JP4621620B2 - Elevator group management system, method and program - Google Patents

Description

  The present invention relates to an elevator group management system, method and program.

  The elevator group management system aims to improve the operation efficiency of the elevator and improve the service to passengers. With the group management algorithm, an optimal elevator car is created for hall calls that occur at any time on any floor in the building. Assign and control its operation. When the group management algorithm assigns an elevator car to a hall call that has occurred, in addition to the service situation for all hall calls currently occurring in the entire building, the car call situation, the current location of the car, and the car Car information such as the number of passengers is added. The elevator group management control policy adopted by the elevator group management system is to improve average elevator performance in consideration of the entire building, and the system is configured to allocate elevator cars according to this control policy.

  In selecting an elevator car to be assigned to a hall call from among a plurality of elevator cars, each elevator car is evaluated. In the conventional elevator group management system, each evaluation term in the evaluation function used for this evaluation is weighted by a weight parameter corresponding to the traffic demand and the control target. That is, when selecting an elevator car to be assigned to a hall call from among a plurality of elevator cars, the weighting parameter for each evaluation term is changed to an evaluation function for evaluating each elevator car according to traffic demand and control target.

  A technique using a neural network as a weight parameter adjuster of the evaluation function is proposed in Japanese Patent Laid-Open No. 2-226425. This technology allows the neural network to learn the elevator performance when the car assignment control is performed with a certain weight parameter, and outputs the predicted performance at that time by inputting the traffic demand and the weight parameter. The neural network input layer is composed of traffic demand and weight parameters, and the output layer is composed of elevator performance. When the traffic demand and weight parameters are input, the predicted performance at that time is output. Playing a role. A weight parameter is selected based on the output prediction performance.

  Further, for example, a technique for predicting elevator performance based on simulation (simulation experiment) when an elevator simulator is installed in an elevator group management system and control is performed with an arbitrary weight parameter in an arbitrary traffic demand This is proposed in Japanese Patent No. 2000-318938.

This technology also describes that a request for a control target based on a user's sensitivity is received and elevator group management control is performed in response to the request.
Japanese Patent Application No. 2-226425 JP 2000-318938 A

  While tracking the traffic demand and the control target by adjusting the weight parameter described above can improve the average performance considering the entire building, it is possible to improve the average performance between the floors. It is difficult to realize control with smaller granularity. As for the current building structure, building forms such as a company-occupied building are decreasing, a plurality of tenants are moving into the building, and more complex building forms are being replaced. In such a form, there are many changes in the movement between specific floors according to the time zone, and the conventional technology that improves the average performance of the entire building performs group management control corresponding to the building form I can not say. Therefore, instead of improving the average performance of the entire building, it is desirable to provide an elevator group management system that can set elevator control targets in units of combination of starting and destination floors, that is, OD (Origin Destination) units. ing.

  The present invention provides an elevator group management system, method, and program for realizing group management control toward an elevator control target set in combination units of a departure floor and a destination floor.

The elevator group management system as one aspect of the present invention includes:
An elevator group management system that executes group management control for assigning an elevator car to a hall call that occurs at an arbitrary time on an arbitrary floor in a building,
A traffic demand detector that detects traffic demand in the building and generates traffic demand data holding traffic demand for each combination of a departure floor and a destination floor;
An elevator control target storage unit that stores elevator control target data that holds an elevator control target value for each combination of a departure floor and a destination floor;
A virtual hall call list storing at least one or more virtual hall call information representing a virtual hall call, elevator control result data holding an elevator control result for each combination of the departure floor and the destination floor, and the departure floor And a case data storage unit that stores a plurality of case data corresponding to traffic demand data that holds traffic demand for each combination of destination floors,
The traffic demand data is received from the traffic demand detection unit, the elevator control target data is received from the elevator control target storage unit, and the received traffic demand data and the elevator control target data are used in the case data storage unit. A virtual hall call list acquisition unit that performs similarity calculation with each case data, selects the case data based on the result of the similarity calculation, and acquires a virtual hall call list included in the selected case data;
A car call determining unit for determining a derived car call derived from each of the virtual hall calls shown in the acquired virtual hall call list;
Information on hall calls that have been assigned to each elevator car and that have not been answered is stored as assigned hall call information, and information on unanswered car calls that have occurred in each elevator car is stored as unanswered car call information. A basket stop information storage unit;
A hall call detector for detecting hall calls in the building;
A car call prediction unit for predicting a derived car call derived from the detected hall call;
Using the allocated hall call information and the unanswered cage call information, a floor patrol schedule for each elevator car is generated, and each virtual hall call indicated in the acquired virtual hall call list is determined. Each derived car call, the hall call detected by the hall call detecting unit, and the predicted derived car call are added to each of the floor patrol schedules by a plurality of additional patterns, and added by the additional pattern. Calculating an evaluation value of the additional pattern based on each of the floor patrol schedules, selecting an additional pattern based on the evaluation value calculated from each of the additional patterns, and detecting the detected hall call from the selected additional pattern. A car assignment calculation unit for determining an elevator car to be assigned to
Is provided.

The elevator group management method as one aspect of the present invention includes:
An elevator group management method for executing group management control for assigning an elevator car to a hall call generated at an arbitrary time on an arbitrary floor in a building,
A traffic demand detecting step for detecting traffic demand in the building;
A traffic demand data generation step for generating traffic demand data holding traffic demand for each combination of a departure floor and a destination floor;
An elevator control target setting step for setting elevator control target data holding an elevator control target value for each combination of the departure floor and the destination floor;
A virtual hall call list storing at least one or more virtual hall call information representing a virtual hall call, elevator control result data holding an elevator control result for each combination of the departure floor and the destination floor, and the departure floor And a preparation step of preparing a database storing a plurality of case data corresponding to traffic demand data holding traffic demand for each combination of destination floors,
Using the generated traffic demand data and the set elevator control target data, perform similarity calculation with each case data in the database, and select the case data based on the result of the similarity calculation A virtual hall call list acquisition step of acquiring a virtual hall call list included in the selected case data;
A car call determining step for determining a derived car call derived from each of the virtual hall calls included in the acquired virtual hall call list;
Step of storing hall call information assigned to each elevator car and unanswered as assigned hall call information, and information of unanswered car call generated in each elevator car as unanswered car call information When,
A hall call detecting step for detecting a hall call in the building;
A car call prediction step for predicting a derived car call derived from the detected hall call;
Using the allocated hall call information and the unanswered cage call information, a floor patrol schedule for each elevator car is generated, and each virtual hall call indicated in the acquired virtual hall call list is determined. Each derived car call, the hall call detected by the hall call detection step, and the predicted derived car call are added to each floor patrol schedule by a plurality of additional patterns, and added by the additional pattern. Calculating an evaluation value of the additional pattern based on each of the floor patrol schedules, selecting an additional pattern based on the evaluation value calculated from each of the additional patterns, and detecting the detected hall call from the selected additional pattern. A car assignment calculation step for determining an elevator car to be assigned to
Is provided.

The elevator group management program as one aspect of the present invention,
An elevator group management program for causing a computer to execute group management control for assigning an elevator car to a hall call generated at an arbitrary time on an arbitrary floor in a building,
A traffic demand detecting step for detecting traffic demand in the building;
A traffic demand data generation step for generating traffic demand data holding traffic demand for each combination of a departure floor and a destination floor;
An elevator control target setting step for setting elevator control target data holding an elevator control target value for each combination of the departure floor and the destination floor;
A virtual hall call list storing at least one or more virtual hall call information representing a virtual hall call, elevator control result data holding an elevator control result for each combination of the departure floor and the destination floor, and the departure floor And a preparation step of preparing a database storing a plurality of case data corresponding to traffic demand data holding traffic demand for each combination of destination floors,
Using the generated traffic demand data and the set elevator control target data, perform similarity calculation with each case data in the database, and select the case data based on the result of the similarity calculation A virtual hall call list acquisition step of acquiring a virtual hall call list included in the selected case data;
A car call determining step for determining a derived car call derived from each of the virtual hall calls shown in the acquired virtual hall call list;
Step of storing hall call information assigned to each elevator car and unanswered as assigned hall call information, and information of unanswered car call generated in each elevator car as unanswered car call information When,
A hall call detecting step for detecting a hall call in the building;
A car call prediction step for predicting a derived car call derived from the detected hall call;
Using the allocated hall call information and the unanswered cage call information, a floor patrol schedule for each elevator car is generated, and each virtual hall call indicated in the acquired virtual hall call list is determined. Each derived car call, the hall call detected by the hall call detection step, and the predicted derived car call are added to each floor patrol schedule by a plurality of additional patterns, and added by the additional pattern. Calculating an evaluation value of the additional pattern based on each of the floor patrol schedules, selecting an additional pattern based on the evaluation value calculated from each of the additional patterns, and detecting the detected hall call from the selected additional pattern. A car assignment calculation step for determining an elevator car to be assigned to
Is executed by the computer.

