Computer-implemented method for automatically monitoring a logistics network, monitoring system, computer program, and method for training a neural network

Abstract

The invention relates to a computer-implemented method for automatically monitoring a logistics network for handling postal items (1), having the following steps: (a) receiving or determining network properties of the logistics network, (b) receiving or determining at least one actual operating value of the logistics network, and (c) determining an anomaly value on the basis of the at least one actual operating value and the network properties.

Description

[0001] Computer-implemented method for automatically monitoring a logistics network, monitoring system, computer program and method for training a neural network

[0002] The invention relates to a computer-implemented method for automatically monitoring a logistics network for handling mail items, a monitoring system, a computer program and a method for training a neural network for anomaly detection in a logistics network for mail items.

[0003] A logistics network for mail items can be used to process mail items. For example, such logistics networks are provided in mail and parcel centers. Furthermore, a logistics network can also comprise several mail and parcel centers and, if necessary, other areas of the logistics chain up to a recipient. In operation, such a logistics network must process a large number of mail items in a very short time. Logistics networks can be individually designed and adapted to different areas of application. Different logistics networks can vary in structure and scope. The variation can depend, for example, on the respective provider or company operating the logistics network, the country in which the logistics network is located, the hierarchy of a system in which the logistics network is located, the size of the logistics network, and other factors.Accordingly, different logistics networks can each have different problem areas, such as critical connections between individual components of the logistics network or in inbound or outbound processes. An important aspect of logistics network operation can be monitoring the utilization and functionality of the logistics network and individual sections of it. Monitoring can include locating problems, such as backlogs, and resolving any identified issues.

[0004] Due to the diverse and individual design of various logistics networks, it is difficult to establish general monitoring rules, and monitoring must usually be adapted to the specific logistics network. Currently, logistics networks are therefore usually monitored individually by specially trained experts. These experts must consider complex interrelationships, including the respective configuration, properties, and parameters of the logistics network, as well as their analysis. Problems are identified, for example, by visually inspecting individual sections of the logistics network. This often requires a significant amount of manual work. To maintain an overview, status reports and dashboards are often used for manual monitoring, which are created and analyzed by the experts.A key criterion for effective and efficient monitoring is the experts' extensive experience with logistics networks in general, as well as with the specific logistics network in use. Furthermore, the logistics network can also be dynamic. More specifically, a logistics network can change based on external influences or changes (e.g., material flows). This makes it particularly difficult for a human to gain a complete overview of the logistics network.

[0005] The disadvantage is that newly commissioned logistics networks often have to be optimized step by step through experience gained during operation and that effective operation depends on the experience of the experts.

[0006] The object of the present invention is therefore to create a way to reduce the dependence on the experience of the operators of such logistics networks for efficient operation. It would also be desirable to be able to put new logistics networks into operation with fewer or more quickly overcome individual initial problems compared to the current, often common situation.

[0007] This object is achieved by a method having the features of claim 1, by a monitoring system having the features of claim 36, by a computer program having the features of claim 37, and by a method having the features of claim 38. Further features, advantages, and embodiments emerge from the dependent claims, the description, and the figures.

[0008] According to one aspect of the invention, a computer-implemented method for automatically monitoring a logistics network for handling mail items is provided. The method may comprise the following steps:

[0009] (a) Receiving or determining network properties of the logistics network,

[0010] (b) receiving or determining at least one actual operating value of the logistics network, and

[0011] (c) determining an anomaly value based on the at least one actual operating value and the network characteristics.

[0012] The method can be carried out, for example, by a control unit of a monitoring system. The control unit can be designed to carry out the method steps of the method. Furthermore, the control unit can be designed to receive and process required data and to send commands. Mailpieces can include mail items, parcels, letters, mailing envelopes, and the like. Network properties can include properties relating to a structure of the logistics network. The network properties can be transmitted from elements of the logistics network to the control unit. The logistics network can include a transport system designed to transport mailpieces. Furthermore, the logistics network can include a manipulation station for mailpieces. Here, the mailpieces can be gripped, rotated, sorted, stored, accelerated, and / or decelerated.Such manipulation stations can have controllers, actuators, and sensors. The at least one actual operating value can comprise current operating parameters of the logistics network. The actual operating value can comprise operating parameters of at least one manipulation station and / or at least one transport system of the logistics network. The at least one actual operating value can be determined by being recorded by a recording device. Additionally or alternatively, the actual operating value can also be obtained by a data query. The at least one actual operating value can be determined by being determined based on at least one other output value, in particular at least one current operating parameter and / or at least one operating parameter recorded in the past. For example, actual operating values ​​or output values ​​can be recorded by the recording device and sent to the control unit or the monitoring system.The detection device can be, for example, a camera, a scanner, or a tactile sensor. The detection device can be configured to detect the network properties and / or the at least one actual operating value. The anomaly value can be a binary value. For example, an anomaly value confirming an anomaly can be determined as soon as a maximum permitted load is exceeded at a point in the logistics network, e.g., at a node, an edge, or a route (see explanations below). For example, the anomaly value can include the value "a problem exists" or "a problem does not exist." Accordingly, the anomaly value can only assume two states (e.g., 1 or 0). A problem can be a defect, an overload, and / or incorrect identification of mailpieces. A problem can be, for example, a hardware problem, such as a defective motor, missing power supply, or defective transmission components.For example, an identification may be incorrect due to a defective or dirty barcode reader or a defective or dirty camera. A defect can also be dirt on a transport path, e.g. on a conveyor belt, or a blockage or backflow on a transport path. A defect can also be a control problem caused by a defective or faulty control system. A defect can be a defective insertion point. A insertion point is, in particular, a hardware element that aligns a mailpiece or places it in a designated location. A designated location can be, for example, a tray on a sorting unit. A insertion point can be defective, for example, if a light barrier is defective, so that mailpieces are inserted incorrectly, or if a motor is defective. An overload can, for example, be an occupied destination node or an occupied end point.The anomaly value can have a variety of anomaly levels. Additionally or alternatively, the anomaly value can have a local reference to a section, a single element, or a group of elements of the logistics network. For example, the anomaly value can have the local reference "relates to sorting item 1." Different types of anomaly values ​​can be provided for different elements or sections. Advantageously, by determining or receiving the network properties, the logistics network can be automatically recognized and, for example, its structure can be automatically determined.

