Sorting method for matrix sorter

Abstract

The invention relates to a method for sorting piece goods (10) in a conveyor system. The piece goods (10) are guided along conveyor tracks and each receive a sorting value (25, 26) which determines the order of the piece goods after sorting. The piece goods are conveyed from a common feed path (4) through multiple sorting stages (1, 2, 3), wherein the piece goods, in each sorting stage (1, 2, 3), are directed onto one of multiple alternative sorting paths (1a, 2a, 3a). The sorting paths are assigned on the basis of the sorting values (25, 26) which are evaluated by a sorting algorithm. Before being conveyed into the sorting stages (1, 2, 3), the sorting values (25, 26) are determined according to the number of piece goods (10) to be sorted and the sorting algorithm such that, in each sorting stage (1, 2, 3), the maximum number of piece goods (10) on each sorting path corresponds to a predefined upper limit and / or the minimum number of piece goods (10) corresponds to a predefined lower limit.

Description

[0001] Sorting methods for matrix sorters

[0002] The present invention relates to a sorting method for conveyed unit loads. In particular, the invention relates to a method for sorting unit loads which are guided and conveyed along spatially defined conveyor paths.

[0003] Conveyor systems with corresponding conveyor tracks are known in various designs, for example, as systems for the overhead conveying of unit loads. In overhead conveyor systems, the unit loads are conveyed suspended from transport rails along conveyor tracks or placed in pockets or suspended fixtures and conveyed along the transport rails together with them. In other conveyor systems, unit loads are conveyed lying down on conveyor belts or rollers along conveyor tracks.

[0004] A corresponding overhead conveyor system is disclosed, for example, in DE 10 2018 209 266 A1. The overhead conveyor system described therein allows for order-oriented provision of individual items at a packing station. The introduction of a parking area for the temporary storage of individual items for a subsequent order prevents the mixing of goods from different orders at the packing station. The device comprises several conveyor lanes, which are delimited by infeed and outfeed openings. Automated unloading points and separating elements enable targeted sorting and allocation of the individual items to different orders.

[0005] To enable the demand-oriented transport of unit loads, sorting devices are also implemented in rail-based conveyor systems. Due to the serial transport along the conveyor tracks, the unit loads must be guided via switches or separating devices so that the order of the unit loads can be changed.

[0006] Document EP 3 630 655 Al describes a conveyor system in which a sorting stage is located after a buffer zone. This stage arranges the goods in the desired sequence for a picking workstation. "Matrix sorters" have become established as particularly efficient sorting devices for such conveyor systems. Matrix sorting makes it possible to arrange batch-picked parts into a precise, piece-by-piece sequence.

[0007] As part of the sorting process, the sorting criteria are defined before the actual sorting begins. Each conveyed item is assigned a sorting value (key). The order of the sorting values ​​determines the desired sequence of the items after sorting. Following the conveying of the items along a common infeed conveyor, sorting takes place through multiple sorting stages. In each sorting stage, each item is conveyed onto one of several alternative (parallel) sorting lanes. After the items to be sorted have been conveyed onto the sorting lanes, they are merged from the sorting lanes onto a downstream common collection lane after each sorting stage. The assignment of the items to one of the alternative sorting lanes is carried out in each sorting stage by evaluating the sorting value of each item using a sorting algorithm.The conveyance of each item to its respective sorting lane is then controlled by switches, depending on the evaluation of the sorting algorithm. After passing through all sorting stages, the sorted items are conveyed sequentially to a common output lane. Matrix sorters are frequently configured with three sorting stages, each with six sorting lanes. This is primarily due to a practical and economic consideration of system costs, installation space for the sorting system, and the number of items to be sorted. However, there is no fundamental limitation regarding the number of sorting stages and their respective sorting lanes that matrix sorters can have.

[0008] Matrix sorters most often use a radix sorting algorithm. The radix sorting algorithm is a sorting method that sorts numbers in several sorting stages according to different digits (or radices). The sorting typically proceeds from the least significant to the most significant digit. Since this algorithm is generally well-known and documented, its specific operation will not be discussed here. Furthermore, the present invention is not limited to the application of a specific algorithm, even though the radix algorithm is a very suitable and preferred algorithm for implementing the invention. In this application, the term "radix sorting algorithm" is used to encompass all implementations and, where applicable, variations of this sorting algorithm.

[0009] Matrix sorters always aim for efficient conveying and sorting. This implies the best possible utilization of the conveying systems as well as the optimization of the conveying speed. Failures due to overloading or jams must be avoided. Since matrix sorters are physical structures that require space, there is a desire to design them according to specific needs. On the other hand, the limited capacity of the sorting lanes on which the unit loads are buffered for sorting must always be taken into account. Both the length of the sorting lanes and the dimensions of the conveyed unit loads are determining parameters. When conveying unit loads with variable dimensions, for example, on an overhead conveyor, an increase in the dimensions of the unit loads in the conveying direction can lead to a reduction in the capacity of individual sorting lanes in individual sorting stations.The capacity of individual sorting lanes can be reduced to such an extent in certain sorting stages that sorting is no longer possible due to physical space requirements. This occurs when the length of the sorting lane is less than the sum of the dimensions of the individual items to be fed into the lane. To avoid such problems, the design of sorting lanes is often based on the largest dimensions of the individual items that can still be processed, which entails a corresponding increase in space requirements and costs. This is also the case even if these dimensions are processed only very rarely.