  ADVANTAGE OF THE INVENTION According to this invention, the group management control toward the elevator control target set in the combination unit of the departure floor and the destination floor is realizable.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a schematic configuration of an elevator group management system according to an embodiment of the present invention.

  The elevator group management unit 0 has at least two elevator cars (elevator car 11, elevator car 12,..., Elevator car 1N). ) Group management control.

  The elevator group management unit 0 is largely provided with two modules. One is a car assignment operation processing unit (car assignment operation unit) 01 that determines assignment of a car to a hall call, and the other is a plurality of parameters (virtual call lists) used in the car assignment operation processing unit 01. This is a basket assignment calculation processing parameter selection unit 02 for selecting from the virtual call list. The basket allocation calculation processing unit 01 determines a basket allocation for the hall call by performing a basket allocation calculation process based on the virtual call list selected by the basket allocation calculation parameter selection unit 02.

  The basket allocation calculation processing unit 01 performs a basket allocation calculation for a hall call generated at an arbitrary time in an arbitrary floor in the building based on a parameter (virtual call list) passed from the basket allocation calculation parameter selection unit 02. Processing is performed to determine the basket to be allocated for the hall call. The car assignment calculation parameter selection unit 02 selects a virtual call list to be used by the car assignment calculation processing unit 01 from a plurality of virtual call lists according to the traffic demand to be applied to the hall call and the elevator control target. To do.

  Hereinafter, the car assignment calculation processing unit 01 and the car assignment calculation parameter selection unit 02 will be described in detail.

  First, the basket allocation calculation processing unit 01 will be described.

  Passengers who wish to board the elevator in the building press the hall call button 2X (X = 1, 2,..., M) and request the elevator group management unit 0 to dispatch the elevator car (hall call). Do. “Vehicle allocation” means that an elevator car is allocated in response to a hall call. The elevator car dispatch request is detected by the hall call detection unit 04 in the form of “generation of a new hall call”. The detected hall call information includes the hall call occurrence time, hall call occurrence floor, hall call direction (UP or DOWN), and the like.

  The cage information detection unit 05 detects the cage information of each cage such as “in-car load”, “cage travel direction”, “car speed”, and “car current position”.

  The car call detection unit 06 detects the generation of a car call when the car call button (112-1N2) provided in each car (11-1N) is pressed by an elevator passenger, and detects the car call detected. The information is passed to the basket stop information storage unit 03. The car call information includes, for example, a destination floor (destination floor) and car identification information, and may further include the time of occurrence of the car call.

  The car stop information storage unit 03 stores the car call information passed from the car call detection unit 06. Further, when the basket allocation calculation processing unit 01 determines the allocation of the car to the hall call, the basket stop information storage unit 03 receives the determined basket identification information and the basket allocation target from the basket allocation calculation processing unit 01. The set with the hall call information is received and stored as hall call assignment information (cage identification information, hall call occurrence time, hall call occurrence floor, hall call direction). The hall call allocation information may include an allocation time at which allocation to the hall call was performed. When the processing for an unanswered call (unanswered hall call, unanswered cage call) is completed, the cage stop information storage unit 03 deletes information about the call (hall call allocation information, basket call information). An unanswered hall call is a hall call that has already been assigned to a car but has not yet been dispatched. An unanswered car call is a car call that has not yet stopped at the target floor. Depending on the group management processing system, for example, a basket other than the basket originally assigned to the hall call arrives first due to the occurrence of the car call, etc., and passengers waiting for the floor arrive first. There is a case where a so-called “basket call first arrival” is generated. In such a case, the hall call assignment information for the hall call may not be deleted.

  The basket allocation calculation processing unit 01 calculates the hall call information from the hall call detection unit 04 at a timing at which the hall call information can be calculated (for example, when calculation for another hall call is being performed) Etc.) after receiving. Upon receiving the hall call information, the car assignment calculation processing unit 01 receives information stored in the car stop information storage unit 03 and information detected by the car information detection unit 05, respectively.

  The basket allocation calculation processing unit 01 performs the basket allocation calculation processing using the received information and the virtual call list selected by the basket allocation calculation parameter selection unit 02, and is detected by the hall call detection unit 04. The basket to be assigned to the hall call made is determined. When the basket allocation calculation processing unit 01 determines the basket to be allocated to the hall call, the basket control unit 1X1 (X = 1, X) corresponding to the basket 1X (X = 1, 2,..., N). 2,..., N) is notified of the hall call assignment. Specifically, the basket allocation calculation processing unit 01 issues a stop instruction (including the generation floor (stop floor) and the direction of the hall call) to the determined basket.

  The basket control unit 1X1 has a hall call already assigned to the basket 1X (X = 1, 2,..., N) and a basket call button 1X2 (X = 1, 2,... N in the basket 1X. ) And the current car position information and the information on the car state defined by the system (stopped, moving, door open, etc.) The car is autonomously controlled in accordance with the determined order of the floor of the car.

  The basket allocation calculation processing unit 01 includes a search calculation processing unit 011, a search calculation data storage unit 012, and a model evaluation unit 013 for performing a basket allocation calculation process.

  The search calculation processing unit 011 is based on the parameters (virtual call list) determined by the parameter determination unit 025 in the basket allocation calculation parameter selection unit 02 and the information from the basket stop information storage unit 03 and the basket information detection unit 05. Search calculation processing (basket allocation calculation processing) is performed.

  Here, the virtual call list will be described. FIG. 4 is a conceptual diagram of a virtual call list.

  The virtual call list is a list of virtual call sets. Here, N virtual call sets are listed. The virtual call set includes hall calls that do not actually occur in the future and cage calls that are derived therefrom. More specifically, the virtual call set includes a call set ID, virtual hall call information (virtual hall call information), and cage call information derived from the virtual call (derived car call information). including. The virtual hall call information includes at least “occurrence time (current time is 0 second)”, “occurrence floor”, and “occurrence direction (hall call direction)”. The derived basket call information includes at least one “target floor (destination floor)”. Here, the virtual call list stores a plurality of virtual call sets, but the virtual call list may store only a plurality of virtual hall call information instead of the virtual call sets. In this case, the search calculation processing unit 011 performs processing for obtaining derived cage call information from each virtual hall call information at the time of search calculation processing or when a new virtual call list is selected. That is, the search calculation processing unit 011 may include a cage call determining unit that obtains derived cage call information from each virtual hall call information. The derived cage call information may be obtained, for example, by inputting virtual hall call information into a function prepared in advance or may be obtained randomly.

  The virtual call list includes virtual data between floors where demand is expected to increase, especially between floors where service is expected to improve, based on past operation record data and tenant occupancy. Include multiple call sets. That is, a large number of future virtual calls are generated between specific ODs by using this virtual call list, thereby performing a basket allocation calculation process considering the call generation status between future ODs. In this embodiment, by using a virtual call list in this way and generating a large number of future virtual calls between specific ODs, it is possible to control the allocation of a car that focuses on the specific ODs.

  In this elevator group management system, a plurality of such virtual call lists are prepared as shown in FIG. The parameter determination unit 025 in the car assignment calculation parameter selection unit 02 to be described later selects an appropriate virtual hall call list from the plurality of virtual hall call lists according to the traffic demand and the elevator control target. The data is transferred to the search calculation processing unit 011 in the allocation calculation processing unit 01.

