Systems and methods for controlling elevator group movement

Abstract

The present disclosure provides a system and method for controlling the movement of an elevator group. The method includes accepting a current request for service by the elevator group, accepting a partial travel path of a person's movement through an environment serviced by the elevator group, and obtaining future elevator request probabilities. The method further includes processing the partial travel path with a neural network trained to estimate a weighted combination of probability density functions indicative of the person's arrival time distribution, and generating a set of possible future requests that jointly represent the future elevator request probabilities and the arrival time distribution. The method further includes optimizing a schedule for the elevator group to correspond to the current request and the set of possible future requests, and controlling the elevator group according to the schedule.

Description

[Technical Field]

[0001] TECHNICAL FIELD This disclosure relates generally to elevators, and more particularly to systems and methods for controlling the movement of elevator groups. [Background technology]

[0002] Elevators or elevator groups are installed in buildings containing multiple floors. Users use the elevator groups to travel from one floor to another in the building. Elevator scheduling is an important aspect for ensuring efficient operation of the elevator group, for example, to reduce user waiting times. Typically, current elevator requests are received from users, and a schedule for controlling the elevator group is determined based on the current elevator requests. However, to determine an optimal schedule for the elevator group that minimizes the average waiting time of all users, it is necessary to consider future elevator requests. To consider future elevator requests, the arrival times of potential future passengers should be predicted. The arrival time refers to the amount of time it may take for a user to reach the elevator group.

[0003] Some approaches use model-based methods to predict arrival times. Model-based methods use either queuing theory or agent-based modeling to build a simulator of user movement. The models used by model-based methods make assumptions about the flow, speed, and density of user movement. Because the models used by model-based methods are generative in nature, sampling user movement paths from the models is relatively straightforward. However, sampling movement paths is time-consuming and expensive. Furthermore, performance degrades as the simulation period increases. Furthermore, accurate calibration of the physical model can be time-consuming and tedious.

[0004] Therefore, there remains a need for methods and systems for optimal scheduling and control of elevator fleets. Summary of the Invention

[0005] An objective of some embodiments is to provide a system and method for predicting future passenger arrival times for elevators in an elevator group. Specifically, an objective of some embodiments is to predict a probability distribution of future passenger arrival times. An objective of some embodiments is to obtain probabilities of future passenger elevator requests (i.e., future elevator requests) and determine a continuation of the future passenger arrival stream based on the obtained future elevator request probabilities and the probability distribution of predicted future passenger arrival times. Additionally, an objective of some embodiments is to determine a schedule for the elevator group based on the continuation of the future passenger arrival stream, where the schedule optimizes a performance metric by minimizing the average waiting time (AWT) of all passengers.

[0006] Typically, current elevator requests from passengers for service by the elevator group are received, and a schedule for the elevator group is determined based on the current elevator requests. However, to determine an optimal schedule for the elevator group, it is desirable to consider future elevator requests in addition to the current elevator requests. For example, a first passenger on the sixth floor of a building requests service from the elevator group to the first floor of the building. The elevator car assigned to serve the first passenger is currently located on the fifteenth floor. In the next five seconds, a second passenger on the tenth floor also requests service to the first floor. In this case, the elevator scheduler may determine in advance such arrival information associated with the future passenger, i.e., the second passenger, and determine a schedule such that the elevator car on the fifteenth floor may wait five seconds until the future passenger, i.e., the second passenger, requests service to the first floor. Upon receiving the request from the second passenger, the elevator car may first pick up the second passenger on the tenth floor and then pick up the first passenger on the sixth floor, so that both passengers may arrive at the first floor together. Thus, it is desirable to further optimize passenger AWT by taking into account future elevator demands.

[0007] To account for future elevator requests, it is necessary to determine the probability that a passenger will arrive at the elevator for elevator service and the probability of the amount of time it may take for the passenger to reach the elevator. The probability that a passenger will arrive at the elevator for elevator service is sometimes referred to as the future elevator request probability. The amount of time it may take for the passenger to reach the elevator is called the arrival time.

[0008] The problem of determining future elevator request probabilities is referred to as the future service request problem. The problem of determining arrival time probabilities is referred to as the arrival time prediction problem. Some embodiments are based on the recognition that the future service request problem and the arrival time prediction problem can be considered as separate and independent problems. For example, the probability that a passenger will arrive at the elevator is determined independently, and the arrival time probability is also predicted independently, assuming that the passenger will arrive at the elevator. Because the future service request problem and the arrival time prediction problem are considered as separate problems, each problem can be solved with a different approach.

[0009] For example, in some embodiments, a neural network can be used to predict a passenger's arrival time given the passenger's current location and destination. However, given the current location and destination, a passenger may take multiple paths to reach their destination (e.g., a passenger may first go to some intermediate location (e.g., a restroom) from their current location and then head to the elevator, which may take slightly longer than going directly to the elevator). In order to accurately predict arrival times, such multi-modality of arrival times needs to be taken into account. Some embodiments are based on the recognition that a probability distribution of arrival times can be predicted to take into account multi-modality of arrival times.

[0010] To that end, some embodiments of the present disclosure disclose a neural travel-time mixture model (NTMM) for predicting the probability distribution of arrival times. The NTMM is trained to estimate a weighted combination of probability density functions, such that the weighted combination of probability density functions represents the arrival time distribution of passengers arriving at an elevator via one of multiple routes. The weighted combination of probability density functions predicted by the NTMM is advantageous because the resulting distribution can be used to estimate conditional cumulative distribution functions and generate samples, in addition to estimating probability densities.

