CN114548298B

CN114548298B - Model training method, traffic information processing method, device, equipment and storage medium

Info

Publication number: CN114548298B
Application number: CN202210178922.6A
Authority: CN
Inventors: 李福樑
Original assignee: Apollo Zhilian Beijing Technology Co Ltd; Apollo Zhixing Technology Guangzhou Co Ltd
Current assignee: Apollo Zhilian Beijing Technology Co Ltd; Apollo Zhixing Technology Guangzhou Co Ltd
Priority date: 2022-02-25
Filing date: 2022-02-25
Publication date: 2023-06-23
Anticipated expiration: 2042-02-25
Also published as: CN114548298A

Abstract

The disclosure provides a model training method, a traffic information processing method, a device, equipment and a storage medium, and relates to the technical field of artificial intelligence such as intelligent traffic, deep learning and the like. The specific implementation scheme is as follows: acquiring an acquisition sample acquired at a road intersection; the acquisition samples comprise traffic signal lamp phase information of a road intersection, intersection traffic flow information and intersection road facility information; converting the collected sample into collected sample input data according to a preset information data conversion mode; expanding the acquired sample input data to generate training sample input data; and training a prediction model to be trained according to the training sample input data to obtain a trained prediction model, wherein the prediction model is used for outputting signal lamp phase recommendation information according to the intersection traffic flow information and the intersection road facility information. The prediction model generated by the embodiment of the disclosure can provide efficient and accurate signal lamp phase recommendation information for the road intersection.

Description

Model training method, traffic information processing method, device, equipment and storage medium

Technical Field

The disclosure relates to the field of computer technology, and in particular to the technical field of artificial intelligence such as intelligent transportation and deep learning.

Background

Traffic signal control is an important means for urban road intersection control, and plays a key role in maintaining road traffic order, guaranteeing traffic safety and improving traffic running efficiency. The design of the signal phase scheme of the traffic signal lamp is an important content of the intersection signal control scheme formulation work. The phase setting scheme is a structural framework of the intersection signal control scheme, which not only determines the running order of traffic flow in the intersection, but also plays a decisive role in the traffic efficiency of the intersection.

Disclosure of Invention

The present disclosure provides a model training method, a traffic information processing method, a device, equipment and a storage medium.

According to an aspect of the present disclosure, there is provided a model training method including:

acquiring an acquisition sample acquired at a road intersection; the collected samples comprise traffic signal lamp phase information, intersection traffic flow information and intersection road facility information of the road intersection;

converting the collected sample into collected sample input data according to a preset information data conversion mode;

expanding the acquired sample input data to generate training sample input data;

And training a prediction model to be trained according to the training sample input data to obtain a trained prediction model, wherein the prediction model is used for outputting signal lamp phase recommendation information according to intersection traffic flow information and intersection road facility information.

According to another aspect of the present disclosure, there is provided a traffic information processing method including:

acquiring intersection traffic flow information and intersection road facility information of a road intersection;

inputting the intersection traffic flow information and the intersection road facility information into a prediction model to obtain phase recommendation information, wherein the prediction model is a trained prediction model provided by any one embodiment of the disclosure.

According to another aspect of the present disclosure, there is provided a model training apparatus including:

the sample collection module is used for obtaining a collection sample obtained by collection at a road intersection; the collected samples comprise traffic signal lamp phase information, intersection traffic flow information and intersection road facility information of the road intersection;

the conversion module is used for converting the acquired samples into acquired sample input data according to a preset information data conversion mode;

the training sample input data module is used for expanding the acquired sample input data to generate training sample input data;

The training module is used for training a prediction model to be trained according to the training sample input data to obtain a trained prediction model, and the prediction model is used for outputting signal lamp phase recommendation information according to intersection traffic flow information and intersection road facility information.

According to another aspect of the present disclosure, there is provided a traffic information processing apparatus including:

the information acquisition module is used for acquiring intersection traffic flow information and intersection road facility information of the road intersection;

the phase prediction module is used for inputting the intersection traffic flow information and the intersection road facility information into a prediction model to obtain phase recommendation information, and the prediction model is a trained prediction model provided by any one embodiment of the disclosure.

According to another aspect of the present disclosure, there is provided an electronic device including:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of the embodiments of the present disclosure.

According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method of any of the embodiments of the present disclosure.

According to another aspect of the present disclosure, there is provided a computer program product comprising a computer program/instruction which, when executed by a processor, implements the method in any of the embodiments of the present disclosure.

According to the technology disclosed by the invention, the training of the prediction model can be performed according to the actually collected collection samples of the road intersection, so that the phase recommendation information of the traffic signal lamp can be generated, and a more reasonable phase scheme is provided.

It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification.

Drawings

The drawings are for a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:

FIG. 1 is a schematic diagram of a model training method according to an embodiment of the present disclosure;

FIG. 2 is a schematic diagram of a model training method according to another embodiment of the present disclosure;

FIG. 3 is a schematic diagram of a model training method according to yet another embodiment of the present disclosure;

FIG. 4 is a schematic diagram of a model training method according to yet another embodiment of the present disclosure;

FIG. 5 is a schematic diagram of a model training method according to yet another embodiment of the present disclosure;

FIG. 6 is a schematic diagram of a model training method according to yet another embodiment of the present disclosure;

FIG. 7 is a schematic diagram of a model training method according to yet another embodiment of the present disclosure;

FIG. 8 is a schematic diagram of a model training method according to yet another embodiment of the present disclosure;

FIG. 9 is a schematic diagram of a traffic information processing method according to an embodiment of the present disclosure;

FIG. 10 is a schematic diagram of a model training method according to yet another embodiment of the present disclosure;

FIG. 11 is a schematic diagram of a model training method according to an example of the present disclosure;

FIG. 12 is a schematic illustration of an intersection according to an example of the present disclosure;

FIG. 13 is a schematic illustration of an intersection according to another example of the present disclosure;

FIG. 14 is a schematic diagram of a model training method according to another example of the present disclosure;

FIG. 15 is a schematic view of an intersection road numbering scheme according to an example of the present disclosure;

FIG. 16 is a schematic illustration of an intersection lane numbering scheme according to another example of the present disclosure;

FIG. 17 is a schematic diagram of a model training method according to yet another example of the present disclosure;

FIG. 18 is a schematic diagram of a generative model structure in accordance with an example of the present disclosure;

FIG. 19 is a schematic diagram of a discriminant model structure according to an example of the present disclosure;

FIG. 20 is a schematic diagram of a decision tree generation flow in accordance with another example of the present disclosure;

FIG. 21 is a schematic illustration of a pruning process according to another example of the present disclosure;

FIG. 22 is a schematic diagram of a model training and prediction process according to another example of the present disclosure;