  When the search calculation processing unit 011 receives the hall call information from the hall call detection unit 04, the search calculation processing unit 011 generates the derived cage call information by predicting the cage call derived from the hall call of the hall call information. That is, the search calculation processing unit 011 may include a car call prediction unit that predicts a derived car call from a hall call. Derived basket call information may be obtained, for example, by inputting hall call information into a function prepared in advance, or may be obtained randomly. The search calculation processing unit 011 uses a combination of the received hall call information and the generated derived cage call information as a call set, and an allocation in which the virtual call list determined by the parameter determination unit 025 is added after the call set. Generate a target call list. FIG. 6 shows an example of the allocation target call list in which the call set is added to the virtual call list No. 1 shown in FIG.

  The search calculation processing unit 011 is an optimal or sub-optimal basket allocation pattern (basis temporary allocation pattern) for a plurality of call sets included in the allocation target call list (a virtual call set is also simply referred to as a call set), that is, described later. A search calculation process for finding a plurality of call set addition patterns to the model data to be executed is executed. The search calculation processing unit 011 is based on the basket temporary allocation pattern found by the search calculation processing, and the cage 1X (X = 1, 2,..., N) to which the hall call detected by the hall call detection unit 04 should be allocated. And the hall controller 1X1 of the specified basket 1X is notified of the hall call assignment.

  The search calculation processing by the search calculation processing unit 011 is controlled by a calculation control unit (not shown). This calculation control unit manages the calculation processing time, and the search calculation processing unit 011 performs a search repeatedly until the search calculation time limit managed by the calculation control unit is reached. When all the processes are completed before the search calculation time limit is reached, an optimal basket temporary allocation pattern is found. On the other hand, when all the processes are not completed before the search calculation time limit is reached, an optimal (ie, sub-optimal) temporary basket allocation pattern is found within the range in which the processes are performed. The sub-optimal basket temporary allocation pattern may or may not match the optimal basket temporary allocation pattern. It is assumed that the search calculation time limit can be arbitrarily changed by setting the calculation control unit.

  FIG. 9 is a flowchart showing an outline of the entire search calculation process.

  First, in step A1, derived car call information is generated from hall call information generated at an arbitrary floor at an arbitrary time, a call set that is a combination of the hall call information and the derived car call information, and a parameter determination unit 025. An allocation target call list is created in combination with the already determined virtual call list.

  Next, in step A2, the car information from the car information detecting unit 05 (for example, the current position of each car (the floor position where the car can start decelerating when moving) and the car stop information storage unit 03). Based on the car call information and hall call assignment information from, the floor patrol schedule of each car at the current time is obtained. Then, model data that is a set of floor patrol schedules for each cage obtained is created. The model data created here is referred to as initial model data.

  Here, the model data will be described.

  The model data is data that collectively represents the floor patrol schedule for each cage as described above. The floor tour schedule for each cage can be estimated at high speed by performing a simple simulation (simulation) based on the current location of each cage, the cage call information of each cage, the hall call allocation information of each cage, etc. is there.

  FIG. 7 shows an example of model data.

  The row data is data relating to the stop of the car to the floor (hereinafter referred to as the car stop data). “Data No” is an ID for identifying the basket stop data. The “car number” is a car ID assigned to each car, and in this example, an example in which there are two cars is shown. “Floor” refers to the ID of the floor where the basket stops. “Direction” means UP hall call stop if UP, DN hall stop if DN, C call stop if C. “Occurrence time” indicates the time at which a hall call that causes a car stop occurs, based on the current time (= 0). “Arrival time” represents the scheduled time when the basket arrives at the floor with the current time (= 0) as a reference. “Departure time” represents the scheduled time when the basket leaves the floor with the current time (= 0) as a reference. “Previous stop floor data No.” is a data number indicating the car stop data for the previous stop, and if it is −1, it indicates that there is no previous stop floor. “Next stop floor data No” is a data number indicating the car stop data for the floor to be stopped next. If it is −1, it indicates that there is no next stop floor. The value of the “occurrence time” column in the car stop data indicating the car call stop is set to 0, but this does not mean that the car call stop occurred at the current time. That is, as will be described later, the model evaluation unit 013 evaluates the model data that is sequentially updated. In the present embodiment, the evaluation value includes, as an evaluation value, a non-response time, a (car arrival time to an arbitrary floor, and its arbitrary floor) The time difference from the time at which the hall call occurred that causes the car to stop. The calculation of this unanswered time does not require the time at which the cage call occurred. Therefore, in the case of the cage call stop, 0 is entered in the occurrence time column for convenience.

  In the example of FIG. 7, for the sake of simplicity, the time required to move between the first floor is 5 seconds, the stop time of the cage accompanying the floor stop is the hall call, and both the car calls are 5 seconds. The time is being calculated. In practice, these values are determined according to the elevator specifications. The order of traveling around the floor is expressed by tracing “previous stop floor data No” and “next stop floor data No”, and the model data has a data structure like a so-called linked list. Here, as a rule for creating model data, when hall call stop and basket call stop occur on the same floor, hall call stop is processed after the cage call stop processing. This corresponds to the fact that in a normal elevator system, it is normal for a passenger to get on after getting off. Such processing is an example, and the present invention is not limited to such an example.

  Returning to FIG. 9, in step A3, the initial model data created in step A2 is set at the head of the search list. The search list is a list that can store a plurality of model data. In step A2, initial model data is set at the head of such a search list. The search list is stored in the search calculation data storage unit 012 together with a storage list to be described later.

  In step A4, based on the initial model data set in the search list, the search calculation processing unit 011 and the model evaluation unit 013 optimize or quasi-optimize a plurality of call sets included in the allocation target call list. A search calculation process for finding an allocation pattern (basic allocation pattern) of a basket is performed.

  FIG. 10 is a flowchart showing in detail the flow of processing performed in step A4.

  In step B1, it is checked whether the search list is empty. If the search list is empty (YES in step B1), the process in step A4 is terminated, and the process proceeds to step A5 in FIG. If the search list is not empty (NO in step B1), model data is extracted (copied) from the top of the search list (step B2), and the process proceeds to step B3.

  In Step B3, it is checked whether or not all the call sets in the assignment target call list have been provisionally assigned (added) to the model data extracted in Step B2. If all call sets have been provisionally assigned (YES in step B3), the process proceeds to step B4.

  In step B4, the model data is stored in the storage list, and the model data is deleted from the search list, and the process returns to step B1. Here, the storage list is a list for storing model data for which provisional allocation of all call sets has been completed. The saving list can store a plurality of model data.

  On the other hand, in step B3, if there is a call set that has not been temporarily assigned in the extracted model data (NO in step B3), the process proceeds to step B5.

  In step B5, a call set N that has not been provisionally allocated (not added) to the extracted model data is acquired from the allocation target call list.

  In step B6, the extracted model data is searched for a place where a hall call included in the acquired call set N can be added (inserted).

  In step B7, the extracted model data is searched for a place where the derived basket call included in the call set N can be inserted. In step B7 and step B6, the place where insertion is possible is a place that satisfies the physical restrictions and the restrictions on the operation rules of the cage that must be kept to a minimum so as not to cause discomfort to the passengers.

  In step B8, a copy of the extracted model data is generated.

  In step B9, the call set N (a hall call and a derived car call) is inserted into this copy to update the model data.

  Here, step B6 to step B9 will be described in more detail using specific examples.

  Using the model data shown in Fig. 7 as an example, the new hall call "Upstairs hall call on the 5th floor" and its derived basket call "Kago call to the 6th floor" are temporarily assigned to the first car, and the model data is updated. An example will be described.

  FIG. 8 is a diagram for explaining a process of inserting “5th floor upward hall call” and “6th floor hall call” into the model data shown in FIG.

  In FIG. 8 (A), the model data shown in FIG. 7 is searched for a place where “5th floor upward hall call” can be inserted, and in FIG. 8 (B), the derived basket call “cargo to the 6th floor”. A place where “call” can be inserted is searched, and these are inserted into the searched place as shown in FIG. At this time, the arrival time and departure time, etc. of the floor that stops after the stop by the “5th floor upward hall call” are changed, so these arrival time and departure time depend on the moving time of the elevator and the car stop time. Etc. are corrected. FIG. 11 shows a state in which “5th floor upward hall call” and “6th floor hall call” are inserted into the model data of FIG. 7 to update the model data.

