FI121595B - Lift system - Google Patents

Description

FIELD OF THE INVENTION

The invention relates to control of an elevator group. In particular, the invention relates to a method, hardware and software product for utilizing symmetry in the allocation of passenger calls to elevator group elevators.

BACKGROUND OF THE INVENTION

10 The elevator system of the building constitutes a transportation system designed to transport passengers from their departure floors to their target floors in a safe, comfortable and efficient manner. In large buildings such as skyscrapers and shopping malls, elevators 15 are often arranged in groups to improve transport capacity. The elevators in the elevator group are placed close to each other and have common calling devices in the elevator lobby. Conventionally, in elevator groups, the call is triggered by "up / down" call buttons 20 located near the elevators, by means of which the passenger indicates the desired direction of travel. When the serving elevator reaches the called floor, the passenger enters the elevator car and indicates the desired target floor using the car control buttons on the elevator car. Increasingly 25 new elevator groups are using so-called elevator cars. target call panels. They allow each passenger to identify the target floor o ς already in the elevator lobby, and in this case the elevator car does not have to be called separately to the destination.

CVJ

V floor for traveling. In addition, the destination invitation panel 30 enables the number of ir ir passengers involved in the invitation or other

CL

for more information. In this case, under the control of the elevator group c \ i ......

g is more information about passengers as compared to penny systems, since the working layers, target layers and number of passengers 2 of the invitations to be allocated are known when selecting the elevator baskets for the passengers.

Among many different tasks, the basic function of group lift control for a lift group is to allocate invitations from passengers. The purpose of the allocation is to distribute the invitations to be served to the elevators of the elevator group so that one or more performance indicators of the elevator group are as good as possible. The indicators can be, for example, the average waiting time of passengers, the average travel time, the number of stops of the elevators, the energy consumption of the elevator group or a combination thereof. Some of these goals may be contradictory. The different allocation options 15 are compared against each other based on their goodwill values.

The goodness values are scalar and are calculated by a cost function formed by the selected metrics. Thus, the objective of the allocation is to select as the allocation solution the solution that optimizes the value of the 20 cost functions. Invitation allocation strategies can be divided into two groups based on the timing of the allocation assignment. In continuous allocation allokointitehtävä settled continuously at certain time intervals, for example every 25 side have stepped-Nin. In this case, the call to be allocated is permanently affixed to the elevator when the serving elevator car is sufficiently close to the passenger's departure floor. This means that the elevator car serving the waiting passenger i ™ on the departure floor can be replaced with the elevator car of the other elevator in subsequent assignments, if a better allocation solution is found. In the immediate allo- cation, the allocation task is solved only when a new call is registered, that is, the traveler's call is immediately assigned to the elevator at the time of the call, and this allocation cannot be changed later.

3

The allocation tasks are traditionally solved by some heuristic based mathematical optimization algorithm. These optimization algorithms typically initially define a valid initial solution 5, which iteratively improves iteratively by repeating certain operations for each solution until some end condition is fulfilled. One such method is described in US5907137, which discloses an up-down allocation strategy for elevator group 10 with up / down call buttons where the allocation tasks are solved by a genetic algorithm.

The prior art solution methods for allocating passenger calls, as described above, are computationally intensive, especially in large elevator groups, and require a large amount of processor power, making group control easily an expensive and complex control system. Because the computation time for the optimal allocation wheel 20 has to be limited for practical reasons, it is possible that the optimal allocation solution will not have time to calculate within a given time, which may result in lower transport capacity of the elevator group and longer travel times. Prior art optimization algorithms often perform unnecessary computation if there are several alternative optimal or near-optimal solutions. For example, if the mutual i ou symmetry of the elevators in certain situations is not taken into account, the evaluation of structurally similar solutions of substantially equal goodness value will cause unnecessary computation cc, limiting or at least slowing down the finding of an optimal allocation solution.

co o O) o o w 35 4

PURPOSE OF THE INVENTION

It is an object of the present invention to overcome the drawbacks of the prior art solutions described above. Furthermore, the object of the invention is to achieve one or more of the following objectives: - to speed up and improve the finding of an optimal allocation solution, - to enable larger elevator groups, 10 - a solution applicable to elevator groups with call up based on traditional up / down call buttons or both, a solution that is suitable for both continuous and direct call allocation, - a solution that can easily be installed on existing elevator groups as well.

