Method for computer-aided determination of one or more operating configurations of a conveyor system for unit loads

Abstract

The invention relates to a method for the computer-aided determination of one or more operating configurations of a conveyor system (100) for unit loads (P1, P2, ..., P7), wherein the conveyor system (100) comprises a plurality of partial conveyor sections (C1, C2, ..., C5) along which the unit loads (P1, P2, ..., P7) are transported on respective conveying means (1, 2, ..., 5) of the partial conveyor sections (C1, C2, ..., C5), wherein each partial conveyor section (C1, C2, ..., C5) is assigned a speed (v1, v2, ..., v5) of the conveying means (1, 2, ..., 5) of the respective partial conveyor section (C1, C2, ..., C5), wherein to determine a respective operating configuration: - the velocities (v1, v2, ..., v5) of all sub-conveyor sections (C1, C2, ..., C5) are modeled by a set of parameters (PS) such that the velocity (v1, v2, ..., v5) of a respective sub-conveyor section (C1, C2, ..., C5) is represented by a time-dependent periodic function (F1, F2, ..., FN) comprising one or more parameters (a i , b i , c i , d i ) is described; - parameter values ​​optimized for the respective operating configuration using an optimization procedure (OPT) opt ) for the set of parameters (PS), wherein the optimization procedure (OPT) is based on an optimization goal of a high quality measure (FT) and within the optimization procedure (OPT) the quality measure (FT) is repeatedly determined by means of a physical simulation (SIM) of the conveyor system (100) or by means of a function.

Description

[0001] The invention relates to a method for the computer-aided determination of one or more operating configurations of a conveyor system for unit loads.

[0002] In industrial facilities, such as logistics systems, conveyor systems for handling unit loads are regularly used. These systems transport unit loads, such as packages in a parcel sorting facility. A conveyor system comprises numerous individual conveyor sections with corresponding conveying units on which the unit loads are transported. The speeds of the individual conveying units are individually adjustable.

[0003] Traditionally, a control system based on expert knowledge is used to adjust the appropriate speeds of the conveying elements in a conveyor system. This system also takes into account sensor readings acquired during operation. The sensors might, for example, include multiple light beams that detect when individual items are located on a section of the conveyor.

[0004] Conventional conveyor system controls are designed to transport as many individual items as possible through the conveyor system in a short time. Nevertheless, optimal operation of the conveyor system is not always guaranteed. This is partly because the sensor readings recorded during operation often do not provide a reliable indication of optimal conveyor system performance.

[0005] The object of the invention is to determine one or more operating configurations of a conveyor system for unit loads using a computer-aided method without taking sensor values ​​into account.

[0006] This problem is solved by the method according to claim 1. Further developments of the invention are defined in the dependent claims.

[0007] The method according to the invention serves for the computer-aided determination of one or more operating configurations of a conveyor system for unit loads. The conveyor system comprises a plurality of successive sub-conveyor sections along which the unit loads are transported from an input to an output of the conveyor system on respective conveying elements (such as conveyor belts) of the sub-conveyor sections. Each sub-conveyor section is assigned an individually adjustable speed at which the conveying element of the respective sub-conveyor section is driven.

[0008] The successive conveyor sections do not necessarily have to connect directly to one another. For example, it is also possible that a chute without a conveying system is provided between two adjacent conveyor sections. Such chutes are often used in conveyor systems whose purpose is, among other things, to separate the transported goods so that there are no, or as few as possible, stacks of goods at the conveyor system's exit.

[0009] In the method according to the invention, the velocities of all partial conveying sections are modeled by a set of parameters to determine a respective operating configuration. The modeling is designed such that the velocity of a respective partial conveying section is described by a time-dependent periodic function comprising one or more parameters, wherein the parameter(s) belong to the set of parameters. In other words, the set of parameters comprises all parameters of the time-dependent periodic functions that model the velocities of the partial conveying sections.

[0010] According to the invention, optimized parameter values ​​for a given set of parameters are determined using an optimization method for each operating configuration, wherein the optimization method is based on the optimization goal of a high performance metric. Within the framework of the optimization method, the performance metric is repeatedly determined as a function of parameter values ​​from the set of parameters, either by means of a physical simulation of the conveyor system or by means of a function that models a physical simulation of the conveyor system.

