telecommunications network as described herein may comprise a plurality of devices for transmitting data within the network. The telecommunications network may include telephone networks, computer networks, financial transaction networks, satellite networks, and any other interconnected network based on data transmission. The telecommunications network may comprise a plurality of data transfer and data acquisition devices that operatively communicate directly or indirectly with each other. The telecommunications network may also comprise devices for data processing and control. In some aspects, the telecommunications network may comprise a series of interconnected components capable of transmitting and receiving data, e.g., an Aircraft Communications Addressing and Reporting System (ACARS) system for communication between aircraft and ground control stations. In some aspects, the telecommunications network may comprise a series of interconnected devices that transmit data only to a central acquisition hub, e.g., a network for remote weather sensing.

[0095] A component or node in a telecommunications network may comprise a data transfer or acquisition device. A component or node in a telecommunications network may be associated with an address that identifies the location of the device in the network. In some aspects, a data transfer or acquisition device may be associated with a fixed address. In other aspects, a data transfer or acquisition device may have a variable address that reconfigures at random or regular intervals. A component or node may comprise a data transmission device, such as a sensor or a wireless transmitter. A component or node may comprise a data receiving device, such as a pager, a television, a radio, a printer, or an answering machine. A component or node may comprise a device capable of transmitting and receiving data, such as a fax machine, a switchboard, an Ethernet hub, a wireless router, a computer, a landline phone, a wireless phone, a communications satellite, or a modem.

[0096] A telecommunications network may achieve operational connectivity for data transmission via electrical or electromagnetic signaling. In some aspects, connectivity in a telecommunications network may be achieved via wired connections, such as Ethernet cables, coaxial cables, fiber optic cables, or telephone lines. In some aspects, connectivity in a telecommunications network may utilize electromagnetic transmission equipment for data transfer, such as transmitters, receivers, satellite dishes, waveguides, antennas, and repeaters. In some aspects, data transmission may occur via transmission of packets. In other aspects, data may be transmitted in a streaming or continuous manner. In some aspects, data transmission may be governed by a communication protocol, such as TCP / IP, IPX / SPX, X.25, or AX.25.

[0097] Telecommunications networks may undergo events that create topological rearrangements of components or nodes and connections within the network. In some aspects, an event may comprise the addition or removal of a device to or from the network. In other aspects, an event may comprise the alteration or reallocation of an address of a component of the network. In some aspects, an event may comprise a physical or non-physical disturbance in the network. A physical disturbance may include an outage due to a power line disruption or the electromagnetic effects of a solar storm. A non-physical disturbance may include an erroneous or failed transmission. Events in telecommunications networks may be temporary, such as an outage, or permanent, such as the removal of a device from the network.

[0098] The telecommunications network may be a mesh network. A mesh network is a network in which nodes of the network connect directly, dynamically, and non-hierarchically to other nodes in the network. The non-hierarchical arrangement of nodes prevents the failure of one node from affecting a large portion of the network and allows the nodes to cooperate with each other to efficiently route data to and from clients. Mesh networks may be "self-healing" mesh networks in that they may automatically reconfigure to avoid transmitting data through a particular node when that node fails, e.g., when an event occurs at that node. Data about how a self-healing mesh network automatically reconfigured itself can be used to localize the event.

[0099] Some telecommunications networks may comprise computer networks. A computer network may comprise multiple computer devices and devices under the control of one or more computers. A computer network may be a closed system or an open system. An open computer network may comprise one or more connections to other computer networks. Computer network systems may include personal area networks, local area networks, wireless local area networks, wide area networks, system area networks, carrier private networks, and virtual private networks.

[0100] A computer network may comprise multiple computers or nodes, comprising interconnected devices. A computer network may include components such as desktop computers, laptop computers, handheld or tablet computers, wireless telephone devices, modems, routers, hubs, servers, printers, monitors, touch screens, and scanning devices. A computer network may also include hardware components, such as robots or programmable logic controller (PLC) controllers, that perform tasks under the control of the network. The components of a computer network may be involved in a variety of tasks, including data transfer, data acquisition, control, signaling, calculation, processing, and display. Components in a computer network, such as a display screen, may play a passive role. Components in a computer system, such as a process controller, may play an active role.

[0101] Computer networks may have operational connectivity through wired or wireless connections. Wired may include Ethernet cables, USB cables, flash drives, pin connectors, and any other form of connection that allows for the transfer of data between components of a computer network. Computer networks may be connected by wireless networks. Wireless connectivity may include hubs, routers, and any other devices necessary for the transmission of electromagnetic signaling between computer devices.

[0102] Computer networks may undergo events that create a topological rearrangement of components or nodes and connections within the network. In some aspects, an event may comprise the addition or removal of a computing device to or from the network. In other aspects, an event may comprise the alteration or reallocation of an address of a component within the network. In some aspects, an event may comprise a physical or non-physical failure within the network. A physical failure may include any event associated with a hardware disruption or disconnection, such as a printer jamming. A non-physical failure may include any event associated with a failure in data transmission, acquisition, processing, or output. A non-physical failure may include a failure to transmit data, such as an outage in a server or router. A non-physical failure may also include events associated with malicious activity, such as the removal of a networked computer due to ransomware, the forced isolation of a computing device infected with malicious code, or the overloading of one or more components with large amounts of data transmission during a distributed denial of service attack. Events in telecommunications networks may be temporary, such as an outage, or permanent, such as the removal of a device from the network.

