Elevator control method and system based on big data processing
Through the elevator control method based on big data processing, the elevator docking scheme is optimized using the generative adversarial network and graph convolutional network, the problem of inefficiency in traditional elevator systems during peak hours is solved, and the passenger's riding experience is significantly improved.
Patent Information
- Application Number
- CN202510516335.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-06-20
- Filing Date
- 2025-04-23
- Publication Date
- 2025-05-23
AI Technical Summary
Traditional elevator systems operate inefficiently during peak hours, resulting in long waits and crowded ride experiences for passengers.
The elevator control method based on big data processing is adopted, by obtaining elevator monitoring video, determining the docking floor and the number of elevators for each elevator, calculating the minimum and maximum intervals of the parking space of each elevator, and using a generative adversarial network to generate multiple docking plans, and finally determining the target docking plan through the graph convolution network.
It improves the operating efficiency of elevators during peak hours and improves passengers' riding experience.
Smart Images
Figure CN120024764A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of elevators, and in particular to an elevator control method and system based on big data processing. Background Art
[0002] As an indispensable vertical transportation tool in high-rise buildings, elevators play a vital role in ensuring the efficiency and convenience of personnel flow within the building. Especially in crowded areas such as commercial office buildings, large shopping malls and apartments, the operating efficiency of elevators directly affects people's travel experience.
[0003] During rush hour, a large number of people arrive at the elevator almost at the same time, resulting in a surge in demand for elevators. Traditional elevator systems often adopt simple parking strategies, such as stopping at each floor, which can lead to serious efficiency bottlenecks during peak hours, causing passengers to wait for a long time and experience crowded riding.
[0004] Therefore, how to improve the operating efficiency of elevators during peak hours and improve passengers' riding experience is an urgent problem to be solved. Summary of the invention
[0005] The main technical problem solved by the present invention is how to improve the operating efficiency of elevators during peak hours and improve the riding experience of passengers.
[0006] According to a first aspect, the present invention provides an elevator control method based on big data processing, comprising: acquiring a surveillance video of an elevator; determining a stopping floor and a number of enabled elevators for each person waiting for the elevator using an elevator personnel information determination model based on the surveillance video of the elevator; determining a minimum stop interval for each elevator and a maximum stop interval for each elevator based on the stopping floor and the number of enabled elevators for each person waiting for the elevator; generating multiple stop plans using a generative adversarial network based on the stopping floor of each person waiting for the elevator, the number of enabled elevators, the minimum stop interval for each elevator, and the maximum stop interval for each elevator, wherein each of the multiple stop plans includes multiple stop floors for each elevator; and determining a target stop plan based on the multiple stop plans.
[0007] Furthermore, determining a target docking plan based on the multiple docking plans includes: Based on the multiple stop plans and the stop floors of each person waiting for the elevator, the average waiting time of each person waiting for the elevator in each stop plan and the total power consumption of the elevator are determined; based on the average waiting time of each person waiting for the elevator in each stop plan and the total power consumption of the elevator, a target stop plan is determined.
[0008] Furthermore, the method of determining a target docking plan based on the average waiting time of each person waiting for the elevator in each docking plan and the total power consumption of the elevator includes: constructing a graph structure, the graph structure including multiple nodes and multiple edges between the multiple nodes, each node representing a docking plan, the node features of each node including multiple docking floors of each elevator in each docking plan, the average waiting time of each person waiting for the elevator, and the total power consumption of the elevator, and the edges between the nodes represent the similarity between the docking plans; and processing the graph structure based on a graph convolutional network to determine the target docking plan.
[0009] Furthermore, the elevator personnel information determination model is a door control cycle unit.
[0010] Furthermore, the input of the generative adversarial network is the stopping floor of each person waiting for the elevator, the number of enabled elevators, the minimum stopping interval of each elevator, and the maximum stopping interval of each elevator, and the output of the generative adversarial network is multiple stopping plans.