[0010] The US patent 2002 / 053535 discloses a sorting method for objects with multiple sorting runs between two blocks of parking areas. The known method describes a mathematical capacity calculation and aims for a uniform filling of the parking areas. The known solution represents a reactive approach to capacity control, in which the system parameters are adjusted to the available capacity.

[0011] The publication EP 3 666 690 Al describes a picking procedure for individual items in overhead conveyor systems with an order buffer and time-interval-based order processing. The known system takes into account the sorting performance and the maximum permissible number of transport bags per order list to ensure trouble-free operation.

[0012] Document US 2019 / 039834 describes a matrix sorter with multiple sorting stages and sorting lines for sorting objects of various sizes in space-constrained environments. The known system considers the maximum length and number of objects per sorting line for hardware dimensioning and addresses capacity problems through mechanical scaling by means of additional sorting stages and lines.

[0013] Prior art approaches to optimizing sorting systems are reactive in nature. The disadvantage of these reactive systems is that they implement capacity limitation as a downstream control function.

[0014] The present invention aims to operate matrix sorters in a more efficient manner.

[0015] The problem described above is solved by a method having the features of claim 1.

[0016] The solution of the invention consists of improving the control of the matrix sorter without requiring any structural modifications. This is achieved by determining the sorting values ​​(keys) for the individual items in a manner according to the invention at the beginning of a sorting process, i.e., before the items are conveyed into the first sorting stage of the matrix sorter. According to the invention, predictive optimization of the sorting values ​​ensures a uniform utilization of the sorting lanes in all sorting stages without requiring any structural modifications to the matrix sorter. The invention utilizes the deterministic calculability of the item distribution in matrix sorters. The sorting values ​​are determined according to the invention depending on the number of items to be sorted and depending on the sorting algorithm.The selection of sorting values ​​according to the invention ensures that in each sorting stage and for all sorting lanes, the maximum number of items on each sorting lane corresponds to a predetermined upper limit and / or the minimum number of items on each sorting lane corresponds to a predetermined lower limit. In contrast to known reactive, order-based, and hardware-oriented approaches, the present invention provides a proactive, algorithmic optimization method which, through the systematic calculation or selection of specifically optimized sorting values, ensures a uniform utilization of all sorting lanes in all sorting stages of a matrix sorter. A key feature of the present invention lies in the systematic decoupling of the target position and the sorting value while simultaneously maintaining the desired sorting sequence.While known methods use the sorting values ​​as a direct, sequential mapping of the desired target order (e.g., sorting values ​​1, 2, 3, ..., n for n units), the sorting values ​​according to the invention form a mathematically optimized sequence of numbers that deviates from the sequential target position but preserves the order. An "order-preserving sequence" means that the order relation between the sorting values ​​corresponds to the order relation between the desired target positions of the units. Mathematically expressed: If unit A is to be placed before unit B in the desired sorting order, then the sorting value assigned to unit A must be smaller than the sorting value assigned to unit B. The sorting values ​​determine the order of the units after sorting, i.e., their target position in the sequential order of the units conveyed by the matrix sorter.The desired target position is determined by the sorting objectives. For example, the target position can take into account specific characteristics of the individual items (e.g., the postal code, a shipping service provider assigned to the item, or an assigned packing station). If a target position is defined for each item in the sorting process, the sorting values ​​are assigned to the items according to the invention, based on the order of the target positions. In ascending order of the target position values, the sorting values ​​determined according to the invention are also assigned to the items in ascending order. If the items are to be sorted, for example, by postal code, the smallest sorting value can be assigned to the item with the smallest postal code, the second smallest sorting value to the item with the next (e.g., larger or identical) postal code, and so on.up to the parcel with the largest postal code, to which the highest sorting value is assigned.

[0017] The inventive effect is achieved by using sorting values ​​with specific properties. These sorting values ​​are derived from a set of sorting values ​​whose use in the sorting algorithm ensures the fulfillment of the aforementioned criteria for maximum and / or minimum utilization of the sorting lanes. This results in optimized conveying of the unit loads through the matrix sorter, as described below. Optimized conveying according to the invention includes improved utilization of the matrix sorter's sorting lanes, since the use of all sorting lanes to a predetermined extent is ensured. Over- or under-utilization of individual sorting lanes is thus avoided. The inventive solution is based on the understanding that the distribution of unit loads onto the sorting lanes in deterministic sorting algorithms is mathematically calculable and therefore proactively optimizable.Instead of relying on downstream capacity control or hardware expansion, proactive control of the unit load distribution takes place before the sorting process begins, through the use of mathematically optimized sorting values.

[0018] While a sorting value is determined for each item according to the invention, the subsequent ordering of the sorting values ​​takes place according to the unchanged algorithm and its usual operating procedure. The difference according to the invention compared to the prior art therefore lies in the control of the matrix sorter without functionally changing the algorithm.