[0011] Furthermore, in some embodiments, the future elevator request probability is determined by using an attention-based destination predictor having an encoder / decoder structure. The attention-based destination predictor may be implemented using a transformer architecture-based neural network. Alternatively, in some embodiments, the future elevator request probability is determined by using a recurrent neural network (RNN). Additionally or alternatively, in some embodiments, the future elevator request is obtained directly. For example, in an organization's office space, a user may be provided with a user interface at their seat position (or in their cubicle) for submitting an elevator service request. The user may submit the elevator service request via the user interface.

[0012] The determined future elevator request probabilities and the probability distribution of predicted arrival times can be used to determine an optimal schedule for the elevator group. However, it is difficult to directly incorporate the determined future elevator request probabilities and the probability distribution of predicted arrival times when determining the optimal schedule. Some embodiments are based on the recognition that a sequence of future passenger arrival streams (also referred to as a "set of possible future requests") can be generated based on the determined future elevator request probabilities and the probability distribution of predicted arrival times. Furthermore, an optimal schedule may be determined based on the sequence of future passenger arrival streams. Furthermore, the elevator group is controlled according to the optimized schedule to accommodate both current and future elevator requests.

[0013] Accordingly, one embodiment discloses a control system for controlling the movement of an elevator group, the control system comprising at least one processor and a memory having instructions stored thereon. The instructions cause the at least one processor of the control system to accept one or more current elevator requests for service by the elevator group, accept partial travel paths of a person's movement through an environment serviced by the elevator group, obtain future elevator request probabilities, and process the partial travel paths with a neural network trained to estimate a weighted combination of probability density functions. The weighted combination of probability density functions indicates an arrival time distribution of the person arriving at the elevator group by one of a plurality of paths through the environment. The instructions further cause the at least one processor of the control system to generate a set of possible future requests jointly representing the future elevator request probabilities and the arrival time distributions, optimize a schedule for the elevator group to correspond to the one or more current elevator requests and the set of possible future requests, and control the elevator group according to the schedule.

[0014] Accordingly, another embodiment discloses a method for controlling the movement of an elevator group. The method includes accepting one or more current elevator requests for service by the elevator group, accepting partial travel paths of a person's movement through an environment serviced by the elevator group, obtaining future elevator request probabilities, and processing the partial travel paths with a neural network trained to estimate a weighted combination of probability density functions. The weighted combination of probability density functions indicates an arrival time distribution of the person arriving at the elevator group by one of a plurality of paths through the environment. The method further includes generating a set of possible future requests that jointly represent the future elevator request probabilities and the arrival time distributions, optimizing a schedule for the elevator group to correspond to the one or more current elevator requests and the set of possible future requests, and controlling the elevator group according to the schedule.

[0015] Accordingly, yet another embodiment discloses a non-transitory computer-readable storage medium containing a processor-executable program for performing a method for controlling the movement of an elevator group. The method includes accepting one or more current elevator requests for service by the elevator group, accepting partial travel paths of a person's movement through an environment serviced by the elevator group, obtaining future elevator request probabilities, and processing the partial travel paths with a neural network trained to estimate a weighted combination of probability density functions. The weighted combination of probability density functions indicates an arrival time distribution of the person arriving at the elevator group by one of multiple paths through the environment. The method further includes generating a set of possible future requests that jointly represent the future elevator request probabilities and the arrival time distributions, optimizing a schedule for the elevator group to correspond to the one or more current elevator requests and the set of possible future requests, and controlling the elevator group according to the schedule.

[0016] The presently disclosed embodiments are further described with reference to the accompanying drawings, which are not necessarily to scale, emphasis instead generally being placed upon illustrating the principles of the presently disclosed embodiments. [Brief explanation of the drawings]

[0017] [Figure 1A] FIG. 1 is a schematic diagram illustrating control of elevator bank movement according to an embodiment of the present disclosure. [Figure 1B] FIG. 1 is a block diagram of a control system for controlling the movement of an elevator group according to an embodiment of the present disclosure. [Figure 1C] FIG. 2 illustrates a graph showing an arrival time distribution and a sample empirical probability density function (PDF) derived from the arrival time distribution, according to an embodiment of the present disclosure. [Figure 2] FIG. 1 illustrates training of a neural travel time mixture model (NTMM), according to some embodiments of the present disclosure. [Figure 3A] FIG. 1 illustrates a layout of floors in a building with travel paths used to train an NTMM, according to some embodiments of the present disclosure. [Figure 3B] FIG. 1 illustrates retraining of a trained NTMM according to some embodiments of the present disclosure. [Figure 4A] FIG. 1 is a block diagram of a transformer architecture-based neural network according to some embodiments of the present disclosure. [Figure 4B] FIG. 1 is a schematic diagram illustrating a Long Short-Term Memory (LSTM) based neural network, according to some embodiments of the present disclosure. [Figure 5] FIG. 1 is a schematic diagram of a building floor plan, according to some embodiments of the present disclosure. [Figure 6] FIG. 1 is a flow diagram of a method for controlling the movement of an elevator group according to an embodiment of the present disclosure. [Figure 7] FIG. 1 is a schematic diagram illustrating a computing device for implementing the methods and control systems of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0018] (Description of the embodiment) In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. However, it will be apparent to those skilled in the art that the present disclosure may be practiced without these specific details. In other instances, devices and methods are shown in block diagram form solely to avoid obscuring the present disclosure.