  In FIG. 11, the new basket stop data is added to the end of the model data. From FIG. 8 (B), since the new “5th floor hall call” is inserted after “5th floor basket call” indicated by data No4, “next stop floor data No” of data No4 is updated to 16, “No. of previous stop floor data” = 4 of data No. 16 indicating “5th floor upward hall call”. Similarly, “No. 5th floor hall call” indicated by data No. 16 will be the new “Kag call to 6th floor”, so “No. of next stop floor data No.” = 17 in data No. 16 and “No. Floor data No ”= 16,“ Next stop floor data No ”= 5. In addition, since the “arrival time” and “departure time” of the subsequent car stop data (No. 5 to No. 9) change due to the insertion of the new car stop data, depending on the car floor movement time and the car floor stop time, These basket stop data are also updated. In this way, when a new call (hall call, basket call) is inserted, the “arrival time” and “departure time” of the car stop data after the place where it was inserted is updated, and “ The “stop floor data No” and “next stop floor data No” are also updated.

  Returning to FIG. 10, as described above, in step B9, the call data set N (a hall call and a derived basket call) is inserted into the extracted copy of the model data to update the model data.

  In the next step B10, the evaluation value of the updated model data is calculated. Here, the evaluation value is the sum of the unanswered times (the time from the hall call occurrence time to the arrival of the cage) of each hall call included in the model data. However, the evaluation value may be “maximum unanswered time”, “service time (time from arrival at the floor to arrival at the target floor for the hall call)”, or the like, but is not limited thereto.

  In step B11, the updated model data is added to an empty place in the search list.

  In step B12, each model data in the search list is sorted in the order of high evaluation (in this case, the evaluation value is small).

  In step B13, it is determined whether or not processing has been performed for all locations where the hall call of the call set N can be inserted in the extracted model data.

  If there is a part that has not yet been inserted (NO in step B13), the process proceeds to step B14 to search for another part into which the hall call in the call set N can be inserted. Then, the process returns to step B7, and a search is made for a place where the cage call in the call set N can be inserted.

  If it is determined in step B13 that processing has been performed for all locations where the hall call of the call set N can be inserted (YES in step B13), the process proceeds to step B15 and processing is performed for all assignable carts. It is determined whether or not. As a basket that cannot be assigned, there is a cage in which the floor where the hall call is generated corresponds to a non-stop floor. If the processing has not been completed for all the baskets, the process proceeds to step B16, the next basket is selected, and the process returns to step B6.

  On the other hand, if the processing has been completed for all the baskets (YES in step B15), the extracted model data is deleted from the search list, the process proceeds to step B17, and whether or not the search calculation time limit has been reached is determined. judge. If the search calculation time limit has not been reached (NO in step B17), the process returns to step B1. If the search calculation time limit has been reached (YES in step B17), the search calculation process is terminated, and the process proceeds to step A5 in FIG. In this way, when the search calculation time limit is reached and the search calculation time limit is reached, even if the temporary allocation of all the call sets for all the baskets is not completed, the search is terminated in the middle, so that in the elevator group management Real time realization is realized.

  In step B17, when the search calculation time limit is reached (YES in step B17), the process proceeds to step A5 as shown in FIG. 9, and the model data in the storage list is sorted in ascending order of evaluation value.

  In the next step A6, it is determined whether or not the storage list is empty.

  If the storage list is not empty (NO in step A6), the process proceeds to step 7, the first model data is acquired from the storage list, and the process proceeds to step A9.

  If the storage list is empty (YES in step A6), the process proceeds to step A8, the first model data in the search list (sorted in ascending order of evaluation values) is acquired, and the process proceeds to step A9.

  In step A9, a basket to which a hall call generated in an arbitrary floor at an arbitrary time is specified from the model data acquired in step A7 or step A8, and the search calculation process ends.

  Instead of step A6, step A7, and step A8, the evaluation value of the top model data in the list for saving is compared with the evaluation value of the top model data in the search list, and model data with a good evaluation value is obtained. May be.

  In the above, in the model data sorting based on the evaluation values executed in step A5 and step B12, when the evaluation values of the model data are compared as they are, the evaluation values of the model data with a small number of hall calls that are temporarily allocated are obtained. This will improve the accuracy of the comparison. Therefore, for example, the following may be performed so that the comparison of evaluation values can be performed properly even between model data having different numbers of temporarily assigned hall calls. That is, an increase in evaluation value when a call set is temporarily assigned to one model data is stored, and an average of the increase values is obtained every time a call set is temporarily assigned. This average is used for adjustment as an estimated value. Specifically, at the time of comparison, a value obtained by multiplying the evaluation value of the model data with the fewer temporarily assigned call sets by the difference between the number of call sets and the average value of the increased values is added. The addition result is used as an evaluation value. The closer the estimated value (predicted value) is to the evaluation value increment when the call set is actually temporarily allocated, the more appropriate the search proceeds. As described above, a method of performing search by assigning priorities (evaluation) to search nodes (model data) using estimated values is called “A search” and is used in the field of artificial intelligence.

  Here, in order to deepen the understanding of the search calculation processing described with reference to the flowcharts of FIGS. 9 and 10, supplementary explanation of the search calculation processing will be given with reference to FIGS. 12 to 17.

  In the following supplementary explanation, a case where there are two elevator cars (No. 1 and No. 2) will be described as an example.

  The hall call to be assigned that occurred at any time and on any floor (the hall call detected by the hall call detection unit 04) is “2nd floor upward hall call”, and the “4th floor upward hall” Suppose there are two types of calls: “call” and “7th floor hall call” In other words, a total of three hall calls are stored in the allocation target call list. Here, for simplicity of explanation, the cage call derived from each hall call is not considered.

  First, as shown in FIG. 12, model data (initial model data) representing the current set of floor patrol schedules for each car is generated (step A2), and initial model data is set at the top of the search list (step A2). Step A3).

  Next, as shown in FIG. 13, a series of processing from Step B1 to Step B12 is performed.

  First, since initial model data exists in the search list (NO in step B1), the top model data, that is, initial model data is extracted from the search list (step B2). Since all hall calls in the allocation target call list have not been added to the initial model data (NO in step B3), the “second floor upward hall call” not added to the initial model data is acquired from the allocation target call list. (Step B5). Model data 1 in the case where “2nd floor upward hall call” is assigned to the first car is created as the initial model data (steps B6 to B9), and the evaluation value of the model data 1 is calculated (step B10). It is assumed that 50 evaluation values are obtained as a result of the calculation. The created model data 1 is stored in the search list (step B11), and the model data in the search list is sorted in the order of good evaluation values (in ascending order of evaluation values) (step B12).

  Since the model data for the case where the “second-floor upward hall call” is assigned to Unit 2 has not yet been created (YES in Step B13, NO in Step B15), as shown in FIG. (Step B16, Steps B6 to B9). Then, the evaluation value of the model data 2 is calculated (step B10), the model data 2 is stored in the search list (step B11), and the model data in the search list is sorted in ascending order of the evaluation value (step B12). . Assume that the evaluation value of the model data 2 is 60.

  Since processing has been completed for all the baskets for “upstairs hall call on the second floor” (YES in step B15), the initial model data is deleted from the search list, and the search calculation time limit has not been reached (step B17). NO) returns to step B1.

  Since the search list is not empty (NO in step B1), the top model data is extracted from the search list (step B2). Since the search list is sorted in ascending order of evaluation value in step B12, model data 1 with the best evaluation is extracted as shown in FIG. Since model data 1 is not model data in which all hall calls in the allocation target call list have been added (NO in step B3), the hall call that has not been added from the allocation target call list, here, “the 4th floor upward hall call” Is acquired (step B5).

  Then, a series of processes as shown in FIGS. 13 and 14 are performed. As shown in FIG. 15, model data 1-1 in which “4th floor hall call” is assigned to Unit 1 and “4th floor” is assigned to Unit 2. Model data 1-2 assigned “upward hall call” is created from model data 1. Assume that the evaluation values of model data 1-1 and model data 1-2 are calculated, and 40 is obtained as the evaluation value of model data 1-1 and 70 is obtained as the evaluation value of model data 1-2. Model data 1-1 and model data 1-2 are stored in the search list and sorted in ascending order of evaluation value. If the model data 1 is deleted from the search list and the time limit has not been reached (NO in step B17), the process returns to step B1.