[0019] The invention is based on the predictability of the distribution of unit loads with known sorting values ​​across the sorting lanes of all sorting stages. The conveying paths along the stages of the matrix sorter are determined and predictable when the algorithm is known. The invention utilizes this predictability of conveyance along the sorting lanes of all sorting stages to select the sorting values, using a predefined sorting algorithm, such that the aforementioned conditions regarding the utilization of the sorting lanes are met in all stages. For example, a suitable selection of sorting values ​​makes it possible to achieve an even distribution of unit loads across the sorting lanes of all sorting stages. An even distribution here means that the distribution is approximated as closely as possible to a perfectly even distribution, whereby deviations in the number of units of +1 / -1 across the sorting lanes are possible.However, a perfect uniform distribution cannot be achieved with a number of individual items that is not a multiple of the number of parallel sorting lanes in a sorting stage, so the aforementioned deviations are accepted as optimal uniform distribution.

[0020] Within the scope of the invention, sorting values ​​can be temporarily assigned to the individual items in addition to other identifiers assigned to the item in the conveyor system's control system. For example, a sorting value can be assigned to each item solely for the purpose of passing through the matrix sorter, with an assignment between the sorting value and the item's other identification information being stored for the purpose of carrying out the sorting. The sorting value can then be discarded after the item has completed its passage through the matrix sorter.

[0021] Determining the sorting values ​​is possible in a variety of ways within the scope of the invention. Depending on the sorting algorithm and the implementation of the invention, the sorting values ​​can be discretely determined for any number of individual items using computational formulas. This approach is particularly suitable if the sorting algorithm allows for the simple derivation of such formulas, as illustrated below with the radix sorting algorithm.

[0022] Alternatively, simulations can be performed for groups of sort values. This involves simulating the sorting process in a matrix sorter with the intended number of sorting levels and lanes for systematically or randomly selected values. Based on the simulation results, groups of sort values ​​are selected that achieve the desired distributions across all sorting levels. This comprehensive simulation approach is universally suitable for any type of algorithm, as it does not require analysis of the algorithm itself; instead, the selection of sort values ​​is based solely on the algorithm's output. However, comprehensive simulations may require time-consuming calculations.Therefore, within the scope of the invention, it is possible and advantageous to perform such calculations only once and to store the determined groups of sorting values ​​in relation to the number of group elements, in order to access them later during operation of the process without further simulation processes. A complete database in tabular form can be created through simulation, which assigns a suitable set of sorting values ​​to a given number of items to be sorted for carrying out the process according to the invention. Depending on the chosen method, it is therefore possible to use stored sorting values ​​or to perform live calculations during the assignment of the sorting values ​​when the items are fed into the matrix sorter. The composition of the set of sorting values ​​is essential, not the time of their generation.Accordingly, in each sorting run (sorting batch), depending on the number of items in the sorting run, a group of sorting values ​​is individually assigned to the items according to their target positions. This results in an optimization of the conveyor paths by the matrix sorter with regard to the desired utilization of the sorting lanes. Thus, conventional matrix sorters and their controls can be used unchanged, provided that the assignment of sorting values ​​according to the invention is ensured.

[0023] The solution according to the invention thus enables deterministic optimization of the sorting lane utilization without modifying the hardware configuration. Through the sorting value generation and assignment to individual items according to the invention, the available sorting capacity is utilized optimally, thereby avoiding both overloads of individual sorting lanes and underutilization of the overall system. This leads to a significant increase in the throughput capacity and operational stability of existing matrix sorters. The technical advantages of the solution according to the invention lead to an improvement in system performance: While reactive approaches according to the prior art can only react to capacity problems after they have occurred, the preventive optimization according to the invention prevents the emergence of uneven distributions from the outset.Compared to order-based systems, microscopic single-piece optimization enables significantly more precise control of capacity utilization. Compared to hardware-oriented solutions, increased capacity is achieved without requiring additional space or mechanical components.

[0024] The use of specifically calculated or selected sorting values ​​according to the invention thus represents a cost-effective and efficient approach, as it enables for the first time a systematic, preventive and software-based optimization of matrix sorter performance.

[0025] In a preferred embodiment of the invention, the upper limit of the items on each sorting lane is determined such that it corresponds to the rounded integer value resulting from dividing the number of items to be sorted by the number of sorting lanes in a sorting stage. Alternatively or additionally, the lower limit is determined such that it corresponds to the rounded integer value resulting from dividing the number of items to be sorted by the number of sorting lanes.

[0026] These limits ensure the most even distribution of the goods possible and minimize the risk of overcrowding of sorting lanes and unused capacity on other sorting lanes.

[0027] It is particularly preferred that a matrix sorter is used in the process in which the number of sorting lanes is identical in all sorting stages.

[0028] This uniformity simplifies the mechanical design and control logic of the sorting system, which facilitates both implementation and maintenance, and also optimizes utilization across all sorting stages.

[0029] In a preferred embodiment of the invention, a radix sorting algorithm is used as the sorting algorithm.