[0019] As used in this specification and claims, the terms "for example," "for example," and "such as," as well as the verbs "comprise," "have," and "include," and each of their other verb forms, when used in conjunction with a list of one or more components or other items, should be construed as open-ended, meaning that the list should not be considered as excluding other additional components or items. The term "based on" means based at least in part on. Furthermore, it should be understood that the style and terminology used herein are for purposes of description and should not be regarded as limiting. Any headings used herein are for convenience only and have no legal or limiting effect.

[0020] FIG. 1A illustrates a schematic diagram illustrating control of the movement of an elevator group 115, according to an embodiment of the present disclosure. The elevator group 115 may include one or more elevator cars, such as car 117a and car 117b. The elevator group 115 may be installed in an environment, such as a building, including multiple floors, such as floor 119a, floor 119b, floor 119c, floor 119d, floor 119e, and floor 119f. An objective of some embodiments is to provide a control system for determining an optimal schedule for the elevator group 115, where the optimal schedule optimizes a performance metric by minimizing the average waiting time (AWT) of all passengers. An objective of some embodiments is to provide a control system for controlling the elevator group 115 based on the optimal schedule. Such a control system is described below in FIG. 1B.

[0021] 1B shows a block diagram of a control system 121 for scheduling and controlling elevator group 115 according to an embodiment of the present disclosure. Control system 121 includes a processor 123 and a memory 125. Processor 123 may be a single-core processor, a multi-core processor, a computing cluster, or any number of other configurations. Memory 125 may include random access memory (RAM), read-only memory (ROM), flash memory, or any other suitable memory system. Additionally, in some embodiments, memory 125 may be implemented using a hard drive, an optical drive, a thumb drive, an array of drives, or any combination thereof.

[0022] The processor 123 receives a partial path of travel 101 of a person (who may be a potential passenger) moving through an environment served by the elevator group 115. The partial path of travel 101 of the person's movement may correspond to a portion of a path the person covers while going from an origin to a destination. The partial path of travel 101 may depend on the path taken by the person. For example, the partial path of travel 101 may be straight, curved, or zigzag, depending on the path taken by the person. The environment served by the elevator group 115 may be an area on one of multiple floors (119a-119f). In one embodiment, the processor 123 may capture the person's movement using sensors (e.g., cameras, etc.) associated with the environment to determine the partial path of travel 101.

[0023] Additionally, processor 123 accepts current elevator requests 111 from passengers for service by elevator group 115 and determines a schedule for elevator group 115 based on the current elevator requests. However, to determine an optimal schedule for elevator group 115, it may be desirable to consider future elevator requests in addition to current elevator requests 111. For example, assume that a first passenger on the sixth floor of a building requests service from the elevator group to the first floor of the building. The elevator car assigned to service the first passenger is currently located on floor 15. In the next five seconds, a second passenger on floor 10 also requests service to the first floor. In this case, the elevator scheduler may determine in advance such arrival information associated with the future passenger, i.e., the second passenger, and determine a schedule such that the elevator car on floor 15 may wait five seconds until the future passenger, i.e., the second passenger, requests service to the first floor. If a request from a second passenger is received, the elevator car may first pick up the second passenger at floor 10, and then pick up the first passenger at floor 6, so that both passengers arrive together at floor 1. In this manner, it may be desirable to further optimize the passenger AWT to take future elevator requests into account.

[0024] To account for future elevator requests, it is necessary to determine the probability that a passenger will arrive at elevator bank 115 for elevator service and the probability of the amount of time it may take for the passenger to reach elevator bank 115. The probability that a passenger will arrive at elevator bank 115 for elevator service is sometimes referred to as the future elevator request probability. The amount of time it may take for a passenger to reach elevator bank 115 is called the arrival time.

[0025] The problem of determining future elevator request probabilities is referred to as the future service request problem. The problem of determining arrival time probabilities is referred to as the arrival time prediction problem. Some embodiments are based on the recognition that the future service request problem and the arrival time prediction problem can be considered as separate and independent problems. For example, the probability that a passenger will arrive at the elevator bank 115 is determined independently, and the arrival time probability is also predicted independently, assuming that the passenger will arrive at the elevator bank 115. Because the future service request problem and the arrival time prediction problem are considered as separate problems, each problem can be solved with a different approach.

[0026] For example, in some embodiments, a neural network can be used to predict a passenger's arrival time given the passenger's current location and destination. However, given the current location and destination, the passenger may take multiple paths to reach their destination (e.g., a passenger may first go to some intermediate location (e.g., a restroom) from their current location and then head to the elevator, which may take slightly longer than going directly to the elevator). In order to accurately predict arrival times, such multi-modality of arrival times needs to be taken into account. Some embodiments are based on the recognition that a probability distribution of arrival times can be predicted to take into account multi-modality of arrival times.

[0027] To that end, some embodiments of the present disclosure disclose a neural network 103 trained to estimate a weighted combination of probability density functions that represent the arrival time distribution of passengers arriving at an elevator via one of multiple paths. A processor 123 processes the partial travel path 101 using the neural network 103 to estimate a weighted combination of probability density functions that represent the arrival time distribution 105 of a person arriving at an elevator group 115 via one of multiple paths. The weighted combination of probability density functions predicted by the neural network 103 is advantageous because the resulting distribution can be used to estimate conditional cumulative distribution functions and generate samples, in addition to estimating probability densities.

[0028] Additionally, processor 123 obtains future elevator request probabilities 107. In one embodiment, to obtain future elevator request probabilities 107, processor 123 processes partial travel path 101 using an attention-based destination predictor having an encoder / decoder structure. The attention-based destination predictor may be implemented using a transformer architecture-based neural network. Alternatively, in some embodiments, to obtain future elevator request probabilities 107, processor 123 processes partial travel path 101 using a recurrent neural network (RNN).