SUMMARY OF THE INVENTION

The method, hardware and software product of the invention are characterized in what is set forth in the characterizing part of claims 1, 5 and 8. Other embodiments of the invention are characterized in what is set forth in the other claims. Inventive embodiments are also provided

CM

and in the drawings of this application. The inventive content of the application may also be defined in a manner different from that described in the claims below. The inventive content may

CL

The invention may also be the result of several separate inventions, especially if the invention is considered in the light of the expressed or implicit subtasks or the benefits or classes of benefits achieved. In this case, some of the attributes contained in the claims below may be redundant for individual inventive ideas. Aspects of various embodiments of the invention may be applied within the scope of the inventive concept with respect to other embodiments.

The essential content of the present invention relates to a method for solving the allocation task which takes into account the symmetry of the elevators in the elevator group. The Allo-10 assembly function is symmetric if two or more elevators are in the same state, ie symmetrical. At the time of allocation, the status of each elevator is determined based on, for example, the position, speed, direction, acceleration, door position, occupancy, car commands in the car 15 and / or calls already assigned to the car. Two or more lifts shall be interpreted as symmetrical if their spaces are sufficiently similar. At the time of allocation, the method of the invention identifies symmetrical lifts of an elevator group 20 and, if symmetrical lifts are present, takes into account symmetry by limiting the search space for the assignment task to a subset that does not include isomorphic solutions with respect to each other. If there are no symmetrical lifts, the allocation problem can be solved by conventionally known and solving methods, o δ

(M

The solution to the allocation problem starts with a valid initial solution of the selected subset, which is iteratively enhanced by mathematical optimization.

X

£ using the algorithm. During the iteration, it is taken care of i- that after each operation a new solution is heard

OJ

o continuing to the selected subset by returning the solution σ>, g to that subset if necessary. For example, an existing optimization algorithm such as a genetic algorithm can be used to solve the allocation task. The solution can be restored to the selected subset by permutating the indexes of symmetric his-5s so that the solution becomes lexogographically small or large. It should be noted that the solution to be optimized may also comprise a plurality of solution options instead of a single solution, as discussed below in the context of genetic algorithms. In this case, each of the set solutions must be returned to the selected subset as described above.

The present invention also provides apparatus 15 for solving an allocation task in an elevator system comprising a plurality of elevators as well as paging devices for recording passenger invitations. The apparatus comprises an interface for receiving information needed to determine elevator conditions, based on which the equipment determines the symmetrical elevators of the elevator group before solving the allocation task. If symmetric elevators are determined, the equipment limits the allocation task to a search space subset that does not contain isomorphic rat-25s. The group control of the elevator group, o δ, can advantageously be used as equipment

(M

The present invention also provides a software product for solving an allocation task in an elevator group comprising a plurality of elevators,

X

£ to record passenger invitations f, and means for identifying the elevators of the elevator group

CVJ

§ for sex. The software product comprises the program code which is a cd ....

while running the information processing device, it is arranged to determine the symmetric lifts of the elevator group based on the elevator states before solving the allocation task, and to limit the search space of the allocation task to a subset that does not include isomorphic allocation solutions if symmetric lifts exist. The software product is advantageously stored in the group control of the elevator group.

In one embodiment of the invention, the genetic algorithm is used as an optimization algorithm for the allocation task. In this embodiment, each solution of the allocation task generated by the selection and / or crossover and / or mutation of the genetic algorithm is returned to the selected subset.

