Freight transport planning and management method and system based on intelligent traffic light OD information examination

An information inspection and management method technology, applied in the field of intelligent transportation, can solve the problems of inability to accurately predict the next time period, inability to manage road traffic, inability to predict and monitor the state of trucks, etc., to achieve accurate data and avoid road congestion. Effect

Pending Publication Date: 2018-08-03
ZHEJIANG FONDA CONTROL TECH
0 Cites 3 Cited by

AI-Extracted Technical Summary

Problems solved by technology

[0003] Although the development trend of the intelligent transportation system is very good, there are still some problems. Because there are more and more trucks on the road, it is impossible to accura...
View more

Method used

In the present invention, adopted semi-supervised learning (semi-supervised learning) algorithm to carry out pattern recognition, after finishing above-mentioned information dimensionality reduction and clustering, existing identification pattern prototype, and can obtain specific by the reverse demarcation of information The information belongs to the mode to guide and predict the subsequent traffic planning, but on the one hand, the traffic OD information flow is not a fixed data set, and on the other hand, in order to obtain more accurate and specific traffic modes for analysis and prediction, this application uses semi The supervised learning algorithm (Semi-supervised learning) performs pattern recognition, and in semi-supervised learning, part of the training data is labeled, and the other part is unlabeled, and the number of unlabeled data is often greater than the number of labeled data. For the acquisition of labeled data sets, historical traffic data with labels can be used, but it is also necessary to accurately calibrate the traffic OD information through manual labeling in the early stage. Therefore, it is necessary to calibrate the traffic OD information,...
View more

Abstract

The invention discloses a freight transport planning and management method based on intelligent traffic light OD information examination. The freight transport planning and management method comprisesthe steps of obtaining freight transport information of each passed intersection, wherein the transportation information comprises transportation video information and truck information of a passed truck; extracting key frame images of a transportation video of the passed truck to obtain related information and calibrating OD information of the passed truck; processing the OD information of the passed truck, building a fright transport management model to locate the position of the truck in real time and predict the next position to obtain conditions of the truck and all road sections and predict the conditions of the truck and all the road sections, and according to the conditions of the road sections, dispatching the truck. By means of the method, not only can the road conditions of thenext duration be more precisely predicated, but also the truck can be managed and monitored in many ways according to the OD information of the passed truck; moreover, according to the road conditions, the truck is dispatched, the occurrence of road congestion is avoided, the data is precise, and the method is convenient to implement and feasible.

Application Domain

Detection of traffic movementForecasting +2

Technology Topic

TruckImage processing +5

Image

  • Freight transport planning and management method and system based on intelligent traffic light OD information examination
  • Freight transport planning and management method and system based on intelligent traffic light OD information examination

Examples

  • Experimental program(1)