[0029] The future elevator request probabilities 107 and the arrival time distribution 105 can be used to determine an optimal schedule for the elevator fleet 115. However, it is difficult to directly incorporate the future elevator request probabilities 107 and the arrival time distribution 105 when determining the optimal schedule. Some embodiments are based on the recognition that a set of possible future requests can be generated based on the future elevator request probabilities 107 and the arrival time distribution 105. To do so, the processor 123 generates a set of possible future requests 109 that jointly represents the future elevator request probabilities 107 and the arrival time distribution 105. An example of generating the set of possible future requests 109 is described below.

[0030]

number

[0031] Additionally, processor 123 optimizes elevator group schedule 113 to correspond to current elevator demand 111 and a set of possible future demands 109. Processor 123 controls elevator group 115 according to optimized schedule 113.

[0032] The arrival time distribution 105 estimated by the neural network 103 and the empirical probability density function (PDF) of the samples derived from the arrival time distribution 105 are shown in FIG. 1C.

[0033] FIG. 1C illustrates a graph 127 showing arrival time distribution 105 and a sample empirical PDF 129 derived from arrival time distribution 105, according to some embodiments of the present disclosure.

[0034] In one embodiment, the neural network 103 trained to estimate a weighted combination of probability density functions representing the arrival time distribution 105 is a neural traveltime mixture model (NTMM). The NTMM is trained based on a dataset as described below in FIG. (NTMM training)

[0035] 2 illustrates training of an NTMM 207, according to some embodiments of the present disclosure. The NTMM 207 is trained on a dataset 201 that includes origin and destination coordinates 203 for different people and durations 205 of travel paths corresponding to different people. Additionally or alternatively, the dataset 201 includes durations of travel paths corresponding to the same person at different times. The travel paths used to train the NTMM 207 are described below in FIG. 3A.

[0036] FIG. 3A illustrates a layout 300 of a floor 301 in a building with travel paths used to train the NTMM 207, according to some embodiments of the present disclosure. In some embodiments, each travel path, such as travel path 305 and travel path 311, may be represented by a sequence of tessellation indices. Each travel path may have its own origin and destination. For example, travel path 305 includes origin 303 and destination 307, and travel path 311 includes origin 309 and destination 313. The tessellation indices may be obtained by discretizing a range of location coordinates, which may represent longitude and latitude associated with the tessellation index corresponding to floor 301. In another embodiment, the travel path may include a travel path that includes both location information and timing information. Furthermore, each travel path may be represented by a sequence of tuples of tessellation indices, each tuple may include a tessellation index obtained by discretizing a range of location coordinates and a prediction period.

[0037] As shown in FIG. 3A , the representation of travel paths is simplified by discretizing the coordinates of locations on floor 301 into tessellation indices 1-2500, collectively designated 315. The space of floor 301 is discretized into a 50×50 tessellation, and each travel path is converted into a sequence of tessellation locations. Such travel paths may serve to mimic the movement of a person throughout the floor during a certain time of day, and the destinations visited by that person may represent actual indoor destinations primarily accessed, such as restrooms, stairs, elevators, and break rooms. Mathematically, the travel paths are given as follows:

[0038]

number

[0039]

number

[0040]

number

[0041]

number

[0042]

number

[0043]

number

[0044]

number

[0045]

number

[0046]

number

[0047]

number

[0048] To this end, the hyperparameters in the objective function are (a) the number of components (K), (b) the regularization penalty (λ), (c) the number of nodes in the two dense layers, and (d) the learning rate.

[0049] Some embodiments are based on the recognition that the travel paths of different people may change over time. In other words, the movement of people in an environment changes over time. Therefore, the trained NTMM 207 needs to be retrained on new datasets, e.g., movement patterns for which the hyperparameters are no longer optimal.

[0050] 3B illustrates retraining of the trained NTMM 207 according to some embodiments of the present disclosure. An updated NTMM is obtained by further retraining the NTMM 207, which was trained on dataset 201, on a new dataset 317. The new dataset 317 includes updated travel path durations corresponding to different people. Because the NTMM 207 is retrained using previous hyperparameters, expensive hyperparameter search is not performed each time the NTMM 207 is retrained. In this way, the NTMM 207 is trained in a computationally inexpensive manner.

[0051] The NTMM 207 estimates a weighted combination of probability density functions that describe the arrival time distribution 105. Based on the arrival time distribution and the future elevator request probabilities 107, the processor 123 generates a set of possible future requests 109, as described below. (Generating a set of possible future requirements)

[0052]

number

[0053]

number

[0054]

number

[0055] Furthermore, the allocation is determined by minimizing the AWT over the contiguous set of candidate partitions, for which either an exact or approximate algorithm can be applied.

[0056]

number

[0057]

number

[0058]

number

[0059] Additionally, in one embodiment, to obtain future elevator request probabilities 107, processor 123 processes partial travel paths 101 using a transformer architecture-based neural network, which is described below in FIG. 4A.

[0060] 4A shows a block diagram 400 illustrating components of a Transformer architecture-based neural network 435 according to some embodiments of the present disclosure. The Transformer architecture-based neural network 435 may include an encoder 413 including a stack of N encoder blocks and a decoder 431 including a stack of N decoder blocks. These may further include stacks of self-attention and fully connected feedforward layer components, in an aspect where the Transformer architecture-based neural network 435 uses positional coding and attention mechanisms that enable parallelization.