[0011] According to a second aspect, the present invention provides an elevator control system based on big data processing, comprising: an acquisition module for acquiring a monitoring video of an elevator; A personnel information determination module, used to determine the stopping floor and the number of enabled elevators for each person waiting for the elevator based on the monitoring video of the elevator using an elevator personnel information determination model; An interval determination module, used to determine the minimum stop interval of each elevator and the maximum stop interval of each elevator based on the stop floor of each elevator waiting person and the number of enabled elevators; A stop plan generation module, configured to generate a plurality of stop plans using a generative adversarial network based on the stop floor of each elevator waiting person, the number of enabled elevators, the minimum stop interval of each elevator, and the maximum stop interval of each elevator, wherein each of the plurality of stop plans includes a plurality of stop floors of each elevator; The target docking plan determining module is used to determine the target docking plan based on the multiple docking plans.
[0012] Furthermore, the target docking plan determination module is also used for: Based on the multiple stop plans and the stop floors of each person waiting for the elevator, determine the average waiting time of each person waiting for the elevator in each stop plan and the total power consumption of the elevator; A target parking plan is determined based on the average waiting time of each person waiting for the elevator in each parking plan and the total power consumption of the elevator.
[0013] Furthermore, the target docking plan determination module is also used for: Constructing a graph structure, wherein the graph structure includes a plurality of nodes and a plurality of edges between the plurality of nodes, wherein each node represents a docking plan, and the node features of each node include a plurality of docking floors of each elevator in each docking plan, an average waiting time of each person waiting for the elevator, and a total power consumption of the elevator, and the edges between the nodes represent the similarity between the docking plans; The graph structure is processed based on a graph convolutional network to determine a target docking plan.
[0014] Furthermore, the elevator personnel information determination model is a door control cycle unit.
[0015] Furthermore, the input of the generative adversarial network is the stopping floor of each person waiting for the elevator, the number of enabled elevators, the minimum stopping interval of each elevator, and the maximum stopping interval of each elevator, and the output of the generative adversarial network is multiple stopping plans.
[0016] The present invention provides an elevator control method and system based on big data processing, the method comprising acquiring a monitoring video of the elevator; determining a stopping floor and the number of enabled elevators for each person waiting for the elevator using an elevator personnel information determination model based on the monitoring video of the elevator; determining a minimum stop interval for each elevator and a maximum stop interval for each elevator based on the stopping floor and the number of enabled elevators for each person waiting for the elevator; generating multiple stop plans using a generative adversarial network based on the stopping floor of each person waiting for the elevator, the number of enabled elevators, the minimum stop interval for each elevator, and the maximum stop interval for each elevator, wherein each of the multiple stop plans includes multiple stop floors for each elevator; and determining a target stop plan based on the multiple stop plans. The method can improve the operating efficiency of elevators during peak hours and improve the riding experience of passengers. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 A flowchart of an elevator control method based on big data processing provided by an embodiment of the present invention; Figure 2 A schematic diagram of a process for determining a target docking plan provided by an embodiment of the present invention; Figure 3 A schematic diagram of an elevator control system based on big data processing provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0018] In an embodiment of the present invention, there is provided Figure 1 An elevator control method based on big data processing is shown, and the elevator control method based on big data processing includes steps S1 to S5: Step S1, obtaining the surveillance video of the elevator.
[0019] The elevator surveillance video refers to the video recorded by the camera installed outside the elevator. The elevator surveillance video can be used to monitor the situation in the elevator waiting area.
[0020] Step S2, based on the monitoring video of the elevator, use the elevator personnel information to determine the stop floor and the number of enabled elevators for each person waiting for the elevator.
[0021] The elevator personnel information determination model is a gate control loop unit. The input of the elevator personnel information determination model is the monitoring video of the elevator, and the output of the elevator personnel information determination model is the stop floor and the number of enabled elevators for each person waiting for the elevator.