[0013] By including at least one actual operating value when determining an anomaly value, automated problem detection can advantageously be enabled. Dependence on the individual knowledge of one or more experts can thus be reduced. The potentially centralized recording of network properties, which was previously not common in the prior art, enables rapid consideration and adaptation to the respective structure of a logistics network. This can be made possible, in particular, by considering both the actual operating value and the network properties when determining the anomaly value. The method according to the invention can advantageously be applied to many different logistics networks. For example, capacities or runtimes of parcel sorting systems or parcel networks can be monitored.

[0014] The network properties preferably comprise a topological structure of the logistics network. The topological structure can be indicative of the structure of the logistics network. The topological structure can define the location and optionally the type of individual elements, e.g., processing units, of the logistics network. The type of individual elements can designate how the elements handle the mail items, in particular how the mail items are processed at these elements. Handling in this sense can, for example, include scanning, separating, or sorting the mail items. The topological structure can be constructed analogously to a network topology as is known in the art. The topological structure can be a type of map of the logistics network. The topological structure can reflect not only the location of specific elements, but also properties of the elements.In addition, the topological structure can also reflect the operating characteristics of elements. This makes it advantageous to use the topological structure to automatically monitor a logistics network.

[0015] The topological structure preferably comprises edges and nodes. The nodes can be waypoints in the logistics network via which mailpieces can be transported. The points can represent an element in the logistics network at which the mailpieces are handled. The edges can be connections between the nodes, on which mailpieces can be transported between two nodes each. Provision can be made to identify the edges and nodes based on a detection of at least one mailpiece. For example, provision can be made to detect a passage of the at least one mailpiece through the logistics network and to infer the nodes and edges based on the path of the at least one mailpiece or to derive the nodes and edges therefrom. Edges and nodes can advantageously represent a simple way of structurally defining the logistics network.A monitoring system can be designed to automatically detect or infer the edges and nodes.

[0016] The logistics network preferably has at least one processing unit for processing mailpieces. The at least one processing unit can comprise an unloading unit, an identification unit, a conveying unit, a singling unit, a separating unit, an insertion unit, and / or a sorting unit. The at least one processing unit can be provided at a node. The unloading unit can be an exit point of the logistics network, at which a mailpiece can leave the logistics network or at which the mailpiece can be transferred to another section of the logistics network. The unloading unit can alternatively also be referred to as a drop-off point. The insertion unit can be an entry point of the logistics network, at which a mailpiece can be inserted into the logistics network or at which the mailpiece can be inserted from one section of the logistics network to another section of the logistics network.The insertion unit can alternatively also be referred to as a bullet point. The separation unit can in particular be designed to forward or be able to forward mail pieces in different directions. For example, the separation unit can be a separating switch. The separation unit can also be referred to as a diverter. The singling unit can be designed to create a predefined minimum distance between several mail pieces, in particular in the direction of a transport path of the mail pieces. The sorting unit can be designed to sort mail pieces according to defined criteria. For example, it can be provided to sort mail pieces according to a categorization of their type, e.g. parcel / letter or according to weight classes, and / or their destination. The sorting unit can, for example, be part of a sorting insert of the logistics network.The conveyor unit can be configured to enable transport of the mailpiece. The identification unit can be configured to identify mailpieces. The identification unit can, for example, be or comprise a scanning point and / or a detection device. For example, the identification unit can be configured to detect the presence or passage of a mailpiece. Additionally or alternatively, the identification unit can be configured to determine a type of mailpiece or information associated with the mailpiece. The information associated with the mailpiece can, for example, be an intended destination or a weight of the mailpiece. The monitoring system can be configured to use the identification unit to determine the network properties and / or the at least one actual operating value.The logistics network preferably comprises a source node and a destination node, with at least one edge being provided between the source node and the destination node. The logistics network can comprise a plurality of source nodes and / or a plurality of destination nodes. The source node and the destination node can be dynamically assigned to the at least one mail item. For example, the source node can be an insertion unit via which the mail item is inserted into the logistics network. Alternatively, the source node can be a current location of the mail item, which is dynamically adapted to the respective current location of the mail item. The destination node can be a designated starting point, for example an unloading unit, via which the mail item is to leave the logistics network and / or to which the mail item is to be transported.Within the scope of the method, it can be provided that the route of the mailpiece between a source node and a destination node is adjusted depending on the anomaly value. This can advantageously enable a flexible adjustment of the transport or transport route of the mailpiece to the at least one actual operating value.

[0017] Preferably, the network properties are determined automatically by a monitoring system. The monitoring system can be configured to determine the network properties via at least one detection device. Additionally or alternatively, the monitoring system can be configured to retrieve the network properties from a dispatch unit. Advantageously, this embodiment can provide a particularly simple and time-efficient way of determining the network properties. In particular, it can create independence from the expertise of relevant specialists.