[0061] Some embodiments are based on the realization that the attention mechanism of the Transformer architecture-based neural network 435 allows for modeling dependencies of elements in a sequence without considering their distance, and that such advantages of the Transformer architecture-based neural network 435 are not limited to speech-related applications but also apply to extended destination prediction tasks.

[0062] As shown in FIG. 4A, first, an input movement path 401 (such as a partial movement path 101) passes through an embedding and position coding block 403, where the input token is encoded by an embedding layer. modeldimensional vector, and the input tokens can be discrete elements. Furthermore, order information about the elements of the input movement path 401 is obtained using positional encoding and then combined with the embedding vector to obtain a new vector by an addition operation. Furthermore, the new vector can be fed into the multi-head attention block 407. Attention in the encoder 413 can be used as a means to refer to other tokens in the input movement path 401 when attempting to encode a particular token.

[0063] The multi-head attention block 407 may include an array of scaled dot-product attention components, and an attention function may be calculated based on the data matrix X. The data matrix may be projected using three different matrices learned during training, resulting in projection matrices Q, K, and V, which represent the query, key, and value, respectively.

[0064]

number

[0065] In some embodiments, the multi-head attention mechanism involves computing multiple attention functions using different learned projections to yield improved representation performance.

[0066] Additionally, the summing and normalizing block 409 in the encoder 413 may include residual connections along with layer normalization (RLN), which may combine the inputs to the multi-head attention block 407 with the outputs of the multi-head attention block 407, sum and normalize them. The resulting elements may further pass through a feedforward sublayer 411 and subsequently through another summing and normalizing block 415 that includes RLN connections upon exiting the encoder 413. The summing and normalizing block 415 may combine the inputs to the feedforward layer 411 with the outputs of the feedforward layer 411, sum and normalize them.

[0067] Additionally, the target output movement path 405 may pass through an embedding and position coding block 417, which may be similar to the embedding and position coding block 403, before entering the masked multi-head attention block 419. The masked multi-head attention block 419 elements may be similar to the multi-head attention block 407 of the encoder 413, except that the target output movement path 405 is masked so that the masked multi-head attention block 419 may only reference the preceding output movement path positions. Additionally, a summing and normalizing block 421, which includes residual combinations with RLNs, may combine the inputs to the masked multi-head attention block 419 with the outputs of the masked multi-head attention block 419, sum them, and normalize them.

[0068] After the summation and normalization block 421, there may be another multi-head attention block 423. The multi-head attention block 423 may be similar to the multi-head attention block 407 of the encoder 413, except that the multi-head attention block 423 uses information from the encoder output as an additional input in the calculation of attention.

[0069] Additionally, another summing and normalizing block 425 may combine the inputs to the multi-head attention block 423 with the outputs of the multi-head attention block 423, sum them, and normalize them. The resulting elements may further pass through a feedforward sublayer 427 and subsequently through another summing and normalizing block 429 upon exiting the decoder 431. The summing and normalizing block 429 may combine the inputs to the feedforward layer 427 with the outputs of the feedforward layer 427, sum them, and normalize them.

[0070] External to the decoder 431 may be a linear layer and softmax block 433 that may take the output of the decoder 431 and create a logical vector for all possible output elements. Additionally, the linear layer and softmax block 433 may output future elevator request probabilities 107 by performing a softmax operation to transform the values ​​in the logical vector.

[0071] Alternatively, the processor 123 processes the partial travel path 101 using an RNN to obtain future elevator request probabilities 107. Examples of RNNs may include, but are not limited to, long short-term memory (LSTM)-based neural networks, bidirectional long short-term memory (BiLSTM)-based neural networks, etc.

[0072] FIG. 4B illustrates an LSTM-based neural network 437 according to an embodiment of the present disclosure. The LSTM-based neural network unit 437 includes an input gate, a forget gate, and an output gate. Each of these gates corresponds to a "standard" neuron in a feedforward (or multi-layer) neural network; i.e., each gate calculates a weighted sum activation (using an activation function). Furthermore, it, ot, and ft represent the activations of the input gate, output gate, and forget gate, respectively, at time step t. The exit arrows from memory cell c to gates i, o, and f represent peephole connections. The peephole connections indicate the contribution of memory cell c's activation at time step t-1, i.e., ct-1. In other words, gates i, o, and f calculate their activations at time step t (i.e., it, ot, and ft, respectively) while also taking into account the activation of memory cell c at time step t-1, i.e., ct-1. The single left-to-right arrow emanating from memory cell c indicates c rather than a peephole connection, and the symbol "x" represents element-wise multiplication between inputs. Additionally, LSTM-based neural networks 437 include the application of a differentiable function (such as a sigmoid function) to the weighted sum.

[0073] In operation, the first LSTM unit receives the partial travel path 439 and outputs the future elevator request probability 107 by sending the hidden state vector ht to the next LSTM unit and finally to the softmax layer.

[0074] Alternatively, in one embodiment, processor 123 receives elevator service requests from people via user devices and generates a set of possible future requests that jointly represent elevator service requests and arrival time distributions, such an embodiment being described below in FIG.

[0075] 5 illustrates a schematic diagram of a floor 500 of a building, according to some embodiments of the present disclosure. The floor 500 may include an elevator group 115, one or more conference rooms 501, one or more cubicles 505, and a restroom 507. In some embodiments, a user device may be provided in each conference room and cubicle. For example, a user device 503 may be provided in the conference room 501. The user device 503 may include a user interface for submitting an elevator service request for service by the elevator group 115. A person in the conference room 501 may submit an elevator service request via the user interface of the user device 503. Alternatively, the person may submit an elevator service request via a user device such as a smartphone, laptop, or the like.

