CN113159372A - Conv1D + LSTM-based multi-step traffic flow prediction method - Google Patents

Conv1D + LSTM-based multi-step traffic flow prediction method Download PDF

Info

Publication number
CN113159372A
CN113159372A CN202110121994.2A CN202110121994A CN113159372A CN 113159372 A CN113159372 A CN 113159372A CN 202110121994 A CN202110121994 A CN 202110121994A CN 113159372 A CN113159372 A CN 113159372A
Authority
CN
China
Prior art keywords
traffic flow
conv1d
data
lstm
inputting
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110121994.2A
Other languages
Chinese (zh)
Inventor
郑文
焦乐乐
张越
康洛耀
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Taiyuan University of Technology
Original Assignee
Taiyuan University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Taiyuan University of Technology filed Critical Taiyuan University of Technology
Priority to CN202110121994.2A priority Critical patent/CN113159372A/en
Publication of CN113159372A publication Critical patent/CN113159372A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/01—Detecting movement of traffic to be counted or controlled
    • G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a Conv1D + LSTM-based multi-step traffic flow prediction method, which comprises the following steps: collecting available traffic flow data and carrying out preliminary screening; performing data preprocessing on the screened data, for example, repairing abnormal points of the data, removing noise in the data, performing normalization processing on the data, and the like; inputting the processed data into a one-dimensional convolution Conv1D to extract the characteristics of the traffic flow and simultaneously extracting the influence of external factors on the traffic flow; inputting the features extracted by the Conv1D layer into an LSTM layer for learning so as to predict traffic flow values of a plurality of time intervals in the future; and (4) processing the predicted value generated by the model in an anti-normalization mode. The method combines external factors such as weather, time information, holidays and the like, utilizes Conv1D to model time characteristics, cycle characteristics and local relevant characteristics of traffic flow data, and sends the extracted characteristics to an LSTM to perform multi-step prediction. The invention effectively realizes the multi-step prediction of the traffic flow by fully considering the influence of external factors on the traffic flow.