[0103] A social network as described herein may comprise multiple users connected through a relationship framework. A social network may build the relationship framework based on a number of factors, such as geography, profession, nationality, political preferences, family ties, friendship ties, or any other factor that distinguishes the components of the network. A social network may have a classification system that includes various levels of classification, including corporate, family, individual, or combinations thereof.

[0104] In some aspects, the components of the social network may comprise individual users of the social network. In other aspects, the components of the social network may comprise a collection of individual users that are grouped together within a unifying classification, such as a common employer, alumni status, family, hobbies, or any other basis for classification. In some aspects, individual users in a social network may be connected to other users through multiple forms of classification.

[0105] Connectivity in a social network may be self-defined or system-defined. In some aspects, users of a social network platform may choose how they are categorized and may establish connectivity to other users in the network accordingly. In other aspects, computer algorithms may form connections between individual users or components based on acquired or user-supplied data. Connectivity between components in a social network may be permanent or severable, either by users or by governing computer algorithms.

[0106] An event in a social network may include any change in connectivity within the network. For an individual user, an event may be tied to a life event such as birth, marriage, death, divorce, enrollment, graduation, employment, promotion, firing or layoff, illness, or any other personal event that alters the relationships between individual users in a social network. At the corporate level, connectivity may be altered by events such as mergers, acquisitions, divestitures or subsidiary relationships, turnover, changes in leadership, reorganization, or capacity, e.g., class changes in a school. Event occurrences in a social network may be highly random, such as the formation of new friendships, or highly predictable, such as scheduled employee work availability.

[0107] An oil and gas production network as described herein may include drilling assets, refining assets, and pipeline assets. Drilling assets may include platforms, drilling rigs, drill bits, casings, pumps, and the like. Drilling assets may be onshore or offshore assets. Refining assets may include any machinery or equipment typically found in a refinery plant, including distillation units (e.g., crude distillation units and vacuum distillation units), catalytic reformers, hydrocrackers, treaters (e.g., amine treaters, merox treaters, hydrotreaters, and the like). Pipeline assets may include pipelines, pumps, compressors, heat exchangers, valves, and the like.

[0108] Nodes in an oil and gas production network may be drilling and refining assets. Nodes may be connected by pipeline assets and other supply chain infrastructure, e.g., roads, rails, sea routes, and vehicles that use such roads, rails, and sea routes. An event in an oil and gas production network may be a mechanical or electrical failure. For example, an event may be a failure of a drilling, refining, or pipeline asset. Additionally or alternatively, an event may be a weather event, e.g., weather that prevents drilling or causes a disruption in the supply chain.

[0109] Drilling, refining, and pipeline assets may be equipped with sensors that collect data about such assets. The sensors may be thermometers, pressure gauges, flow meters, accelerometers, magnetometers, and the like. Data collected by such sensors can be used by the algorithms described herein to predict and localize events in oil and gas production networks.

[0110] A supply chain network as described herein includes multiple physical locations, routes between the physical locations, vehicles that navigate the routes, and personnel working at the physical locations and operating the vehicles. Supply chain networks often rely on coordination among multiple entities, e.g., business organizations, operating in many countries across thousands or millions of square miles. Supply chain networks are configured to efficiently transport goods from one or more sources to multiple destinations.

[0111] Nodes in a supply chain network can include supplier manufacturing plants, assembly plants, local distribution centers, regional distribution centers, and destinations, such as residential or commercial customers. Branches in a supply chain network can include roads, railroads, sea routes, and flight paths that connect together the nodes in the supply chain network.

[0112] A supply chain network also includes the vehicles that transport goods on the routes of the supply chain network, and the people who operate the vehicles and work at factories and distribution centers. Vehicles in a supply chain network can include trucks, trains, ships, planes, and the like. Typically, one or more people operate each vehicle. Manufacturing factories and distribution centers also typically employ a significant number of people to facilitate the production process, the organization of goods, and the loading of goods onto the vehicles.

[0113] An operator of a supply chain network can collect data about the supply chain network from various sources. For example, the operator can collect weather data from Doppler radar sensors or publicly available sources, traffic data from vehicles in the supply chain network or from publicly available sources such as Google Maps or Waze, vehicle data, such as mileage or maintenance records of vehicles in the chain network, and manufacturing equipment data. Because a supply chain network can span hundreds or thousands of miles and multiple countries, it is generally not feasible to collect data about every node, branch, and vehicle in the supply chain network.

[0114] Events, i.e., failures, in a supply chain network can disrupt the supply chain network and cause costly shipping delays. For example, for a commercial customer, a disruption in the supply chain network can cause a reduction in production because, for example, component parts of a product are not available. For this reason, it can be beneficial to predict and prevent events before they occur.

[0115] Events in a supply chain network may include weather events or similar natural disasters, such as hurricanes, snowstorms, floods, forest fires, and the like. Weather events may make routes impassable and cut off the supply chain network into isolated islands. However, operators of the supply chain may not know the exact location of the impassable roads. For example, operators may know that a snowstorm has affected a particular city or region, but may not know the specific roads that are impassable. Operators may also know which roads have been affected by snowstorms in the past. Based on this limited data and using the algorithms disclosed herein, operators can predict which specific roads will be impassable and adjust shipping routes accordingly or send snow removal equipment to those roads.