[0018] The network properties are preferably determined by at least one detection device detecting at least one mailpiece that is processed through the logistics network. The at least one detection device can be arranged on the processing unit or can be designed to detect the at least one mailpiece at the processing unit. The at least one detection device can, for example, be arranged on the identification unit, on the insertion unit and / or on the unloading unit or can be designed to detect the at least one mailpiece at the respective unit. The monitoring system can be designed to derive the network properties from the detection of the at least one mailpiece. For example, the monitoring system can be designed to derive the network properties from a detected processing of the mailpiece, in particular comprising a transport route of the mailpiece.Advantageously, an already occurring processing of mail items can thus be used to determine required and / or useful information. Additionally or alternatively, it can be provided to determine or obtain the at least one actual operating value in a corresponding manner. Furthermore, at least two recording units can also be provided. The at least two recording units can record mail items at different points in the logistics network. This can advantageously make it possible to detect a material flow (e.g., of mail items) and then derive properties for edges from it.

[0019] Preferably, the network properties are recorded in real time while the at least one mail item is being processed in the logistics network. This advantageously enables monitoring during real-time operation of the logistics network, and thus, in particular, a particularly rapid response to errors or problems. Provision can be made to automatically adjust the processing in real time based on the determined anomaly value. In other words, the network properties can be obtained directly as the mail item passes through the logistics network.

[0020] Preferably, the at least one actual operating value of the logistics network relates to at least one mailpiece. It can be provided that the actual operating value is used to monitor the processing of the at least one mailpiece. In particular, the monitoring system can be designed to monitor the transport of the mailpiece and, if necessary, optimize it, particularly depending on the anomaly value. This enables particularly efficient monitoring and optimization of the processing of mailpieces by monitoring an individual mailpiece at a time and optimizing its processing if necessary.

[0021] Preferably, the anomaly value comprises a confirming anomaly value and a negative anomaly value, wherein a confirming anomaly value is indicative of the presence of an anomaly and a negative anomaly value is indicative of the absence of an anomaly. In other words, the anomaly value can comprise a digital statement (i.e., be indicative of only two statements). Advantageously, this can provide a relatively simple binary decision criterion. Provision can be made for initiating a response measure or issuing a command that initiates a response measure in the case of a confirming anomaly value. Provision can be made for not initiating a response measure or issuing a response measure according to which dispatch continues without adjustments in the case of a negative anomaly value.

[0022] Preferably, the at least one actual operating value comprises a number of mail pieces. A number of mail pieces can in particular be a shipment quantity. The number of mail pieces can relate to a specific location, for example a specific node, a specific edge and / or a handling unit. For example, the number of mail pieces can be a number of mail pieces that currently or in a specific time period occur or are handled at the specific location. The specific time period can be a time span. The time span can lie in the past or relate to a projected prediction for the future. Advantageously, a number of mail pieces can be used to detect overload, in particular local overload, in the logistics network.

[0023] Preferably, the at least one actual operating value comprises a transit time of the at least one mailpiece. The transit time can be a measure of how long the mailpiece is on a transport route. A transport route can in particular be the route between a source node and a destination node. Furthermore, the transit time can also be determined between other locations or edges in the logistics network. Provision can be made to determine the anomaly value by comparing the transit time with a threshold value. An anomaly value confirming an anomaly can be present if the threshold value for a maximum transit time is exceeded. Advantageously, it can thus be detected if one or more mailpieces are on the move for an unusually long time, which can be used to identify problems on the transport route.

[0024] The at least one actual operating value preferably comprises a characteristic of a route of at least one mailpiece from a source node to a destination node of the logistics network. The route can be defined by edges and nodes of the logistics network. The characteristic can be, for example, a shipment quantity on the route or a number of mailpieces on the route. The characteristic can be a number of mailpieces on a section of the route. The section can in particular be a node or an edge. The characteristic can be a current average transit time of mailpieces on the route or on a section of the route, e.g. on an edge. The current average transit time can be determined over a defined period of time. The defined period of time can be on the order of a few seconds, e.g. 30 seconds, a few minutes, e.g. 15 minutes, hours, days or weeks. Advantageously, this can thus be specified orThe respective parameter can be monitored specifically. The actual operating value can be configured to include several parameters.

[0025] According to one embodiment, the network properties comprise at least one operating value recorded in the past. In particular, it can be provided that, to determine the anomaly value, the actual operating value is compared with the at least one operating value recorded in the past. Advantageously, changes, in particular errors or problems, in the operating process can thus be detected relatively easily. Preferably, the network properties comprise a target operating value. It can be provided that, when determining the anomaly value, an absolute difference between the actual operating value and the target operating value is determined as the operating value difference. Preferably, the operating value difference can be compared with a predetermined minimum deviation, and the anomaly value can be indicative of whether or not the minimum deviation has been exceeded.Advantageously, unexpectedly large deviations from a norm defined by the target operating value can thus be detected effectively and relatively easily. The target operating value can, for example, be an average operating value determined over a specific period of time. The specific period of time can preferably lie before the time for determining the actual operating value. Provision can be made for the target operating value to be updated regularly according to a predetermined number of defined time intervals, which together make up the specific period of time. For example, a calculation of the target operating value can be based on an average value of several operating values ​​from several time intervals. As time progresses, a new interval can be added to the calculation based on current operating values. Optionally, provision can be made for older time intervals to no longer be taken into account when calculating the average value.In particular, an older interval can be eliminated for each additional interval. The actual operating value can correspond to the current or most recent interval. This advantageously makes it possible to detect when the actual operating value deviates from previous operating values. The minimum deviation can be a specified deviation tolerance from the target operating value, which the actual operating value should not exceed.