[0030] The use of this algorithm is proven and robustly implemented in matrix sorters. Furthermore, the algorithm's functionality allows for the rapid calculation of sorting groups based on the number of items to be sorted, using easily implemented programs, as explained below.

[0031] In a preferred embodiment of the invention, the method is operated with a matrix sorter in which the sorting stages are designed as a plurality of spatially arranged groups of sorting lanes, wherein a collection lane is arranged between each adjacent sorting stage.

[0032] The state of the art referenced above describes such a matrix sorter, which is also the most common design. This arrangement optimizes the time required for the throughput between sorting stages. Freed-up (lower) sorting stages can be used and occupied again for a subsequent sorting operation, even if the individual items from a previous sorting operation are still located in subsequent (higher) sorting stages.

[0033] In an alternative embodiment of the invention, the method is operated with a matrix sorter in which several sorting stages are formed by a single physical group of sorting lanes. These lanes are traversed multiple times by returning the goods from the output side of the sorting lanes to the input lane via a controllable diverter until the predetermined number of sorting stages has been completed. Thus, the same sorting lanes sequentially form different sorting stages. After sorting, the sorted sequence of goods is conveyed to the downstream conveyor system.

[0034] This approach reduces the need for physical space and lowers the costs for additional sorting lanes.

[0035] In a preferred embodiment, the sorting values ​​for each number of items to be sorted are determined by simulating the sorting process with execution of the algorithm and depending on the number of sorting stages and sorting lanes, wherein a set of sorting values ​​from a plurality of sets of sorting values ​​is successively supplied to the simulation and such sets of sorting values ​​are selected as suitable according to the invention which lead to a distribution of the sorting values ​​on the sorting lanes in all sorting stages in accordance with the conditions.

[0036] The simulation of the sorting processes can be easily performed using the algorithm for a given group of sort values. The sort values ​​are fed into the algorithm as input values, and corresponding sorting lanes are determined for each sorting stage. Each sort value has an associated sorting lane in each sorting stage. The resulting distribution of the simulated sorts across the sorting lanes is evaluated, and groups of sort values ​​that do not meet the aforementioned conditions regarding the maximum or minimum number of items per sorting lane are discarded. Those groups of sort values ​​that do meet the conditions can be used, according to the invention, for assigning items to specific goods.

[0037] In this context, it is advantageous if the simulation results are stored in relation to the number of items and, at the start of a sorting process, the sorting values ​​are retrieved from memory depending on the number of items. Since the simulations can be computationally intensive, it is useful to calculate a reserve of permissible sorting values ​​and store them in relation to a number of items to be sorted. This speeds up the subsequent sorting process and reduces the required computing resources. Regarding the quantities of sorting values ​​to be simulated, it is advantageous within the scope of the invention to only supply quantities with numbers of sorting values ​​to a simulation that are actually to be processed in the physical setup of the controlled matrix sorters.A simulation is therefore only meaningful up to a number of sorting values ​​per quantity that would actually be recorded with optimized distribution in the sorting lanes of all sorting stages. In the case of inventory calculations, the simulations must also be performed within the limits of the practically relevant quantities of individual items to be sorted, so that a group of sorting values ​​is available for each incoming quantity to be sorted. Appropriate groups of sorting values ​​can be permanently stored in the control system and also delivered with the entire conveyor system as part of the control software. Retrofitting existing matrix sorters is also easily possible by appropriately upgrading the control systems.

[0038] If the radix sorting algorithm is used, the total number of items to be sorted can be derived from the number of sorting stages and the respective sorting lanes. A matrix sorter with three sorting stages and six sorting lanes each has a theoretical sorting capacity of 6 A 3 = 216 items. Therefore, for such systems, it is necessary to create groups of sort values ​​containing at most 216 elements. For each group size X, a size-sorted array S can then be created, for example. x (y) of the sorting values ​​are stored. If a number X of individual items are then fed to the matrix sorter for sorting, a sorting value S is assigned to each item. x ( 1 ) • • • S x(X) is assigned from the corresponding array. If the sort values ​​are stored in an ordered manner in the array, a suitable sort value can be easily assigned to each item using a running index.

[0039] In a preferred embodiment of the invention, the radix sorting algorithm is used, and the sorting values ​​are formed for each quantity in a number system whose base corresponds to the number of sorting lanes and whose number of digits corresponds to the number of sorting levels. The sorting values ​​are formed such that each of the digits corresponds to each

[0040] A digit with a maximum frequency corresponding to the specified upper limit and / or at each of the digit positions, each digit with a minimum frequency corresponding to the specified lower limit occurs.

[0041] Such a calculation can be performed to generate sorting values ​​usable according to the invention using simple programs. For example, a possible program for a matrix sorter with three sorting stages and six sorting lanes each, i.e., in a base-6 number system, involves the following steps (assuming, by way of example, that at least six individual items are being sorted):

[0042] - Initialize a list with the starting sort values: 000, 111, 222, 333, 444, 555.

[0043] - Create a function that counts the frequency of each digit at each position in the sorted values ​​generated so far.