[0116] Events in a supply chain network can also include vehicle malfunctions. For example, a truck may break down or run out of fuel. Using limited prior data, for example, vehicle maintenance data and data about vehicles that have malfunctioned in the past, and using the algorithms described herein, an operator of the supply chain network can predict vehicles that will malfunction and perform preventive maintenance on such vehicles. Similarly, a supply chain operator can predict equipment malfunctions in a manufacturing plant.

[0117] Events in a supply chain network may also include traffic events. In some cases, an event may make an entire road impassable. In other cases, an accident may only affect certain vehicles. Examples of traffic events include vehicle accidents, construction, and large public events. Traffic events may increase shipping times, resulting in delays in the transportation of goods from a source to a destination.

[0118] Events in the supply chain network may also include worker shortages or strikes. A lack of workers may lead to increased shipping times.

[0119] An event in a supply chain network can be localized to a particular branch, for example a road, sea, rail or flight path, a particular vehicle, a particular factory or distribution center, or a particular supplier.

[0120] Events in a supply chain network, similar to events in a power distribution network, may cause portions of the supply chain network to become isolated from one another. Such isolated portions are referred to herein as "islands." When routes are restored, the islands reconnect. The ability to accurately predict likely failures in a supply chain network and localize failures at the node / route level can help enhance the reliability and timeliness of the supply chain network.

[0121] 3 is a flow chart of an exemplary process 300 for detecting and localizing events occurring in a network using a discrete model. The process 300 can be implemented by one or more appropriately programmed computers configured to collect data about the network, i.e., event data. For example, the process 300 can be implemented by the event allocation server 120 depicted in FIG. 1. For convenience, the process 300 will be described as being implemented by a single server. The process 300 can be adapted or modified for use in detecting and localizing events in any of the network types discussed above, including power distribution networks, telecommunications networks, computer networks, supply chain distribution networks, social networks, oil and gas production networks, and the like.

[0122] The server observes island-level events in the network 310. The network may be an electric power distribution network, a telecommunications network, a computing network, a supply chain network, a social network, an oil and gas production network, or any other type of network.

[0123] An island-level event is an event that occurs in a network that is directly traceable to an island, but not to a component within the island. An island is an isolated partition of a network. An island may be formed when a connection in the network is broken, for example, when a power line in an electric distribution network falls, a component in a telecommunications network or a computing network fails, or a route in a supply chain network becomes impassable. Similarly, two or more islands may merge when a connection in the network is restored, for example, when an electric utility repairs a fallen power line in an electric distribution network, when a failed component in a telecommunications network or a computing network is repaired, or when an impassable route is cleared. Thus, the islands that make up a network can dynamically change by splitting or merging with each other.

[0124] Dynamic changes of an island may involve multiple instances of splitting, merging, or re-merging of an island. In general, splitting an island involves splitting two or more nodes or branches within the island, while merging two or more islands involves merging a node or branch from each of those two or more islands. Stated differently, dynamic changes of an island may involve splitting and / or merging a subset of the individual components of the island. The subsets of individual components may be of the same size, or the subsets of individual components may be of variable size.

[0125] Each island may include one or more individual components. In general, the individual components in an island may be represented as nodes and one or more branches to or from those nodes. More specifically, for example, an island in a power distribution network may include one or more of the following: a power plant, a transmission substation, a high voltage transmission line, a switch, a circuit breaker, a receiving substation, a feeder or distribution line, a local transformer, and a consumer. An island in a computing network may include a subset of the servers, processors, interconnects, wireless routers, and wireless receivers in a distributed computing network. Also, an island in a system of automation equipment in a manufacturing plant may include a subset of such automation equipment. In some embodiments, an island is a smaller partition, e.g., a partition that includes only a subcomponent of the components listed above. For example, an island in a power distribution network may include only a subset of the generators in a power plant or a subset of the transformers in a transmission substation. Any level of abstraction is possible.

[0126] In general, an event is a change in the operational state of one or more individual components in a network. For example, an event may be a failure, removal, upgrade, modification, or other similar change of a component. Examples of failures include a severed branch, a hardware failure, a disruption of a data transfer path, and other similar events that impede the performance of the network. A removal may involve the intentional removal, either temporary or permanent, of a component from the network, e.g., temporary removal of a component for maintenance. An upgrade may be the addition of a new component to the network or a naturally occurring transient change in output from a component that involves variable operation. Also, a modification may involve the substitution of one or more components for one or more other components. In some embodiments, an event may create a bottleneck when flow between two components or nodes becomes inhibited by a lack of connectivity.

[0127] One example of an event in an electrical distribution network is the opening or closing of a breaker switch in the electrical distribution network, the breaker switch being associated with one or more of the nodes and / or branches of the electrical distribution network. Other examples of events in electrical distribution networks include broken generators, downed transmission and distribution lines, transformer faults, circuit breaker overloads, and the like. Examples of events in distributed computing systems include hard drive failures, memory overloads, race conditions, and the like.

[0128] An event may cause other events, or even a cascading series of events. For example, a downed power line in an electrical distribution network may overload a circuit breaker in another part of the electrical distribution network. In some cases, an event may cause an entire island or the entire distribution network to fail.