[0026] Preferably, a reaction measure can be implemented based on the anomaly value. The reaction measure can be designed, in particular, to correct an anomaly corresponding to the anomaly value. It can be provided that the reaction measure is only implemented if the anomaly value confirms an anomaly, or that the reaction measure "make no change" is implemented if the anomaly value negates an anomaly. Advantageously, this allows for an appropriate response to any detected problems.

[0027] According to one embodiment, several different actual operating values ​​are determined for at least one of the mailpieces, each of which relates to this mailpiece, and are compared with a target operating value. Thus, several aspects can advantageously be taken into account within the scope of monitoring.

[0028] The network properties preferably comprise at least one distance measure. The distance measure can in particular be recorded or have been recorded before the actual operating value. The distance measure can, for example, be an average running time or average throughput time of mail pieces. In this case, the distance measure can be indicative of a running time of a mail piece from a specific node or a specific edge to another node or another edge. A running time can in particular be a time that a mail piece is in transit in the logistics network for processing or the total time it is in transit between a source node and a destination node. A throughput time can be a time in which a mail piece is at a node or on an edge or between two consecutive nodes. The actual operating value can be a current operating value corresponding to the distance measure, e.g. a current running time or a current throughput time.In particular, it may be provided that the anomaly value is determined by comparing the distance measure with the actual operating value.

[0029] Preferably, the network properties include an average transit time of mail items recorded in the past. According to one embodiment, it can be provided that, upon receiving or determining network properties, transit times of mail items are recorded by a monitoring system, and wherein the average transit time is determined before the actual operating value is determined. Preferably, if the actual operating value deviates toward longer transit times above the average transit time, a confirmatory anomaly value is determined. The confirmatory anomaly value can, in particular, be confirmatory in such a way that the necessity of a response measure is confirmed.

[0030] According to one embodiment, the network properties comprise a distance measure for a plurality of available combinations of nodes in the logistics network. In other words, the distance measure can be determined for multiple routes, each comprising a different combination of nodes. As a reaction measure, alternative routes for the mailpiece can be calculated based on a comparison of different distance measures. Depending on the anomaly value, an optimized or at least improved route for the mailpiece can thus be determined.

[0031] According to one embodiment, the average transit time is recorded for a plurality of combinations of nodes in the logistics network. Advantageously, this can, for example, be used to determine or determine an alternative route from a plurality of recorded routes, in particular by using the plurality of combinations as options.

[0032] Preferably, in step (c) or in the step for determining an anomaly value, the actual operating value is examined for a deviation from a transport rule. The transport rule can, for example, be a standard runtime including a tolerance range, whereby the tolerance range must not be exceeded.

[0033] According to one embodiment, the transport rule states that a mail item should not be transported through the logistics network for longer than a predetermined maximum total throughput time. Advantageously, it can therefore be quickly detected if transport of the mail item or multiple mail items takes longer than expected, which can in particular indicate problems in the logistics network. The actual operating value preferably comprises a route of the mail items or of at least one of the mail items through the logistics network. For example, for a mail item or items, it can be recorded which nodes the mail item or items pass through. The route can be deduced from the nodes passed through. For example, the route can include information about a destination node and a source node. For example, it can be determined whether a route between the destination node and the source node has an unusual property.An unusual property can, for example, be unusual in that it deviates from previously recorded routes between the destination node and the source node. An unusual property can, for example, include an unusual sequence of nodes on the route between the destination node and the source node, compared to previously recorded routes. An unusual sequence can, for example, include an unusually high number of nodes passed between the destination node and the source node. An unusual sequence can, for example, include an unusual transit time, in particular a particularly long transit time, between the destination node and the source node. Preferably, the actual operating value comprises a total transit time of at least one mail item through the logistics network. For example, it can be provided that the total transit time is compared with a target transit time in order to determine the anomaly value.

[0034] Preferably, the transport rule stipulates that a mailpiece may not follow circular routes. Circular routes may occur if a problem or error prevents the mailpiece from being forwarded. A problem may be an overcrowded transport route or terminal, preventing the mailpiece from being transported further to a destination node. An error may be incorrect forwarding, particularly forwarding in the wrong (always the same) direction, e.g., at a separation unit or at a diverter. For example, a blocked destination node of a route or sub-route may result in one or more mailpieces being recirculated. A sub-route may, for example, be a circular route of a sorting unit. A blocked destination node, e.g., an exit point, of the sorting unit may result in recirculation on the circular route.As a result, mail items, in particular, may take longer to process through the sorting unit. Advantageously, this embodiment allows such a problem or error to be detected.

[0035] Preferably, the anomaly score is determined based on the number of nodes passed by a mail item. The transport rule may state that no node may occur twice within a route. This would be expected, for example, with a circular route. A duplicate node may indicate an inefficient route.

[0036] Preferably, the anomaly value is determined based on the type of node passed by a mailpiece. For example, passing a certain type of node too frequently, e.g., passing a singulation unit too frequently, may indicate an inefficient route. Advantageously, such an inefficient route can thus be identified.

[0037] The logistics network preferably comprises at least one anomaly detector, wherein the at least one anomaly detector is configured to determine an anomaly. Advantageously, the anomaly detector can be configured to automatically detect an anomaly value relating to a mailpiece at at least one of a node, an edge, and a route. Preferably, the actual operating value is transmitted to the at least one anomaly detector, wherein the at least one anomaly detector is configured to determine the anomaly value. An anomaly can be an incorrect or incorrect handling of a mailpiece. Advantageously, the anomaly detector can be a way to automatically determine an anomaly or the anomaly value. The anomaly detector can generally be configured to perform the determination of an anomaly value according to any of the embodiments described herein.