[0044] - Generate a new sort value by favoring the least frequent digit for each position. If the digits are evenly distributed, a random digit is chosen. The digits 0, 1, 2, 3, 4, 5 are allowed.

[0045] - Check if the newly generated sort value already exists in the list. If so, try using an alternative digit (the second most frequent) for one of the positions without compromising the uniqueness of the sort value.

[0046] - Add the new, unique sort value to the list.

[0047] Repeat these steps until the desired number of sort values ​​is reached.

[0048] Depending on the implementation of the algorithm in the matrix sorter's control system, the determined sort values ​​can subsequently be converted into another number system, e.g., the decimal system. It is advantageous if the specified upper limit depends on the dimensions of all or selected matrix elements.

[0049] General cargo is determined.

[0050] When the adjustment of the upper limit and the conveying are linked to the physical dimensions or weight of the individual items, the matrix sorter exhibits improved adaptability. This adaptability allows the system to respond to different sizes and shapes of the items, thus expanding the applicability of the sorting process. For example, in logistics centers with overhead conveyors, changes can be made as needed from conveying flat items (e.g., outerwear) to bulkier items (e.g., shoes in boxes), with the control system according to the invention ensuring optimized conveying and utilization at all times.

[0051] In a further development of the invention, the number of unit loads conveyed onto the feed track is specified depending on the dimensions of all or selected unit loads.

[0052] This adaptation also allows the system to react to different sizes and shapes of the individual items, which expands the applicability of the sorting process.

[0053] The method according to the invention is typically designed in its general application to remove all sorting lanes from all sorting stages. However, within the scope of the invention, it is also possible to selectively remove individual sorting lanes from the sorting process. This can be achieved by simulation or by determining suitable sorting values. This allows the number of items to be sorted to be reduced, so that the matrix sorter can be operated with fewer sorting lanes temporarily. Consequently, only those sorting values ​​are used that prevent any material from being conveyed onto the sorting lanes that are being removed. In the event of a defect in individual sorting lanes, the matrix sorter can continue operating at a reduced capacity while the defect in the affected sorting lane is repaired.Once the defect is repaired, the system can revert to sorting with a larger number of sorting values, thus reusing the capacity of the repaired sorting lane. For example, a matrix sorter with three sorting stages and six sorting lanes each can be temporarily converted into a matrix sorter with three stages and five sorting lanes each, operating at a reduced sorting capacity. Simply assigning the sorting values ​​is sufficient to effect the desired change. This can be done at any time, even during operation, between two batches of items being sorted.

[0054] The invention will now be explained with reference to the accompanying drawing.

[0055] Figures 1a, 1b and 1c schematically show the distribution of unit loads in a matrix sorter during a sorting process according to the state of the art;

[0056] Figures 2a, 2b and 2c schematically show the distribution of unit loads in a matrix sorter during a sorting process according to a first embodiment of the invention;

[0057] Figure 3 schematically shows a tabular assignment of target position of unit loads to sorting values ​​according to the first embodiment of the invention;

[0058] Figure 4 schematically shows the sorting according to the first embodiment of the invention in a representation with indication of the sorting values;

[0059] Figures 1a, 1b and 1c illustrate the structure and

[0060] Sequence of a sorting process in a matrix sorter according to the prior art. The illustrated matrix sorter, which is also used with the method according to the invention and has the same physical structure, has sorting stages 1, 2, 3, each of which in turn has groups of six sorting lanes 1a, 2a, and 3a, respectively. The matrix sorter is designed as a component of a conveying system, to which further conveying components with conveying sections can be assigned upstream and downstream. The sorting lanes in all sorting stages 1 to 3 are numbered from 0 to 5 in the illustration. From an upstream part Fl of the conveying system, unit loads 10 are conveyed into the matrix sorter via an infeed lane 4. Each of the sorting lanes aa, 2a, 3a has a controllable switch on the input side (in the figures aa, 1b and 1c on the left) which is coupled to a control system of the matrix sorter.Unit loads 10 conveyed via infeed conveyor 4 are conveyed onto one of the six sorting conveyors la by controlling the switches. The assignment of an incoming unit load to one of the six sorting conveyors la is carried out according to a sorting algorithm.

[0061] Each unit item is assigned a sorting value. This sorting value, within the total number of sorting values ​​assigned to the units, indicates the position the unit item assigned to that sorting value should occupy in the sequence of conveyed units after passing through the matrix sorter. For example, the sorting values ​​are sorted in descending or ascending order by the sorting algorithm.

[0062] The sorting algorithm is implemented in the control system, and the control system controls the diverters for conveying the unit loads according to the algorithm's outputs. In this prior art example, as well as in the embodiment of the invention described later, a radix sorting algorithm is used. In this prior art method, the sorting values ​​are determined by assigning a target position to the unit loads based on certain properties. Subsequently, numbers corresponding to the target position are assigned to the unit loads as sorting values. The positions of the unit loads along the conveyor track in the matrix sorter are known at all times, so the control unit makes the diverter settings for conveying them onto one of the six sorting lanes.In a matrix sorter with three sorting stages (S=3) and six sorting lanes (M=6) in each of the stages, 216 unit items (6. A 3) are sorted in a single pass. The radix sorting algorithm, for a number K of items to be sorted, triggers the control of the switches, which determines the distribution of the items 10 onto one of the sorting lanes laa according to the following formula: m s = ( (P _ l) div M s-1 ) mod M with m s : Target sorting lane [0..M-1] in sorting stage s [1..S] M: Number of sorting lanes per sorting stage div: Integer division mod: Modulo p: Desired sorting position of the part within a pass with K parts (1 < p < K) .