FIG. 23 is a schematic illustration of a process of expanding capacity according to another example of the present disclosure;

FIG. 24 is a schematic diagram of a model training apparatus according to an embodiment of the present disclosure;

FIG. 25 is a schematic diagram of a model training apparatus according to another embodiment of the present disclosure;

FIG. 26 is a schematic diagram of a model training apparatus according to yet another embodiment of the present disclosure;

FIG. 27 is a schematic diagram of a model training apparatus according to yet another embodiment of the present disclosure;

FIG. 28 is a schematic diagram of a model training apparatus according to yet another embodiment of the present disclosure;

FIG. 29 is a schematic diagram of a model training apparatus according to yet another embodiment of the present disclosure;

fig. 30 is a schematic view of a traffic information processing apparatus according to still another embodiment of the present disclosure;

FIG. 31 is a schematic diagram of a pruned decision tree according to an example of the present disclosure;

FIG. 32 is a block diagram of an electronic device used to implement the model training method of an embodiment of the present disclosure.

Detailed Description

Exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

The embodiment of the disclosure first provides a model training method, as shown in fig. 1, including:

step S11: acquiring an acquisition sample acquired at a road intersection; the acquisition samples comprise traffic signal lamp phase information of a road intersection, intersection traffic flow information and intersection road facility information;

step S12: converting the collected sample into collected sample input data according to a preset information data conversion mode;

step S13: expanding the acquired sample input data to generate training sample input data;

step S14: and training a prediction model to be trained according to the training sample input data to obtain a trained prediction model, wherein the prediction model is used for outputting signal lamp phase recommendation information according to the intersection traffic flow information and the intersection road facility information.

In this embodiment, the collected sample collected at the intersection may be a collected sample obtained by performing information collection at the intersection of an actual urban road, or may be a collected sample obtained by performing information collection at any intersection with traffic lights.

In this embodiment, a real road intersection may correspond to a collected sample. Alternatively, a real road intersection may correspond to more than one collected sample.

In this embodiment, the intersection road may include an intersection where two or more roads intersect, such as a cross, a herringbone, a T-intersection (T-intersection), etc. In this embodiment, one sample may be obtained for each straight entry lane for a cross. For example, as shown in fig. 12, the cross includes a north-south entrance and a east-west entrance, and one collection sample may be obtained for each of the north-south entrance and the east-west entrance.

For the T-shaped intersection shown in fig. 13, one collection sample can be obtained for three entrance ways, or the collection samples of the three entrance ways can be integrated into one collection sample.

The intersection facility information may include any facility information related to road traffic provided at the intersection. Such as road canalization information, traffic light facility information, etc. The channeling traffic is characterized in that traffic lines are drawn on roads or traffic is divided by green sheets, so that vehicles with different properties and different speeds can travel along a specified direction without interfering with each other like water flow in channels. The road canalization information can be information such as traffic signs, marks and/or traffic islands arranged at the plane intersections, and can be method information for guiding traffic and pedestrians to travel on the roads.

In this embodiment, the collected sample may be data originally collected, including video, website data, document data, image data, and the like. Or the data obtained by converting the original collected data, such as text information or other types of characters obtained by converting videos, website data, document data, image data and the like. For example, the collected sample table can be generated by summarizing the image, video and other data of the road intersection, and the table comprises traffic signal lamp phase information, intersection traffic flow information, intersection road facility information, corresponding sub-categories and specific contents corresponding to the sub-categories.

In this embodiment, the collected samples are converted into the collected sample input data according to a preset information data conversion mode, which may be the data that can be identified by the prediction model according to a set conversion mode.

The method comprises the steps of expanding acquired sample input data to generate training sample input data, wherein the data quantity of the acquired sample input data can be expanded, the data quantity is increased, and more training sample input data are constructed according to limited acquired sample input data.

Training the prediction model to be trained according to the training sample input data to obtain a trained prediction model, which may include constructing the prediction model to be trained according to the training sample input data, and training the prediction model to be trained by using the training sample input data to obtain the trained prediction model.

In this embodiment, training of the prediction model can be performed according to the actually collected collection samples of the road intersection, so that phase recommendation information of the traffic signal lamp can be generated, and a more reasonable phase scheme is provided.

In one embodiment, as shown in fig. 2, the preset information data conversion modes include a preset numbering mode and a preset calculating mode; according to a preset information data conversion mode, converting the collected sample into collected sample input data, including:

step S21: according to a preset numbering mode, converting the traffic signal lamp phase information and intersection road facility information into numbering data;

step S22: according to a preset calculation mode, calculating traffic flow index data according to intersection traffic flow information;

step S23: and acquiring acquired sample input data according to the number data and the traffic flow index data.

In this embodiment, the converting the traffic signal phase information and the intersection road facility information into the number data according to the preset numbering mode may include converting the traffic signal phase information into the number data according to the preset numbering mode; and converting the intersection road facility signal information into serial data according to a preset serial number mode. The method can further comprise the step of converting the relevant combination of the traffic signal lamp phase information and the intersection road facility information into serial data according to a preset serial number mode.

The numbering data may be integer data, decimal data, special symbol data.

In a specific implementation manner, the step S21 and the step S22 may be performed in any order, or may be performed simultaneously.

In this embodiment, acquiring the acquired sample input data according to the number data and the traffic flow index data may include taking the number data and the traffic flow index data as the acquired sample input data.

In actual operation, each actual intersection can acquire related traffic light phase information, traffic flow information and intersection specific road facility information. Thus, the collected sample input data may include multiple sets, and each set of collected sample input data may correspond to a specific intersection, i.e., a set of traffic flow index data, intersection asset information numbering data, and traffic light phase information numbering data.

In this embodiment, the collected samples are converted into the collected sample input data, so that the collected samples of a non-data type can be converted into the input data of a data type, and training of the prediction model to be trained is achieved.

In one embodiment, according to a preset calculation mode, calculating traffic flow index data according to intersection traffic flow information includes:

determining intersection type information corresponding to intersection traffic flow information;

and calculating traffic index data according to the intersection type information, the intersection road corresponding to the intersection traffic information and the intersection traffic information.

In this embodiment, the intersection type information may include specific types of intersections, for example, in actual road traffic, there may be intersections, t-intersections, and the like in general. For further specific differentiation, the crossroads, t-intersections, etc. may also be divided into further sub-categories depending on the angle.

In other implementations, the T-junctions may be further partitioned, such as into Y-junctions or T-junctions.