[0038] Preferably, the anomaly detector is configured to automatically determine a minimum deviation based on network characteristics and / or detection data from previous mailpieces. The anomaly detector checks whether the actual operating value exceeds the minimum deviation and, if an excess is detected, issues an excess signal. Detection data from previous mailpieces can be part of the network characteristics. For example, the detection data can be throughput times recorded in the past. Advantageously, the anomaly detector can be configured to detect anomalies based on historical experience from the logistics network.

[0039] The anomaly detector is preferably designed to use a threshold-based method to detect the minimum deviation. A threshold-based method can be a method according to which the anomaly value is determined by comparing the actual operating value with a threshold operating value or with a target operating value range. A target operating value range is a range within which the actual operating value should lie. In particular, an actual operating value that lies outside the target operating value range can be classified as an anomaly. The threshold operating value is in particular a threshold value that the actual operating value must not exceed or, alternatively, which the actual operating value must not fall below. Accordingly, depending on the embodiment, exceeding or falling below this value can be classified as an anomaly. Target operating value range orThe threshold operating value can, for example, be understood as a definition of a tolerated deviation from an ideal or normal value. The actual operating value can, in particular, be the transit time of at least one mail item.

[0040] Preferably, the at least one anomaly detector comprises a neural network trained with training data. The neural network can preferably be a deep neural network. A deep neural network is defined in particular by comprising a plurality of intermediate layers between an input layer and an output layer. The neural network can have been trained with training data comprising the network properties of the logistics network, in particular nodes and edges of the logistics network, and actual operating values, e.g., runtimes, as input. Reference data can be anomaly values ​​assigned to the respective input. The anomaly values ​​for training can, for example, be manually determined by logistics experts. Alternatively, archived logs from logistics networks can be used for training, for example. The neural network can optionally be trained to predict future behavior of actual operating values.For example, actual operating values ​​from the same mail items at different points in time in the past can be used as training data for this purpose. A neural network can be a particularly good way to achieve flexible adaptation to different logistics networks. Alternatively, the neural network can also be implemented as hardware, for example, with fixed connections on a chip or other computing unit. The computing unit that can execute the method according to the invention can be any computing unit such as a CPU (Central Processing Unit) or GPU (Graphics Processing Unit). The computing unit can be part of a computer, a cloud, a server, a mobile device such as a laptop, tablet computer, mobile phone, smartphone, etc.

[0041] Preferably, the training data includes annotated time points for anomalies. The annotation (i.e., the annotated time points) characterize the respective normal or anomalous status of a component in the logistics network (e.g., a plant component) with a so-called label (classification), which can be advantageously used for training machine learning processes (so-called supervised ML).

[0042] Preferably, the at least one anomaly detector is designed to use a cluster-based method to determine the anomaly value. Cluster-based methods can be unsupervised ML methods in which, for example, groups of similar data points are formed using feature vectors based on a distance measure. In this case, data points can also be assigned to clusters that are not clearly defined. In this case, one speaks of outliers or anomalies. This offers the advantage that results can also be generated in cases where ambiguous measured values ​​are available. In this case, an optimal result can be approximated by repeated repetitions. Preferably, the at least one anomaly detector is designed to use a statistical method to determine the anomaly value. For example, a variance of historical operating values ​​can be used to determine a target operating value range orto specify an upper or lower threshold. For example, the anomaly detector can be designed to continuously record operating values ​​and use the operating values ​​to determine a standard deviation or a predefined multiple of a standard deviation, and to classify an exceedance of the standard deviation as an anomaly.

[0043] In other words, statistics can be calculated based on the observed values ​​(e.g., mean, median, variance, standard deviation, distribution, etc.). These statistics can then be compared with the respective actual values ​​to define and detect an anomaly.

[0044] Preferably, the at least one anomaly detector comprises a cyclic detector configured to detect cyclic behavior of the at least one actual operating value during successive measurements by the cyclic detector. Cyclical behavior can occur when operating values ​​change cyclically or in a similarly recurring manner. For example, some operating values ​​may be typical for certain times of day or for different days of the week and therefore recur cyclically. These time-dependent patterns can thus advantageously be taken into account during the evaluation.

[0045] Preferably, the at least one anomaly detector is designed to monitor delivery times of mail items. In particular, the anomaly detector can be designed to monitor delivery times as generally described herein within the scope of the method. Preferably, adjacent nodes, edges and / or routes are aggregated in order to determine a starting point of the anomaly. Aggregating can mean that the adjacent nodes, edges and / or routes are summarized into sections. Provision can be made to determine anomaly values ​​for the summarized sections. If it is determined that a section has an anomaly, provision can be made to divide the section back into its individual parts, i.e. in particular into individual nodes, edges and / or routes, and to determine an anomaly value specifically for each individual part. For example, a throughput time through a section can be determined.If the runtime for this section is unusually long, the runtime can be determined individually for each node in this section to determine the exact cause of the longer runtime. Aggregating can advantageously simplify the problem and allow for the application of a less complex algorithm, for example, saving computation time.

[0046] Furthermore, recorded values ​​can also be aggregated within different time intervals. For example, all measurements within a 5-minute interval can be summarized to calculate anomalies based on the aggregate. This reduces the amount of data and provides efficient analysis.