[0076] Additionally, processor 123 generates a set of possible future requests that jointly represent the elevator service requests received via user device 503 and arrival time distribution 105. Processor 123 optimizes a schedule for elevator fleet 115 to correspond to current elevator requests 111 and the generated set of possible future requests. Based on the optimized schedule, processor 123 controls elevator fleet 115.

[0077] The overall method for controlling the movement of elevator bank 115 is described below in FIG.

[0078] FIG. 6 illustrates a block diagram of a method 600 for controlling the movement of an elevator bank 115 according to an embodiment of the present disclosure.

[0079] At block 601, method 600 includes receiving one or more current elevator requests 111 from one or more passengers. For example, one or more passengers may issue a request for an elevator car (e.g., car 117a or car 117b) by pressing or touching an up or down button associated with elevator bank 115.

[0080] At block 603, method 600 includes accepting a partial travel path (e.g., partial travel path 101) of a person's movement through an environment served by elevator fleet 115. At block 605, method 600 includes obtaining a future elevator request probability, e.g., future elevator request probability 107.

[0081] At block 607, method 600 includes processing the partial travel paths with a neural network trained to estimate a weighted combination of probability density functions that represent the arrival time distribution of a person arriving at an elevator by one of multiple paths in the environment. In one embodiment, the neural network corresponds to NTMM 207 described above in FIG. 2.

[0082] At block 609, method 600 includes generating a set of possible future requests that jointly represent future elevator request probabilities and arrival time distributions. At block 611, method 600 includes optimizing a schedule for elevator fleet 115 to accommodate the current requests and the generated set of possible future requests.

[0083] At block 613, the method 600 includes controlling the elevator bank 115 according to the optimized schedule.

[0084] 7 is a schematic diagram illustrating a computing device 700 for implementing the method 600 and control system 121 of the present disclosure. The computing device 700 includes a power supply 701, a processor 703, a memory 705, and a storage device 707, all connected to a bus 709. Furthermore, a high-speed interface 711, a low-speed interface 713, a high-speed expansion port 715, and a low-speed expansion port 717 may be connected to the bus 709. Additionally, a low-speed connection port 719 is connected to the bus 709. Furthermore, an input interface 721 may be connected to an external receiver 723 and an output interface 725 via the bus 709. The receiver 727 may be connected to an external transmitter 729 and a transmitter 731 via the bus 709. An external memory 733, an external sensor 735, a machine 737, and an environment 739 may also be connected to the bus 709. Furthermore, one or more external input / output devices 741 may be connected to the bus 709. A network interface controller (NIC) 743 may be adapted to connect to a network 745 via bus 709. Among other things, data or other data may be rendered on a third-party display device, a third-party imaging device, and / or a third-party printing device external to computing device 700.

[0085] Memory 705 may store instructions executable by computing device 700, as well as any data that may be utilized by the methods and systems of this disclosure. Memory 705 may include random access memory (RAM), read-only memory (ROM), flash memory, or any other suitable memory system. Memory 705 may be one or more volatile memory units and / or one or more non-volatile memory units. Memory 705 may also be another form of computer-readable medium, such as a magnetic disk or optical disk.

[0086] The storage device 707 may be adapted to store supplemental data and / or software modules used by the computing device 700. The storage device 707 may include a hard drive, an optical drive, a thumb drive, an array of drives, or any combination thereof. Additionally, the storage device 707 may include a computer-readable medium such as a floppy disk device, a hard disk device, an optical disk device, or an array of devices including a tape device, a flash memory or other similar solid-state memory device, or a storage area network or other configuration of devices. The instructions may be stored on an information carrier. When executed by one or more processing devices (e.g., processor 703), the instructions perform one or more methods, such as those described above.

[0087] The computing device 700 may optionally be linked through a bus 709 to a display interface or user interface (HMI) 747 adapted to connect the computing device 700 to a display device 749 and a keyboard 751. The display device 749 may include, among other things, a computer monitor, a camera, a television, a projector, or a mobile device. In some implementations, the computing device 700 may include a printer interface for connecting to a printing device. The printing device may include, among other things, a liquid inkjet printer, a solid ink printer, a large-scale commercial printer, a thermal printer, a UV printer, or a dye sublimation printer.

[0088] The high-speed interface 711 manages bandwidth-intensive operations of the computing device 700, and the low-speed interface 713 manages low-bandwidth-intensive operations. Such an allocation of functions is merely an example. In some implementations, the high-speed interface 711 can be coupled to memory 705, a user interface (HMI) 747, a keyboard 751, and a display 749 (e.g., through a graphics processor or accelerator), and can be coupled to a high-speed expansion port 715 that can accept various expansion cards via a bus 709. In one implementation, the low-speed interface 713 is coupled to storage 707 and a low-speed expansion port 717 via a bus 709. The low-speed expansion port 717, which can include various communication ports (e.g., USB, Bluetooth, Ethernet, wireless Ethernet), can be coupled to one or more input / output devices 741. The computing device 700 can be connected to a server 753 and a rack server 755. The computing device 700 can be implemented in several different forms. For example, the computing device 700 may be implemented as part of a rack server 755 .

[0089] The description provides only exemplary embodiments and is not intended to limit the scope, applicability, or configuration of the disclosure. Rather, the following description of exemplary embodiments will provide those skilled in the art with an enabling description for implementing one or more exemplary embodiments. Various changes are contemplated that may be made in the function and arrangement of elements without departing from the spirit and scope of the disclosed subject matter as set forth in the appended claims.