[0129] The server can "observe" an event by collecting data that defines the event. Generally, the server cannot immediately identify the specific component in the network that caused or was the source of the event. This is because the server does not receive data from all nodes and branches in the network. For example, it would be impractical and expensive to outfit the entire power distribution network, which may include thousands of electrical devices, hundreds of miles of transmission and distribution lines, and thousands of consumers, with the necessary sensors. Similarly, it would be impractical to outfit the entire supply chain network, which may span multiple countries and cover thousands or millions of square miles of land, sea, and airspace. However, the server can determine the approximate area in the network where the event occurred based on data collected at a subset of nodes in the network. For example, the server can receive power line and distribution line data from a subset of power lines. The data can include line temperature and vibration, instantaneous and average current and voltage measurements, and the like. Similarly, a subset of the transformers, switches, circuit breakers, and electric meters in the power distribution network may be equipped with sensors that transmit data to the server. The sensors may communicate with the server via a wired or wireless network.

[0130] By comparing data from different nodes across the network, the server can generally attribute or assign an event to a particular island. The server can also use the data to map the network, especially when the server does not have a pre-existing map of the network or when an event changes the configuration of the network. Mapping the network can involve identifying separate partitions of the network, i.e., islands, and their boundaries. Islands may be identifiable and distinguishable from one another based on one or more characteristics or properties of the data the server receives. For example, in the context of an electrical distribution network, islands may be identifiable and distinguishable from one another based on unexpected voltage levels in certain parts of the network.

[0131] The server provides data describing island-level events to the discrete model (220). In general, a discrete model is a model that differentiates a system into parts. In the context of this disclosure, the parts are the individual components that make up the island.

[0132] The discrete model is configured, i.e., trained, to use the data to detect, isolate, and localize events at the individual component level. That is, the discrete model is configured to determine individual components within an island that are involved in or are the source of an event. For example, a particular distribution line within an electric power distribution network may be the source of an event if the line fails, e.g., buckles under the weight of heavy snow during a snowstorm. Using island-level data describing the event, the discrete model is configured to identify the distribution line as the source of the event. As another example, a server memory within a distributed computing network may be the source of an event if the memory exceeds capacity and disrupts a data path within the distributed computing network. Using island-level data describing the event, e.g., data identifying a data path that is disrupted within a distributed computing system, the discrete model is configured to identify the memory as the source of the event. Training of the discrete will be described in more detail with reference to FIG. 4.

[0133] The discrete model is configured to detect one or more local events at the individual component level when one or more of the islands are dynamically changing. Assigning or attributing events to individual components is important for this very reason, namely, because islands may split, merge, and re-merge, making it difficult to track which island an event belongs to.

[0134] In some embodiments, the discrete model is a machine learning latent variable model. A latent variable is a variable that is not directly observed, but instead is inferred from other observed variables. A classic example of a latent variable is general intelligence. Although general intelligence cannot be directly observed and measured, it can be known by other observed variables, such as standardized test scores, high school or college grade point averages, subjective assessments, or combinations thereof. Another example of a latent variable is general health. Again, general health cannot be directly measured, but it can be known by several other observed variables, including body mass index, resting heart rate, blood pressure, cholesterol levels, and the like.

[0135] In the context of the networks described herein, the components involved in or that are the source of a particular event are generally unobservable and therefore latent variables that can be modeled. Although it is theoretically possible to observe the individual components that cause an event, it is impractical to do so for reasons explained above. However, both (i) the components involved in a particular event and (ii) the probability that a particular component is the source of a given event can be inferred from other observable data organized into a latent variable model.

[0136] A machine learning latent variable model is a statistical model that (i) models latent variables such as those described above and (ii) is generally refined over time through training in an unsupervised learning environment. In the context of the networks described in this disclosure, the machine learning latent variable models can be implemented at the component level. That is, the machine learning latent variable models can each include a latent variable that corresponds to the probability that a local event occurs in each of the individual components.

[0137] For convenience, the machine learning latent variable model will be referred to simply as the latent variable model. Training of the latent variable model will be described in more detail with reference to FIG.

[0138] Latent variable models generally operate under the assumption that the observable variables are simply the result of their location on the latent variables, and that the observable variables do not have any commonality after controlling for the latent variables. This is known as local independence. For example, a latent variable model of an electrical distribution network may operate under the assumption that a voltage imbalance is simply the result of the particular island on which the component that caused the event is located.

[0139] One example of a latent variable model is a mixture model. A mixture model is a probabilistic model for representing the existence of subpopulations within a larger population. That is, a mixture model can be used to make statistical inferences about the nature of subpopulations given only observations on the entire population, without subpopulation identification information. The model does not require that the observed data identify the subpopulation to which an individual observation belongs. In the context of the networks described herein, the observed island-level events may be the larger population of the mixture model, and the individual components involved in the events may be subpopulations that exist but are not observed.

[0140] In a classical implementation, a mixture model is used to model house prices. In general, different types of houses in different neighborhoods have widely different prices, but, for example, prices of a particular type of house in a particular neighborhood tend to cluster. One possible model of such prices is a model that assumes that (i) prices are accurately described by a mixture model with k distinct components, (ii) each component has a normal distribution, and (iii) each component specifies a particular combination of house type / neighborhood. Fitting this model to observed prices, for example by using an expectation maximization algorithm, would tend to cluster the prices according to house type / neighborhood and reveal the spread of prices within each house type / neighborhood.