Description

Conv1D + LSTM-based multi-step traffic flow prediction method
Technical Field
The invention relates to the technical field of traffic flow prediction, in particular to a Conv1D + LSTM-based multi-step traffic flow prediction method.
Background
Traffic flow prediction is an important research area in current intelligent traffic management systems. In recent years, with the increasing progress of urbanization, traffic congestion has caused economic, social and environmental problems in many cities around the world. Therefore, traffic prediction becomes an important research field in current intelligent traffic management systems, and the emphasis is to predict traffic to alleviate congestion. By referring to the traffic flow prediction result, related departments not only can process traffic by adopting a corresponding traffic management strategy, but also can provide travel suggestions for pedestrians so as to ensure better liquidity and less congestion. Therefore, it is a main objective of traffic managers to accurately predict road traffic flow in real time, which is a problem to be solved urgently.
At present, although researchers at home and abroad have achieved certain results in the aspect of traffic flow prediction, the following 2 problems still exist. First, most of these traffic flow prediction methods implement only single-step prediction and do not implement multi-step prediction (i.e., only predict the next time interval and do not predict multiple time intervals). The actual multi-step prediction of the traffic flow has more practical significance. First, the multi-step traffic flow indicates a change trend of traffic conditions for a certain period in the future, which is advantageous to avoid an impulsive traffic scheduling response due to temporary fluctuations. From the aspect of dynamic decision, it is far more important than obtaining the traffic condition at the current moment, because the traffic flow demand prediction result at the current moment is often short-time and is easy to cause extra traffic scheduling. Secondly, the methods only use the historical traffic flow data, and do not take the influence of external factors on the traffic flow into consideration. Actually, weather, holidays, date and time information and the like easily influence the traffic flow, and numerous researches show that extracting external factors influencing the traffic flow and determining the relevance of the external factors and the traffic flow are of great significance for improving the accuracy of traffic flow prediction.
Disclosure of Invention
The invention aims to solve the technical problem that the multi-step traffic flow prediction method based on Conv1D + LSTM aims at overcoming the defects in the prior art.
The technical scheme adopted by the invention for solving the technical problems is as follows: constructing a multi-step traffic flow prediction method based on Conv1D + LSTM, comprising the following steps:
step 1: collecting available traffic flow data and carrying out preliminary screening;
step 2: carrying out data preprocessing on the screened traffic flow data;
and step 3: inputting the preprocessed traffic flow data into a one-dimensional convolution Conv1D to extract traffic flow data characteristics, and extracting the influence of external factors on the traffic flow data;
and 4, step 4: inputting the features extracted by the Conv1D layer into an LSTM layer for learning so as to predict traffic flow values of a plurality of time intervals in the future;
and 5: and (4) carrying out anti-normalization processing on the traffic flow value generated by the model as a prediction result.
The mode of carrying out data preprocessing on the traffic flow data in the step 2 comprises the following steps: and repairing abnormal points of the data, removing noise in the data and normalizing the data.
Wherein, the step of inputting the preprocessed traffic flow data into the one-dimensional convolution Conv1D to extract the traffic flow data features in the step 3 comprises the following steps: the temporal, periodic and locally relevant characteristics of traffic flow data were modeled with Conv1D in conjunction with weather, time information and external factors of holidays.
Wherein, in the step of inputting the characteristics extracted by the Conv1D layer into the LSTM layer for learning so as to predict the traffic flow values of a plurality of time intervals in the future,
the LSTM calculation is as follows:
ft＝σ(Wf·[ht-1,Xt]+bf)
it＝σ(Wi·[ht-1,Xt]+bi)
Figure BDA0002922489350000021
Figure BDA0002922489350000031
ot＝σ(Wo·[ht-1,Xt]+bo)
Figure BDA0002922489350000032
wherein, Wf,Wi,WoWeight matrices, b, each representing a corresponding gatef,bi,boRepresenting offset terms, symbols "
Figure BDA0002922489350000034
"denotes multiplication by element, σ denotes the activation function,
Figure BDA0002922489350000033
indicating the state of the cell at the current time t network input.
In step 5, the predicted value generated by the inverse normalization processing model is as follows:
X＝x″×(max(x)-min(x))+min(x)
wherein x "is a normalized value of the model prediction; and X is a predicted true value of the model.
Compared with the prior art, the multi-step traffic flow prediction method based on Conv1D + LSTM has the following technical effects: first, the one-dimensional convolution and long-short term memory network are integrated. One-dimensional convolution is used to capture time, period and locally relevant features of traffic flow data to mine deep features of traffic flow and to efficiently utilize traffic flow time series data. The LSTM uses the features extracted by the one-dimensional convolution to carry out multi-step prediction on the traffic flow. Second, the Conv1D + LSTM model adequately accounts for the effects of external factors on traffic flow. The result shows that the prediction precision of the method reaches 91% compared with other methods.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a block diagram of the LSTM model according to the present patent
Fig. 2 is a schematic overall framework diagram of a multi-step traffic flow prediction method based on Conv1D + LSTM provided by the invention.
Detailed Description
For a more clear understanding of the technical features, objects and effects of the present invention, embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
As shown in fig. 2, the present invention designs a multi-step traffic flow prediction method based on Conv1D + LSTM, including:
step 1: collecting available traffic flow data and carrying out preliminary screening;
step 2: carrying out data preprocessing on the screened traffic flow data;
and step 3: inputting the preprocessed traffic flow data into a one-dimensional convolution Conv1D to extract traffic flow data characteristics, and extracting the influence of external factors on the traffic flow data;
and 4, step 4: inputting the features extracted by the Conv1D layer into an LSTM layer for learning so as to predict traffic flow values of a plurality of time intervals in the future;
and 5: and (4) carrying out anti-normalization processing on the traffic flow value generated by the model as a prediction result.
The mode of carrying out data preprocessing on the traffic flow data in the step 2 comprises the following steps: and repairing abnormal points of the data, removing noise in the data and normalizing the data.
Wherein, the step of inputting the preprocessed traffic flow data into the one-dimensional convolution Conv1D to extract the traffic flow data features in the step 3 comprises the following steps: the temporal, periodic and locally relevant characteristics of traffic flow data were modeled with Conv1D in conjunction with weather, time information and external factors of holidays.
Wherein, in the step of inputting the characteristics extracted by the Conv1D layer into the LSTM layer for learning so as to predict the traffic flow values of a plurality of time intervals in the future,
the LSTM calculation is as follows:
ft＝σ(Wf·[ht-1,Xt]+bf)
it＝σ(Wi·[ht-1,Xt]+bi)
Figure BDA0002922489350000041
Figure BDA0002922489350000042
ot＝σ(Wo·[ht-1,Xt]+bo)
Figure BDA0002922489350000043
wherein, Wf,Wi,WoWeight matrices, b, each representing a corresponding gatef,bi,boRepresenting the bias term, the symbol "°" representing multiplication by element, σ representing the activation function,
Figure BDA0002922489350000051
indicating the state of the cell at the current time t network input.
In step 5, the predicted value generated by the inverse normalization processing model is as follows:
X＝x″×(max(x)-min(x))+min(x)
wherein x "is a normalized value of the model prediction; and X is a predicted true value of the model.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (5)