In another possible implementation, the cross or t-intersection may also be classified as whether it extends in one of the north and south directions. For example, a cross in the forward southwest northwest direction may be classified into a first category and a cross in the non-forward southwest direction may be classified into a second category. The direction of the T-junction can be divided according to the angle of the road. For example, the three roads of the T-shaped intersection can be divided into a third category by just adjacent T-shaped intersections with an included angle of about 120 degrees, two of the three roads form a straight line, and the other one is perpendicular to the other two, so that the T-shaped intersection is divided into a fourth category.

In this embodiment, the intersection road corresponding to the intersection traffic flow information may include a category for dividing the entrance channels of the respective intersections according to the direction information of the intersection road. For example, for a cross, the north-south inlets may be given a first number and the east-west inlets may be given another number.

In this embodiment, the traffic flow data is converted into the traffic flow index, so that the traffic flow value is expressed in a unified manner, and dimension unification is performed with other types of collected sample information.

In one embodiment, the intersection asset information includes: intersection type information, left-turning lane widening information, right-turning lane channelizing information, left-turning waiting area information, pedestrian crossing facility information, left-turning motor vehicle signal lamp group type, right-turning motor vehicle signal lamp group type information and non-motor vehicle signal lamp group type; as shown in fig. 3, the method for converting the traffic signal phase information and the intersection road facility information into the number data according to the preset numbering mode includes:

step S31: according to a preset numbering mode, the intersection road facility information is converted into a digital number, and the traffic signal lamp phase information is converted into a phase vector;

Step S32: the numbers are combined into facility vectors;

step S33: and splicing the phase vector and the facility vector, and taking the obtained spliced vector as serial number data.

In this embodiment, the preset numbering scheme may be a preset numbering table.

Converting intersection asset information into a digital number according to a preset numbering mode and converting traffic signal lamp phase information into a phase vector can comprise converting intersection asset information into a digital number according to a preset numbering mode and converting traffic signal lamp phase information into a phase vector according to a preset or randomly selected coding mode.

In practical applications, traffic signal phase information may include a variety of information. For various traffic signal phase information, the numbers can be numbered according to numbers ranging from 1 to 100 or any range. If the phase information of each traffic signal lamp is an integer, the phase information of the traffic signal lamp can be converted into binary numbers, octal numbers, hexadecimal numbers and the like through a binary coding mode, an octal coding mode, a hexadecimal coding mode and the like, and the converted numerical values are directly used as phase vectors.

In this embodiment, the numeric numbers are combined into the facility vector, and the numeric numbers may be converted into the vector according to a given encoding scheme. The number may be directly used as one of the multidimensional vector numbers. For example, if the multidimensional vector includes [ x1, x2, x3 … ], the number may be at least one of x1, x2, and x 3.

In this embodiment, the intersection road facility information and the traffic signal lamp phase information are expressed in a vectorization manner, so that the prediction model to be trained can learn the effective information of the actual sample from the vector, and a more accurate recommended information prediction result can be generated after the training process is finished.

In one embodiment, as shown in fig. 4, expanding the collected sample input data to generate training sample input data includes:

step S41: performing dimension expansion processing on the numbered data to obtain a first vector;

step S42: performing linear rectification calculation on the first vector to obtain a second vector;

step S43: according to the second vector, training sample input data is generated.

The dimension expansion processing is performed on the numbered data, and the numbered data in one dimension or two dimensions can be converted into a multidimensional vector.

The first vector is subjected to a linear rectification calculation, which may be a calculation of the first vector using a linear rectification function (Rectified Linear Unit, reLU). The linear rectification function, which may be referred to as a modified linear element, is an activation function (activation function) commonly used in artificial neural networks, and may generally refer to a nonlinear function represented by a ramp function and its variants.

In this embodiment, generating the training sample input data according to the second vector may include using the second vector as the training sample input data or using the second vector as a portion of the training sample input data.

In this embodiment, through dimension expansion and linear rectification calculation, the content of the expanded data can be controlled while the data volume is expanded, so that the expanded data meets the requirement of randomness and maintains a certain consistency with the original real sample data.

In one embodiment, as shown in fig. 5, generating training sample input data from the second vector includes:

step S51: performing linear activation treatment on the second vector to obtain a third vector;

step S52: sampling the content corresponding to the traffic signal lamp phase information in the third vector to obtain a fourth vector;

step S53: training sample input data is determined based on the fourth vector and the collected sample input data.

In this embodiment, performing the linear activation processing on the second vector may include calculating the second vector by using a linear activation function, so as to implement the linear activation processing.

In a specific implementation, the linear activation function employed may include: at least any one of Sigmoid (S-shaped growth curve) function, tanh (hyperbolic tangent) function, reLU function, and the like.

In this embodiment, through linear activation and sampling operations, data meeting the requirements can be selected from new data generated in the expansion process, so as to form data of an expansion sample, and increase of the sample data volume is realized.

In this embodiment, the sampling of the content corresponding to the traffic light phase information in the third vector may be performed by performing a random or other set sampling operation on the generated third vector with more than necessary dimension data, and determining a fourth vector, where the content targeted by the sampling operation is the content corresponding to the traffic light phase information.

In one embodiment, the predictive model is a decision tree model; as shown in fig. 6, the prediction model is a decision tree model; the model training method further comprises the following steps:

step S61: obtaining a sample subset according to the training sample input data, wherein the sample subset comprises at least one of the training sample input data;

step S62: constructing nodes of a decision tree model according to the sample subset;

step S63: and obtaining a prediction model to be trained according to the nodes of the decision tree model.

Deriving the sample subset from the training sample input data may include taking a portion of the training sample input data as the sample subset.

In this embodiment, the sample subset includes at least one of the training sample input data, which may mean that training itself includes at least one group of the training sample input data, where each group corresponds to information about facilities, traffic, phase, and the like of an intersection.

According to the sample subset, constructing nodes of the decision tree model, namely constructing nodes by taking the intersection road types corresponding to the sample subset and the traffic signal lamp phase information corresponding to the intersection traffic flow information as decision references of the nodes of the decision tree model.

For example, a sample subset includes a set of training sample input data, the type of an intersection is A, the traffic flow information of the intersection is B, and the traffic signal lamp phase information is C, so that according to the sample subset, the constructed node can provide signal lamp phase information suggestion decision of C for the intersection with the type of the intersection being A and the traffic flow information of the intersection being B.

In this embodiment, the nodes of the decision tree model are constructed by training the sample subset of the sample input data, so that the decision tree model from the root node to the leaf node can be formed, and an accurate prediction result is provided for the phase prediction of the traffic signal lamp.

In one embodiment, as shown in fig. 7, obtaining a sample subset from training sample input data includes:

Step S71: obtaining a plurality of characteristics according to training sample input data; each feature corresponds to a traffic signal lamp phase information or an intersection road facility information;

step S72: determining information gain values of each of the plurality of features on training sample input data; the information gain value is determined according to conditional experience entropy of the characteristics on the training sample input data;

step S73: and dividing the training sample input data according to the information gain value to obtain a sample subset.