[0141] This same approach is applicable to the networks described herein. One possible model of events in a network is one that assumes that (i) island-level events can be traced to one of k different components within the island, (ii) each component has a normal distribution, and (iii) each component of the model defines a physical component within the island. Fitting this model to observed island-level events, for example by using an expectation-maximization algorithm, will tend to cluster the observed events by the components involved in the event and reveal the likelihood that a given event originated from a particular component.

[0142] Another example of a latent variable model is a Bayesian model. In general, a Bayesian model is a probabilistic graphical model that represents a set of variables or features and their conditional dependencies. A Bayesian model can utilize a directed acyclic graph (DAG) to represent the variables or features. Each variable or feature can be represented by a node in the Bayesian model. In one example, a connection, link, or edge can exist between a first node and a second node if one node is influenced by or dependent on the other. If the first node is influenced by or dependent on the second node, a connection can be directed from the second node to the first node. If the second node is influenced by or dependent on the first node, a connection can be directed from the first node to the second node.

[0143] A Bayesian model can represent probabilistic relationships between island-level events. In one embodiment, when limited data about island-level events is available, a Bayesian model can be used to determine, infer, or predict the individual components involved in the island-level events.

[0144] The inference of the individual components involved in the event can be based on at least some data provided to the Bayesian model. For example, some island-level event data can be provided to the Bayesian. In some implementations, inferring the individual components in the island involved in the event includes applying a maximum a posteriori estimation process to the Bayesian model to obtain a plurality of energy consumption values ​​for a plurality of energy consumption sources associated with the user. In general, with respect to the Bayesian network model, the maximum a posteriori estimation process can correspond to the mode of the posterior distribution. The maximum a posteriori estimation process can be utilized to obtain a point estimate of an unobserved quantity, e.g., the probability that a particular component is involved in the event, based on empirical data, e.g., island-level data from previous events. In one embodiment, one or more of the probabilities that an individual component causes a given event in the island are unknown. The maximum a posteriori estimation process can determine, infer, or estimate a most likely value for one or more unknown probabilities based on available information, such as limited island event data. Thus, discrete probabilities can be determined, inferred, or estimated using the maximum a posteriori estimation process. It should be understood that many variations are possible. Also, in some embodiments, inference of multiple individual component probabilities can be performed without deploying additional sensors in the distribution network.

[0145] 3, the server receives (330) detections of local events at the component level and island level from the discrete model, which may be any of the models described above. The detections can be used to perform predictive maintenance on the components. In some implementations, the process 300 is used to predict component-level events before they occur and perform maintenance on the corresponding components.

[0146] 4 is a flow chart of an exemplary process 400 for training a latent variable model, for example, one or more of the latent variable models described with reference to FIG. 3. Specifically, the process 400 corresponds to an expectation-maximization algorithm. The expectation-maximization algorithm is an unsupervised learning algorithm in that the training examples do not need to be labeled. In fact, since latent variable models model unobserved variables, it is often impractical or impossible to provide labels for the training examples. The process 400 can be implemented by the training module 127 depicted in FIG. 1.

[0147] As a preliminary note, the expectation-maximization algorithm is an iterative method for finding maximum likelihood or maximum a posteriori estimates of parameters in a statistical model, where the model depends on unobserved latent variables. In the context of networks described herein, the unobserved latent variables are the individual components that cause island-level events, and the parameters are the probabilities that an individual component is the source of a given island-level event.

[0148] The maximum a posteriori probability estimate is an estimate of an unknown quantity that is equal to the mode of the posterior distribution of that unknown quantity. Maximum a posteriori probability can be used to obtain point estimates of unobserved quantities based on empirical data, e.g., island-level data. It is closely related to the method of maximum likelihood estimation, but employs an extended optimization objective that incorporates a prior distribution over the quantity one wishes to estimate (quantifying the additional information available through prior knowledge of the relevant events). Maximum a posteriori estimation can therefore be viewed as a regularization of maximum likelihood estimation.

[0149] Expectation-maximization iterations alternate between an expectation step, which creates a function for the expectation of the log-likelihood evaluated using current estimates for the parameters, e.g., the component probabilities, and a maximization step, which calculates the parameters that maximize the expected log-likelihood determined on the expectation step. These parameter estimates are then used to determine the distribution of the latent variables in the next expectation step.

[0150] In other words, it uses prior knowledge of relevant events to model a probability distribution that future events will occur. In the context of the networks described herein, this means that the probability that each component in the island is involved in or is the source of an island-level event is pre-computed or pre-estimated. The pre-computed probabilities are, in effect, probability distributions. When an event is observed, the model infers the probability that a particular component caused the event, and the pre-computed probability distribution is updated based on the inference and new events.

[0151] The training module estimates the prior probability of each component on the island causing an event (410). In general, the prior probability is a probability distribution over the observed variables. The prior probability can be a priori known, actual, or estimated probabilities based on previous event data.

[0152] A training module implemented on the server observes events on the islands (420). As mentioned above, observing the events involves collecting island-level data that defines the events. However, that data does not include the components to which the events should be assigned, which are the latent variables being modeled.

