Parking lot free parking space predicting method, apparatus, electronic device and storage medium

Patent metadata

Field	Value
Publication number	US-11574259-B2
Application number	US-202017021831-A
Country	US
Kind code	B2
Filing date	Sep 15, 2020
Priority date	Jan 23, 2020
Publication date	Feb 7, 2023
Grant date	Feb 7, 2023

Abstract

A parking lot free parking space predicting method, apparatus, electronic device and storage medium are provided. The method comprises: building a parking lot association graph for parking lots in a region to be processed; as for any parking lot i, performing the following processing respectively: determining local space correlation information of the parking lot i at a current time according to environment context features of the parking lot i and neighboring parking lots which are in the parking lot association graph and connected to the parking lot i through edges; determining time correlation information of the parking lot i at the current time according to the local space correlation information, and predicting free parking space information of the parking lot i at at least one future time step according to the time correlation information of the parking lot i at the current time.

First claim

What is claimed is:

1. A computer-implemented parking lot free parking space predicting method, comprising:
building a parking lot association graph for parking lots in a region to be processed, each junction therein representing a parking lot, and connecting any two parking lots meeting a predetermined condition through edges;
aggregating, by at least one processor, environment context features of neighboring parking lots according to weights of edges between the neighboring parking lots and a parking lot i to obtain a representation vector of the parking lot i at a current time;
pre-training, by the at least one processor, a graph attention neural network model using the environment context features of the neighboring parking lots and free parking space information;
pre-training, by the at least one processor, a gated recurrent neural network model according to the representation vector of the parking lot i at the current time;
for a parking lot i, performing the following processing respectively:
determining, by the at least one processor, local space correlation information of the parking lot i at a current time based on the graph attention neural network model according to environment context features of the parking lot i and neighboring parking lots which are in the parking lot association graph and connected to the parking lot i through edges;
determining, by the at least one processor, time correlation information of the parking lot i at the current time based on the gated recurrent neural network model according to the local space correlation information; and
predicting, by the at least one processor, free parking space information of the parking lot i at at least one future time step according to the time correlation information of the parking lot i at the current time;
wherein the determining local space correlation information of parking lot i at a current time based on a graph attention neural network model comprises:
for the parking lot i, executing the following predetermined processing:
for each neighboring parking lot, determining a weight of edge between the neighboring parking lot and the parking lot i at the current time according to the environment context features of the neighboring parking lots and parking lot i at the current time, respectively;
incrementing execution times by one, the execution times being initially zero;
if the execution times are equal to a second predetermined threshold, selecting the representation vector as the local space correlation information of the parking lot i at the current time, otherwise, selecting the representation vector as the environment context features of the parking lot i, and executing the predetermined processing repeatedly,
wherein a weight α ij of the edge between any neighboring parking lot j and parking lot i is expressed as

α

i
⁢
j

=

exp
⁡
(

c

i
⁢
j

)

∑

k
∈

N
i

exp
⁡
(

c

i
⁢
k

)

;

where c ij =Attention(W a x i , W a x j ); Attention represents a graph attention mechanism; N i represents the number of neighboring parking lots; x i represents the environment context feature of the parking lot i at the current time; x j represents the environment context feature of neighboring parking lot j at the current time; W a represents a model parameter obtained by pre-training, or the representation vector x′ i =σ(Σ j∈N α ij W a x j ),
where N i represents a number of the neighboring parking lots; x j represents the environment context feature of any neighboring parking lot j among N i neighboring parking lots at the current time; α ij represents a weight of the edge between the neighboring parking lot j and parking lot i at the current time; W a represents a model parameter obtained by pre-training; σ represents an activation function, and
wherein the method further enhancing the accuracy of subsequent prediction results comprises:
training the graph attention neural network model and the gated recurrent neural network model by selecting N l parking lots with real-time sensors as sample parking lots, building annotation data based on historical free parking space information of the sample parking lots, performing training optimization based on the annotation data, and minimizing an objective function O;
where the objective function

O
=

1

τ
⁢

N
l

⁢

∑

i
=
1

N
l

∑

j
=
1

τ

(

y
ˆ

i

t
+
j

-

y
i

t
+
j

)

2

,

where N l is a positive integer greater than 1; y i t+j represents real free parking space information of any

Assignees

• Beijing Baidu Netcom Sci & Tech Co Ltd

Inventors

• Liu Hao
• Zhang Weijia
• Xiong Hui

Classifications

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06Q10/06 Primary

  Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N3/045

  Combinations of networks · CPC title