Elevator Hall People Flow Prediction Without Costly Cameras
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing elevator systems struggle to accurately predict the number of people who will arrive at an elevator hall and their destination floors without the need for costly equipment like cameras.
Innovation Solution
A computer-based method using simulation and machine learning to predict future elevator traffic by calculating past elevator data, creating virtual scenarios, and developing conversion models to estimate the number of people arriving at elevator halls.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If cameras or other expensive sensing equipment are installed in elevator halls to directly measure the number of people appearing, then measurement precision of people flow is improved, but device complexity and installation cost increase significantly
Solution Approach 1:
The patent creates a virtual copy of the elevator hall environment through simulation, where a virtual elevator hall and virtual people are modeled based on actual elevator operation data. Instead of directly measuring physical people with cameras, the system creates and analyzes a virtual replica that mirrors real-world conditions, achieving accurate people flow measurement without expensive sensing equipment.
Solution Approach 2:
The patent replaces the mechanical/optical sensing system (cameras, sensors) with an information processing system. By substituting physical measurement devices with computational models that process elevator operation data, the system achieves people flow measurement through data analysis rather than direct physical sensing.
2Measurement precision
If machine learning models are trained directly on limited sensor data from elevators alone, then device complexity is kept low, but measurement precision of people flow prediction deteriorates due to insufficient data
Solution Approach 1:
The patent performs preliminary actions by pre-generating comprehensive simulation data that covers various elevator operation scenarios before actual prediction is needed. This simulation data is created in advance and stored, providing a rich training dataset that would otherwise require extensive real-world data collection and processing.
Solution Approach 2:
The patent introduces simulation data as an intermediary between the limited sensor data and the machine learning model training. The simulation data acts as a bridge that enriches the training dataset, allowing the model to learn from a broader range of scenarios without requiring direct installation of additional sensors or extensive real-world data collection.
3Measurement precision
If real-time prediction of people flow is implemented using complex models, then measurement precision is improved, but processing time and computational resources increase
Solution Approach 1:
The patent performs model training and conversion model generation in advance as preliminary actions. By pre-training the machine learning model with comprehensive simulation data and generating conversion models beforehand, the system reduces the computational burden during real-time operation, enabling fast and accurate predictions without requiring complex real-time processing.
Data Source
AI summary
A people flow prediction system and method are provided, in which a first conversion model for converting virtual getting in and out data before a certain time point into a 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 based on the number of people who appear and the virtual getting in and out data. A prediction model is learned based on the number of people who appear converted by the second conversion model, and a 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.