[0047] The method preferably further comprises the step of performing at least one reaction measure based on the anomaly value. It can be provided that the reaction measure is only performed for some anomaly values, in particular for anomaly values ​​that indicate a problem, an error, or an anomaly. Anomaly values ​​that confirm normal operation generally do not require a reaction. In such a case, the reaction measure can aim to ensure that no change is necessary or that normal operation continues as before. Advantageously, the reaction measure can be used to react automatically to problems or anomalies, whereby faulty operation can be corrected more quickly. Preferably, the reaction measure comprises rerouting a mail item to an alternative route.This can be particularly advantageous if it is determined based on the actual operating value that an alternative route enables a shorter delivery time. Alternatively, it can be determined that a current route is overloaded, while the alternative route still has more capacity. The alternative route can be selected based on whether it enables a shorter throughput time for the mailpiece to the destination node than the planned route. The alternative route can be selected based on whether it has a lower planned mailpiece volume than the planned route and, optionally, whether the delivery time via the alternative route does not exceed a predetermined additional delivery time compared to the originally planned route. Advantageously, it can thus be achieved that the delivery time is automatically optimized in real time by being able to react to the current state of the logistics network.Preferably, for a current mailpiece, a check is carried out at a plurality of nodes of the logistics network, each of which comprises a separation unit, as the mailpiece passes through to determine whether there is an anomaly on the planned route. If an anomaly is detected, the mailpiece is rerouted as a reaction measure at the respective current node of the plurality of nodes at which the mailpiece is currently located. Preferably, the alternative route is calculated in accordance with the proviso that the transit time of the mailpiece is minimized. Provision can be made to minimize the transit time for a mailpiece based on the determined actual operating value. Preferably, the transit time of the mailpiece along the alternative route is calculated based on transit times of mailpieces in a predetermined period of the past.For example, an average transit time of mail items in a just-passed period of defined length can be determined in order to determine the transit time of a route or multiple routes. It can be provided that if there are no recorded transit times for a route in the determined period, an average transit time of the last known time interval in which a recorded transit time is available is used. Preferably, the alternative route is calculated in such a way that nodes or sub-routes from which no signals from mail items were received in a predetermined previous period are not part of the alternative route. It can happen that no mail items were traveling on these nodes or sub-routes during the period and therefore no signals were received. This can be because the corresponding nodes or sub-routes are experiencing a disruption. It can therefore be advantageous to disregard corresponding routes.Alternatively, it may preferably be provided to check whether nodes or sub-routes for which no mail item signals have been received are clear or if there is a disruption. If the route is clear, an average transit time from the last known time interval can be used, for example, by using a recorded transit time.

[0048] The actual operating value is preferably an estimated future load of nodes, edges and / or routes of the logistics network via which at least one piece of mail is to be transported or processed as planned. For example, a history of several known current pieces of mail on their respective routes can be projected into the future in order to enable a prediction or estimation of a future load. In other words, for example, information from past load volumes can be used to determine the actual operating value in order to estimate the volume of as yet unknown pieces of mail in the future. Analogous to determining an anomaly value based on the current status, it can be provided to estimate an anomaly value for the future. Advantageously, this allows any future problems such as backlogs, overflow or end point fill levels at the destination node to be identified at an early stage.Provision can be made to initiate a reaction measure before the problem has actually occurred. A reaction measure can, for example, be the rerouting of at least one mail item to an alternative route. Alternatively, a reaction measure can, for example, be the short-term provision of additional capacity, in particular transport capacity, e.g., additional nodes or edges. Preferably, the estimation of the future load can be carried out using a prediction model based on machine learning. A further aspect of the invention is a monitoring system comprising a control unit, wherein the control unit is configured to carry out a method as described herein for monitoring the logistics network. All advantages and features of the method for automatic monitoring can be transferred analogously to the monitoring system and vice versa.The monitoring system preferably comprises a recording device for recording operating values ​​of a logistics network, wherein the recording device is connected to the control unit and is configured to transmit recorded operating values ​​to the control unit. The recording device can, for example, be integrated into a node of the logistics network. The recording device can, for example, comprise a camera or a scanner, e.g., a barcode scanner. The monitoring system can preferably comprise an output interface connected to the control unit for outputting a response measure. The output interface can, for example, be connected to the nodes of the logistics network. The control unit can be configured to initiate a rerouting of mail items. The control unit can be configured to output a message to operating personnel or monitoring personnel of the logistics network.

[0049] A further aspect of the invention is a logistics network comprising a monitoring system as described herein. All advantages and features of the method for automatic monitoring and the monitoring system can be applied analogously to the logistics network, and vice versa.

[0050] According to a further aspect of the present invention, a computer program is provided, comprising instructions which, when the program is executed by a computing unit, cause the computing unit to carry out the method according to one of the preceding embodiments.

[0051] The invention also relates to a computer-readable medium containing instructions that, when executed by a computing unit, cause the computing unit to execute the inventive method, in particular the above-mentioned method or the further development method. Such a computer-readable medium can be any digital storage medium, for example a hard disk, a server, a cloud or a computer, an optical or magnetic digital storage medium, a CD-ROM, an SSD card, an SD card, a DVD, or a USB or other memory stick.

[0052] According to a further aspect of the present invention, a method for training an artificial neural network for anomaly detection in a logistics network for mail items is provided, comprising the steps of:

[0053] Receiving input training data, namely network properties of the logistics network and at least one actual operating value of the logistics network,

[0054] Receiving initial training data, namely at least one anomaly value. Training can involve training a neural network for the first time or retraining an existing neural network. Variables within the neural network can be adjusted so that input training data matches the initial training data. The variables can cause the transformation of the input data into the output data. Once the variables have been adjusted (i.e., the neural network has been trained), an anomaly value can be output when network properties of the logistics network are entered along with at least one actual operating state. A deployed neural network can also be further trained during operation by marking incorrect outputs as such and adjusting the neural network variables accordingly.