1. A Conv1D + LSTM-based multi-step traffic flow prediction method is characterized by comprising the following steps:
step 1: collecting available traffic flow data and carrying out preliminary screening;
step 2: carrying out data preprocessing on the screened traffic flow data;
and step 3: inputting the preprocessed traffic flow data into a one-dimensional convolution Conv1D to extract traffic flow data characteristics, and extracting the influence of external factors on the traffic flow data;
and 4, step 4: inputting the features extracted by the Conv1D layer into an LSTM layer for learning so as to predict traffic flow values of a plurality of time intervals in the future;
and 5: and (4) carrying out anti-normalization processing on the traffic flow value generated by the model as a prediction result.
2. The Conv1D + LSTM-based multi-step traffic flow prediction method according to claim 1, wherein the means for preprocessing the traffic flow data in step 2 comprises: and repairing abnormal points of the data, removing noise in the data and normalizing the data.
3. The Conv1D + LSTM-based multi-step traffic flow prediction method according to claim 1, wherein the step of inputting the pre-processed traffic flow data into a one-dimensional convolution Conv1D to extract the traffic flow data features in step 3 comprises: the temporal, periodic and locally relevant characteristics of traffic flow data were modeled with Conv1D in conjunction with weather, time information and external factors of holidays.
4. The Conv1D + LSTM-based multi-step traffic flow prediction method according to claim 1, wherein in the step of inputting Conv1D layer extracted features to LSTM layer for learning to predict traffic flow values for a plurality of time intervals in the future,
the LSTM calculation is as follows:
ft＝σ(Wf·[ht-1,Xt]+bf)
it＝σ(Wi·[ht-1,Xt]+bi)
Figure FDA0002922489340000021
Figure FDA0002922489340000022
ot＝σ(Wo·[ht-1,Xt]+bo)
Figure FDA0002922489340000023
wherein, Wf,Wi,WoWeight matrices, b, each representing a corresponding gatef,bi,boRepresenting offset terms, symbols
Figure FDA0002922489340000024
Meaning multiplication by element, sigma denotes the activation function,
Figure FDA0002922489340000025
indicating the state of the cell at the current time t network input.
5. The Conv1D + LSTM-based multi-step traffic flow prediction method according to claim 1, wherein in step 5, the inverse normalization process model generates prediction values:
X＝x″×(max(x)-min(x))+min(x)
wherein x "is a normalized value of the model prediction; and X is a predicted true value of the model.
CN202110121994.2A 2021-01-29 2021-01-29 Conv1D + LSTM-based multi-step traffic flow prediction method Pending CN113159372A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110121994.2A CN113159372A (en) 2021-01-29 2021-01-29 Conv1D + LSTM-based multi-step traffic flow prediction method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110121994.2A CN113159372A (en) 2021-01-29 2021-01-29 Conv1D + LSTM-based multi-step traffic flow prediction method