  Since the search list is not empty (NO in step B1), the top model data is extracted from the search list (step B2). Since the search list is sorted in ascending order of evaluation value in step B12, model data 1-1 (evaluation value 40) having the highest evaluation is extracted as shown in FIG. Since model data 1-1 is not model data in which all hall calls in the allocation target call list are added (NO in step B3), an unadded hall call from the allocation target call list, “7th floor downward hall call” Is acquired (step B5).

  Then, as shown in FIG. 16, model data 1-1-1 in which “No. 7 floor downward hall call” is assigned to Unit 1 and model data 1-1-2 in which “No. 7 floor downward hall call” is assigned to Unit 2. Is created from model data 1-1. Then, the evaluation values of the model data 1-1-1 and the model data 1-1-2 are calculated, and 55 as the evaluation value of the model data 1-1-1, and 65 as the evaluation value of the model data 1-1-2. Suppose that it was obtained. Model data 1-1-1 and model data 1-1-2 are stored in the search list and sorted in ascending order of evaluation values. If the model data 1-1 is deleted from the search list and the time limit has not been reached (NO in step B17), the process returns to step B1.

  Since the search list is not empty (NO in step B1), model data 2, model data 1-2, model data 1-1-1, and model data 1-1-2 are stored in the search list. Therefore, the top model data is extracted from the search list (step B2). Since the search list is sorted in ascending order of evaluation value in step B12, model data 1-1-1 (evaluation value 55) having the best evaluation is extracted as shown in FIG.

  Since the extracted model data 1-1-1 is model data in which all hall calls in the allocation target call list have been allocated (YES in step B3), the model data 1-1-1 is stored. Move to the list and delete the model data 1-1-1 from the search list (step B4).

Since the search list is not empty (NO in step B1), the top model data, that is, model data 2 having the highest evaluation is acquired from the search list (step B2).

  Based on the acquired model data 2, create model data 2-1 when assigning the `` 4th floor upward hall call '' to the car 1 and model data 2-2 when assigning to the car 2; Evaluation values of the created model data 2-1 and 2-2 (not shown) are calculated. Then, the model data 2-1 and 2-2 are stored in the search list, the model data in the search list is sorted in ascending order of evaluation value, and the model data 2 is deleted.

  If the search calculation time limit is reached (YES in step B17), the process proceeds to step A5. In step A5, the model data in the storage list is sorted in ascending order of evaluation values. Since the saving list is not empty (NO in step A6), that is, the model data 1-1-1 is stored in the saving list, the first model data in the saving list is acquired (step A7). That is, the model data 1-1-1 that is the model data having the best evaluation is acquired from the storage list. Then, from the acquired model data 1-1-1, the car number to which the “second floor upward hall call” to be assigned is assigned is obtained. In the model data 1-1-1, “second floor upward hall call” is assigned to the first car, so “first car” is acquired as a search calculation result (step A9). Thus, the search calculation process is completed.

  Next, returning to FIG. 1, the basket allocation calculation parameter selection unit 02 will be described.

  The car assignment calculation parameter selection unit 02 includes an elevator control result detection unit 021, a traffic demand detection unit 022, an elevator performance case totaling unit 023 that creates performance case data, and a performance case storage unit (case data storage unit) that stores performance case data ) And a parameter determination unit (virtual hall call list acquisition unit) that searches the performance case data group in the performance case data storage unit 024 and determines a virtual call list (the virtual call list is a part of the performance case data) ) 025.

  The traffic demand detecting unit 022 receives data from the hall call detecting unit 04, the car call detecting unit 06, and the car information detecting unit 05, and the elevator 1 (11, 12,..., 1N) generated in the building. Detect traffic demand. The traffic demand detection unit 022 creates traffic demand data that holds the detected traffic demand in OD units. That is, an OD table representing traffic demand is created. For example, the OD table may be normalized so that the sum of each cell is 1. For the method of detecting traffic demand in units of OD (Origin Destination) by the traffic demand detection unit 022, refer to the descriptions of JP-B-7-106845, JP-B-5-71513, JP-B-1-15473, etc. Can be. The traffic demand data may be created for a specified period, for each time zone specified in advance, or for a predetermined time unit. Moreover, you may create the traffic demand data which calculated the traffic demand of the past same time slot | zone (for example, averaged). In addition, traffic demand data can be created using any method.

Here, the OD table will be briefly described.
For quantitative analysis of traffic flow, there is a case where an array table having a starting point as a row and a destination as a column is used. This sequence table is generally called an OD table. The OD table can identify traffic from any starting point to any destination and is used for a specific period, purpose, or other classification. In the OD table of the n × m array, n corresponds to the number of departure places and m corresponds to the number of destinations. When the number of starting points and destinations is the same, n = m, resulting in a square OD table. Such an OD table is used for elevator group management of the present embodiment. The departure point corresponds to the departure floor and the destination corresponds to the destination floor.

  Here, the “floor” in the OD table and the “floor” at the time of designating the OD may represent a high-level concept of a floor in which a plurality of specific floors are combined into one. It may represent the smallest unit, ie a single floor. A plurality of specific floors may correspond to restaurant streets, store children, etc. across different floors. In the OD table, different floors do not overlap each other. In addition, an upper concept floor that represents a plurality of specific floors together and a single floor may be mixed in the OD table.

  Next, the elevator control result detection unit 021 will be described. Prior to this, the elevator control target will be described. The elevator control target is stored in the elevator control target storage unit 3. An example of the elevator control target is shown in FIG.

  FIG. 2 shows three elevator control targets, each having a table format of 5 rows × 5 columns. This table corresponds to an OD (Origin Destination) table, where the row item represents the departure floor of the elevator and the column item represents the destination floor of the elevator. Each OD table corresponds to a different control target, and each target value is set for each OD. In the present embodiment, three elevator control targets such as “waiting time (time from arrival at the elevator hall to arrival of the car)”, “service time”, and “long waiting rate” are settable items. A control target value for each OD is set. For example, in the elevator control target “waiting time”, the control target value for moving from the A floor to the B floor is set to 25 (unit: sec (second)). This means that "the average waiting time for elevator passengers going from the A floor to the B floor is 25 seconds". The elevator control target items set using such an OD table are not limited to the above-mentioned “waiting time”, “service time”, and “long waiting rate”, and the number of items is arbitrarily at least one or more. is there. For example, control target items such as “boarding time”, “number of passengers”, “car passing interval time”, “maximum car passing interval time”, and “number of times of car passing” may be used. FIG. 2 shows one control target of “waiting time”, “service time”, and “long waiting rate”, but a plurality of control items may be stored in the elevator control target storage unit 3.

  FIG. 3 is a diagram for explaining an example of setting the control target value set in the OD table.

  The OD control target can be set by a specific numerical value, or can be a numerical range or condition. In the following description, numerical values, numerical ranges, conditions, and the like are collectively referred to as “OD control target values”. For example, as shown in FIG. 3 (A), the setting that “the average waiting time for elevator passengers from the A floor to the B floor should be 25 seconds or less”, as shown in FIG. 3 (B), “The average waiting time for elevator passengers going from the A floor to the B floor is 25 seconds or more and 30 seconds or less”, as shown in FIG. 3C, “any average for the A floor to the B floor” A setting such as “waiting time is sufficient” is possible. Also, as shown in FIG. 3D, “A floor to B floor is 25 seconds”, “B floor to C floor is 30 seconds to 40 seconds”, “D floor to B floor is Numerical values, numerical ranges, and conditions may be mixed, such as “30 seconds or less” and “E floor to A floor are optional”.

  In order to change the value of the elevator control target after delivery of the elevator, the elevator control target storage unit 3 may be provided with an elevator control target value setting GUI. Using this GUI, for example, the elevator control target value can be changed during operation of the elevator, or the elevator control target value can be changed during an elevator stop such as at night or during maintenance.

  The elevator control result detection unit 021 detects the elevator control result (elevator operation result) of each car within the time when the traffic demand is detected by the traffic demand detection unit 022. And the elevator control result detection part 021 produces the elevator control result data which totaled the elevator control result of each cage | basket in OD unit about the elevator control item which can be set as an elevator control target at least. The elevator control result data created by the elevator control result detection unit 021 is data corresponding to the service result (actual result) for the elevator passenger. For example, “waiting time” “service time” “long waiting rate” “hall call” The value of “no response time” is expressed in OD units.