[0055] Individual features or embodiments can be combined with other features and other embodiments to form new embodiments. All configurations and advantages then also apply to the new embodiments. All advantages mentioned with regard to the method also apply analogously to the device, and vice versa. In particular, all advantages and features of the method for automatic monitoring, the monitoring system, the logistics network, and the computer program can be transferred analogously to the method for training a neural network, and vice versa.

[0056] Further advantages and features of the present invention will become apparent from the following description with reference to the figures. Individual features disclosed in the illustrated embodiments may also be used in other embodiments, unless expressly excluded. They show:

[0057] Fig. 1 is a flowchart of a method for automatically monitoring a logistics network for handling mail items according to an embodiment of the invention,

[0058] Fig. 2 shows a representation of a logistics network with a monitoring system according to an embodiment of the invention and

[0059] Fig. 3 is a schematic representation of a logistics network according to an embodiment of the invention.

[0060] Fig. 1 shows a flowchart of a method for automatically monitoring a logistics network for handling mail items 1 according to one embodiment of the invention. The method can be carried out, for example, by a central control unit 20 (see Fig. 2) or a central monitoring unit. In a first step 101, network properties of the logistics network are received or determined. The network properties can preferably comprise a topological structure. The topological structure can comprise nodes 5 and edges 6 (see Fig. 2), wherein the edges 6 are in particular connections between nodes 5. It can be provided to define a source node 7 and a destination node 8 for each edge 6, wherein the source node 7 can in particular be a starting point of a route of a mail item 1 to be transported and the destination node 8 can in particular be a destination point of the route of the transported mail item 1.The topological structure can comprise topological properties, such as a recorded shipment quantity, a distance measure or a recorded time measurement of delivery times of the mail items 1. The topological properties can be recorded, for example, using one or more recording devices. The one or more recording devices can be arranged, for example, at recording points of a sorting system in the logistics network. The recording devices can be arranged, for example, at scanning, insertion or drop points or also at other points in the logistics network. Provision can be made to record the topological properties in real time. For example, time units or time intervals can be defined and the topological properties are recorded for each time unit or each time interval.In addition, for at least one mail item 1, preferably for several or each mail item 1, its route through the logistics network can be recorded. In particular, various possible routes, preferably all possible routes on which mail items 1 are transported, can thus be recorded. In particular, the topological properties on these routes can also be recorded and optionally assigned to the routes. In this step, operating values ​​can also be determined over a certain period of time, in particular continuously, wherein the operating values ​​can, for example, include topological properties. Target operating values ​​can be determined from the recorded operating values, for example by calculating averages of operating values ​​over a certain period of time, e.g. over a week.

[0061] In a further step 102, at least one actual operating value of the logistics network is received or determined. An actual operating value is in particular an operating value that is present or determined at the respective current point in time. The actual operating value can, for example, be a currently determined delivery time of a mail item 1. The actual operating value can alternatively or additionally also include a route intended for a current mail item 1. In a modified or additional variant of this step 103, the actual operating value can also be an estimate of the future based on current values, e.g. based on current operating values. In particular, a load of nodes 5, edges 6 and / or routes of the logistics network in the future can be estimated. In a further step 104, an anomaly value is determined based on the at least one actual operating value and the network properties. In a simple embodiment, the anomaly value can, for example,B. assume the two values ​​"anomaly exists" and "anomaly does not exist". This can be done, for example, by comparing the actual operating value with the target operating value, whereby the target operating value can be used in particular to define a threshold value, exceeding which is automatically assessed as an anomaly by the actual operating value. This step can be carried out, for example, by an anomaly detector. Alternatively or additionally, the actual operating value can be compared with a defined rule. For example, a rule can state that mail items 1 should not take circular routes, i.e. in particular no routes where individual nodes 5 are passed through more than once. A violation of such a rule can automatically be assessed as an anomaly.

[0062] In an optional further step 104, it may be provided to implement at least one reactive measure based on the anomaly value. The reactive measure may, in particular, be designed to correct a detected anomaly.

[0063] Fig. 2 shows a representation of a logistics network with a monitoring system according to an embodiment of the invention. The monitoring system comprises a control unit 20 which is configured to carry out a method for automatically monitoring a logistics network for processing mailpieces 1 as described herein. The logistics network comprises a plurality of edges 6 and nodes 5 for processing the mailpieces 1. The nodes 5 can, for example, be a sorter (i.e. sorting unit) or sub-points in the sorter such as drop-off points 10 or identification units 11. Furthermore, the nodes 5 can comprise insertion units 12 and unloading units 13. The edges 6 are the connections between the various nodes 5. In particular, the edges 6 are represented by conveyors in this example.At the various nodes 5, detection devices (not shown) are arranged, which can detect the mail items 1 and operating values ​​as they pass through. The detection devices are connected to the control unit 20 and transmit detected operating values ​​or information from which operating values ​​can be derived to the control unit 20. The control unit 20 is particularly designed to initiate a response measure when an anomaly is detected.

[0064] In a further embodiment not shown, the logistics network comprises one or more branches (e.g., flow splitters) or transfer points of multiple circular sorters and redundant routes. Analogous to the above embodiment, the points can again be defined at stations (e.g., branches, transfer points, etc.) or at subpoints of a station. The edges can then connect the points.

[0065] Fig. 3 shows a schematic representation of a logistics network according to one embodiment of the invention. Between a source node 7 and a destination node 8, there are several possible routes that a mail item 1 can take to reach the destination node 8 via several nodes 5 and edges 6.

