# US 2021/0276824 A1 · People Flow Prediction Method and People Flow Prediction System > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2021/0276824 A1 - Document ID: US20210276824A1 - Status: Published application - Kind: A1 - Application number: 17/255,835 - Filed: 2019-05-10 - Published: 2021-09-09 - Earliest claimed: 2018-06-26 - Assignees: Hitachi, Ltd. - Applicants: Hitachi, Ltd. - Inventors: Yu KITANO; Akinori ASAHARA; Naoki SHIMODE; Nobuo SATO - Web page: https://patentgrep.com/patent/US20210276824A1 ## CPC - B66B 1/3476: {Load weighing or car passenger counting devices (B66B5/14 takes precedence)} - B66B 5/0037: {Performance analysers} - B66B 1/24: Control systems with regulation, i.e. with retroactive action, for influencing travelling speed, acceleration, or deceleration - B66B 1/3446: {Data transmission or communication within the control system} - B66B 3/002: {Indicators} - B66B 5/0012: {Devices monitoring the users of the elevator system} - B66B: ELEVATORS; ESCALATORS OR MOVING WALKWAYS ({apparatus for raising or lowering persons on theatrical stages or the like A63J5/12} ; funicular railbound systems with rigid ground-supported tracks and cable traction, e.g. cliff railways, B61B9/00; arrangements of ammunition handlers in vessels B63G3/00; hoists, lifts, or conveyors for loading or unloading in general B65G; braking or detent devices controlling normal movements of winding drums or barrels B66D; ship-lifting devices E02C; garages for many vehicles with mechanical means for lifting vehicles E04H6/12; hoists for feeding ammunition or projectiles to launching apparatus or to loading mechanisms F41A9/00) ## Abstract On-site getting in and out data including the number of people who got in the elevator in the past is created, virtual getting in and out data including the number of people who get in the elevator is created by making people who arrive at a landing of the elevator virtually appear and simulating operation of the elevator on the basis of the number of people who appear, a first conversion model for converting the virtual getting in and out data before a certain time point into the number of people who appear after the certain time point and a second conversion model for converting the virtual getting in and out data after a certain time point into the number of people who appear before the certain time point are created on the basis of the number of people who appear and the virtual getting in and out data, a prediction model for predicting the number of people who appear after a certain time point from the number of people who appear before the certain time point is learned on the basis of the number of people who appear converted by the second conversion model, and the number of people who appear after a certain time point is predicted from the on-site getting in and out data before the certain time point by using the first conversion model and the prediction model. ## Claims ### Claim 1 A people flow prediction method executed by a computer system having a processor and a storage device connected to the processor, the method comprising: a number of people getting in and out calculating procedure in which the processor calculates a number of people who got in an elevator in a past on a basis of information of a sensor installed in the elevator and creates on-site getting in and out data including the number of people calculated; a simulation data creating procedure in which the processor creates virtual getting in and out data including at least a number of people who get in the elevator by making a person who arrives at each of landings of the elevator in order to use the elevator virtually appear and simulating operation of the elevator on a basis of a number of people who appear; a first conversion model creating procedure in which the processor creates a first conversion model for converting the virtual getting in and out data before a certain time point into the number of people who appear after the certain time point on a basis of the number of people who appear and the virtual getting in and out data; a second conversion model creating procedure in which the processor creates a second conversion model for converting the virtual getting in and out data after a certain time point into the number of people who appear before the certain time point on a basis of the number of people who appear and the virtual getting in and out data; a prediction model learning procedure in which the processor predicts a prediction model for predicting the number of people who appear after a certain time point from the number of people who appear before the certain time point on a basis of the number of people who appear converted by the second conversion model; and a predicting procedure in which the processor predicts the number of people who appear after a certain time point from the on-site getting in and out data before the certain time point by using the first conversion model and the prediction model. ### Claim 2 (depends on 1) The people flow prediction method according to claim 1, wherein in the first conversion model creating procedure, the processor calculates a first number of people who appear by applying the first conversion model to the on-site getting in and out data, in the second conversion model creating procedure, the processor calculates a second number of people who appear by applying the second conversion model to the on-site getting in and out data, and in the prediction model learning procedure, the processor learns a prediction model for predicting the second number of people who appear after a certain time point from the first number of people who appear before the certain time point. ### Claim 3 (depends on 1) The people flow prediction method according to claim 1, wherein the on-site getting in and out data further includes at least one of operation performed for a call button of the elevator on each floor, operation performed for a destination floor button in the elevator, and arrival frequency of the elevator on each floor, and