[0153] Using the available island-level event data and prior probability distributions, the training module performs a rigorous posterior analysis to determine the probability that the observed event was caused by each component on the island (430). The posterior inference is an estimate of an unknown quantity equal to the mode of the posterior distribution, which is the conditional probability assigned to a latent variable, e.g., an individual component within an island involved in an island-level event, after any relevant evidence or background has been taken into account.

[0154] The training module updates 440 the prior probabilities based on new event data and inference.

[0155] Steps 420-440 are repeated 450 until convergence, i.e., until the latent variable distributions change by less than some threshold amount when new events are observed.

[0156] Figure 5a depicts a one-dimensional example of island splitting and merging in a network. At time T=1, the network has three islands: A, B, and C. Island A is composed of nodes 1 and 2, island B is composed of node 3, and island C is composed of nodes 4 and 5.

[0157] At time T=2, due to an event between nodes 1 and 2 at time T=1, node 2 splits from node 1 and merges with node 3. As a result, a new island D is created from node 1 only, and a new island E is created from nodes 2 and 3. Node C remained the same.

[0158] At time T=3, due to an event between nodes 2 and 3 at time T=2, node 2 re-merges with node 1, and node 3 merges with nodes 4 and 5. As a result, island A is reformed, and nodes 3, 4, and 5 constitute island F, which is a new

[0159] At time T=4, due to an event at node 4 at time T=3, node 3 is split from node 4, causing all of the original islands A, B, and C to be reformed.

[0160] FIG. 5b depicts the hypothetical probability that an event occurs at each node at each time.

[0161] The present disclosure provides a computer system programmed to implement the methods of the present disclosure.

[0162] FIG. 6 illustrates a computer system 601 that is programmed or otherwise configured to implement the methods provided herein.

[0163] The computer system 601 can coordinate various aspects of the present disclosure, such as, for example, obtaining an inventory data set comprising a plurality of inventory variables, applying a trained algorithm to the inventory data set to generate a prediction of a variable having future uncertainty, and applying an optimization algorithm to the inventory data set to optimize a plurality of inventory variables. The computer system 601 can be a user's electronic device or a computer system located remotely relative to the electronic device. The electronic device can be a mobile electronic device.

[0164] The computer system 601 includes a central processing unit (CPU, also referred to herein as "processor" and "computer processor") 605, which may be a single-core or multi-core processor, or multiple processors for parallel processing. The computer system 601 also includes memory or memory locations 610 (e.g., random access memory, read-only memory, flash memory), an electronic storage unit 615 (e.g., hard disk), a communication interface 620 (e.g., network adapter) for communicating with one or more other systems, and peripheral devices 625, such as cache, other memory, data storage devices, and / or electronic display adapters. The memory 610, the storage unit 615, the interface 620, and the peripheral devices 625 communicate with the CPU 605 through a communication bus (solid lines), such as a motherboard. The storage unit 615 may be a data storage unit (or data repository) for storing data. The computer system 601 may be operatively coupled to a computer network ("network") 630 using the communication interface 620. The network 630 may be the Internet, an intranet and / or an extranet, or an intranet and / or an extranet in communication with the Internet.

[0165] Network 630 is, in some cases, a telecommunications and / or data network. Network 630 can include one or more computer servers, which may enable distributed computing, such as cloud computing. For example, one or more computer servers may enable cloud computing via network 630 ("cloud") to perform various aspects of the analysis, calculation, and generation of the present disclosure, such as, for example, obtaining an inventory data set comprising a plurality of inventory variables, applying a trained algorithm to the inventory data set to generate a forecast of a variable having future uncertainty, and applying an optimization algorithm to the inventory data set to optimize a plurality of inventory variables. Such cloud computing may be provided, for example, by cloud computing platforms such as Amazon Web Services (AWS), Microsoft Azure, Google Cloud Platform, and IBM Cloud. Network 630 may, in some cases, implement a peer-to-peer network, which may enable devices coupled to computer system 601 to behave as clients or servers with computer system 601.

[0166] CPU 605 can execute sequences of machine-readable instructions, which may be embodied in a program or software. The instructions may be stored in a memory location, such as memory 610. The instructions can be directed to CPU 605, which can subsequently program or otherwise configure CPU 605 to implement the methods of the present disclosure. Examples of operations performed by CPU 605 can include fetch, decode, execute, and writeback.

[0167] The CPU 605 may be part of a circuit, such as an integrated circuit. One or more of the other components of the system 601 may be included in the circuit. In some cases, the circuit is an application specific integrated circuit (ASIC).

[0168] The storage unit 615 can store files such as drivers, libraries, and saved programs. The storage unit 615 can store user data, such as user preferences and user programs. The computer system 601 can include one or more additional data storage units that are external to the computer system 601, such as located on a remote server that communicates with the computer system 601 through an intranet or the Internet in some cases.

[0169] Computer system 601 can communicate with one or more remote computer systems through network 630. For example, computer system 601 can communicate with a user's remote computer system. Examples of remote computer systems include a personal computer (e.g., a portable PC), a slate or tablet PC (e.g., Apple® iPad®, Samsung® Galaxy Tab), a phone, a smartphone (e.g., Apple® iPhone®, Android-enabled devices, Blackberry®), or a personal digital assistant. A user can access computer system 601 via network 630.