[0022] The Gated Recurrent Unit (GRU) is used to process sequence data and timing information. The Gated Recurrent Unit includes three components: a memory unit, an update gate, and a reset gate. The Gated Recurrent Unit can better capture the long-term dependencies in the surveillance video sequence data of the elevator through the gating mechanism, so as to more accurately determine the stop floor and the number of enabled elevators for each person waiting for the elevator. The elevator personnel information determination model can determine the identity of each person waiting for the elevator based on the surveillance video of the elevator, and thus determine the department floor according to the working floor corresponding to the identity of each person waiting for the elevator. For example, the elevator personnel information determination model compares the identified user image with the user image in the database to determine the user's identity. For example, if the identified user is an employee of the accounting department, and the system knows that the accounting department is located on the 8th floor, then the stop floor is determined to be the 8th floor.
[0023] In some embodiments, the elevator personnel information determination model includes a user identification layer and an elevator quantity determination layer. The input of the user identification layer is the monitoring video of the elevator, the output of the user identification layer is the identity of each person waiting for the elevator and the stop floor of each person waiting for the elevator, the input of the elevator quantity determination layer is the stop floor of each person waiting for the elevator, and the output of the elevator quantity determination layer is the number of enabled elevators.
[0024] Dividing the model into two layers allows for modular design, with each layer focusing on handling specific tasks, making it easier to develop, test, and optimize them separately. For example, the user identification layer can focus on image recognition and identity verification, while the elevator quantity determination layer can focus on elevator scheduling and resource allocation.
[0025] Step S3, determining the minimum stop interval of each elevator and the maximum stop interval of each elevator based on the stopping floor of each elevator waiting person and the number of activated elevators.
[0026] In some embodiments, the minimum stop interval and the maximum stop interval of each elevator can be determined by querying the elevator interval preset table. The elevator interval preset table includes the stopping floor of each person waiting for the elevator, the number of enabled elevators, and the corresponding minimum stop interval and the maximum stop interval of each elevator. The elevator interval preset table can be constructed artificially based on historical data.
[0027] Step S4, using a generative adversarial network to generate multiple stop plans based on the stop floors of each elevator waiting person, the number of enabled elevators, the minimum stop interval of each elevator, and the maximum stop interval of each elevator, each of the multiple stop plans includes multiple stop floors of each elevator.
[0028] The Generative Adversarial Network (GAN) includes a generator and a discriminator. The input of the GAN is the stopping floor of each person waiting for the elevator, the number of enabled elevators, the minimum stop interval of each elevator, and the maximum stop interval of each elevator. The output of the GAN is multiple stop plans, each of which includes multiple stop floors for each elevator.
[0029] In the elevator control system based on big data processing, the minimum and maximum intervals for each elevator to stop are determined first, and then multiple stop plans are generated using a generative adversarial network. The reason is that the minimum and maximum intervals provide operating rules for elevator stops, ensuring that the elevator service is neither too frequent and inefficient, nor too long to affect the passenger experience. These intervals serve as constraints for the operation of the system, helping the generator to take into account actual operating restrictions when generating stop plans.
[0030] During the training process of the generative adversarial network, the generator attempts to generate the optimal solution that satisfies all constraints, while the discriminator provides feedback to guide the generator on how to improve.
[0031] The optimization of elevator stop plans can be regarded as a generation problem. The generative adversarial network framework is suitable for this type of problem and can generate multiple candidate plans close to the optimal solution.
[0032] Step S5: determining a target docking plan based on the multiple docking plans.
[0033] In some embodiments, the Figure 2 To determine the target docking plan, Figure 2 A schematic diagram of a process for determining a target docking plan provided by an embodiment of the present invention, Figure 2 The method comprises steps S21 and S22: Step S21, based on the multiple stop plans and the stop floors of each person waiting for the elevator, determine the average waiting time of each person waiting for the elevator in each stop plan and the total power consumption of the elevator; In some embodiments, the average waiting time of each person waiting for the elevator and the total power consumption of the elevator in each stop plan can be determined by a deep neural network model. The deep neural network may include multiple processing layers, each processing layer is composed of multiple neurons, and each neuron performs a matrix transformation on the data. The parameters used by the matrix can be obtained through training. The input of the deep neural network model is the multiple stop plans and the stop floors of each person waiting for the elevator, and the output of the deep neural network model is the average waiting time of each person waiting for the elevator in each stop plan and the total power consumption of the elevator. The deep neural network model can learn the complex relationship between stop plans, passenger behavior and elevator operation from a large amount of historical data. The deep neural network model can capture and simulate nonlinear relationships through multiple hidden layers, which is crucial for predicting waiting time and power consumption.