  The elevator control result in OD units can be detected using the hall call detection unit 04, the car call detection unit 06, the car information detection unit 05, and the car stop information storage unit 03. For example, “occurrence time” and “occurrence floor” represented by hall call information, “car allocation time” included in hall call assignment information, “car arrival time”, “door open time”, “door close time” represented by the car information Passenger's “waiting time”, “service time”, “long waiting rate” sequentially based on “target floor” represented by “car stop time” “car load” and car call information of actual car call derived from hall call , And “unanswered time” and “maximum unanswered time” of the hall call. By summing up this in units of OD, the “elevator control result in units of OD” can be obtained.

  The elevator performance example totaling unit 023 includes “OD unit traffic demand data” generated by the traffic demand detection unit 022, “OD unit elevator control result data” (service result) generated by the elevator control result detection unit 021, The “virtual call list identifier” indicating the virtual call list used in the basket allocation calculation processing unit 01 during the period in which these data are acquired is aggregated as one data set (the same for all hall calls during this period). Virtual call list is used). A data set number is issued to this data set, and is passed to the performance case storage unit 024 as performance case data. The performance case storage unit 024 stores the passed performance case data. The above situation is shown in the lower part of FIG. Moreover, the data format of performance example data is shown in FIG. As an elevator control result, “waiting time”, “service time”, and “long waiting rate” are tabulated in OD units. It is assumed that the data set number is issued in the performance case storage unit 024 when it is recorded in the performance case storage unit 024. However, the numbering place is not limited to this. In this way, for example, at every predetermined time, the performance example data is generated by the elevator performance case totaling unit 023, and the generated performance case data is stored in the performance case storage unit 024.

  Further, the elevator performance case totaling unit 023 can also acquire performance case data using the elevator simulation experiment unit 026. That is, the elevator performance example totaling unit 023 generates virtual traffic demand in OD units (the traffic demand detecting unit 022 may be used at this time), and the elevator simulation experiment unit 026 Based on the virtual hall call list and elevator specification information given in advance, an elevator operation simulation is performed. The elevator performance example totalization unit 023 uses the “OD unit traffic demand data”, “OD unit elevator control result data” (predicted service result), and “virtual call list identifier” as the performance example as a result of the elevator operation simulation. The data is recorded in the performance case storage unit 024 as data. The elevator performance example totaling unit 023 performs a simulation using the elevator simulation experiment unit 026 when the elevator is closed, for example.

  In the above description, both the actual measurement and the simulation are shown as the method for calculating the performance example data. However, the performance example data may be calculated only by the actual measurement or only by the simulation.

  The parameter determination unit 025 determines a virtual call list to be used by the basket allocation calculation processing unit 01 from a plurality of virtual call lists. This determination is performed, for example, for each time slot specified in advance (for example, each time slot when there are three time slots A, B, and C). The parameter determination unit 025 searches the performance case storage unit 024 based on the search key described below to search for performance case data that matches the search key, and selects the virtual call list included in the searched performance case data. The virtual call list to be used in the basket allocation calculation processing unit 01 is determined.

  Here, the search key consists of two data. The first data is traffic demand data. The parameter determination unit 025 receives traffic demand data from the traffic demand detection unit 022. For example, the traffic demand detection unit 022 passes an average of traffic demand data (for the past 10 days, etc.) in the same time period in the past to the parameter determination unit 025. In addition, the traffic demand may be obtained by calculating the passenger generation interval for the same time zone in the past and letting passengers generated in the building arrive at Poisson. The parameter determination unit 025 uses the traffic demand data passed from the traffic demand detection unit 022 as the first data in the search key. The second data of the search key is elevator control target data. The parameter determination unit 025 receives from the elevator control target storage unit 3. The elevator control target storage unit 3 sets, for example, elevator control target data (three elevator control target data of “waiting time”, “service time”, and “long waiting rate”) corresponding to a time zone for determining a virtual call list as parameters. The data is passed to the determination unit 025. The elevator control target storage unit 3 may pass all elevator control target data set in itself. In this case, for example, the setting of the elevator control target storage unit 3 may be appropriately changed (change the elevator control target data to be used) using the user interface according to the time zone.

  The parameter determination unit 025 uses a set of “traffic demand data” and “elevator control target data” as a search key, and searches the performance case storage unit 024 for performance case data that matches the search key. That is, the virtual call list in which the feasibility of the elevator group management control for achieving the elevator control target is confirmed based on the operation example (actual result and prediction) is searched. The traffic demand data in the search key is aggregated for each OD, and the traffic demand data in the performance example data set is also aggregated for each OD. Similarly, the elevator control target in the search key and the elevator control result in the performance example data have the same items (“waiting time”, “service time”, “long waiting rate”), and are counted for each OD. It is possible to compare.

  Hereinafter, a specific example of the search will be described with reference to FIG.

  The parameter determination unit 025 compares the traffic demand data and the elevator control target data included in the search key with the performance case data stored in the performance case storage unit 024. That is, the similarity calculation between “traffic demand data in OD” from the traffic demand detection unit 022 and traffic demand data stored in the “OD traffic demand data field” included in the performance example data i, “elevator control target” Data ”and the elevator control result data aggregated in OD units stored in the“ OD unit elevator control result data field ”of the performance example data i are executed. Based on the results of the similarity calculation, the performance case data having the highest similarity is searched from the performance case data stored in the performance case storage unit 024. Then, the virtual hall call list identifier stored in the “virtual hall call list identifier field” in the retrieved performance example data is acquired. The parameter determination unit 025 passes the acquired virtual hall call list identifier to the search calculation processing unit 011 in the basket allocation calculation processing unit 01. The basket allocation calculation processing unit 01 uses the virtual hall call list corresponding to this “virtual hall call list identifier” to execute the basket allocation calculation process. The parameter determination unit 025 may specify a virtual hall call list corresponding to the virtual hall call list identifier, and pass the specified virtual hall call list to the basket allocation calculation processing unit 01.

  FIG. 21 is a flowchart showing a series of procedures for determining a virtual call list.

  First, traffic demand data in units of OD in the building is acquired by the traffic demand detection unit 022 (step C1). On the other hand, elevator control target data is acquired from the elevator control target storage unit 3 (step C2).

  Next, the case similarity calculation process (the traffic demand similarity calculation process and the traffic demand similarity calculation process and the OD unit traffic demand data acquired in step C1) and the elevator control target data acquired in step C2 are used. Performance similarity calculation processing) is performed.

  Here, first, the traffic demand data, that is, the traffic demand data detected in step C1, and the OD traffic demand data stored in the OD traffic demand data field of the performance case data stored in the performance case storage unit 024 (hereinafter referred to as the traffic demand data) In order to avoid confusion, the traffic demand similarity calculation process with “OD traffic case data” is executed (step C3).

  The performance example data is narrowed down by the traffic demand similarity calculation process in step C3. For the narrowed-down data, the performance similarity calculation process with the elevator control target data acquired in step C2 is executed, and the most similar performance case data is acquired (step C4).

  Then, the virtual call list identifier stored in the virtual call list identifier field of the most similar performance case data in step C4 is acquired (step C5).

  Thereafter, the acquired virtual call list identifier is transferred to the car assignment calculation processing unit 01 (step C6), and the car assignment calculation processing unit 01 uses the virtual call list corresponding to the virtual call list identifier to perform the car assignment calculation operation. Processing is executed (step C7).

  Here, the similarity calculation processing in step C3 and step C4 shown in FIG. 21 will be described in detail. In the following description, “OD traffic demand element data” is element data forming OD traffic demand data acquired from the traffic demand detection unit 022 and is data indicating traffic demand from the i-th floor to the j-th floor. “OD traffic case element data” is element data that forms the OD traffic demand data (OD traffic case data) stored in the OD traffic demand data field of the performance case data. It is data showing an example of traffic demand. The “elevator control target element data” is element data forming the elevator control target data, and is data indicating a control target value from the i-th floor to the j-th floor in the elevator control target item T. The “elevator control result element data” is element data forming “elevator control result data” stored in the “OD unit elevator control result data field” of the performance example data. It is the data which shows the elevator control result to the j-th floor.

  First, the traffic demand similarity calculation process performed at step C3 is demonstrated.