[0170] Methods as described herein may be implemented using machine (e.g., computer processor) executable code stored on electronic storage locations of computer system 601, such as, for example, on memory 610 or electronic storage unit 615. The machine executable or machine readable code may be provided in the form of software. During use, the code may be executed by processor 605. In some cases, the code may be read from storage unit 615 and stored on memory 610 for quick access by processor 605. In some circumstances, electronic storage unit 615 may be omitted and machine executable instructions are stored on memory 610.

[0171] The code can be pre-compiled and configured for use with a machine having a processor adapted to execute the code, or can be compiled during run-time. The code can be provided in a programming language that can be selected to allow the code to be executed in a pre-compiled or as-compiled manner.

[0172] Aspects of the systems and methods provided herein, such as computer system 601, can be embodied in programming. Various aspects of the technology can be considered as a "product" or "article of manufacture" typically in the form of machine (or processor) executable code and / or associated data carried on or embodied in some type of machine-readable medium. The machine executable code can be stored on an electronic storage unit, such as a memory (e.g., read-only memory, random access memory, flash memory) or a hard disk. A "storage" type medium can include any or all of the tangible memory of a computer, processor, or equivalent, or its associated modules, such as various semiconductor memories, tape drives, disk drives, and the like, that can provide non-transitory storage at any time for software programming. All or portions of the software may be communicated from time to time over the Internet or various other telecommunications networks. Such communication may, for example, enable loading of the software from one computer or processor to another, for example, from a management server or host computer to a computer platform of an application server. Thus, other types of media that may carry software elements include optical, electrical, and electromagnetic waves, such as those used across physical interfaces between local devices, through wired and optical fixed networks, and via various air links. Physical elements that carry such waves, such as wired or wireless links, optical links, or the like, may also be considered to be media carrying the software. As used herein, unless limited to non-transitory tangible "storage" media, terms such as computer or machine "readable medium" refer to any medium that participates in providing instructions to a processor for execution.

[0173] Thus, a machine-readable medium such as a computer executable code may take many forms, including, but not limited to, a tangible storage medium, a carrier wave medium, or a physical transmission medium. Non-volatile storage media include optical or magnetic disks, such as any of the storage devices in any computer or equivalent, such as may be used to implement, for example, the databases shown in the figures. Volatile storage media include dynamic memory, such as the main memory of such a computer platform. Tangible transmission media include coaxial cables, copper wire and optical fibers, including the wires that make up a bus in a computer system. Carrier wave transmission media may take the form of electric or electromagnetic signals, or acoustic or light waves, such as those generated during radio frequency (RF) and infrared (IR) data communications. Common forms of computer readable media thus include, for example, a floppy disk, a flexible disk, a hard disk, a magnetic tape, any other magnetic medium, a CD-ROM, a DVD or DVD-ROM, any other optical medium, punch cards, paper tape, any other physical storage medium with a pattern of holes, RAM, ROM, PROM and EPROM, FLASH-EPROM, any other memory chip or cartridge, a carrier wave that transmits data or instructions, a cable or link that transmits such a carrier wave, or any other medium from which a computer may read programming code and / or data. Many of these forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to a processor for execution.

[0174] The computer system 601 can include or communicate with an electronic display 635 that includes a user interface (UI) 640. Examples of user interfaces (UI) include, but are not limited to, graphical user interfaces (GUI) and web-based user interfaces. For example, the computer system can include a web-based dashboard (e.g., GUI) configured to display, for example, the BOM to a user.

[0175] The methods and systems of the present disclosure can be implemented using one or more algorithms. The algorithms can be implemented using software responsive to execution by the central processing unit 605. The algorithms can, for example, obtain an inventory data set comprising a plurality of inventory variables, apply a trained algorithm to the inventory data set to generate a prediction of a variable having future uncertainty, and apply an optimization algorithm to the inventory data set to optimize the plurality of inventory variables.

[0176] Although the description has been given with respect to certain embodiments thereof, these specific embodiments are merely illustrative and not limiting, and the concepts illustrated in the examples may be applied to other examples and implementations.

[0177] Although preferred embodiments of the present invention have been shown and described herein, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. The present invention is not intended to be limited by the specific examples provided herein. Although the present invention has been described with reference to the above specification, the description and illustration of the embodiments herein are not meant to be construed in a limiting sense. Numerous variations, changes, and substitutions will now occur to those skilled in the art without departing from the present invention. Furthermore, it should be understood that all aspects of the present invention are not limited to the specific depictions, configurations, or relative proportions described herein, which depend upon a variety of conditions and variables. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed in practicing the present invention. It is therefore contemplated that the present invention also covers any such alternatives, modifications, variations, or equivalents. The following claims define the scope of the present invention, and it is intended that methods and structures within the scope of these claims and their equivalents be covered thereby.

Claims

1. 1. A computer-implemented method for event allocation, comprising: Mapping a network comprising a plurality of islands, the plurality of islands being capable of dynamically changing by splitting or merging one or more of the plurality of islands, the network comprising an electricity distribution network, each of the plurality of islands comprising one or more components; using a discrete model to predict and / or detect and localize a component of the one or more components of the island that is responsible for an island event in the island; Including, the island event is associated with a component event, the component event (i) occurring in the component and (ii) causing the splitting or merging of the one or more islands; the discrete model is a machine learning latent variable model; the machine learning latent variable model comprises latent variables corresponding to the component events in the component; The method, wherein the machine learning latent variable model is configured to receive event data describing the island event associated with the component event, and process the event data based on an assumption that the latent variable corresponding to the component event caused the island event.