The solution according to the invention provides several advantages over the prior art solutions. With the solution of the invention, under symmetry, the search space of the allocation task is smaller and comprises only non-isomorphic solutions. This results in a shorter computation time and more evaluation of structurally different solutions, with a higher probability of finding a better allocation solution than traditional methods. The passenger will benefit from this in terms of reduced travel times and better transport services in general. Cost-effective implementation of larger elevator groups is also possible because of the lower computing capacity of the equipment needed to solve the allocation task compared to conventional solutions. The solution is also cost-effective because it is software-based and does not require

X

£ installation of additional equipment for elevator group control, which also applies to existing elevator groups. The solution according to the invention can also advantageously be charged for different types of elevator systems, such as elevator groups 0X1 35, where the call output is based on 8 up / down call buttons, destination call panels or both, as well as elevator groups having single carriageways, both.

5 PHOTO LIST

The present invention will be described in the following paragraph by way of exemplary embodiments, wherein: Figure 1 illustrates an elevator group in which the method of the invention has been applied, Figure 2 shows an example of a symmetry-based allocation task, Figures 3a, 3, b, 3c 4 shows steps of the method according to the invention, 15 shows another example of a symmetry-based allocation task, and Figures 6a-6h show some solutions of the allocation problem of Figure 5 in connection with a genetic algorithm.

DETAILED DESCRIPTION OF THE INVENTION

The following defines the meaning of the terms used in this patent application: o - destination call panel: An elevator group call device that allows a passenger to already provide a destination call outside the elevator car, i £ - destination call: a call made using a destination call panel, 1 - that the target layer cc and any other call-related information, oj such as the number of passengers traveling the call 30, oo - external call: A call made using the up / down call buttons, including information about the base layer if 9 calls were received and any directional information , downlink), - basket command: In the elevator car, the call given by the basket buttons, 5 - the call: Any of the above calls, - allocation task: an optimization task to allocate invitations to elevator cars, i.e., to select an allocation with the best value a within the given optimization criteria, 10 - Goodness value: A scalar goodness value calculated for allocation, on the basis of which different allocation solutions can be compared with each other. The value of good is determined by a cost function that takes into account the constraints of elevators and elevator group control 15 and the quantities to be optimized, symmetric allocation task: Allocation task involving symmetric lifts, - symmetric lifts: Lifts that are in the same state. The status of the elevator depends, for example, on the position, speed, direction, acceleration, door position, cage command, fill rate and / or calls already allocated to the elevator car 20. A set of symmetrical lifts may consist of several subsets of symmetrical lifts in which all the lifts in one subset are in the same state, but the lifts in the different subsets are in a different state,

(M

c \ j - isomorphic solution: Two solutions are isomorphic to each other if another solution is obtained by rotating the indices of two symmetric h x x 30 sin, the oj - lexicographic order: The sequence of number sequences, where - ° the order of the numbers determines the order of the number sequences 10. For example, 112 is lexicographically smaller than 113, symmetry criterion: Specifies the conditions under which elevators are interpreted as either symmetrical or non-symmetrical with one another. Two or more elevators may be interpreted as symmetrical, even if their states are not exactly the same, but based on symmetry criteria close enough to each other. Symmetry criteria can be determined on a case-by-case basis and refined, e.g. whether there are single or multiple elevators in the elevator group.