First, it is stored in “OD traffic demand data” that is one of the data included in the search key and “OD traffic demand data field” of m pieces of performance case data stored in the performance case storage unit 024. The similarity calculation with the “OD traffic case data” is performed as follows, for example.
(C3-1): For “OD traffic demand data” and “OD traffic example data”, the distance of each data between the same OD, that is,
OD traffic demand element data-OD traffic example element data |
Calculate For example, for the first to second floors between the same OD, the data value indicating the traffic demand from the first floor to the second floor of the OD traffic demand data, and the data value indicating the OD cases from the first floor to the second floor of the OD traffic case data And calculate the distance.
(C3-2): The process of (C3-1) is performed between all the ODs, and the sum (Manhattan distance) or the sum of squares is obtained.
(C3-3): (C3-2) is performed on all the OD traffic case data stored in the performance case storage unit 024, and n OD traffic cases having the smallest sum (Manhattan distance; or sum of squares) are selected. Select data. Thereby, n pieces of performance case data are selected.

  Next, the performance similarity calculation process performed in step C4 will be described.

The “elevator control result data” stored in the “OD unit elevator control result data field” of each of the n performance example data selected in (C3-3), and another data included in the search key The performance similarity calculation with a certain “elevator control target data” is performed as follows, for example.
(C4-1): The similarity between the same item and the same OD is calculated for “elevator control target data” and “elevator control result data”. For example, between the same OD (from the first floor to the second floor), the control target value of the item “waiting time” of the “elevator control target data” and the control result value of the item “waiting time” of the “elevator control result data” Calculate similarity. And the similarity calculated between each OD is totaled. A method of calculating the similarity here will be described later.
(C4-2): The process of (C4-1) is performed for all the control target items T, and the total sum of the similarities obtained from the control target items T is obtained. However, it is not always necessary to calculate the similarity for all control target items. For example, when it is desired to execute the target control that focuses only on the waiting time, the similarity calculation process can be omitted for other service times and long waiting rates.
(C4-3): The process of (C4-2) is performed on the elevator control result data in the n pieces of performance case data selected in (C3-3), and the most similar performance case data is selected. Then, the virtual call list identifier stored in the “virtual call list identifier field” included in the selected performance case data is acquired.

  Here, a specific example of the performance similarity calculation method in (C-4) will be described with reference to FIG.

FIG. 22 shows an example of the elevator control target and the elevator control result data. As described above, there are several patterns (e.g., setting a control target with one value, setting a control target with a range, etc.) in an elevator control target setting method. In any pattern, the similarity is calculated by comparing both values for each OD. Here, i (= 1, 2, 3) is the departure floor and j (= 1, 2, 3) is the destination floor. Further, the value set as the elevator control target is G _{i, j} , the data value of the elevator control result data is S _{i, j,} and the similarity in OD units is F _{i, j} . At this time, the value of the performance similarity can be calculated by the calculation shown in the following equation. Hereinafter, calculation examples of the performance similarity for each setting pattern will be shown.

(Elevator control target setting pattern 1)
When the elevator control target of a certain OD is set by one value (for example, 25 sec), the similarity of this OD is calculated as the following equation (1).

In FIG. 22, when i = 1 and j = 2, the set value of the elevator control target is G _1,2 = 25. This means that “the elevator control target from the first floor to the second floor is 25 seconds.” When i = 1 and j = 2, the value of the performance similarity is F _1,2 = | 25−35 | = 10. When i = 3 and j = 2, F _3,2 = | 20-30 | = 10.

(Elevator control target setting pattern 2)
When an elevator control target of a certain OD is set by a range having an upper limit (for example, 35 seconds or less), the performance similarity of this OD is calculated as the following equation (2).

In FIG. 22, when i = 1 and j = 3, the set value of the elevator control target is G _1,3 ≦ 30. This means that “the control target of the elevator from the first floor to the third floor may be 30 seconds or less.” The similarity when i = 1 and j = 3 is F ₁ from the above equation _{. 3} = 0.

As understood from the equation (2), when the value stored in the elevator control result data is equal to or smaller than the value set in the elevator control target, the similarity is 0 as described above. On the other hand, when the value stored in the elevator control result data is larger than the value set in the elevator control target, for example, when S _1,3 = 40, the performance similarity is calculated from the above equation by F _1,3 = | 30-40 | = 10.

(Elevator control target setting pattern 3)
When an elevator control target of a certain OD is set by a range having a lower limit value and an upper limit value (for example, 30 sec or more and 40 sec or less, etc.), the similarity degree of the OD is calculated as the following equation (3).

In FIG. 22, when i = 2 and j = 3, the set value of the elevator control target is 30 ≦ G _2,3 ≦ 40. This means that “the control target of the elevator from the second floor to the third floor may be 30 seconds or more and 40 seconds or less”, and the similarity when i = 2 and j = 3 is F from the above equation. _2,3 = 0.

If the value stored in the elevator control result data is outside the range set by the elevator control target, for example, if S _2,3 = 25, the performance similarity F _2,3 = | 30- If 25 | = 5 and S _2,3 = 45, then F _2,3 = | 40−45 | = 5.

(Elevator control target setting pattern 4)
When an elevator control target of a certain OD is arbitrarily set, the similarity of this OD is calculated as the following equation (4).

In FIG. 22, when i = 3 and j = 1, the set value of the elevator control target is G _2,3 = ∀. This means that the control target of the elevator from the third floor to the first floor is arbitrary, and the performance similarity when i = 3, j = 1 is unconditionally F _3,1 = 0. .

From the above, the value of the performance similarity in the example of FIG. 22 is F _1,2 + F _1,3 + F _2,1 + F _2,3 + F _3,1 + F _3,2 = 30
Is calculated. As the performance similarity value calculated in this way is smaller, it is determined that the elevator control target set in OD units is more similar to the elevator control result data in OD units.

  As described above, this embodiment focuses on a more micro-level viewpoint such as elevator passenger transport performance in OD units, rather than focusing on improving the passenger transport performance of elevators at the macro level of the entire building. This makes it possible to improve the elevator control performance in detail.

  Specifically, “OD traffic demand” for a certain period in the building, “identifier of the virtual call list used during that period”, and “elevator control result for each OD during that period” are used as performance example data. A database is stored in the case storage unit 024. The parameter determination unit 025 searches the database using a set of “elevator control target” and “OD traffic demand” as a search key during the car assignment calculation process, and performs elevator group management control satisfying “elevator control target”. A virtual call list whose feasibility has been confirmed is acquired, and this virtual call list is used for actual elevator group management control.

  Thus, by using the virtual call list that satisfies the elevator control target set in OD units, the elevator performance can be improved in OD units. Therefore, more detailed elevator group management control can be realized as compared with the conventional elevator group management system which improves the average elevator operation performance of the entire building.

  Note that the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. In addition, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

The block diagram which shows schematic structure of the elevator group management system concerning embodiment of this invention. The figure which shows the example of an elevator control target. The figure explaining the example of a setting of the control target value set to OD table. The conceptual diagram of a virtual call list. The figure which shows a some virtual call list. The figure which shows the example of an allocation object call list. The figure which shows an example of model data. The figure which shows a mode that model data is updated. The flowchart which shows the outline of the whole search calculation process. The flowchart which showed the flow of the process performed by step A4 of FIG. 9 in detail. The figure which shows the state which updated model data. The figure explaining supplementary of search calculation processing. The figure explaining supplementary of search calculation processing. The figure explaining supplementary of search calculation processing. The figure explaining supplementary of search calculation processing. The figure explaining supplementary of search calculation processing. The figure explaining supplementary of search calculation processing. The figure which shows the data format of performance example data. The figure which shows a mode that a performance case memory | storage part memorize | stores performance case data. The figure which shows the mode of the search performed by the parameter determination part 25. FIG. FIG. 9 is a flowchart showing a series of procedures for determining a virtual call list. The figure explaining the specific example of the performance similarity calculation method.