[0063] In the example shown, the position in each of the sorting levels is: m s = ( (p-1) div 6 s “ 1 ) mod 6

[0064] In this analysis, the lanes are numbered starting from 0, which facilitates understanding of the following explanations of the invention. In the exemplary embodiment shown in Figures 1a to 1c, according to the prior art, with 18 items to be sorted and sorting values ​​from 1 to 18, the distributions on the sorting lanes 1a are as shown.

[0065] In sorting stage 1, the unit loads 10 are evenly distributed across the six sorting lanes 1a. Subsequently, the unit loads are consolidated on the collection lane 5, with the sorting lanes 1a being emptied onto collection lane 5 sequentially. In sorting stage 2, the unit loads 10 are distributed according to the sorting values ​​and the radix sorting algorithm, resulting in the distribution shown in Figure 1b. Even at this stage, the prior art method leads to an uneven utilization of the sorting lanes 2a. After conveying via collection lane 6 to sorting stage 3, the unit loads are arranged on a single sorting lane 3a. From there, the sorted unit loads are conveyed to a downstream section F2 of the conveying system.

[0066] Even if, in reality, larger numbers of individual items are sorted in one sorting process and the distribution onto the lanes is then possibly less concentrated, the illustration shows that with this conventional application of the sorting method, sorting lanes can become overloaded or, in many cases, underutilization of the matrix sorter must be accepted.

[0067] Figures 2a, 2b, and 2c show the same matrix sorter with the same number of sorting stages 1, 2, 3 and the same number of sorting lanes 1a, 2a, 3a in each sorting stage. However, the sorting shown according to the first embodiment of the invention results in the sorting lanes 1a, 2a, 3a being utilized evenly across all sorting stages. As in the previous example according to the prior art, 18 unit loads are conveyed into the sorting system; however, the unit loads are shown here with significantly larger dimensions. The method according to the invention allows significantly larger unit loads to be sorted with the same number, for example, as described in the prior art, since an equal utilization of the sorting lanes is ensured at all times.In this exemplary embodiment, the predefined sorting values ​​ensured that the maximum number of individual items on each sorting lane corresponds to three in all stages. Dividing the number of individual items (a total of 18) by the number of sorting lanes (six each) yields the aforementioned result. A corresponding result would also be achieved by setting the minimum number of individual items for each sorting lane to three. The essential difference between the inventive method, as illustrated in Figures 2a, 2b, and 2c, and the conventional method lies in the specific selection of the sorting values.In this exemplary embodiment according to the invention, the sorting values ​​are not assigned to the individual items as a simple sequence of increasing numbers, but are specified in such a way that the even distribution on the sorting lanes is achieved with unchanged application of the radix sorting algorithm.

[0068] Figure 3 illustrates the corresponding allocation procedure, which is carried out before the individual items enter the first sorting stage 1. Column 20 on the left shows an ascending sequence of numbers from 1 to 18. These values ​​represent, in their order, the desired target positions of the individual items after sorting. The direct use of these values ​​from column 20 as sorting criteria leads to the sorting according to Figures 1a, 1b, and 1c and is state of the art.

[0069] However, in this invention, these values ​​are not used as sorting values. Instead, sorting values ​​are either calculated or looked up in a table, resulting in the desired distribution of the individual items across the sorting lanes. In this example, the two rightmost columns, 25 and 26, represent the corresponding sorting values ​​that lead to the desired distribution. The rightmost column, 26, displays sorting values ​​in the decimal system, while the middle column, 25, shows the corresponding values ​​in the sensary system (base 6). The use of sorting values ​​where the digits are uniformly distributed across all positions in the sensary system is evident when examining column 25. This corresponds to the allocation of lanes according to these digits, leading to an even distribution of the individual items across the respective lanes.This leads to a significant improvement in the utilization of the railway lines and even the targeted control of the utilization of individual lines. First, a desired order for the goods to be sorted can be determined, as in the prior art, which essentially corresponds to the order in the left column of Figure 3. Then, each item is assigned a sort value from the right column (depending on the required numerical base of the sorting algorithm, either 25 or 26). The assignment is rank-based: The item intended for position 1 in the desired order receives the lowest sort value. The item for position 2 receives the second lowest sort value, and so on. The radix sorting algorithm performs the sorting based on this sort value. It should be noted that a corresponding assignment table can be stored for any number K of items (in this example, for K=18).In the case of the example shown with three sorting levels and six sorting lanes each, for example 216 tables would be stored, which allow an assignment of the desired sorting position to a sorting value for any desired number K <= 216.