Figure 1 shows an elevator group 100 applying the method of the invention. The elevator group comprises two elevators E1, E2, and a group control 110 which allocates passenger calls to the elevators, i.e., solves the allocation task. Elevator baskets 121 have korean command buttons 122 for issuing commands. The elevator group serves floors 1-10, in which lounges F1-F10 20 are provided with paging devices 123 for giving calls to the elevator group. The Lobby F1 Calling Device is of the type of a Destination Calling Panel, whereas the Lobby F2-F10 has Call Up Devices having up / down call buttons, i.e. so called. a hybrid solution. The type of paging devices 25 may also be selected in a manner other than that shown in Figure 1. Elevator baskets 121 are equipped with o-basket scales 124 and door photocells 125.

o J

persons who move to the elevator car during stops

(M

V and detecting departing passengers. Using δ 30 information from rivets and door photocells, the number of passengers in each elevator is updated

CL

sikorissa. The number of passengers in different elevator car co can be taken into account when defining symmetrical lifts, o

CD

o

o

C \ 11

The group control 110 comprises a software product 112 according to the invention for solving the allocation task, which when executed in the application environment takes into account the symmetry that may exist in the allocation task. Further, the group control comprises an interface 113 for receiving information needed to determine elevator states. Based on the status of the elevators obtained through the interface, symmetric elevators of the elevator group are determined at the time of allocation. Since the same interface used by group control to control elevators can be utilized to read the status information, no additional interface is needed to implement the solution according to the invention.

Figure 2 illustrates an elevator group allocation task for allocating invitations to the C1-C6 elevator group elevators E1-E6. In this example, each call comprises one passenger. The elevators in the elevator group are similar and serve floors F1-F13.

20 Elevators E1-E5 are stationary and empty. Elevator baskets for elevators E1-E3 are located on floor F1 and elevators baskets for elevators E4-E5 on floor F3, respectively. In other words, the elevators E1-E3 are in the same state with each other and E4-E5 are in the same state with each other. It can be seen from Fig. 2 25 that the elevators E1 and E4 are not in the same state, so that the elevators E1-E3 form one subset of symmetrical elevators and the elevators E4-E5 another subset of symmetric elevators. Elevator E6 is stationary i on floor F2 and has two basket commands on floors i £ 30 F5 and F13. All invited passengers are 1 going to the lowest floor, F1.

CL

(M

Figures 3a, 3b and 3c show three different solutions for the allocation task according to Fig. 2, S1, S2 and S3. In the carriage SI (Fig. 3a), the calls issued by the passengers 12 are allocated as follows: call C1 to elevator E3, call C2 to elevator E6, call C3 to elevator E5, call C4 to elevator E4, call C5 to elevator E1 and invitation C6 to elevator E2. In Figures 3a, 3b, 3c, the reinforced horizontal lines separate the symmetric elevator sets from one another so that the elevators E1-E3 are symmetrical with each other and the elevators E4-E5 are symmetrical with each other. Because elevators E1-E3 and E4-E5 are symmetric subsets, alternative allocation solutions of equivalent cost function (goodwill) can be determined by re-indexing the symmetric elevators. For example, by swapping the alloying solution SI (Figure 3a), the elevators E1 and E2 induce a solution that is otherwise the same as SI, but the call C5 is allocated to the elevator E2 and the call C6 to the elevator E1, i.e. the invitations of the elevators E1 and E2 are exchanged. by changing the indexes of elevators E1 and E2. The total number of alternative isomorphic allocation solutions is (3 * 2 * 1) * (2 * 1) = 12, because the symmetrical elevators E1-E3 can be indexed in 3! 20 (three-fold) differently and independently of them symmetrical elevators E4-E5 can be indexed 2! (multiplied by two) differently.

If the symmetry of the elevators is not taken into account in this example case, the optimization algorithm of the allocation task performs unnecessary computation when evaluating isomorphic solutions. The optimization algorithm can be enhanced by limiting the solution search space to a subset i £! wherein the allocation solutions are non-^ 30 isomorphic to each other. For example, the delimitation can be done such that the search focuses on a subset in which the lexicographically smallest or largest lexicographies of the lifts are symmetric. The allo-

C/O

The solution solutions S2 and S3 are examples of lexicography 35 for the largest and largest solution for solution SI.