Through the information gain value, the prediction model to be trained can learn the relation between the traffic flow information and the phase information from the actual sample, and meanwhile, the influence of the actual phase information on the traffic flow can be learned from the traffic flow information, so that a more accurate prediction result is provided in the prediction stage.

In this embodiment, a sample subset is determined according to the information gain value, and then a prediction model to be trained is constructed according to the sample subset, so that the prediction model can fully maintain information useful for a prediction result in training sample input data, and the training effect of the prediction model to be trained is improved.

In one embodiment, as shown in fig. 8, dividing the training sample input data according to the information gain value to obtain a sample subset includes:

Step S81: determining target characteristics according to the information gain value;

step S82: determining a dividing point of the training sample input data according to the target characteristics;

step S83: and dividing the training sample input data according to the dividing points to obtain sample subsets.

In this embodiment, the segmentation points of the training sample input data may include segmentation points of the training sample input data. I.e. the division points for dividing the training sample input data corpus into respective subsets.

In one possible implementation, the target features may be multiple. According to the target characteristics, determining the segmentation point of the training sample input data can be to combine multiple target characteristics and determine the segmentation point of the training sample input data.

Specifically, for example, the target features may include an a feature of an intersection road type (such as an intersection in the north-south direction), a b feature of intersection facility information, a c feature of intersection traffic flow information, and a d feature of intersection phase information, and a, b, c, d may be combined into a set of dividing points. The target features are classified into one group and the non-target features are classified into another group. The features corresponding to the number numbers (or indexes) on one side of the target feature may be divided into one group, and the features corresponding to the number numbers (or indexes) on the other side of the target feature may be divided into another group.

In this embodiment, the segmentation of the training sample input data is performed according to the target feature, so as to construct a binary decision tree according to the training sample input data, and improve the prediction effect of the prediction model.

In one embodiment, training a prediction model to be trained to obtain a trained prediction model includes:

calculating a loss value corresponding to a leaf node of the decision tree;

and deleting the leaf nodes under the condition that the loss value accords with the preset pruning condition.

In this embodiment, pruning conditions may correspond to optimization conditions of the decision tree. If the decision given by the leaf node corresponding to the decision tree does not accord with the given phase information operation efficiency of the road intersection, the leaf node can be deleted.

Deleting the leaf node if the loss value meets a preset pruning condition may include deleting content corresponding to the pruning condition in the leaf node if the loss value meets the preset pruning condition. For example, three phase suggestions are given in a leaf node of a decision tree, one of the three phase suggestions meets pruning conditions, and the phase suggestion corresponding to the pruning conditions in the leaf node is deleted.

In the embodiment, the optimization of the decision tree model is realized by pruning and deleting the nodes which are not suitable for prediction in the decision tree model.

In one embodiment, the loss value is calculated according to the number of leaf nodes of the decision tree, the number of training sample input data of the leaf nodes corresponding to the traffic signal phase information, and the number of training sample input data of the leaf nodes.

In this embodiment, the number of leaf nodes of the decision tree, the number of training sample input data of the leaf nodes corresponding to the traffic signal phase information, and the number of training sample input data of the leaf nodes can be reasonably determined, so that the prediction model to be trained can be effectively trained.

The embodiment of the disclosure also provides a traffic information processing method, as shown in fig. 9, including:

step S91: acquiring intersection traffic flow information and intersection road facility information of a road intersection;

step S92: and inputting intersection traffic flow information and intersection road facility information into a prediction model to obtain phase recommendation information, wherein the prediction model is a trained prediction model provided by any one embodiment of the disclosure.

In this embodiment, the step of inputting intersection traffic flow information and intersection road facility information into the prediction model to obtain the phase recommendation information may include searching corresponding nodes in the decision tree according to the intersection traffic flow information and the intersection road facility information, and finally obtaining sub-nodes according to the nodes at each level.

For example, the intersection traffic flow information corresponds to X kinds of index data, each kind of index data corresponds to a first level sub-node, the intersection road facility information corresponds to Y seed information, and each kind of seed information corresponds to a first level sub-node. The method comprises the steps of searching corresponding sub-nodes according to traffic flow information corresponding to a certain road intersection, searching a series of sub-nodes according to the sub-nodes corresponding to road facility information of the intersection, searching a searching path from a root node to a leaf node, which is formed by all the sub-nodes, in a decision tree, finally searching the leaf node, and obtaining phase recommendation information recorded in the leaf node as phase recommendation information corresponding to the road intersection.

In this embodiment, the trained prediction model provided by the embodiment of the present disclosure may be used to predict, and more reasonable phase recommendation information may be provided according to the displayed road traffic facility information and traffic flow information.

In an example of the present disclosure, there is provided a method for recommending urban road intersection traffic signal control phase setting scheme, as shown in fig. 10, including the steps of:

s101: a number of actual intersection samples of the urban road are collected. The sample information comprises four major parts of intersection traffic organization canalization, traffic signal control facilities, traffic flow characteristics and current signal phase setting schemes.

S102: and classifying and marking traffic organization canalization information and traffic signal control facility information of each sample intersection.

S103: and calculating the traffic flow characteristic index of the sample intersection.

S104: the signal phase setting scheme of each sample intersection is classified and numbered. Considering the common 10 phase setting scheme types, the collected intersection samples will be classified according to the phase setting scheme of table 3 below, wherein the right turn arrow indicates the case where the right turn traffic is controlled by the dedicated direction indicator light set.

Table 3 intersection phase setting plan type and numbering

S105: based on the collected intersection sample set, the sample capacity is expanded using a Generated Antagonism Network (GAN) algorithm.

S106: the simulated intersection samples generated based on the real intersection samples and the GAN are trained and a binary decision tree for phase setting scheme recommendation is generated.

S107: pruning is carried out on the generated decision tree T. The number of leaf nodes of the tree T is |T|, T is the leaf nodes of the tree T, and the leaf nodes have Nt intersection patternsThe method comprises the steps that Ntk samples with the phase scheme type Ck are provided, k=1, 2, …, K, alpha is larger than or equal to 0 and is a regularization parameter, and the loss function learned by the decision tree is

S108: and recommending an intersection phase setting scheme by using the pruned decision tree T alpha.

In this example, the intersection phase setting scheme type of table 3 may be determined according to the actual lane line in the road. The dashed line may represent a default scheme.

Further, in steps S102 and S103, the intersection features are split into two parts, i.e., an organization canalization, a static feature of the signal control facility and a dynamic feature of the traffic flow characteristic, wherein the static feature limits the selection space of the signal phase scheme, and the dynamic feature has a constraint effect on the preference of the phase scheme.