[0070] As described above, the sorting values ​​can be determined using a simulation for storage in the table. For this purpose, the sorting process is calculated for every possible combination of sorting values. Subsequently, only those combinations of sorting values ​​that result in the desired distribution are stored. In the example mentioned, for the desired sorting of 100 individual items in the decimal system, a selection of 100 numbers would be made, ranging from 0 to 215 (or 1 to 216), which would then be subjected to a simulated sorting process. Determining a suitable combination that ensures the desired distribution of the items on the sorting lanes is achieved by simultaneously or alternatively considering two criteria: a minimum number of items on each sorting lane and a maximum number. Alternatively, the values ​​can also be calculated discretely.This can occur particularly when, as in this example, it is possible to consider the situation in a different number system with a different base. It can be seen from Figure 3, middle column, that a corresponding sequence of numbers, in which the allowed digits correspond in their frequency to the intended distributions at each position of the numbers, can be generated using a suitable algorithm.

[0071] Figure 4 illustrates the correspondence between the occupancy, as shown in Figures 2a, 2b, and 2c along the sorting process, and the corresponding sorting values. The unsorted unit loads 10 are represented here by their sorting values, which correspond to the sorting values ​​from Figure 3, middle column. In the first sorting stage, the unit loads with a 0 as the third digit are sorted onto sorting lanes 0. The subsequent sorting lanes are then fed with unit loads that have the corresponding digit as the last digit. The unit loads 10 are then conveyed from sorting stage 1 via the collection track 5 to sorting stage 2. There, the unit loads 10 are sorted and assigned to the respective sorting lanes according to the digit in the second position of the sorting values.The individual items are then conveyed via collection line 6 to sorting stages 3 and distributed there onto the sorting lanes according to the number at the top. The sorted items are then provided in their respective sorting lanes 3a and discharged into the downstream conveyor system F2. The items are discharged sequentially, starting with lane 0 of sorting lanes 3a. This results in the sorted order shown in the lower right of the downstream conveyor system F2.

[0072] The illustration also demonstrates that each sorting lane has a consistent utilization throughout the entire sorting process.

[0073] The invention is fundamentally transferable to other sorting methods using matrix sorters. The selection of sort values ​​is particularly easy to understand in the radix sorting algorithm presented here, since switching to a number system with a corresponding base reveals the regularities of the sort values. Simulation of other sorting algorithms is also possible, provided they operate deterministically. Since a simulation to determine suitable sort values ​​typically only needs to be performed once for each group size (number of items to be sorted), the computational effort for the subsequent execution of the sorting process is negligible. Thus, corresponding reference tables can also be determined for matrix sorters with different numbers of sorting stages and / or different numbers of sorting lanes on each sorting stage.

Claims

Patent claims 1. Method for sorting unit loads (10), wherein the unit loads are guided along conveyor tracks in a conveyor system, wherein each of the unsorted unit loads (10) is assigned a sorting value (25, 26), wherein an order of the sorting values ​​(25, 26) of all unit loads (10) determines a sequence of the unit loads (10) after sorting, wherein the unit loads (10) are conveyed sequentially from a common feed track (4) to sorting through a plurality of sorting stages (1, 2, 3), wherein the unit loads (10) are conveyed by means of controllable switches onto one of a plurality of alternative sorting tracks (1a, 2a, 3a), wherein the unit loads (10) are gathered from the plurality of alternative sorting tracks (1a, 2a, 3a) between the sorting stages (1, 2, 3) onto a collection track (5, 6), wherein the assignment of the Unit goods (10) in each of the sorting stages (1, 2, 3) to one of the alternative sorting lanes (1a, 2a, 3a) takes place,by evaluating the sorting value (25, 26) of each of the unit loads (10) in a deterministic sorting algorithm and controlling the switch control for conveying to the alternative sorting lanes (1a, 2a, 3a) in each of the sorting stages (1, 2, 3) depending on the evaluation, wherein after passing through a predetermined number of sorting stages (1, 2, 3) the sorted unit loads (10) are conveyed one after the other onto a common output lane (7), characterized in that before conveying the unit loads (10) into the sorting stages, (1, 2, 3) by a computer-implemented Optimization procedure depending on the number of items to be sorted (10) and depending on the sorting algorithm used, the sorting values ​​(25, 26) are determined depending on the sorting algorithm, wherein the sorting values ​​(25, 26) form a sequence of numbers that deviates from the sequential target position but maintains order, which, when the sorting algorithm is executed, leads to the uniform utilization distribution in which, in each of the sorting stages (1, 2, 3) for all sorting lanes (1a, 2a, 3a), the maximum number of items (10) on each sorting lane corresponds to a predetermined upper limit and / or the minimum number of items (10) on each sorting lane (1a, 2a, 3a) corresponds to a predetermined lower limit.

2. The method according to claim 1, comprising the following steps: a) determining the total number (K) of the items (10) to be sorted, b) determining a desired sorting sequence of the items (10) based on sorting criteria, c) calculating or retrieving an optimized group of sorting values ​​(25, 26) from a data storage, wherein the number of sorting values ​​corresponds to the total number (K), d) assigning the optimized sorting values ​​(25, 26) to the items (10) according to the desired sorting sequence, wherein the assignment of the sorting values ​​(25, 26) to the items (10) is carried out by a monotonic mapping in which the order of the desired target positions of the items corresponds to the order of the assigned sorting values, e) performing the sorting with the assigned optimized sorting values ​​(25, 26).