2. The method of claim 1 , wherein the machine learning latent variable model comprises a prior probability that the component causes the component event.

3. The method of claim 1 , wherein the dynamic changes to the islands include multiple instances of splitting, merging, or re-merging the one or more islands.

4. The method of claim 1 , wherein the one or more components include a plurality of components, and the dynamic change of the island includes splitting or merging a subset of the plurality of components.

5. The method of claim 4 , wherein the subset of components is the same size.

6. The method of claim 4 , wherein the subset of the plurality of components is of variable size.

7. The method of claim 2 , wherein the prior probabilities are iteratively updated using an expectation-maximization (EM) algorithm.

8. 10. The method of claim 1, further comprising: performing predictive maintenance at an individual component level based on the prediction and / or detection and localization of the component responsible for the island event.

9. The method of claim 1 , wherein the multiple islands correspond to multiple distribution feeders.

10. the one or more components of the island correspond to one or more breaker switches of a distribution feeder of the plurality of distribution feeders; The method of claim 9 , wherein the component event comprises an opening or closing of the one or more breaker switches in the power distribution network.

11. The method of claim 1 , wherein the components in the network are arranged in a geometric configuration selected from the group consisting of a one-dimensional configuration, a two-dimensional configuration, and an irregular configuration.

12. The method of claim 1 , wherein the components in the network are arranged in a two-dimensional configuration, the two-dimensional configuration being a rectangular configuration, a radial configuration, or a spoke and hub configuration.

13. The method of claim 1 , further comprising identifying the islands of the network based on the event data by mapping the network.

14. The method of claim 2 , wherein the machine learning latent variable model is configured to perform rigorous posterior inference based on the prior probabilities to assign the one or more local events to the individual components.

15. The method of claim 14 , wherein the assignment of the one or more local events to the individual components comprises determining a probability estimate that each local event will occur in each of the corresponding components.

16. 16. The method of claim 15, wherein the machine learning latent variable model is configured to use the probability estimates and iteratively update prior or prior probabilities until convergence.

17. The method of claim 2 , wherein the prior probabilities comprise initially known, actual, or estimated probabilities based on previous event data.

18. The method of claim 16 , wherein the prior or prior probabilities are iteratively updated using an expectation-maximization (EM) algorithm.

19. The method of claim 1 , further comprising: performing predictive maintenance at the individual component level based on the one or more detected local events.

20. The method of claim 1 , wherein the power distribution network comprises a plurality of power distribution feeders.

21. 21. The method of claim 20, wherein a plurality of nodes are associated with the plurality of distribution feeders, the plurality of nodes being connected via a plurality of branches corresponding to connections associated with each feeder.

22. 22. The method of claim 21 , wherein the one or more events comprise an opening or closing of one or more breaker switches in the power distribution network, the one or more breaker switches associated with one or more of the plurality of nodes and / or the plurality of branches.

23. The method of claim 1 , wherein the components in the network are arranged in a geometric configuration selected from the group consisting of a one-dimensional configuration, a two-dimensional configuration, and an irregular configuration.

24. The method of claim 1 , wherein the components in the network are arranged in a two-dimensional configuration, the two-dimensional configuration being a rectangular configuration, a radial configuration, or a spoke and hub configuration.

25. The method of claim 13 , wherein mapping the network includes identifying unexpected voltage levels in individual components of the network.

26. 1. A system for event allocation, comprising: a processor in a server in communication with a network comprising a plurality of islands, the network comprising a power distribution network; A memory in communication with the processor, the memory storing instructions that, when executed by the processor, cause the processor to: Mapping the network including a plurality of islands, the plurality of islands being capable of dynamically changing by splitting or merging one or more of the plurality of islands, each of the plurality of islands comprising one or more components; providing event data describing island events occurring in islands of the network to the discrete model; using the discrete model to predict and / or detect and localize one or more local events in individual components of the one or more components of the island that are responsible for the island event; A memory and Equipped with the island event is associated with a component event, the component event (i) occurring in the component and (ii) causing the splitting or merging of one or more of the plurality of islands; the discrete model is a machine learning latent variable model; the machine learning latent variable model comprises latent variables corresponding to the component events in the component; The system, wherein the machine learning latent variable model is configured to receive the event data describing the island event associated with the component event, and process the event data based on an assumption that the latent variable corresponding to the component event caused the island event.

27. A non-transitory computer readable medium having instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform an event allocation method, the method comprising: Mapping a network comprising a plurality of islands, the network comprising a power distribution network, the plurality of islands each comprising one or more components, the plurality of islands being capable of dynamically changing by splitting or merging one or more islands; providing event data describing island events occurring in islands of the network to the discrete model; using the discrete model to predict and / or detect and localize a component of the one or more components of the island that is responsible for the island event; Including, the island event is associated with a component event, the component event (i) occurring in the component and (ii) causing the splitting or merging of one or more of the plurality of islands; the discrete model is a machine learning latent variable model; the machine learning latent variable model comprises latent variables corresponding to the component events in the component; A non-transitory computer-readable medium, wherein the machine learning latent variable model is configured to receive the event data describing the island event associated with the component event, and process the event data based on an assumption that the latent variable corresponding to the component event caused the island event.

Images