[0034] Step S22, determining a target parking plan based on the average waiting time of each elevator waiting person in each parking plan and the total power consumption of the elevator.
[0035] In some embodiments, a target docking plan can be determined based on the average waiting time of each person waiting for the elevator in each docking plan and the total power consumption of the elevator through a graph convolutional network. For example, a graph structure is constructed, the graph structure includes multiple nodes and multiple edges between the multiple nodes, each node represents a docking plan, the node features of each node include multiple docking floors of each elevator in each docking plan, the average waiting time of each person waiting for the elevator, and the total power consumption of the elevator, the edges between the nodes represent the similarity between the docking plans, and the graph structure is processed based on the graph convolutional network to determine the target docking plan.
[0036] Graph Convolutional Network (GCN) is a neural network architecture specifically designed to process graph structured data. GCN is designed to process irregular graph data. The input of the graph convolutional network is the average waiting time of each person waiting for the elevator in each docking plan and the total power consumption of the elevator. The output of the graph convolutional network is the target docking plan.
[0037] The optimization of elevator stop plans involves complex relationships between multiple factors, such as the stop floor, waiting time, and total power consumption of each elevator. There are nonlinear and complex interactions between these factors, which can be better captured through graph structures. Each node represents a stop plan, and the node features include various information about the plan, such as the stop floor, waiting time, and total power consumption of each elevator. The graph structure can naturally represent the complex relationships between elevator stop plans, including similarities and differences. These node features can be processed and learned in the graph convolutional network to better understand the characteristics of each stop plan. The edges in the graph represent the similarity between the stop plans, that is, the degree of association between different stop plans. The graph convolutional network can learn this similarity information, so that the mutual influence between different plans can be taken into account when optimizing the target stop plan.
[0038] In some embodiments, the similarity between the docking solutions may be calculated by cosine similarity.
[0039] The input of the graph convolutional network is a graph structure, which includes multiple nodes and multiple edges between the multiple nodes. Each node represents a docking plan. The node features of each node include the multiple docking floors of each elevator in each docking plan, the average waiting time of each elevator waiting person, and the total power consumption of the elevator. The edges between the nodes represent the similarity between the docking plans. The output of the graph convolutional network is the target docking plan.
[0040] Based on the same inventive concept, Figure 3 A schematic diagram of an elevator control system based on big data processing provided by an embodiment of the present invention, the elevator control system based on big data processing includes: An acquisition module 31 is used to acquire the surveillance video of the elevator; A personnel information determination module 32, for determining the stopping floor and the number of enabled elevators for each person waiting for the elevator using an elevator personnel information determination model based on the monitoring video of the elevator; An interval determination module 33 is used to determine the minimum stop interval of each elevator and the maximum stop interval of each elevator based on the stop floor of each elevator waiting person and the number of enabled elevators; A stop plan generating module 34 is used to generate a plurality of stop plans using a generative adversarial network based on the stop floor of each elevator waiting person, the number of enabled elevators, the minimum stop interval of each elevator, and the maximum stop interval of each elevator, wherein each of the plurality of stop plans includes a plurality of stop floors of each elevator; The target docking plan determining module 35 is used to determine a target docking plan based on the multiple docking plans.