[0011] The quality measure is determined in advance with regard to the intended purpose of the conveyor system. Determining a suitable quality measure falls within the scope of expert knowledge. Examples of quality measures are given below. The physical simulation used in the method according to the invention is based on a simulation model or simulation program that was created beforehand and with which the conveyor system can be simulated based on physical laws. Creating a suitable simulation program for a conveyor system is known per se or falls within the scope of expert knowledge. For example, the well-known development environment Unity can be used. ® to create a suitable simulation program for carrying out corresponding physical simulations of a conveyor system.

[0012] The inventive method is based on the finding that the operation of a conveyor system can be very well modeled by a periodic change in the speeds of the conveying means of the partial conveyor sections.

[0013] In this process, optimized parameters for corresponding periodic functions are determined using a simple optimization procedure, taking a quality measure into account, without incorporating current sensor values. The conveyor system can then be controlled based on these parameterized periodic functions by adjusting the conveying speeds according to these parameters.

[0014] In a preferred embodiment of the method according to the invention, the optimization method is initialized with a plurality of value instances of the set of parameters, wherein the value instances each define parameter values ​​for the set of parameters and the parameter values ​​of each value instance are at least partially different from the parameter values ​​of every other value instance in the plurality of value instances. Thus, a plurality of parameterizations for the set of parameters is initially specified. The optimization method then varies the parameter values ​​of the plurality of value instances depending on the optimization goal in order to find a value instance with the highest possible performance level.

[0015] In a preferred embodiment, the optimization method is a gradient-free optimization method, i.e., an optimization method in which the optimization does not depend on the gradient of a function to be optimized. Any gradient-free optimization method known from the prior art can be used. In a preferred variant, the gradient-free optimization method used is particle swarm optimization. However, a known evolutionary optimization method, such as CMA-ES optimization (CMA-ES = Covariance Matrix Adaptation Evolution Strategy), can also be used.

[0016] Instead of a gradient-free optimization method, a gradient-based optimization method is used in a further embodiment of the invention, wherein the gradient-based optimization method uses the function that models the physical simulation of the conveyor system to determine the gradient. In this case, this function is configured to be differentiable.

[0017] The function that models the physical simulation of the conveyor system can be derived from the physical simulation in various ways. Preferably, the function is a neural network that has been trained on data derived from the physical simulation.

[0018] As mentioned above, the quality measure considered in the optimization process can be selected differently depending on the intended use of the conveyor system. In one variant, the quality measure is designed such that it is higher the fewer stacks of individual items leave the conveyor system's output in a predetermined time interval during the physical simulation, provided that each stack of items contains at least two items stacked on top of each other. This quality measure is used in conveyor systems where individual items are to be separated.

[0019] In another variant, the quality measure is designed such that it is higher the less frequently, within a predetermined time interval in the physical simulation, the event occurs in which a unit load rotates at least once around its axis during transport along the partial conveyor sections. This quality measure ensures that damage to the transported unit loads is kept to a minimum. This quality measure is preferably also used in a conveyor system for singulating unit loads.

[0020] In another variant, the quality measure is designed such that it increases with the number of individual items leaving the conveyor system's output within a predetermined time interval in the physical simulation. This quality measure ensures a high throughput of individual items through the conveyor system.

[0021] The previously described variants of quality measures can also be combined if necessary. In other words, a single quality measure can also realize several of the functional relationships defined above.

[0022] In a further preferred embodiment of the method according to the invention, for at least one partial conveying section and preferably each partial conveying section, the parameter(s) of the time-dependent periodic function include the period of the function and / or the amplitude of the function and / or a time offset in the function and / or a velocity offset in the function.

[0023] The time-dependent periodic function for the individual conveying sections can be defined differently depending on the design. In one variant, the time-dependent periodic function for at least one, and preferably every, individual conveying section is a sine function. However, the time-dependent periodic function can also be another type of periodic function, such as a sawtooth function.