Further, in step S105, the sample capacity is enlarged by using the generation countermeasure network algorithm, so that on one hand, enough samples are ensured to train an effective decision tree; on the other hand, the characteristic value of the intersection of the sample set is enriched, and the distribution of samples of different phase scheme types is balanced through sample screening, so that the problem of reduced accuracy of recommending the phase scheme of the decision tree caused by too concentrated distribution of samples of part of phase scheme types is avoided.

In one example of the present disclosure, as shown in fig. 11, collecting a number of actual intersection samples of an urban road includes:

s111: and collecting traffic organization channelized information of the intersection, wherein the traffic organization channelized information comprises an intersection type, a left-turning lane type, a right-turning organization type, a left-turning waiting zone setting condition and a pedestrian crossing organization condition.

S112: and collecting intersection traffic signal control facility information, wherein the intersection traffic signal control facility information comprises a left turning motor vehicle signal lamp group type, a right turning motor vehicle signal lamp group type and a non-motor vehicle signal lamp group type.

S113: and collecting intersection traffic flow characteristic information, wherein the intersection traffic flow characteristic information comprises lane flow and average saturated headway data in a certain period of each traffic flow.

S114: the signal phase setting scheme of the acquisition intersection comprises a signal control phase and phase sequence scheme in a corresponding time period.

In another embodiment of the present disclosure, classifying and marking traffic organization canalization information and traffic signal control facility information of each sample intersection includes:

considering two types of common crossroads and T-shaped crossroads, the crossroads are split into two independent samples according to the orthogonal angle, as shown in fig. 12, and the two independent samples comprise sample 1: corresponds to east-west inlet lane and sample 2: corresponding to the north-south entryway. The T-shaped intersection is regarded as a single integral sample, and 3 inlet channels are respectively numbered, as shown in figure 13, and comprise an inlet channel 1, an inlet channel 2 and an inlet channel 3. The sample intersection traffic organization channeling information and traffic signal control facility information were labeled 0-1 classification according to table 1 below.

TABLE 1 traffic organization canalization and traffic signal control facility information classification tags for intersections

In one example, calculating the sample intersection traffic flow characteristic index, as shown in fig. 14, includes:

s141: and numbering key traffic flows of the crossroad and the T-shaped intersection corresponding to the acquired samples respectively. The numbering modes can be respectively shown in fig. 15 and fig. 16, the number of the crossroad can be m1-m4, and the corresponding number of each entrance (Leg) of the T-shaped intersection can be m1-m3.

S142: and calculating the traffic flow ratio of each strand of key traffic flow of each intersection sample according to the data in the acquired samples. Note that the traffic flow qi is the lane flow (pcu/h/lane) of the traffic flow mi, hi is the saturated headway(s) of the traffic flow mi, and the traffic flow ratio yi=qihi/3600 of the traffic flow mi. Where i may correspond to the number of the m subscript in fig. 15, 16.

S143: and calculating the traffic flow index according to a preset formula. Defining the Difference (DR) index of the traffic demand of each key traffic, namely a bidirectional straight traffic DR1, a bidirectional left-turning traffic DR2, a opposite straight left-turning conflict traffic DR3, and same-direction straight left-turning traffic DR4 and DR5, wherein the values of the indexes can be calculated according to the following table 2:

table 2 differential index calculation method for critical traffic flow passing demand

In one example, as shown in fig. 17, expanding sample capacity with a Generated Antagonism Network (GAN) algorithm based on the collected intersection sample set, comprising:

s171: and constructing a characteristic vector of the acquired real intersection sample. Specifically, each intersection sequentially combines the intersection information classification flag and the traffic characteristic index value into 13-dimensional feature vectors xi= [ Xi (1), xi (2), …, xi (13) ] in the order listed in tables 1 and 2, where Xi represents the feature vector of intersection i.

S172: and constructing a phase scheme type vector of the acquired real intersection samples. Each intersection is one-hot coded according to the phase setting scheme type number of table 3, specifically using a 10-dimensional vector yi= [ Yi (1), yi (2), …, yi (10) ] to represent the phase scheme type, such as vector [1,0, …,0] to represent the phase scheme type 1, vector [0,1,0, …,0] to represent the phase scheme type 2, and so on, where Yi represents the phase scheme type vector of intersection i.

S173: and transversely combining the characteristic vector and the phase scheme type vector of each intersection to form an intersection sample 23-dimensional information vector. I.e. zi= [ xi (1), xi (2), …, xi (13), yi (1), yi (2), …, yi (10) ]= [ Zi (1), zi (2), …, zi (23) ], where Zi represents a sample information vector of intersection i. Where i may represent the number of the intersection. For example, 10 ten thousand intersections are collected, and the number can be 1-100000.

S174: building a countermeasure network (GAN) model. The GAN model includes two parts, a generation model and a discriminant model.

The generated model structure is shown in fig. 18, the generated model structure is input as m-dimensional random prior noise, is unfolded into j-dimension through a full-connection layer 1, is subjected to linear rectification function ReLU, is then input into the full-connection layer 2, outputs 23-dimension tensor, is activated through a nonlinear activation function Tanh, is split into a sample feature vector and a phase scheme type vector, is subjected to sampling through a Gumbil-Softmax (Geng Buer normalization) layer on the data representing the phase scheme type vector in the tensor, and is finally subjected to complete tensor (analog sample) representing intersection sample information based on the sample feature vector and the normalized result.

The structure of the discrimination model is shown in fig. 19, the input of the discrimination model is a real intersection sample or a simulation sample tensor generated by a generation model, the real intersection sample or the simulation sample tensor is unfolded into an l-dimensional tensor through the full-connection layer 1, the l-dimensional tensor is input into the full-connection layer 2 through the linear rectification function ReLU, a 1-dimensional numerical value is output, and the phase recommendation and the corresponding probability result PD are output after the processing of the nonlinear activation function Sigmoid. The α, β in fig. 18, 19 represent intermediate vectors or intermediate data.

S175: training a GAN model by using a real intersection sample set until a model cost function converges, and then generating a simulated intersection sample.

S176: and removing invalid analog intersection samples, wherein samples, of which the phase scheme type is opposite to the traffic organization canalization and signal control facilities of the intersections, in the analog intersection samples generated by the GAN are invalid samples. The judgment conditions are shown in the following table 4:

table 4 invalid intersection sample judgment conditions ("×" indicates that the phase scheme type is invalid)

Further, in step S174, a gummel-Softmax sampling layer is added to the GAN generation model, so as to implement one-hot conversion of the generated phase scheme type vector of the simulated intersection, and not to block back propagation of the model parameter gradient, thereby ensuring that the model parameter can be updated iteratively.