Claims
1. An elevator control method based on big data processing, characterized in that: include: Get the elevator surveillance video; Based on the surveillance video of the elevator, a model for determining the stopping floor and the number of enabled elevators for each person waiting for the elevator is used to determine the elevator personnel information; Determine the minimum stop interval of each elevator and the maximum stop interval of each elevator based on the stop floor of each elevator waiting person and the number of enabled elevators; Generate multiple stop plans using a generative adversarial network based on the stop floor of each elevator waiting person, the number of enabled elevators, the minimum stop interval of each elevator, and the maximum stop interval of each elevator, wherein each of the multiple stop plans includes multiple stop floors of each elevator; A target docking plan is determined based on the multiple docking plans.
2. The elevator control method based on big data processing according to claim 1, characterized in that: Determining a target docking plan based on the multiple docking plans includes: Based on the multiple stop plans and the stop floors of each person waiting for the elevator, determine the average waiting time of each person waiting for the elevator in each stop plan and the total power consumption of the elevator; A target parking plan is determined based on the average waiting time of each person waiting for the elevator in each parking plan and the total power consumption of the elevator.
3. The elevator control method based on big data processing according to claim 1, characterized in that: The method of determining the target parking plan based on the average waiting time of each person waiting for the elevator in each parking plan and the total power consumption of the elevator includes: Constructing a graph structure, wherein the graph structure includes a plurality of nodes and a plurality of edges between the plurality of nodes, wherein each node represents a docking plan, and the node features of each node include a plurality of docking floors of each elevator in each docking plan, an average waiting time of each person waiting for the elevator, and a total power consumption of the elevator, and the edges between the nodes represent the similarity between the docking plans; The graph structure is processed based on a graph convolutional network to determine a target docking plan.
4. The elevator control method based on big data processing according to claim 1, characterized in that: The elevator personnel information determination model is a door control cycle unit.
5. The elevator control method based on big data processing according to claim 1, characterized in that: The input of the generative adversarial network is the stopping floor of each person waiting for the elevator, the number of enabled elevators, the minimum stopping interval of each elevator, and the maximum stopping interval of each elevator. The output of the generative adversarial network is multiple stopping plans.
6. An elevator control system based on big data processing, characterized in that: include: An acquisition module is used to acquire the surveillance video of the elevator; A personnel information determination module, used to determine the stopping floor and the number of enabled elevators for each person waiting for the elevator based on the monitoring video of the elevator using an elevator personnel information determination model; An interval determination module, used to determine the minimum stop interval of each elevator and the maximum stop interval of each elevator based on the stop floor of each elevator waiting person and the number of enabled elevators; A stop plan generation module, configured to generate a plurality of stop plans using a generative adversarial network based on the stop floor of each elevator waiting person, the number of enabled elevators, the minimum stop interval of each elevator, and the maximum stop interval of each elevator, wherein each of the plurality of stop plans includes a plurality of stop floors of each elevator; The target docking plan determining module is used to determine the target docking plan based on the multiple docking plans.
7. The elevator control system based on big data processing according to claim 6, characterized in that: The target docking plan determination module is also used for: Based on the multiple stop plans and the stop floors of each person waiting for the elevator, determine the average waiting time of each person waiting for the elevator in each stop plan and the total power consumption of the elevator; A target parking plan is determined based on the average waiting time of each person waiting for the elevator in each parking plan and the total power consumption of the elevator.
8. The elevator control system based on big data processing according to claim 6, characterized in that: The target docking plan determination module is also used for: Constructing a graph structure, wherein the graph structure includes a plurality of nodes and a plurality of edges between the plurality of nodes, wherein each node represents a docking plan, and the node features of each node include a plurality of docking floors of each elevator in each docking plan, an average waiting time of each person waiting for the elevator, and a total power consumption of the elevator, and the edges between the nodes represent the similarity between the docking plans; The graph structure is processed based on a graph convolutional network to determine a target docking plan.
9. The elevator control system based on big data processing according to claim 6, characterized in that: The elevator personnel information determination model is a door control cycle unit.
10. The elevator control system based on big data processing according to claim 6, characterized in that: The input of the generative adversarial network is the stopping floor of each person waiting for the elevator, the number of enabled elevators, the minimum stopping interval of each elevator, and the maximum stopping interval of each elevator. The output of the generative adversarial network is multiple stopping plans.