Explanation of symbols

0: Elevator group management unit 3: Elevator control target storage unit 01: Basket allocation calculation processing unit 02: Basket allocation calculation parameter selection unit 03: Basket stop information storage unit 04: Hall call detection unit 05: Basket information detection unit 06: Basket Call detection unit 011: Search calculation processing unit 012: Search calculation data storage unit 013: Evaluation model unit 021: Elevator control result detection unit 022: Traffic demand detection unit 023: Elevator performance case totaling unit 024: Performance case storage unit 025: Parameter Determination unit 026: Elevator simulation experiment unit 111 to 1N1: Basket control unit 11 to 1N: Elevator basket 21 to 2M: Hall call button 112 to 1N2: Basket call button

Claims (11)

An elevator group management system that executes group management control for assigning an elevator car to a hall call that occurs at an arbitrary time on an arbitrary floor in a building,
A traffic demand detector that detects traffic demand in the building and generates traffic demand data holding traffic demand for each combination of a departure floor and a destination floor;
An elevator control target storage unit that stores elevator control target data that holds an elevator control target value for each combination of a departure floor and a destination floor;
A virtual hall call list storing at least one or more virtual hall call information representing a virtual hall call, elevator control result data holding an elevator control result for each combination of the departure floor and the destination floor, and the departure floor And a case data storage unit that stores a plurality of case data corresponding to traffic demand data that holds traffic demand for each combination of destination floors,
The traffic demand data is received from the traffic demand detection unit, the elevator control target data is received from the elevator control target storage unit, and the received traffic demand data and the elevator control target data are used in the case data storage unit. A virtual hall call list acquisition unit that performs similarity calculation with each case data, selects the case data based on the result of the similarity calculation, and acquires a virtual hall call list included in the selected case data;
A car call determining unit for determining a derived car call derived from each of the virtual hall calls shown in the acquired virtual hall call list;
Information on hall calls that have been assigned to each elevator car and that have not been answered is stored as assigned hall call information, and information on unanswered car calls that have occurred in each elevator car is stored as unanswered car call information. A basket stop information storage unit;
A hall call detector for detecting hall calls in the building;
A car call prediction unit for predicting a derived car call derived from the detected hall call;
Using the allocated hall call information and the unanswered cage call information, a floor patrol schedule for each elevator car is generated, and each virtual hall call indicated in the acquired virtual hall call list is determined. Each derived car call, the hall call detected by the hall call detecting unit, and the predicted derived car call are added to each of the floor patrol schedules by a plurality of additional patterns, and added by the additional pattern. Calculating an evaluation value of the additional pattern based on each of the floor patrol schedules, selecting an additional pattern based on the evaluation value calculated from each of the additional patterns, and detecting the detected hall call from the selected additional pattern. A car assignment calculation unit for determining an elevator car to be assigned to
Elevator group management system with   The basket allocation calculation unit selects an additional pattern from the additional patterns for which the evaluation values are calculated when the time limit is reached before the calculation of the evaluation values for all the additional patterns is completed. Item 2. The elevator group management system according to Item 1.   3. The elevator group management system according to claim 1, wherein the virtual hall call list acquisition unit performs the virtual hall call list acquisition process for each time period specified in advance. The virtual hall call list stores a plurality of call sets that are a set of the virtual hall call information and derivative basket call information,
The said cage call determination part determines the derived cage call shown in each said call set contained in the said virtual hall call list as a derived cage call derived from each said virtual hall call. The elevator group management system according to 3. The case data storage unit stores identification information of the virtual hall call list as the virtual hall call list,
The virtual hall call list acquisition unit acquires the virtual hall call list from correspondence information in which a virtual hall call list is associated with identification information and the identification information included in the selected case data. The elevator group management system according to any one of claims 1 to 4.   The elevator group management system according to any one of claims 1 to 5, further comprising setting means for setting the elevator control target data in the elevator control target storage unit. The elevator control target storage unit stores a plurality of the elevator control target data,
The said virtual hall call list acquisition part selects the elevator control target data used for the said similarity calculation from several said elevator control target data, The Claim 1 thru | or 5 characterized by the above-mentioned. Elevator group management system. Detecting the control results of each of the elevator cars, adding up the detected control results of each of the elevator cars, and generating elevator control result data holding the elevator control results for each combination of the departure floor and the destination floor, An elevator control result detection unit;
The generated elevator control result data, the traffic demand data holding the traffic demand in the period to be aggregated for each combination of the departure floor and the destination floor, and the virtual hall call list used in the period And a case data generation unit that stores the case data as the case data in the case data storage unit,
An elevator group management system according to any one of claims 1 to 7, further comprising: An elevator simulation experiment unit that generates the elevator control result data holding the elevator control result for each combination of the departure floor and the destination floor by performing the group management control by simulation;
A case data generation unit that stores the generated elevator control result data, the traffic demand data used for the simulation, and the virtual hall call list used for the simulation in the case data storage unit as the case data,
The elevator group management system according to any one of claims 1 to 8, further comprising: An elevator group management method for executing group management control for assigning an elevator car to a hall call generated at an arbitrary time on an arbitrary floor in a building,
A traffic demand detecting step for detecting traffic demand in the building;
A traffic demand data generation step for generating traffic demand data holding traffic demand for each combination of a departure floor and a destination floor;
An elevator control target setting step for setting elevator control target data holding an elevator control target value for each combination of the departure floor and the destination floor;
A virtual hall call list storing at least one or more virtual hall call information representing a virtual hall call, elevator control result data holding an elevator control result for each combination of the departure floor and the destination floor, and the departure floor And a preparation step of preparing a database storing a plurality of case data corresponding to traffic demand data holding traffic demand for each combination of destination floors,
Using the generated traffic demand data and the set elevator control target data, perform similarity calculation with each case data in the database, and select the case data based on the result of the similarity calculation A virtual hall call list acquisition step of acquiring a virtual hall call list included in the selected case data;
A car call determining step for determining a derived car call derived from each of the virtual hall calls included in the acquired virtual hall call list;
Step of storing hall call information assigned to each elevator car and unanswered as assigned hall call information, and information of unanswered car call generated in each elevator car as unanswered car call information When,
A hall call detecting step for detecting a hall call in the building;
A car call prediction step for predicting a derived car call derived from the detected hall call;
Using the allocated hall call information and the unanswered cage call information, a floor patrol schedule for each elevator car is generated, and each virtual hall call indicated in the acquired virtual hall call list is determined. Each derived car call, the hall call detected by the hall call detection step, and the predicted derived car call are added to each floor patrol schedule by a plurality of additional patterns, and added by the additional pattern. Calculating an evaluation value of the additional pattern based on each of the floor patrol schedules, selecting an additional pattern based on the evaluation value calculated from each of the additional patterns, and detecting the detected hall call from the selected additional pattern. A car assignment calculation step for determining an elevator car to be assigned to
Elevator group management method. An elevator group management program for causing a computer to execute group management control for assigning an elevator car to a hall call generated at an arbitrary time on an arbitrary floor in a building,
A traffic demand detecting step for detecting traffic demand in the building;
A traffic demand data generation step for generating traffic demand data holding traffic demand for each combination of a departure floor and a destination floor;
An elevator control target setting step for setting elevator control target data holding an elevator control target value for each combination of the departure floor and the destination floor;
A virtual hall call list storing at least one or more virtual hall call information representing a virtual hall call, elevator control result data holding an elevator control result for each combination of the departure floor and the destination floor, and the departure floor And a preparation step of preparing a database storing a plurality of case data corresponding to traffic demand data holding traffic demand for each combination of destination floors,
Using the generated traffic demand data and the set elevator control target data, perform similarity calculation with each case data in the database, and select the case data based on the result of the similarity calculation A virtual hall call list acquisition step of acquiring a virtual hall call list included in the selected case data;
A car call determining step for determining a derived car call derived from each of the virtual hall calls shown in the acquired virtual hall call list;
Step of storing hall call information assigned to each elevator car and unanswered as assigned hall call information, and information of unanswered car call generated in each elevator car as unanswered car call information When,
A hall call detecting step for detecting a hall call in the building;
A car call prediction step for predicting a derived car call derived from the detected hall call;
Using the allocated hall call information and the unanswered cage call information, a floor patrol schedule for each elevator car is generated, and each virtual hall call indicated in the acquired virtual hall call list is determined. Each derived car call, the hall call detected by the hall call detection step, and the predicted derived car call are added to each floor patrol schedule by a plurality of additional patterns, and added by the additional pattern. Calculating an evaluation value of the additional pattern based on each of the floor patrol schedules, selecting an additional pattern based on the evaluation value calculated from each of the additional patterns, and detecting the detected hall call from the selected additional pattern. A car assignment calculation step for determining an elevator car to be assigned to
Is an elevator group management program that causes the computer to execute.

Images