Example Embodiment

[0041] Example 1:
[0042] A freight planning management method based on intelligent traffic light OD information inspection, such as figure 1 As shown, including the following steps:
[0043] S100. Obtain truck information passing through each intersection, where the traffic information includes traffic video information and truck information passing through the truck;
[0044] S200: Extract key frame image processing on the traffic video of the passing truck to obtain the license plate information of the passing truck, the information of the truck itself and the truck environment information, and calibrate and record the OD information of the passing truck;
[0045] S300. Process the OD information of the passing trucks and establish a freight management model. The freight management model locates the location of the truck in real time and predicts the next location, obtains the condition of the truck and each section, and predicts the condition of the truck and each section, and according to The condition of the road section dispatches trucks.
[0046] More specifically, the truck information passing through each intersection is generated by a video surveillance sensor device, and is transmitted via a communication module set on a traffic light, and the video surveillance sensor device is set on a traffic light at each intersection.
[0047] In step S200, the key frame image processing is performed on the traffic video of the passing truck to obtain the license plate information of the passing truck, the information of the truck itself and the truck environment information, and the OD information of the passing truck is calibrated and recorded. The specific steps include:
[0048] S210. Extracting key frames from all the obtained traffic videos of trucks passing by, removing non-key frames, obtaining relevant information about the passing truck, recording the license plate information of the truck, the color and model of the truck, and the traffic information of the lane where the truck is located;
[0049] S220. Query all intersections the truck passes through according to the license plate information of the truck, and obtain the OD information of the passing truck and perform calibration.
[0050] From the perspective of the hardware technical solution, the present invention is a video monitoring sensor device installed on the traffic light, and the traffic information of the truck is obtained by recording video by the video monitoring sensor device, and is transmitted through the communication module to the computing center After receiving the video data, the computing center extracts the key frame images through the video key frame detection algorithm, obtains vehicle information through key frame time axis sorting and image recognition based on the key point anchor algorithm, and calibrates and records the OD information of the passing vehicles.
[0051] In step S200, the freight management model includes at least a truck management model and a road section management model.
[0052] Based on the freight management model including at least a truck management model and a road section management model, the OD information of the passing trucks is processed to establish a freight management model. The freight management model performs real-time positioning of the truck's location and prediction of the next location to obtain the truck And the conditions of each road section and predict the condition of trucks and various road sections, and the specific steps for dispatching trucks according to the road section conditions are:
[0053] S310. By selecting the vector of the OD information of the truck and the traffic information of the lane where the truck is located, a non-parametric regression prediction algorithm is used to establish a truck management model and a road section management model respectively;
[0054] S320. Input real-time truck related information into the truck management model for analysis and prediction of the next location, obtain the truck and each road section status, and predict the truck and each road section status;
[0055] S330. Manage the trucks according to the conditions of the trucks and various road sections and predict the conditions of the trucks and various road sections, and perform an integrated analysis of the status of each road section to obtain complete overall regional traffic information, according to the overall regional traffic information combined with the truck status Dispatch trucks.
[0056] Through accumulated statistics of previous historical data, a variety of freight transportation models can be established with non-parametric regression prediction algorithms, and the results of pre-planned routes can be obtained through algorithm loop iteration. The specific process is to first select the starting location, speed and time requirements and other freight traffic state vectors, and then calculate and select the number of historical similarity curves, and change the similarity measurement criteria parameters according to different situations, and find the most similar historical data through comparison And the performance is excellent, so as to carry out freight transportation route planning and traffic state prediction.
[0057] The condition of the truck also depends on the OD information and real-time feedback of the intelligent traffic lights. The key frame extraction can be carried out through the video image information of multiple special traffic light nodes on the pre-planned route, which can locate the vehicle in a certain section in real time, and can be based on the freight vehicle passing through. The speed of these road sections is more accurate to predict the arrival time, and the positioning information is fed back to the shipper, receiver, and freight dispatching and control center in real time. This positioning method is accurate to a specific road section, and is more accurate and does not rely on human reporting or The identification does not depend on the equipment level of the freighter or related operations, and it can report in real time even without the influence of the strength of the regional signal. It is suitable for most scenarios and has strong practicability.
[0058] At the same time, by acquiring the traffic light information of the unpassed part of the planned route, real-time monitoring and feedback of special events on the route ahead can be carried out. Once abnormal congestion, accidents, maintenance and other special circumstances occur, the freight dispatching monitoring and control center will re-schedule the route. Plan, give early warning to the freight forwarder and help the freight forwarder make decisions, and minimize the unnecessary extension of freight time.
[0059] Before dispatch, the computing center will analyze and plan the current status of the vehicle, the history of the driver, the type of goods and the route planning route area, etc., and give the driver a targeted safety reminder. During the driving, the driver himself will conduct safety inspections and safety. You can report on sex, and you can take the initiative to call for help when an accident occurs. Secondly, the intelligent traffic light video surveillance will obtain more detailed information and more frequent feedback on the current section of the freight vehicle. Once an accident occurs, the traffic information of the section will inevitably change significantly. When this happens, it will Obtain information from other ways to confirm whether it is a freight vehicle accident. If so, perform rescue and freight loss assessment operations. If not, use the time status information of the freight vehicle passing the D traffic light to determine whether the route needs to be re-planned and how to re-plan the route .
[0060] In the present invention, the key frame image processing is performed on the traffic video of the passing truck to obtain the license plate information of the passing truck, the information of the truck itself and the truck environment information, and the OD information of the passing truck is calibrated and recorded. More specifically, the following processing is implemented , Mainly depends on the use of PCA algorithm (Principal Component Analysis Algorithm) information preprocessing and dimensionality reduction algorithm. The specific operations are as follows: After obtaining the OD information of passing vehicles, perform information dimensionality reduction and clustering processing on it, Information dimensionality reduction processing relies on the PCA method, which is mainly to perform preprocessing such as information compression and information structure simplification under the premise of minimizing information loss, which facilitates subsequent information clustering and pattern recognition, and facilitates the visual presentation of information . In the PCA algorithm, the original data is arranged in rows to form a matrix, and the matrix is ​​standardized to make its mean value zero; find the covariance matrix of the matrix; and then arrange the eigenvectors in descending order of eigenvalues, taking the first k A new matrix is ​​formed by rows; finally the data after dimensionality reduction is obtained. Here, the principal component of the matrix is ​​obtained by sorting the eigenvectors of its covariance matrix according to the size of the corresponding eigenvalues. By analyzing the actually collected data, the singular value decomposition of the data matrix can be used to find The eigenvector of the covariance matrix and the square root of the eigenvalue complete the principal component analysis, or select an appropriate kernel function to project the data to a low-dimensional subspace to achieve dimensionality reduction. In general, an n-dimensional data set can be reduced to a k-dimensional subspace by mapping, where k≤n. In this patent, the variance distribution in different dimensions of the data set is relatively uneven. At this time, the PCA effect is better, and you can get The obtained multi-dimensional traffic OD information (including front and back time information, vehicle information, road section information, environmental information, behavior information and other complex dimensions) is compressed and integrated into two dimensions for visualization and pattern recognition analysis.