in the simulation data creating procedure, the processor creates the virtual getting in and out data further including at least one of the operation for the call button of the elevator on each floor, the operation for the destination floor button in the elevator, and the arrival frequency of the elevator on each floor by simulating operation of the elevator on a basis of the number of people who appear. ### Claim 4 (depends on 1) The people flow prediction method according to claim 1, wherein in the prediction model learning procedure, the processor learns a prediction model corresponding to a time zone having a predetermined attribute, the prediction model for predicting the number of people who appear after a certain time point in the time zone having the predetermined attribute from the number of people who appear before the certain time point on a basis of the number of people who appear calculated by applying the second conversion model to the on-site getting in and out data in the time zone having the predetermined attribute, and in the predicting procedure, the processor uses the first conversion model and the prediction model corresponding to the time zone having the predetermined attribute to predict the number of people who appear after a certain time point in the time zone having the predetermined attribute from the on-site getting in and out data before the certain time point. ### Claim 5 (depends on 4) The people flow prediction method according to claim 4, wherein the time zone having the predetermined attribute is one of a time zone in each day, a predetermined day of a week, and a day corresponding to a predetermined event. ### Claim 6 (depends on 1) The people flow prediction method according to claim 1, wherein in the simulation data creating procedure, the processor calculates distribution of a number of people who get in the elevator on a basis of the on-site getting in and out data, and makes a person who arrives at each of the landings of the elevator in order to use the elevator to virtually appear on the basis of the distribution calculated. ### Claim 7 (depends on 1) The people flow prediction method according to claim 1 further comprising: a procedure in which the processor creates a destination floor prediction model for predicting a destination floor of a person who gets in the elevator on a basis of the on-site getting in and out data; and a procedure in which the processor predicts the number of people who appear on each destination floor on a basis of the destination floor prediction model and the number of people who appear predicted in the predicting procedure. ### Claim 8 (depends on 1) The people flow prediction method according to claim 1 further comprising an image processing procedure in which the processor calculates a number of people included in an image on a basis of the image obtained by photographing one of the landings of the elevator, wherein in the simulation data creating procedure, the processor makes a person who arrives at each of the landings of the elevator in order to use the elevator virtually appear on a basis of the number of people calculated in the image processing procedure. ### Claim 9 (depends on 8) The people flow prediction method according to claim 8, wherein in the simulation data creating procedure, the processor makes a number of people obtained by adding a number of people calculated in a predetermined manner to the number of people calculated in the image processing procedure, as a number of people who arrive at each landing of the elevator in order to use the elevator virtually appear. ### Claim 10 A people flow prediction system comprising: a number of people getting in and out calculation unit which calculates a number of people who got in an elevator in a past on a basis of information of a sensor installed in the elevator and creates on-site getting in and out data including the number of people calculated; a simulation data creation unit which creates virtual getting in and out data including at least a number of people who get in the elevator by making a person who arrives at each of landings of the elevator in order to use the elevator virtually appear and simulating operation of the elevator on a basis of a number of people who appear; a first conversion model creation unit which creates a first conversion model for converting the virtual getting in and out data before a certain time point into the number of people who appear after the certain time point on a basis of the number of people who appear and the virtual getting in and out data; a second conversion model creation unit which creates a second conversion model for converting the virtual getting in and out data after a certain time point into the number of people who appear before the certain time point on a basis of the number of people who appear and the virtual getting in and out data; a prediction model learning unit which learns a prediction model for predicting the number of people who appear after a certain time point from the number of people who appear before the certain time point on a basis of the number of people who appear converted by the second conversion model; and a prediction unit which predicts the number of people who appear after a certain time point from the on-site getting in and out data before the certain time point by using the first conversion model and the prediction model. ### Claim 11 (depends on 10) The people flow prediction system according to claim 10, wherein the first conversion model creation unit calculates a first number of people who appear by applying the first conversion model to the on-site getting in and out data, the second conversion model creation unit calculates a second number of people who appear by applying the second conversion model to the on-site getting in and out data, and the prediction model learning unit learns a prediction model for predicting the second number of people who appear after a certain time point from the first number of people who appear before the certain time point.