3. Method according to claim 1 or 2, wherein — the upper limit corresponds to the rounded integer value resulting from dividing the number of items to be sorted (10) by the number of sorting lanes (1a, 2a, 3a) of a sorting stage (1, 2, 3) results or — where the lower limit corresponds to the rounded integer value resulting from dividing the number of items to be sorted (10) by the number of sorting lanes (1a, 2a, 3a) of a sorting stage (1, 2, 3).

4. Method according to one of the preceding claims, wherein the number of sorting lanes (1a, 2a, 3a) is identical in all sorting stages (1, 2, 3).

5. Method according to any of the preceding claims, wherein a radix sorting algorithm is used as the sorting algorithm.

6. Method according to one of the preceding claims, wherein the sorting stages (1, 2, 3) are designed as a plurality of spatially arranged groups of sorting lanes (1a, 2a, 3a), wherein a collecting lane (5, 6) is arranged between mutually adjacent sorting stages.

7. Method according to any one of claims 1 to 5, wherein several sorting stages are formed by a single group of sorting lanes which are traversed several times by returning the unit loads from the output side of the sorting lanes to the infeed lane until the predetermined number of sorting stages has been traversed.

8. Method according to one of the preceding claims, wherein the sorting values ​​(25, 26) for each number of the items to be sorted are determined by a computer-implemented simulation of the sorting process with execution of the algorithm and depending on the number of sorting stages (1, 2, 3) and the sorting lanes (1a, 2a, 3a), wherein a set of sorting values ​​(25, 26) is successively supplied to the simulation from a plurality of sets of sorting values ​​and such sets of sorting values ​​(25, 26) are selected which lead to a uniform utilization distribution of the sorting values ​​(25, 26) on the sorting lanes (1a, 2a, 3a) in all sorting stages (1, 2, 3).

9. Method according to claim 8, wherein the selected quantities of sorting values ​​(25, 26) of the simulation are stored in relation to the number of unit items (10) and at the start of a sorting process the sorting values ​​(25, 26) are retrieved from the memory depending on the number of unit items.

10. The method of claim 8 or 9, comprising the following steps: a) generating a plurality of candidate sorting value groups, each with K sorting values, where K corresponds to the number of items to be sorted; b) for each candidate sorting value group: simulating the complete sorting process through all sorting stages (1, 2, 3) using the deterministic sorting algorithm; c) quantitatively evaluating the simulated Distribution of individual items on the sorting lanes (1a, 2a, 3a) of all Sorting levels (1, 2, 3) using a uniformity measure, d) Selection of the candidate sorting value group that achieves the best uniform distribution according to the uniformity measure, whereby the simulation procedure is carried out offline and the determined optimal sorting value groups are stored in a data storage in assignment to the respective number of items K.

11. Method according to one of the preceding claims, wherein the radix sorting algorithm is used as the sorting algorithm and wherein the sorting values ​​(25, 26) are formed in a number system whose base is the number of sorting lanes (1a, 2a, 3a) corresponds and whose number of digit positions corresponds to the number of sorting levels (1, 2, 3), wherein the sorting values ​​(25, 26) are formed such that at each of the digit positions each digit occurs with a maximum frequency according to the specified upper limit and / or with a minimum frequency according to the specified lower limit.

12. The method of claim 11, wherein the sorting values ​​(25, 26) for a given number K of items to be sorted are generated by an algorithmic calculation process comprising the steps of: a) initializing a sorting value list for a M-ary number system with S digits, where M corresponds to the number of sorting lanes per sorting stage and S to the number of sorting stages, b) iteratively generating sorting values ​​by systematically constructing digit combinations using a uniform distribution optimization function, which is repeatedly — for each digit position (1 to S) the current frequency distribution of all digits (0 to Ml) in the already generated sort value list is determined, — when constructing each new sort value, for each digit position, selects the digit that has the lowest current frequency at that position, — when there are several digits with the same lowest frequency, a deterministic selection rule is applied which prefers the lexicographically smallest digit or selects a digit at random, — checks the admissibility of each constructed digit combination, whereby a sort value is considered admissible if, by adding it, no digit at any position reaches a frequency greater than the rounded integer value of the division of K by M, c) uniqueness check of each sort value deemed admissible against all sort values ​​already present in the list, whereby in the case of duplicates the next admissible digit combination is determined by systematic modification of the digit selection while maintaining the uniform distribution optimization, d) addition of each unique and admissible sort value to the list and continuation of the iterative generation until the list contains K sort values.

13. Method according to one of the preceding claims, wherein the predetermined upper limit is determined depending on the dimensions of all or selected unit goods (10).

14. Method according to one of the preceding claims, wherein the number of unit loads (10) which are placed on the infeed track (4) The amount of funding provided depends on the dimensions of all or selected items.

Images