[0024] In a further preferred embodiment of the method according to the invention, several operating configurations are determined using a computer, each operating configuration containing a different operating setpoint for the conveyor system. The operating setpoint for each section of the conveyor is contained in the time-dependent periodic function, preferably in combination with another parameter from the set of parameters. In this way, predefined operating specifications for the conveyor system can also be taken into account when determining the corresponding operating configurations. For example, the operating setpoint can define a target value for the number of individual items that leave the conveyor system's output within a predetermined time interval. However, the operating setpoint can also be defined in any other way.

[0025] In a further preferred embodiment of the method according to the invention, the determined operating configuration(s) are stored in a control unit of the conveyor system and the conveyor system is operated by means of the control unit based on the determined operating configuration or on one of the determined operating configurations.

[0026] According to the variant just described, a method for operating a conveyor system for unit loads is thus placed under protection, wherein the operating configuration(s) used for operation are determined with the method according to the invention or one or more of the variants of the method according to the invention described above.

[0027] The invention further relates to a device for data processing, comprising means for carrying out the method according to the invention or one or more preferred variants of the method according to the invention.

[0028] Furthermore, the invention relates to a computer program product with a program code stored on a machine-readable medium for carrying out the method according to the invention or one or more preferred variants of the method according to the invention when the program code is executed on a computer.

[0029] Furthermore, the invention relates to a computer program with program code for carrying out the method according to the invention or one or more preferred variants of the method according to the invention when the program code is executed on a computer.

[0030] Exemplary embodiments of the invention are described in detail below with reference to the accompanying figures.

[0031] They show: Fig. 1 a schematic representation of a conveyor system for which an operating configuration can be determined according to an embodiment of the invention; Fig. 2 a diagram illustrating the steps of an embodiment of the method according to the invention; and Fig. 3 the representation of a time-dependent periodic function, the parameters of which are determined using an embodiment of the invention.

[0032] Fig. Figure 1 shows a schematic side view illustrating an example of a conveyor system 100, the operating configuration of which can be determined using the embodiment of the method according to the invention described below. The conveyor system is controlled by a control unit CO and comprises a total of five sub-conveyor sections C1, C2, C3, C4, and C5. Each sub-conveyor section includes a conveying element in the form of a conveyor belt, which can be moved at an individually adjustable speed. Specifically, sub-conveyor section C1 includes conveyor belt 1, which moves at speed v1; sub-conveyor section C2 includes conveyor belt 2, which moves at speed v2; sub-conveyor section C3 includes conveyor belt 3, which moves at speed v3; sub-conveyor section C4 includes conveyor belt 4, which moves at speed v4; and sub-conveyor section C5 includes conveyor belt 5, which moves at speed v5.Between the partial conveyor section C1 and the lower-lying partial conveyor section C2 there is also a slide R.

[0033] The conveyor system 100 will move individual items in the form of packages in the transport direction T from an input IN to an output OU. Examples are shown in Fig. A total of seven individual items or packages, P1 to P7, are shown. The conveyor system is a so-called layering line, which is frequently used in logistics systems and aims to separate the individual items so that they arrive at output OU individually and are not stacked on top of each other. The separation of the individual items is achieved by the partially inclined conveyor sections C1 to C5 and the chute R.

[0034] Packages arrive at the entrance of the conveyor system, some of them stacked on top of each other. Examples include... Fig. Figure 1 indicates two stacks of packages, one comprising packages P1 to P3 and located on conveyor belt 1, while the other comprising packages P4 and P5 is sliding down chute R. By setting different speeds for the individual conveyor sections C1 to C5, the singulation of the packages can be supported by corresponding decelerations or accelerations between the conveyor sections.

[0035] In the embodiment described here, the aim is to adjust the speeds of the individual conveyor belts so that as few package stacks as possible appear at the output (OU) of the conveyor system. Unlike conventional methods, the speed adjustment of the conveyor belts does not rely on current sensor readings. Instead, the time-varying speeds of the individual conveyor belts are determined in advance as the operating configuration and then used during operation of the conveyor system, without relying on any sensors. The pre-determined operating configuration is stored in the control unit (CO) so that it is available during operation of the conveyor system. Optionally, the control unit (CO) can also have its own computational mechanism for determining the operating configuration.