In one example, the simulated intersection samples generated based on the real intersection samples and the GAN train and generate a binary decision tree for phase setting scheme recommendation, as shown in fig. 20, comprising:

s201: adding a part of effective simulated intersection samples into a real intersection sample set to form a full sample set U, so that the intersection phase scheme type distribution of the sample set U is as uniform as possible; the samples in U are randomly divided into a training sample set D and a test sample set F according to a certain proportion, and the samples of all phase scheme types in U are guaranteed to be contained in both D and F.

S202: a training sample set D is input. Each intersection sample input vector comprises 13 characteristic condition values and phase scheme type numbers, the intersection information characteristic set contained in the sample input vector is B, |D| represents the sample capacity of D, namely the number of samples, the samples in D cover K types of phase scheme types Ck, k=1, 2, …, K, |Ck| is the number of samples of the phase type Ck, and the empirical entropy of the sample set D

S203: if the sample phase type in D is the same type Ck, the decision tree T is a single node tree, and the type Ck is used as a type mark of the node and returns to T.

S204: if it is

T is a single junction tree, and the sample various phase scheme numbers and the sample duty ratio in D are used as class marks of the junction, and returned to T.

S205: otherwise, calculating the information gain of each characteristic condition pair D, and selecting the characteristic Ag with the maximum information gain. The characteristic conditions are of two types, and the information gains are calculated according to the following methods:

(1) For traffic organization canalization information and traffic signal control facility information shown in table 1, which are 0-1 binary variables, D is divided into 2 subsets D1 and D2 according to the value of a certain characteristic condition a, i Di is the number of samples of Di, and the sample set of phase type Ck in Di is Dik The number of samples, | Dik | is Dik, and the conditional empirical entropy of the characteristic condition A to D is

Wherein, the information gain of the characteristic condition a is g (D, a) =h (D) -H (d|a).

(2) For the traffic flow characteristic index shown in table 2, which is a continuous value variable, in order to construct a binary decision tree, a sample set may be divided into two subsets by taking a certain continuous value characteristic condition a and a certain value s thereof in the sample set D as a segmentation variable and a segmentation point: d1 (A, s) = { D|A.ltoreq.s }, D2 (A, s) = { D|A > s }, then traversing the values of the characteristic condition A in all samples in D, solving argmins [ H (D1) +H (D2) ], and obtaining s as an A optimal segmentation point, wherein H (Di) is the empirical entropy of a sample set Di, and then dividing the two subsets according to the segmentation point s, and calculating the information gain of the characteristic condition A according to the method in (1).

S206: if the information gain of Ag is equal to 0, setting T as a single junction tree, taking the sample various phase scheme numbers and the sample duty ratio in D as class marks of the junction, and returning to T.

S207: otherwise, for each value of Ag (if the characteristic condition of continuous value is based on the value range of the dividing point), ai (i=1, 2), dividing D into two non-empty subsets D1 and D2, constructing two sub-nodes, forming a tree T by the nodes and the sub-nodes, and returning to T.

S208: for the ith (i=1, 2) sub-node, using Di as training set and B- { Ag } as feature set, recursively calling steps S202-S207 to obtain sub-tree Ti, returning Ti.

Further, in step S205, the heuristic search method is used to find the optimal segmentation point for the continuous valued feature, and the feature value is divided into two intervals, so as to implement the construction of the binary decision tree.

In one embodiment, as shown in fig. 21, the pruning process includes:

s211: the generated overall tree T and the parameter α are input. The overall tree may be a binary decision tree generated by the method shown in fig. 20.

S212: recursively retracting upward from the leaf nodes of the tree. And setting the overall trees before and after the set of leaf nodes retract to the father node as TA and TB respectively, wherein the corresponding loss functions are C alpha (TA) and C alpha (TB) respectively, if C alpha (TB) is less than or equal to C alpha (TA), pruning is carried out, namely the father node is changed into a new leaf node, and the classification mark of the leaf node is updated.

S213: and returning to and repeating the step S212 until all leaf nodes do not meet pruning conditions, and obtaining the subtree T alpha with the minimum loss function.

Further, in steps S106 and S108, when the decision tree is generated, each phase scheme type and sample ratio included in the subset where each leaf node mark is located can be regarded as the confidence coefficient of selecting different phase schemes for the intersection samples meeting the characteristic conditions corresponding to the leaf node, so that a plurality of phase alternatives can be recommended for the designer.

After the technical scheme is adopted, the technical scheme provided by the disclosed example has at least the following beneficial effects:

according to the method, static conditions and traffic flow state characteristics of various traffic organization facilities of intersections are comprehensively considered, the recommendation of the optimal signal control phase setting scheme is realized through analysis and matching, and compared with the existing phase scheme optimization model method, the information characteristics of the intersections and the types of the phase schemes are more comprehensive.

The method utilizes the intersection sample set to generate the decision tree, and by programming the input and output of the decision tree, a complex mathematical model is not required to be specially established and solved in the practical application process of the phase scheme optimization design, and the intersection optimal phase setting scheme can be conveniently obtained under the condition that a user does not have abundant signal optimization experience.

According to the method, a plurality of phase alternatives and the confidence coefficient thereof are provided for the object intersection, so that a user can flexibly select and apply the phase alternatives according to specific requirements, and a larger operation space is provided for the works such as intersection signal coordination control scheme design and the like.

In another example of the present disclosure, the model training and prediction steps are performed for data of an actual road intersection as shown in fig. 22 as follows:

S221: 159 groups of actual intersection sample data are acquired from cities such as A, B, and the sample information comprises four major parts of intersection traffic organization canalization, traffic signal control facilities, traffic flow characteristics and current signal phase setting schemes.

S222: and classifying and marking traffic organization canalization information and traffic signal control facility information of each sample intersection.

S223: and calculating the traffic flow characteristic index of the sample intersection.

S224: the signal phase setting scheme of each sample intersection is classified and numbered. Wherein the actual samples collected contained 121 sets of intersections and 38 sets of T-type intersections and covered the 9 phase setting plan types in table 3, the number of intersection samples for each phase plan type is shown in table 5 below.

Table 5 number of intersection samples for different phase scheme types

S225: based on the collected intersection sample set, the sample capacity is expanded using a Generated Antagonism Network (GAN) algorithm.

S226: the simulated intersection samples generated based on the real intersection samples and the GAN are trained and a binary decision tree for phase setting scheme recommendation is generated. In order to make the intersection phase scheme type distribution of the decision tree training sample set more uniform, 9×17=153 samples are randomly selected from the effective intersection samples generated by the GAN model and added into the real intersection sample set to form a full sample set. The number of the samples of the selected analog intersection is close to that of the actual samples, and 17 samples are selected for each type of phase scheme. Then, 48 samples are randomly extracted from the whole sample set to form a test sample set, and the rest samples form a training sample set.