[0090] Specific details are provided in the following description to provide a thorough understanding of the embodiments. However, those skilled in the art will understand that the embodiments may be practiced without these specific details. For example, systems, processes, and other elements of the disclosed subject matter may be shown as components in block diagram form in order to avoid obscuring the embodiments in unnecessary detail. In other instances, well-known processes, structures, and techniques may be shown without unnecessary detail in order to avoid obscuring the embodiments. Furthermore, like reference numbers and names in the various drawings indicate like elements.

[0091] Also, particular embodiments may be described as a process that is depicted as a flowchart, a flow diagram, a data flow diagram, a structure diagram, or a block diagram. While a flowchart may describe operations as a sequential process, many of the operations may be performed in parallel or simultaneously. Additionally, the order of operations may be rearranged. A process may be terminated when its operations are completed, but may have additional steps not discussed or included in the diagram. Moreover, not all operations in any specifically described process may occur in all embodiments. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc. When a process corresponds to a function, the termination of the function may correspond to a return of the function to the calling function or the main function.

[0092] Furthermore, embodiments of the disclosed subject matter may be implemented, at least in part, either manually or automatically. The manual or automatic implementation may be performed, or at least assisted, through the use of machines, hardware, software, firmware, middleware, microcode, hardware description languages, or any combination thereof. When implemented in software, firmware, middleware, or microcode, the program code or code segments to perform the necessary tasks may be stored on a machine-readable medium. A processor(s) may perform the necessary tasks.

[0093] The various methods or processes outlined herein may be coded as software executable on one or more processors employing any one of a variety of operating systems or platforms. Additionally, such software may be written using any of a number of suitable programming languages ​​and / or programming or scripting tools, and may be compiled as executable machine language code or intermediate code that runs on a framework or virtual machine. Typically, the functionality of the program modules may be combined or distributed as desired in various embodiments.

[0094] Embodiments of the present disclosure may be embodied as a method, an example of which is provided. The order of operations performed as part of this method may be determined in any suitable manner. Thus, embodiments may be configured to perform operations in an order different from that illustrated, which may include performing some operations simultaneously, even though in the example embodiment they are shown as a sequence of operations.

[0095] Furthermore, embodiments of the present disclosure and the functional operations described herein can be implemented in digital electronic circuitry, tangibly embodied computer software or firmware, computer hardware including the structures disclosed herein and their structural equivalents, or one or more combinations thereof. Furthermore, some embodiments of the present disclosure can be implemented as one or more computer programs, i.e., as one or more modules of computer program instructions encoded on a tangible, non-transitory program carrier for execution by or to control the operation of a data processing apparatus. Furthermore, the program instructions can be encoded on an artificially generated propagated signal, such as a machine-generated electrical, optical, or electromagnetic signal, generated to encode information that is transmitted to a suitable receiving device for execution by the data processing apparatus. A computer storage medium may be a machine-readable storage device, a machine-readable storage substrate, a random or serial access memory device, or one or more combinations thereof.

[0096] According to embodiments of the present disclosure, the term "data processing apparatus" may encompass all types of apparatus, devices, and machines that process data, including, by way of example, a programmable processor, a computer, or multiple processors or computers. The apparatus may include special-purpose logic circuitry, such as an FPGA (field-programmable gate array) or an ASIC (application-specific integrated circuit). In addition to hardware, the apparatus may also include code that creates an execution environment for the computer program, such as code that constitutes processor firmware, a protocol stack, a database management system, an operating system, or one or more combinations thereof.

[0097] A computer program (which may also be called or described as a program, software, software application, module, software module, script, or code) can be written in any form of programming language, including compiled or interpreted languages, or declarative or procedural languages, and can be deployed in any form, such as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program may, but need not, correspond to a file in a file system. A program can be stored in part of a file that holds other programs or data, for example, in one or more scripts stored in a markup language document, in a single file dedicated to the program, or in multiple coordinated files, for example, a file that stores one or more modules, subprograms, or portions of code.

[0098] A computer program can be deployed to be executed on one computer or on multiple computers located at one location or distributed across multiple locations and interconnected by a communications network. Computers suitable for running computer programs may, by way of example, be based on general-purpose or special-purpose microprocessors or both, or any other type of central processing unit. Typically, a central processing unit receives instructions and data from a read-only memory or a random-access memory or both. The essential elements of a computer are a central processing unit for performing or executing instructions and one or more memory devices for storing instructions and data.

[0099] Typically, a computer also includes one or more mass storage devices, such as magnetic, magneto-optical, or optical disks, for storing data, or is operatively coupled to such disks to receive data from, transfer data to, or both. However, a computer need not have such devices. Furthermore, a computer can be embedded in another device, such as a mobile phone, a personal digital assistant (PDA), a mobile audio or video player, a game console, a global positioning system (GPS) receiver, or a portable storage device, such as a universal serial bus (USB) flash drive, to name a few.

[0100] To provide for user interaction, embodiments of the subject matter described herein may be implemented on a computer having a display device, such as a CRT (cathode ray tube) or LCD (liquid crystal display) monitor, for displaying information to the user, and a keyboard and pointing device, such as a mouse or trackball, that allows the user to provide input to the computer. Other types of devices may also be used to provide for user interaction. For example, feedback provided to the user may be any form of sensory feedback, such as visual feedback, auditory feedback, or tactile feedback, and input from the user may be received in any form, including acoustic, speech, or tactile input. Additionally, the computer may provide for user interaction by sending documents to and receiving documents from a device used by the user, for example, by sending a web page to a web browser on the user's client device in response to a request received from the web browser.