[0036] Fig. Figure 2 illustrates the steps carried out in one embodiment of the method according to the invention. These steps are described below in general terms for a conveyor system comprising N sub-conveyor sections, the conveyor belts of which operate at corresponding speeds v. i (i = 1, ..., N) can be moved. The method can be used for the conveyor system of the Fig. 1 can be used.

[0037] According to step S1 of the Fig. 2. The individual velocities v1, ..., v N The conveyor belts of the respective sub-conveyor sections are modeled by time-dependent periodic functions F1, ..., FN. This approach utilizes the understanding that periodic changes in the individual conveyor belt speeds cause accelerations and decelerations between the conveyor belts, leading to the singulation of stacked packages.

[0038] Each periodic function Fi of the individual velocities v i In the embodiment described here, the parameter set θ consists of four parameters a. i , b i , c i and d i described. The periodic function is a sine function, whereby the method according to the invention can also be used for any other periodic functions, such as a sawtooth function.

[0039] Fig. Figure 3 illustrates the sine function used in the embodiment described here to model a corresponding velocity v. i of a conveyor belt. Along the abscissa of the Fig. Figure 3 represents time t, while the ordinate represents the corresponding velocity v. i represented. Mathematically, the velocity v can be expressed as... i out of Fig. 3. Describe as follows: vi(t)=ai∗sin(bi∗t+ci)+di

[0040] The size a i represents the amplitude of the sine function, the quantity b i the period of the sine function, the quantity c i a time offset in the sine function and the size d i a velocity offset in the sine function. These quantities are also shown again in the representation of the Fig. 3 illustrates this.

[0041] By modeling the individual velocities v i with the corresponding parameters a i , b i , c i and d i From a sine function, one obtains a set of parameters, containing four parameters for each velocity. This set of parameters is represented by PS in Fig. 2 is designated.

[0042] Based on the velocities modeled by the parameter set PS, in step S2 the Fig. 2. An optimization OPT was performed. This optimization utilizes a physical simulation (SIM) of the conveyor system. The corresponding computer program that performs the physical simulation was created beforehand. Therefore, the creation of the computer program is not part of the method according to the invention. Creating such a computer program for a conveyor system is known in the art or falls within the scope of what is considered standard practice. For example, the well-known development environment Unity can be used for this purpose. ® can be used.

[0043] The SIM simulation outputs a corresponding quality measure FT (fitness value). This quality measure is appropriately defined depending on the technical purpose of the conveyor system and represents an optimization target for the conveyor line system. For the delaying line of the Fig. The quality measure is defined such that it increases with the number of stacked packages at output OU, which is determined via the simulation SIM. This quality measure takes into account the intended purpose of the conveyor system, namely the singulation of the packages. The appropriate definition of the quality measure falls within the scope of expert judgment. In a modified embodiment, the quality measure is determined by a function that models the aforementioned physical simulation. This function could, for example, be a neural network trained on data derived from the physical simulation.

[0044] Optimization OPT in step S2 of the Fig. 2. Any gradient-free optimization method can be used. Gradient-free means that the optimization does not depend on the gradient of the function being optimized. In a preferred variant, a known particle swarm optimization method is used. Alternatively, other known gradient-free optimization methods, such as evolutionary optimization algorithms, can also be employed. In a modified variant, a gradient-based method can also be used as the optimization OPT. In this case, the goodness-of-fit measure is determined via the function mentioned above, which models the physical simulation and is designed as a differentiable function.

[0045] As part of the implementation of step S2 of the Fig. 2. The optimization OPT is first initialized with a multitude of value instances of the parameter set PS. In other words, a multitude of parameterizations of the parameter set PS with different parameter values ​​are initially generated. Each value instance, or parameterization, represents a particle of the swarm when using particle swarm optimization. Subsequently, each parameterization is evaluated using the simulation SIM; that is, the corresponding conveyor system is simulated based on the respective parameterizations. As a result, the corresponding performance measure FT is obtained for each parameterization. If necessary, the simulation can be run multiple times for each parameterization, in which case the performance measure is the average of the performance measures across all simulations.