Wherein, as shown in fig. 23, the expanding the sample capacity by using the generated countermeasure network (GAN, generative Adversarial Networks) algorithm based on the collected intersection sample set further includes:

s231: a 13-dimensional feature vector is constructed for each real intersection sample acquired.

S232: a 10-dimensional phase scheme type vector is constructed for each real intersection sample acquired.

S233: and transversely combining the characteristic vector and the phase scheme type vector of each intersection to form an intersection sample 23-dimensional information vector.

S234: and constructing a GAN model. The GAN model can be built using tools such as pyrerch and the like according to fig. 18 and 19.

S235: the sample is expanded. Training a GAN model with the set of real intersection samples until the model cost function converges, and then generating 159 x 100 = 15000 simulated intersection samples with the trained generation model.

S236: and eliminating invalid analog intersection samples. The phase scheme type in 15900 analog intersection samples and the invalid samples of the intersection traffic organization canalization and signal control facilities are contradicted, the effective sample rate is about 95 percent.

S237: pruning is carried out on the generated decision tree. The pruning parameter α is taken as 10, and the pruned decision tree is shown in fig. 31, where the leaf nodes are marked with all recommended scheme type numbers and their confidence levels.

S238: and recommending the phase setting scheme for 48 intersections of the test sample set by using the pruned decision tree. The results are shown in Table 6 below. The recommended phase alternative scheme set of all intersection samples comprises actual phase schemes, and the actual phase schemes account for 100%; the recommended phase alternative scheme with the highest confidence coefficient among the 44 intersection samples is consistent with the current scheme, the ratio is 92%, and the test effect is good.

Referring to fig. 31, the root node of the pruned decision tree is of the intersection type. The first level sub-node comprises a combination of a left-turning motor vehicle signal lamp group type and a numerical number corresponding to an intersection type. The third level of non-leaf nodes includes a combination of numerical numbers corresponding to the type of left turn motor vehicle signal light group to straight left conflicting vehicle flows (i.e., DR 3). The fourth level of non-leaf nodes includes a combination of index ranges corresponding to non-motor vehicle signal light group types and object straight left conflicting traffic streams. The fourth level non-leaf node also includes a left turn lane type, a combination of index ranges corresponding to the subject straight left conflicting traffic. The fifth-stage non-leaf node comprises a left-turn waiting zone setting condition and a combination of numerical numbers corresponding to the types of the non-motor vehicle signal lamp groups. The fifth level non-leaf node also includes a co-directional straight left turn traffic stream (i.e., DR 5), a combination of numerical numbers corresponding to non-motor vehicle signal light group types. The sixth level of non-leaf nodes includes a combination of right turn tissue type and left turn pending zone set up. The seventh level of non-leaf nodes includes a combination of numeric numbers corresponding to right turn organization types for straight left turn traffic in the same direction. In the leaf nodes of fig. 31, the numbers in "[ ]", integers represent phases, and percentages represent recommended degrees.

TABLE 6 recommended results for test sample set intersection phase set scheme

/>

In the above table, the numbers in "[ [ ] ]", integers represent phases, and percentages represent recommended degrees.

Under general circumstances, an optimization method for an intersection phase setting scheme mostly builds a mathematical model based on an objective of improving traffic running benefit of an intersection, and solves the model by using a traditional optimization algorithm or an intelligent search algorithm, so that the phase scheme is optimized, only traffic requirements or few specific intersection road conditions are considered, and a relatively complex mathematical model needs to be built and solved in actual application, so that the optimization method has a large limitation. According to the model training method and the traffic information processing method provided by the embodiment of the disclosure, static conditions and traffic flow state characteristics of various traffic organization facilities of the intersections are comprehensively considered, and the rapid recommendation of the optimal phase setting scheme of the intersections is realized by constructing the decision tree, so that the efficiency of the design work of the signal control scheme of the intersections is scientifically and reasonably improved.

The embodiment of the disclosure also provides a model training device, as shown in fig. 24, including:

the sample collection module 241 is configured to obtain a collection sample collected at a road intersection; the acquisition samples comprise traffic signal lamp phase information of a road intersection, intersection traffic flow information and intersection road facility information;

The conversion module 242 is configured to convert the collected sample into collected sample input data according to a preset information data conversion manner;

the training sample input data module 243 is configured to expand the collected sample input data to generate training sample input data;

the training module 244 is configured to train a prediction model to be trained according to the training sample input data, and obtain a trained prediction model, where the prediction model is configured to output signal lamp phase recommendation information according to intersection traffic flow information and intersection road facility information.

In one embodiment, as shown in fig. 25, the preset information data conversion modes include a preset numbering mode and a preset calculating mode; the conversion module includes:

a numbering unit 251, configured to convert the traffic signal lamp phase information and the intersection road facility information into numbered data according to a preset numbering manner;

an index unit 252, configured to calculate traffic index data according to the intersection traffic information according to a preset calculation mode;

the data processing unit 253 is configured to obtain collected sample input data according to the number data and the traffic flow index data.

In one embodiment, the index unit is further configured to:

In one embodiment, the intersection asset information includes: intersection type information, left-turning lane widening information, right-turning lane channelizing information, left-turning waiting area information, pedestrian crossing facility information, left-turning motor vehicle signal lamp group type, right-turning motor vehicle signal lamp group type information and non-motor vehicle signal lamp group type; the numbering unit is also for:

according to a preset numbering mode, the intersection road facility information is converted into a digital number, and the traffic signal lamp phase information is converted into a phase vector;

the numbers are combined into facility vectors;

and splicing the phase vector and the facility vector, and taking the obtained spliced vector as serial number data.

In one embodiment, as shown in fig. 26, the training sample input data module includes:

a first vector unit 261, configured to perform dimension expansion processing on the number input to obtain a first vector;

a second vector unit 262 for performing linear rectification calculation on the first vector to obtain a second vector;

a second vector processing unit 263 for generating training sample input data according to the second vector.

In one embodiment, the second vector processing unit is further configured to:

performing linear activation treatment on the second vector to obtain a third vector;

sampling the content corresponding to the traffic signal lamp phase information in the third vector to obtain a fourth vector;

training sample input data is determined based on the fourth vector and the collected sample input data.

In one embodiment, the predictive model is a decision tree model; as shown in fig. 27, the model training apparatus further includes:

a sample subset module 271 for obtaining a sample subset from the training sample input data, the sample subset comprising at least one of the training sample input data;

a node module 272 for constructing nodes of the decision tree model from the sample subset;

model module 273 is configured to obtain a prediction model to be trained according to the nodes of the decision tree model.