[0061] The clustering process uses the K-means algorithm for clustering. The method is as follows: After the data is reduced in dimensionality, the K-means algorithm is used to complete the cluster analysis of the information. The purpose of the cluster analysis is to find the relationship between the data objects in the data set. Group the data. The greater the similarity within the group and the greater the difference between the groups, the better the clustering effect. Therefore, the choice of distance measurement and objective function is very important. In the present invention, Euclidean is selected. Reid distance is used as a data distance measure, and Sum of the Squared Error (SSE) is used as the objective function of clustering. By pre-designating K centroids (you can use random designation multiple loop calculations or manual designation two ways, It depends on the actual labor and hardware capabilities of the computing center) and performs multiple iterations to cluster the information. After the clustering is completed, it is equivalent to the preliminary completion of pattern recognition classification and data visualization.
[0062] In the present invention, a semi-supervised learning algorithm is used for pattern recognition. After the above-mentioned information dimensionality reduction and clustering are completed, the prototype of the recognition pattern already exists, and the pattern of specific information can be obtained through the reverse calibration of the information In order to guide and predict subsequent traffic planning, on the one hand, traffic OD information flow is not a fixed data set, on the other hand, in order to obtain more precise and specific traffic patterns for analysis and prediction, this application uses semi-supervised learning algorithms (Semi-supervised learning) performs pattern recognition. In semi-supervised learning, part of the training data is labeled and the other part is unlabeled. The amount of unlabeled data is often greater than the amount of labeled data. For the acquisition of the data set, you can use tagged historical traffic data, but you also need to accurately calibrate the traffic OD information through manual tagging in the early stage. Therefore, you need to calibrate the traffic OD information because the data distribution is inevitable It is not completely random. Through some local features of labeled data and the overall distribution of more unlabeled data, you can get acceptable or even very good classification results.
[0063] The present invention also discloses the following technical solutions:
[0064] A freight planning management system based on intelligent traffic light OD information inspection, such as figure 2 As shown, it includes a data acquisition module 100, an analysis and calibration module 200, and a modeling prediction module 300;
[0065] The data acquisition module 100 is configured to acquire truck information passing through each intersection, and the traffic information includes traffic video information and truck information of the truck passing through;
[0066] The analysis and calibration module 200 is used to extract the key frame image processing of the traffic video of the passing truck, obtain the license plate information of the passing truck, the information of the truck itself and the truck environment information, and calibrate and record the OD information of the passing truck;
[0067] The modeling prediction module 300 is used to process the OD information of passing trucks and establish a freight management model. The freight management model locates the truck in real time and predicts the next location, obtains the status of the truck and each road section, and predicts the truck And the conditions of each road section, and dispatch trucks according to the road section conditions.
[0068] Furthermore, the data acquisition module 100 includes a video surveillance sensor device, and the truck information at each intersection of the acquisition path is generated by the video surveillance sensor device, and is transmitted via a communication module set on a traffic light. The monitoring and sensing equipment is set on the traffic lights at each intersection, the communication module is set on the traffic lights at each intersection, and the communication module is used to transmit the truck information passing through each intersection to the analysis and calibration module .
[0069] Here, the communication module is one of the eLTE-IOT wireless module or the eLTE wireless module, and the eLTE-IOT wireless module or the eLTE wireless module is directly integrated in the traffic light.
[0070] In order to achieve more accurate and wider data collection, the video surveillance sensor device is set above the traffic light. With the help of existing equipment, it can be set on the top support column of the traffic light at each intersection. The installation of video surveillance sensing equipment is equipped with a support for the video surveillance sensing equipment on the supporting column, and the video surveillance sensing equipment is fixed on the support. In the present invention, the video surveillance sensing equipment is an infrared night vision camera and a sound sensor, and both the sound sensor and the infrared night vision camera are connected to a communication module. Here, the communication module is a 4G communication module, a sound sensor and an infrared night vision camera. Data information is transmitted through the 4G communication module. The sound sensor is used to sense the horn and passing sound of the vehicle. The sound sensor transmits the data acquired in real time to the analysis and calibration module 200 via the communication module. The analysis and calibration module 200 acquires this sound sensor The data is also processed.
[0071] The analysis and calibration module 200 includes an analysis unit 210 and a calibration unit 220:
[0072] The analysis unit 210 is used to extract key frames from all the obtained traffic videos of trucks passing by, remove non-key frames, obtain information about the trucks passing by, record the license plate information of the truck, the color and model of the truck, and the lane where the truck is located. Traffic information;
[0073] The calibration unit 220 is used to query all the intersections that the truck passes through according to the license plate information of the truck, obtain the OD information of the passing truck, and perform calibration.
[0074] The modeling prediction module 300 includes a modeling unit 310, a prediction unit 320, and a dispatching unit 330. The freight management model includes at least a vehicle management model and a road section management model:
[0075] The modeling unit 310 is configured to select the vector of the OD information of the passing truck and the traffic information of the lane where the truck is located, and use a non-parametric regression prediction algorithm to establish a truck management model and a road section management model respectively;
[0076] The prediction unit 320 is used to input real-time truck related information into the truck management model for analysis and prediction of the next position, to obtain the truck and each road section condition and predict the truck and each road section condition;
[0077] The dispatching unit 330 is used to manage the trucks according to the conditions of the trucks and various road sections and predict the conditions of the trucks and various road sections, and to perform integrated analysis of the status of each road section to obtain complete overall regional traffic information, according to the overall area The traffic information is combined with the truck status to dispatch trucks.
[0078] In order to better implement the freight planning, the entire system also includes a prompt unit, a display unit, and a planning control unit. The prompt unit is used to remind the truck through voice to prompt it to choose a more reasonable route, and the display unit is used to display the system Here, the prompt unit can be a player, and the display unit can be a display screen. The prompt unit and the display unit are respectively connected to the planning control unit, and the planning control unit is connected to the prediction unit 320 and the scheduling unit 330. The planning control unit receives the prediction information from the prediction unit 320, controls the prompt unit to give reminders, and implements the scheduling obtained from the scheduling unit 330 Information and control the display unit to display.