[0101] Embodiments of the subject matter described herein may be implemented in a computing system that includes a back-end component, e.g., a data server, or a middleware component, e.g., an application server, or a front-end component, e.g., a client computer having a graphical user interface or web browser that allows a user to interact with an implementation of the subject matter described herein, or any combination of one or more of such back-end, middleware, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication, e.g., a communications network. Examples of communications networks include local area networks ("LANs") and wide area networks ("WANs"), e.g., the Internet.

[0102] A computing system may include clients and servers. Clients and servers are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a mutual client-server relationship.

[0103] While the present disclosure has been described in terms of certain preferred embodiments, it should be understood that various other adaptations and modifications can be made within the spirit and scope of the disclosure. It is, therefore, the object of the following claims to cover all such variations and modifications that fall within the true spirit and scope of the disclosure.

Claims

1. 1. A control system for controlling the movement of an elevator group, comprising: at least one processor; a memory storing instructions; The instructions may include instructions to the at least one processor of the control system to: accepting one or more current elevator requests for service by the elevator fleet; accepting a partial path of movement of a person moving through an environment served by the elevator group; Obtaining future elevator request probabilities; processing the partial travel paths with a neural network trained to estimate a weighted combination of probability density functions; the weighted combination of probability density functions representing an arrival time distribution of the person arriving at the elevator group by one of a plurality of paths within the environment; The instructions further include for the at least one processor of the control system to: generating a set of possible future requests that jointly represent the future elevator request probabilities and the arrival time distribution; optimizing the elevator group schedule to accommodate the one or more current elevator demands and the set of likely future demands; a control system that controls the elevator group according to the schedule;

2. The control system of claim 1 , wherein the neural network is a Neural Travel-time Mixture Model (NTMM).

3. The control system of claim 2 , wherein the NTMM is trained based on a dataset including origin and destination coordinates and durations of travel paths corresponding to different people.

4. The control system of claim 3 , wherein the durations of the travel paths correspond to the same person at different times.

5. The control system of claim 3 , wherein the processor is further configured to retrain the NTMM based on a new data set including updated travel path durations corresponding to different people.

6. 2. The control system of claim 1, wherein the processor is further configured to process the partial travel path with a transformer architecture-based neural network to obtain the future elevator request probability.

7. 2. The control system of claim 1, wherein the processor is further configured to process the partial travel path with a recurrent neural network (RNN) to obtain the future elevator request probability.

8. The processor further comprises: receiving an elevator service request from the person via a user device; The control system of claim 1 configured to generate a set of likely future requests that jointly represent the elevator service requests and the arrival time distribution.

9. 1. A method for controlling movement of an elevator group, the method comprising: accepting one or more current elevator requests for service by the elevator fleet; receiving a partial travel path of movement of a person moving through an environment served by the elevator group; obtaining future elevator request probabilities; and processing the partial traveled paths using a neural network trained to estimate a weighted combination of probability density functions; the weighted combination of probability density functions representing an arrival time distribution of the person arriving at the elevator group by one of a plurality of paths within the environment; The method further comprises: generating a set of possible future requests that jointly represent the future elevator request probabilities and the arrival time distribution; optimizing the elevator fleet schedule to accommodate the one or more current elevator demands and the set of likely future demands; and controlling the elevator group according to the schedule.

10. The method of claim 9 , wherein the neural network is a neural travel-time mixture model (NTMM).

11. The method of claim 10 , wherein the NTMM is trained based on a dataset including origin and destination coordinates and durations of travel paths corresponding to different persons.

12. The method of claim 11 , wherein the durations of the travel paths correspond to the same person at different times.

13. The method of claim 11 , further comprising retraining the NTMM based on a new dataset including updated travel path durations corresponding to different people.

14. 10. The method of claim 9, wherein the method further comprises processing the partial travel path with a transformer architecture based neural network to obtain the future elevator request probabilities.

15. 10. The method of claim 9, wherein the method further comprises processing the partial travel path with a recurrent neural network (RNN) to obtain the future elevator request probability.

16. The method further comprises: receiving an elevator service request from the person via a user device; and generating a set of likely future requests that jointly represent the elevator service requests and the arrival time distribution.

17. 1. A non-transitory computer-readable storage medium containing a program executable by a processor to perform a method for controlling movement of an elevator group, the method comprising: accepting one or more current elevator requests from one or more passengers for service by the elevator group; receiving a partial travel path of movement of a person moving through an environment served by the elevator group; obtaining future elevator request probabilities; and processing the partial traveled paths using a neural network trained to estimate a weighted combination of probability density functions; the weighted combination of probability density functions representing an arrival time distribution of the person arriving at the elevator group by one of a plurality of paths within the environment; The method further comprises: generating a set of possible future requests that jointly represent the future elevator request probabilities and the arrival time distribution; optimizing the elevator fleet schedule to accommodate the one or more current elevator demands and the set of likely future demands; and controlling the elevator group according to the schedule.

18. 20. The non-transitory computer-readable storage medium of claim 17, wherein the neural network is a neural travel-time mixture model (NTMM).

19. 20. The non-transitory computer-readable storage medium of claim 17, wherein the NTMM is trained based on a dataset including origin and destination coordinates and durations of travel paths corresponding to different persons.

20. 20. The non-transitory computer-readable storage medium of claim 19, wherein the method further comprises retraining the NTMM based on a new dataset including updated travel path durations corresponding to different persons.

Images