13

Figure 4 illustrates the steps of the method of the invention. The method is executed in the application environment at the time of allocation. In step 1, the necessary information is initialized in the optimization al-5 gorithm, such as determining the state of all elevators using interface 113. In step 2, based on the information of the first step, the symmetrical lifts of the elevator group are determined based on the given symmetry criteria. If there are no sym-10 metric elevators, the allocation task will normally be solved using any known solution method (step 4). If symmetric lifts are present, a subset of suitably delimited ones, in which the allocation paths are non-isomorphic to one another, is selected as the search space (step 5). A valid initial solution is then generated that belongs to the selected subset (step 6). A starter solution may also mean a set of starter solutions in this context. Next, the current initial solution is iteratively improved using a suitable optimization algorithm until one of the given termination conditions is fulfilled (steps 7-10). During the iteration, after each operation of the optimization algorithm, it is examined whether the new solution generated during the iteration belongs to the selected subset (steps-25 to 8). If not, the solution is returned to the selected subset by rearranging the indexes of the symmetrical lifts 0, i.e., by permutating the indexes of the symmetric lifts (step 9). In this connection,

(OF

In a cyj operation, the operation may comprise one or more different operations. Step 9 is performed because the operation of the optimization algorithm may take the solution outside the selected subset x £. If the operations of the optimization algorithm are selected so that the

CM

§ rolls out of the selected subset, phase

CD

Q 35 9 does not need to be performed, o

CM

14

Next, the method according to the invention will be described when a genetic algorithm is used as an optimization algorithm for the allocation task. For genetic algorithms reference is made to US 5907137. Figure 5 5 shows the initial situation of an elevator group with four symmetrical elevators E1-E4 and one non-symmetrical elevator E5, and six C1-C6 allocations by passengers. The asymmetrical elevator E5 has two basket commands for floors F5 and F7. For simplicity, only the two chromosomes (allocation variants) f1 and f2 of the genetic algorithm are considered. Initially, a viable initial solution is formed using methods known per se. As a starting point, the following allocation options (chromosomes) are selected: f15 = [E2, E3, E1, E4, E1, E3] and f2 = [E4, E5, E2, E3, E4, E1]. In the kro mosome fl, the call C1 is allocated to elevator E2, the call C2 to elevator E3, etc. In Figures 6a and 6b, chromosomes fl and f2 are shown in tabular form. In Figures 6a-6h, the elevators are divided into symmetrical elevators E1-E4 and non-20 symmetrical elevators E5, separated by a reinforced horizontal line. To obtain chromosomes in the desired subset, permutation is performed on chromosomes f1 and f2 by lexicographically indexing symmetric elevators E1-E4 as shown in Figures 6c and 6d. After permutation, chromosome fl becomes fl '= [E1, E2, E3, E4, E3, E2] (Figure 6c) and chromosome f2 becomes f2' = [E1, E5, E2, E3, E1, E4] (Figure 5 6d) > ·

(M

C \ l £ 30 Next, genetic operations such as x selection, crossing and / or mutation of a new chromo-

CC

to generate a mosquito population. First let's consider cm, for example, the crossover where the last three genes of chromosomes f1 'and f2' are exchanged to give chromosomes f1 '' = [E1, E2, E3, E3, E1, E4] and f2 '' = [E1, E5, E2, E4, E3, E2] shown in tabular form in Figures 6e and 6f, it is seen that f1 '' (Fig. 6e) fits into the selected subset, while f2 '' (Fig. 6f) is not in the selected subset. By performing the change of indices E3 and E4 of the elevators, i.e., the lines E3 and E4 of the elevators 5 are interchanged in Figure 6f, f2 '' is obtained for the selected subset.

In the selection, the best chromosomes are copied to the next generation chromosome population as such, so that they belong to the selected subset without the need for permutation. In a mutation, a random gene is selected from a randomly selected chromosome whose allele, or value, is changed to another allele. For example, if the value of the second gene in Figure 6c is changed from allele E2 to allele E3, the chromosome f1 '' '= [E1, E3, E3, E4, E3, E2] in Figure 6g is obtained which is not part of the selected subset operation as shown in Figure 6h.