Publications (1)

Publication Number Publication Date
CN113159372A true CN113159372A (en) 2021-07-23

Family

ID=76878996

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110121994.2A Pending CN113159372A (en) 2021-01-29 2021-01-29 Conv1D + LSTM-based multi-step traffic flow prediction method

Country Status (1)

Country Link
CN (1) CN113159372A (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108510741A (en) * 2018-05-24 2018-09-07 浙江工业大学 A kind of traffic flow forecasting method based on Conv1D-LSTM neural network structures
CN109658695A (en) * 2019-01-02 2019-04-19 华南理工大学 A kind of multifactor Short-time Traffic Flow Forecasting Methods
CN110070715A (en) * 2019-04-29 2019-07-30 浙江工业大学 A kind of road traffic flow prediction method based on Conv1D-NLSTMs neural network structure
CN110188936A (en) * 2019-05-23 2019-08-30 浙江大学 Short-time Traffic Flow Forecasting Methods based on multifactor spatial choice deep learning algorithm

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108510741A (en) * 2018-05-24 2018-09-07 浙江工业大学 A kind of traffic flow forecasting method based on Conv1D-LSTM neural network structures
CN109658695A (en) * 2019-01-02 2019-04-19 华南理工大学 A kind of multifactor Short-time Traffic Flow Forecasting Methods
CN110070715A (en) * 2019-04-29 2019-07-30 浙江工业大学 A kind of road traffic flow prediction method based on Conv1D-NLSTMs neural network structure
CN110188936A (en) * 2019-05-23 2019-08-30 浙江大学 Short-time Traffic Flow Forecasting Methods based on multifactor spatial choice deep learning algorithm

Similar Documents

Publication Publication Date Title
CN108417033B (en) Expressway traffic accident analysis and prediction method based on multi-dimensional factors
CN102087788B (en) Method for estimating traffic state parameter based on confidence of speed of float car
CN109376906B (en) Travel time prediction method and system based on multi-dimensional trajectory and electronic equipment
CN112270355A (en) Active safety prediction method based on big data technology and SAE-GRU
CN112863182B (en) Cross-modal data prediction method based on transfer learning
CN110348601A (en) A kind of short-term passenger flow forecast method of subway based on two-way shot and long term memory network
CN114463972A (en) Road section interval traffic analysis and prediction method based on ETC portal communication data
CN114048436A (en) Construction method and construction device for forecasting enterprise financial data model
Huang et al. Research on urban modern architectural art based on artificial intelligence and GIS image recognition system
CN115545473A (en) LSTM-based intelligent prediction method for domestic garbage throwing trend
CN110795888A (en) Petroleum drilling risk prediction method
Yu et al. A novel discussion on two long-term forecast mechanisms for hydro-meteorological signals using hybrid wavelet-NN model
CN116542391B (en) Urban area passenger flow volume prediction method based on big data
CN113159372A (en) Conv1D + LSTM-based multi-step traffic flow prediction method
CN116629460A (en) Subway passenger flow prediction method based on ST-RANet model
CN116341391A (en) Precipitation prediction method based on STPM-XGBoost model
CN111275073A (en) Regional people flow analysis method based on mobile phone signaling data
CN114140731B (en) Traction substation abnormality detection method
CN115293293A (en) Rain type fuzzy recognition evaluation method, system and terminal equipment
CN112633579B (en) Traffic flow migration prediction method based on domain countermeasure
CN111143774B (en) Power load prediction method and device based on influence factor multi-state model
CN114418606A (en) Network taxi appointment order demand prediction method based on space-time convolutional network
CN112926670A (en) Garbage classification system and method based on transfer learning
Xiong et al. CLSTAN: convLSTM-based spatiotemporal attention network for traffic flow forecasting
CN104573870A (en) Expressway operating cost forecasting method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20210723