[0046] Using the performance measure FT, the optimization procedure OPT is then used to search for parameterizations with the highest possible performance measure. This search is iterative, with the performance measures for suitably modified parameterizations being determined in each iteration. Since well-known optimization procedures are used, a description of these procedures is omitted. The optimization procedure is then terminated based on a suitable termination criterion. This criterion could, for example, be that the highest overall performance measure among all parameterizations does not change significantly after a certain number of optimization steps.

[0047] The result of step S2 is an optimized parameterization PS. OPT in the form of corresponding values ​​for the parameters a i , b i , c i and d i all speeds v iof the conveyor belts of the conveyor system. The optimized parameterization PS OPT represents a corresponding operating configuration of the conveyor system, which ensures optimized operation of the system with regard to quality.

[0048] In the case of conveyor system 100 of the Fig. 1. The optimized parameterization PS OPT The CO control unit stores the data. The CO control unit then sets the individual speeds of the conveyor belts of the partial conveyor sections based on the time-dependent sine functions with the optimized parameter values.

[0049] In a further development of the method described here, it is also possible to determine several different operating configurations for the corresponding conveyor system. The respective operating configurations are defined for different operating parameters of the conveyor system. These operating parameters are incorporated into the modeling of the periodic functions that describe the conveyor belt speeds and are preferably included in combination with an additional parameter of the periodic function. For example, a target value for the number of individual items leaving the conveyor system's output within a predetermined time interval can be defined as the operating parameter.

[0050] The following is an example function that takes an operating specification into account: vneui(t)=vi(t)+ei∗speed=ai∗sin(bi∗t+ci)+di+ei∗speed

[0051] The speed parameter is defined differently for different operating conditions. This parameter is combined with the further parameter e. i multiplied, whereby this parameter is considered as an additional parameter in the optimization described above. By defining different operating parameters in the form of different values ​​for the speed parameter, it is possible to optimize the process based on the method of Fig. Two different operating configurations can be determined, which can then be used to control the corresponding conveyor system. Depending on the current operating requirements, the appropriate operating configuration is then selected.

[0052] The embodiments of the invention described above offer a number of advantages. In particular, a method for determining one or more operating configurations of a conveyor system is provided that does not require any observations or measurements from the actual operation of the conveyor system. Instead, the corresponding speeds of the conveyor belts of the conveyor system are modeled by time-dependent periodic functions, the parameters of which are determined through optimization in combination with a physical simulation. In this way, a compact representation of the control rules for the conveyor system is created based on only a few parameters.

[0053] Regardless of the grammatical gender of a particular term, persons with male, female or other gender identities are included.

