| CPC G08G 1/202 (2013.01) [G06N 20/00 (2019.01); G06Q 10/04 (2013.01); G06Q 10/06315 (2013.01); G06Q 50/40 (2024.01); H04W 4/023 (2013.01)] | 5 Claims |
|
1. A system for providing a big data-based AI automatic allocation matching service using a taxi demand prediction, the system comprising:
a user terminal comprising a hardware processor configured to:
input a current location and a destination;
transmit a taxi call, to a matching service providing server connected with the user terminal via a network; and
receive a vehicle number and an expected arrival time of a called taxi, from the matching service providing server; and
output the received vehicle number and expected arrival time;
a taxi terminal comprising a hardware processor configured to:
receive a taxi call signal, which is matched with automatic allocation in response to the taxi call transmitted from the user terminal;
start, when an approval event in response to the received taxi call signal is output, a destination guidance to the current location of the user terminal;
transmit an exclusion request signal, which is a GPS signal for requesting exclusion of a corresponding taxi from an empty car list, to the matching service providing server connected with the taxi terminal via the network; and
the matching service providing server comprising a hardware processor configured to:
map at least one taxi call signal on a location, time, and day of week;
store the mapped at least one taxi call signal as an accumulated history log, and build big data;
perform a verification process of the big data with machine learning or deep learning of an artificial neural network having a Convolutional neural network (CNN) structure;
perform data mining on the big data to execute the taxi demand prediction based on the taxi call signal corresponding to the location, time, and day of week;
estimate a kernel density of taxi rides through a taxi call signal and taxi ride data, compare and analyze the kernel density with location data of taxi ranks, to find out whether there is a dense area of taxi rides by using Moran's I analysis, and to find out areas where taxi rides are densely concentrated by using Inverse Distance Weighting and Spatial Weighted Matrix;
execute the taxi demand prediction on the location, time and day of week, based on the estimated kernel density,
transmit a taxi demand amount that is predicted on the location, time and day of week where the taxi demand prediction is executed, to the taxi terminal;
match a single taxi terminal located at a shortest distance from a current location of the user terminal, when the taxi call is received from the user terminal; and
transmit, via the network, the taxi call signal to the matched single taxi terminal, and transmit, via the network, the vehicle number and the expected arrival time to the user terminal when an approval event is received from the taxi terminal,
wherein the processor of the matching service providing server is further configured to:
call the taxi terminal located at the shortest distance from the current location of the user terminal, by using at least one taxi call list and the empty car list as input values to an optimal algorithm for an assignment problem through which an optimal solution for the assignment problem is obtained,
wherein, to increase a speed of obtaining the optimal solution for the assignment problem, calculation of an optimized approximate solution is performed using a simulated annealing, and
wherein the processor of the matching service providing server is further configured to:
perform the taxi demand prediction by: establishing a prediction model with past taxi getting-on and-off data using time series prediction, by using a computer-based algorithm; predicting an expected number of passengers by place, time, and day of the week; visually displaying a predicted result on a map; and figuring out whether weather or events affect the number of passengers, using changes due to the weather or changes due to the events;
perform a pattern analysis by: converting an administrative area to which an actual road name of (X,Y) coordinates belongs into a text value or a numeric code; filtering data that satisfies a specific condition; and calculating a sum of a number of rides of filtered data; and
transmit results of the taxi demand prediction and the pattern analysis to the taxi terminal.
|