PUM

no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.

Similar technology patents

Application management and control method for SSR centralized management platform

InactiveCN108830075ARich sources of dataaccurate data
Owner:ZHENGZHOU YUNHAI INFORMATION TECH CO LTD

Signal lamp with positioning function

PendingCN111899544Aavoid road congestionReduce resource input
Owner:安徽信联建设工程有限公司

Method for automatically monitoring and managing motorized vehicles

InactiveCN103679308Aavoid road congestionComprehensive and accurate computer processing
Owner:HEILONGJIANG KANGGEER TECH

Construction method of reinforced concrete hyperbolic slope roof formwork system

InactiveCN109184194AEasy elevation controlaccurate data
Owner:北京城建七建设工程有限公司

Road repairing device for civil engineering

ActiveCN111364341Aavoid road congestionGood carrying capacity
Owner:邵阳公路桥梁建设有限责任公司

System and method for transmitting aviation flight data and navigation information

ActiveCN102946421Aaccurate dataGuarantee the safety of navigation
Owner:深圳市瑞达飞行科技有限公司

Classification and recommendation of technical efficacy words

  • accurate data
  • avoid road congestion

Road repairing device for civil engineering

ActiveCN111364341Aavoid road congestionGood carrying capacity
Owner:邵阳公路桥梁建设有限责任公司

Method for automatically monitoring and managing motorized vehicles

InactiveCN103679308Aavoid road congestionComprehensive and accurate computer processing
Owner:HEILONGJIANG KANGGEER TECH

Signal lamp with positioning function

PendingCN111899544Aavoid road congestionReduce resource input
Owner:安徽信联建设工程有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products