Claims

[1] Method for the computer-aided determination of one or more operating configurations of a conveyor system (100) for piece goods (P1, P2, ..., P7), wherein the conveyor system (100) comprises a plurality of successive partial conveyor lines (C1, C2, ..., C5) along which the piece goods (P1, P2, ..., P7) are transported from an inlet (IN) to an outlet (OU) of the conveyor system (100) on respective conveyor means (1, 2, ..., 5) of the partial conveyor lines (C1, C2, ..., C5), wherein each partial conveyor line (C1, C2, ..., C5) is assigned an individually adjustable speed (v1, v2, ..., v5) at which the conveyor means (1, 2, ..., 5) of the respective partial conveyor line (C1, C2, ..., C5) is driven, wherein for the determination of a respective operating configuration: - the speeds (v1, v2, ..., v5) of all partial conveyor sections (C1, C2, ..., C5) are modelled by a set of parameters (PS) such that the speed (v1, v2, ..., v5) of a respective partial conveyor section (C1, C2, ..., C5) is modelled by a time-dependent periodic function (F1, F2, ..., FN) comprising one or more parameters (a i , b i , c i , d i ), where the parameter(s) (a i , b i , c i , d i ) belong to the set of parameters (PS); - parameter values ​​(PS) optimized for the respective operating configuration using an optimization procedure (OPT) opt) for the set of parameters (PS), wherein the optimization method (OPT) is based on an optimization objective of a high quality measure (FT) and, within the scope of the optimization method (OPT), the quality measure (FT) is repeatedly determined by means of a physical simulation (SIM) of the conveyor system (100) or by means of a function which models a physical simulation (SIM) of the conveyor system (100), as a function of parameter values ​​of the set of parameters (PS). [2] Method according to claim 1, characterized bythat the optimization method (OPT) is initialized with a plurality of value instances of the set of parameters (PS), wherein the value instances each define parameter values ​​for the set of parameters (PS) and the parameter values ​​of each value instance are at least partially different from the parameter values ​​of every other value instance from the plurality of value instances, wherein the optimization method (OPT) varies the parameter values ​​of the plurality of value instances depending on the optimization objective in order to find a value instance with the highest possible quality measure (FT). [3] Method according to claim 1 or 2, characterized by that the optimization method (OPT) is a gradient-free optimization method, preferably a particle swarm optimization or an evolutionary optimization, in particular a CMA-ES optimization. [4] Method according to claim 1 or 2, characterized bythat the optimization method is a gradient-based optimization method, wherein the gradient-based optimization method uses the function which models the physical simulation (SIM) of the conveyor system (100) to determine the gradient. [5] Method according to one of the preceding claims, characterized by that the function which models the physical simulation (SIM) of the conveyor system (100) is a neural network which is learned based on training data originating from the physical simulation (SIM). [6] Method according to one of the preceding claims, characterized bythat the quality measure (FT) is designed in such a way that it is higher the fewer stacks of piece goods (P1, P2, ..., P7) in the physical simulation (SIM) leave the exit (OU) of the conveyor system (100) in a predetermined time interval, wherein in a stack of piece goods (P1, P2, ..., P7) at least two piece goods are arranged one above the other. [7] Method according to one of the preceding claims, characterized by that the quality measure (FT) is designed in such a way that it is higher the less frequently the event occurs in the physical simulation (SIM) in a predetermined time interval that a piece of goods (P1, P2, ..., P7) rotates at least once around its axis during transport along the partial conveyor lines (C1, C2, ..., C5). [8] Method according to one of the preceding claims, characterized bythat the quality measure (FT) is designed in such a way that it is higher the more piece goods (P1, P2, ..., P7) leave the exit (OU) of the conveyor system in the physical simulation (SIM) in a predetermined time interval. [9] Method according to one of the preceding claims, characterized by that for at least one partial conveyor section (C1, C2, ..., C5) the parameter(s) (a i , b i , c i , d i ) of the time-dependent periodic function (F1, F2, ..., FN) the period (b i ) of the function (F1, F2, ..., FN) and / or the amplitude (a i ) of the function (F1, F2, ..., FN) and / or a time offset (c i ) in the function (F1, F2, ..., FN) and / or a speed offset (d i ) in the function F1, F2, ..., FN). [10] Method according to one of the preceding claims, characterized bythat for at least one partial conveyor section (C1, C2, ..., C5) the time-dependent periodic function (F1, F2, ..., FN) is a sine function. [11] Method according to one of the preceding claims, characterized by that a plurality of operating configurations are determined, each operating configuration containing a different operating specification for the conveyor line system (100), the operating specification for each partial conveyor line (C1, C2, ..., C5) being contained in the time-dependent periodic function, preferably in combination with a further parameter of the set of parameters (PS). [12] Method according to one of the preceding claims, characterized by that the operating specification defines a target value for the number of piece goods (P1, P2, ..., P7) which leave the exit (OU) of the conveyor system (100) in a predetermined time interval. [13] Method according to one of the preceding claims, characterized bythat the determined operating configuration or the determined operating configurations are stored in a control device (CO) of the conveyor system (100) and the conveyor system (100) is operated by means of the control device (CO) based on the determined operating configuration or based on one of the determined operating configurations. [14] A data processing device comprising means for carrying out a method according to any one of claims 1 to 13. [15] Computer program product with a program code stored on a machine-readable carrier for carrying out a method according to one of claims 1 to 13 when the program code is executed on a computer. [16] Computer program with a program code for carrying out a method according to one of claims 1 to 13, when the program code is executed on a computer.

Images