[0066] List of reference symbols:

[0067] I Postal items

[0068] 5 knots

[0069] 6 edge

[0070] 7 source nodes

[0071] 8 target nodes

[0072] 10 Drop-off point

[0073] II Identification Unit

[0074] 12 insertion unit

[0075] 13 Unloading unit

[0076] 20 Control unit

Claims

Claims 1 . A computer-implemented method for automatically monitoring a logistics network for handling mail items (1), comprising the following steps: (a) Receiving or determining network properties of the logistics network, (b) receiving or determining at least one actual operating value of the logistics network, and (c) determining an anomaly value based on the at least one actual operating value and the network characteristics.

2. The method according to claim 1, wherein the network properties comprise a topological structure of the logistics network.

3. The method according to claim 2, wherein the topological structure comprises edges (6) and nodes (5).

4. Method according to one of the preceding claims, wherein the logistics network has at least one processing unit for processing mail items (1).

5. Method according to one of the preceding claims, wherein the logistics network comprises a source node (7) and a destination node (8), wherein at least one edge (6) is provided between the source node (7) and the destination node (8).

6. Method according to one of the preceding claims, wherein the network properties are determined automatically by a monitoring system.

7. Method according to one of the preceding claims, wherein the network properties are determined by at least one detection device detecting at least one mail item (1) which is processed through the logistics network.

8. Method according to one of the preceding claims, wherein the network properties are detected in real time while the at least one mail item (1) is being processed in the logistics network.

9. Method according to one of the preceding claims, wherein the at least one actual operating value of the logistics network relates to at least one mail item (1) and in particular comprises a route of the mail items (1) through the logistics network.

10. The method according to any one of the preceding claims, wherein the anomaly value comprises a confirming anomaly value and a negating anomaly value, wherein a confirming anomaly value is indicative of a presence of an anomaly and a negating anomaly value is indicative of a non-presence of an anomaly.

11. Method according to one of the preceding claims, wherein the at least one actual operating value comprises a number of mail items (1).

12. Method according to one of the preceding claims, wherein the at least one actual operating value comprises a transit time of the at least one mail item (1).

13. Method according to one of the preceding claims, wherein the at least one actual operating value comprises a characteristic of a route of at least one mail item (1) from a source node (7) to a destination node (8) of the logistics network.

14. The method according to any one of the preceding claims, wherein the network properties comprise at least one operating value recorded in the past.

15. The method according to any one of the preceding claims, wherein the network properties comprise a target operating value.

16. The method according to any one of the preceding claims, wherein a response action is performed based on the anomaly value.

17. The method according to any one of the preceding claims, wherein the network properties comprise at least one distance measure.

18. Method according to one of the preceding claims, wherein the network properties comprise an average transit time of mail items (1) recorded in the past.

19. Method according to one of the preceding claims, wherein a confirming anomaly value is determined in the event of a deviation of the actual operating value towards longer running times above the average running time.

20. Method according to one of the preceding claims, wherein for at least one of the mail items (1) a plurality of different actual operating values are determined which relate to this mail item (1) and are each compared with a target operating value.

21. Method according to one of the preceding claims, wherein in step (c) the actual operating value is examined for a deviation from a transport rule.

22. The method according to claim 21, wherein the transport rule comprises that a mail item (1) should not be transported through the logistics network for longer than a predetermined maximum total transit time.

23. The method according to claim 21 or 22, wherein the transport rule includes that a mail item (1) may not traverse circular routes.

24. Method according to one of the preceding claims, wherein the anomaly value is determined based on a number of nodes (5) passed by a mail item (1).

25. The method according to any one of the preceding claims, wherein the anomaly value is determined based on a type of node (5) passed by a mail item (1).

26. The method according to any one of the preceding claims, wherein the logistics network comprises at least one anomaly detector, wherein the at least one anomaly detector is configured to determine an anomaly.

27. The method according to claim 26, wherein the anomaly detector is configured to automatically determine a minimum deviation based on network properties and / or detection data from previous mail pieces (1), wherein the anomaly detector checks whether the actual operating value exceeds the minimum deviation and outputs an exceedance signal if an exceedance is detected.

28. The method according to claim 26 or 27, wherein the anomaly detector is configured to use a threshold-based method to detect the minimum deviation.

29. The method according to any one of claims 26 to 28, wherein the at least one anomaly detector comprises a neural network trained with training data.

30. The method of claim 29, wherein the training data comprises annotated times for anomalies.

31. The method according to any one of claims 26 to 30, wherein the at least one anomaly detector is configured to use a cluster-based method to determine the anomaly value.

32. The method according to any one of claims 26 to 31, wherein the at least one anomaly detector is configured to use a statistical method to determine the anomaly value.

33. The method according to any one of claims 26 to 32, wherein the at least one anomaly detector comprises a cyclic detector configured to detect a cyclic behavior of the at least one actual operating value during successive measurements by the cyclic detector.

34. The method according to any one of the preceding claims, wherein the method further comprises the step of: performing at least one response action based on the anomaly value.

35. The method according to claim 34, wherein the response action comprises rerouting a mail item (1) to an alternative route.

36. Monitoring system comprising a control unit (20), wherein the control unit (20) is configured to carry out a method according to one of the preceding claims for monitoring the logistics network.

37. A computer program comprising instructions which, when the program is executed by a computing unit, cause the computing unit to carry out the method according to any one of claims 1 to 35.

38. A method for training an artificial neural network for anomaly detection in a logistics network for mail items (1), comprising the steps: Receiving input training data, namely network properties of the logistics network and at least one actual operating value of the logistics network, Receiving initial training data, namely at least one anomaly value.

Images