In the genetic algorithm described above, the initial control is initially transferred to a selected subset. This is followed by genetic operations and, if necessary, restoring the chromosomes to a selected subset. Thus, the search focuses on the part of the search space where 25 are the best solutions of goodness value while the search space remains small. As a result, the described algorithm runs faster and produces a better allocation solution under symmetry than the traditional genetic algorithm.

δ X 30 tr

CL

The invention is not limited solely to the foregoing examples, but many modifications are possible within the scope of the inventive idea defined by the claims. Thus, for example, the search space delimitation may be accomplished in some other way than those described above. The invention can also be applied to elevator groups having one or more multi-car lifts as long as the symmetry criterion takes into account the interdependencies between the elevator cars.

5 o δ

(M

(M

δ

X

I do not

CL

(M

C/O

o

σ> o o

(M

Claims (12)

A method for solving an allocation task in an elevator group, which elevator group comprises a number of elevators, callers for registering the calls of the passengers and equipment for identifying the status of the elevator group's elevators, characterized in that the procedure comprises the steps: before each allocation task is resolved, they are determined symmetrical lifts in the lift group on the basis of the status of the lifts; and if there are 15 symmetrical elevators, the search space associated with the allocation task is delimited to a subset that does not contain relative to one another isomorphic solutions. Process according to claim 1, characterized in that the method comprises the steps of: a useful initial solution is created; the solution is iteratively improved by the optimization algorithm for the allocation task, after one or more operations of the optimization algorithm is checked if the solution is still left in the selected subset; and if it is not, the solution is redirected to the selected subset. Method according to claim 1 or 2, characterized in that the solution of the allocation task is returned to the selected subset by permitting the indexes of symmetrical lifts to be (or M) the lexicographically the least or greatest. Method according to any one of claims 1-3, characterized in that as the optimization algorithm for the x allocation task, a genetic algorithm is used, whereby every solution formed by vai and / or crossing co o and / or mutation is returned to the selected subgroup. Device for solving an allocation task in an elevator group 100, comprising elevator group 5 of a number of elevators E1, E2 and callers 123 for recording the passengers' calls, characterized in that the device comprises an interface 113 for receiving data needed to determine the the status of the lifts and that the device is arranged to determine the symmetrical lifts in the elevator group prior to the solution of each allocation task on the basis of the elevators' status and to define the search space associated with the allocation task in a subgroup that does not contain in relation to each other isomorphic isomers. Device according to claim 5, characterized in that the device is designed to check, after one or more operations of the optimization algorithm used in the solution of the allocation task, whether the solution remains in the selected subgroup and if it is not the return of the solution to the selected subgroup. subgroup. Device according to claim 5 or 6, characterized in that the device is the group control 110 of the elevator group. 8. Software for solving an allocation task in an elevator group, which elevator group comprises a number of lifts, callers for registering passengers' calls and equipment for identifying the status of the elevator group's elevators, characterized in that the software comprises program code which is arranged at? 30 execution in a data processing equipment before each allocation task is solved determine the symmetric lifts CM V in the lift group on the basis of the status of the lifts and if o symmetric lifts are defined, the associated allocation task pertaining to each other in a subgroup that does not contain a subset 35 o solutions. CT) § 9. Software according to claim 8, CM characterized in that the software comprises program code 5 which, when executed in a data processing equipment after one or more operations of the optimization algorithm used in the solution of the allocation task, check whether the solution remains in the selected subgroup, and if it is not, return the solution to the selected subset. 10. Software according to claim 8 or 9, characterized in that the software is stored on a memory in the data processing equipment. Software according to any of claims 8 to 10, characterized in that the data processing equipment is the group control of the elevator group. Software according to claim 11, characterized in that the software is subsequently installed in an existing group control for an elevator group. o δ (M (M δ X and CL) M CO O σ> o o {M