In one embodiment, as shown in fig. 28, the sample subset module includes:

a feature unit 281 for obtaining a plurality of features according to training sample input data; each feature corresponds to a traffic signal lamp phase information or an intersection road facility information;

an empirical entropy unit 283 for determining an information gain value of each of the plurality of features for the training sample input data; the information gain value is determined according to conditional experience entropy of the characteristics on the training sample input data;

The dividing unit 283 is configured to divide the training sample input data according to the information gain value to obtain a sample subset.

In one embodiment, the segmentation unit is further configured to:

determining target characteristics according to the information gain value;

determining a dividing point of the training sample input data according to the target characteristics;

and dividing the training sample input data according to the dividing points to obtain sample subsets.

In one embodiment, as shown in fig. 29, the training module includes:

a loss value unit 291, configured to calculate a loss value corresponding to a leaf node of the decision tree according to the training sample input data;

pruning unit 292 configured to delete the leaf node if the loss value meets a preset pruning condition.

The embodiment of the present disclosure further provides a traffic information processing apparatus, as shown in fig. 30, including:

an information acquisition module 301, configured to acquire intersection traffic information and intersection road facility information of a road intersection;

The phase prediction module 302 is configured to input intersection traffic flow information and intersection road facility information into a prediction model, so as to obtain phase recommendation information, where the prediction model is a trained prediction model provided by any one embodiment of the disclosure.

In the technical scheme of the disclosure, the acquisition, storage, application and the like of the related user personal information all conform to the regulations of related laws and regulations, and the public sequence is not violated.

According to embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

FIG. 32 illustrates a schematic block diagram of an example electronic device 320 that may be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the disclosure described and/or claimed herein.

As shown in fig. 32, the device 320 includes a computing unit 321 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 322 or a computer program loaded from a storage unit 328 into a Random Access Memory (RAM) 323. In the RAM 323, various programs and data required for the operation of the device 320 can also be stored. The computing unit 321, ROM 322, and RAM 323 are connected to each other by a bus 324. An input/output (I/O) interface 325 is also connected to bus 324.

Various components in device 320 are connected to I/O interface 325, including: an input unit 326 such as a keyboard, mouse, etc.; an output unit 327 such as various types of displays, speakers, and the like; a storage unit 328, such as a magnetic disk, optical disk, etc.; and a communication unit 329, such as a network card, modem, wireless communication transceiver, etc. The communication unit 329 allows the device 320 to exchange information/data with other devices through a computer network such as the internet and/or various telecommunication networks.

The computing unit 321 may be a variety of general purpose and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 321 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 321 performs the respective methods and processes described above, such as a model training method. For example, in some embodiments, the model training method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 328. In some embodiments, some or all of the computer program may be loaded and/or installed onto device 320 via ROM 322 and/or communication unit 329. When the computer program is loaded into RAM 323 and executed by computing unit 321, one or more steps of the model training method described above may be performed. Alternatively, in other embodiments, the computing unit 321 may be configured to perform the model training method by any other suitable means (e.g., by means of firmware).

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.

Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet.

The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server of a distributed system, or a server incorporating a blockchain.

It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel or sequentially or in a different order, provided that the desired results of the technical solutions of the present disclosure are achieved, and are not limited herein.

The above detailed description should not be taken as limiting the scope of the present disclosure. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure are intended to be included within the scope of the present disclosure.

Claims

1. A model training method, comprising:

Training a prediction model to be trained according to the training sample input data to obtain a trained prediction model, wherein the prediction model is used for outputting signal lamp phase recommendation information according to intersection traffic flow information and intersection road facility information;

the preset information data conversion mode comprises a preset numbering mode and a preset calculation mode; the converting the collected sample into collected sample input data according to a preset information data conversion mode includes:

according to the preset numbering mode, converting the traffic signal lamp phase information and the intersection road facility information into numbering data;

according to the preset calculation mode, calculating traffic flow index data according to the intersection traffic flow information;

acquiring the acquired sample input data according to the number data and the traffic flow index data;

the step of converting the traffic signal lamp phase information and the intersection road facility information into serial number data according to the preset serial number mode comprises the following steps:

according to the preset numbering mode, the intersection road facility information is converted into a digital number, and the traffic signal lamp phase information is converted into a phase vector;

Forming the number into a facility vector;

and splicing the phase vector and the facility vector, and taking the obtained spliced vector as the serial number data.

2. The method according to claim 1, wherein the calculating traffic index data according to the intersection traffic information according to the preset calculation mode includes:

determining intersection type information corresponding to the intersection traffic information;

3. The method of claim 1, wherein the intersection asset information comprises: intersection type information, left-turning lane widening information, right-turning lane channelizing information, left-turning waiting area information, pedestrian crossing facility information, left-turning motor vehicle signal lamp group type, right-turning motor vehicle signal lamp group type information and non-motor vehicle signal lamp group type.

4. The method of claim 1, wherein the augmenting the collected sample input data to generate training sample input data comprises:

performing dimension expansion processing on the numbered data to obtain a first vector;

Performing linear rectification calculation on the first vector to obtain a second vector;

and generating training sample input data according to the second vector.

5. The method of claim 4, wherein the generating training sample input data from the second vector comprises:

performing linear activation processing on the second vector to obtain a third vector;

and determining the training sample input data according to the fourth vector and the acquired sample input data.

6. The method of any of claims 1-5, wherein the predictive model is a decision tree model; the method further comprises the steps of:

obtaining a sample subset according to the training sample input data, wherein the sample subset comprises at least one of the training sample input data;

constructing nodes of a decision tree model according to the sample subset;

and obtaining the prediction model to be trained according to the nodes of the decision tree model.

7. The method of claim 6, wherein the deriving a subset of samples from the training sample input data comprises:

Obtaining a plurality of characteristics according to the training sample input data; each feature corresponds to a traffic signal lamp phase information or an intersection road facility information;

determining an information gain value of each of the plurality of features for the training sample input data; the information gain value is determined according to conditional experience entropy of the characteristic on the training sample input data;

and dividing the training sample input data according to the information gain value to obtain the sample subset.

8. The method of claim 7, wherein the dividing the training sample input data according to the information gain value to obtain the sample subset comprises:

determining a target feature according to the information gain value;

and dividing the training sample input data according to the segmentation points to obtain the sample subset.

9. The method according to claim 6, wherein the training the prediction model to be trained according to the training sample input data, to obtain a trained prediction model, includes:

Calculating a loss value corresponding to a leaf node of the decision tree according to the training sample input data;

and deleting the leaf node under the condition that the loss value accords with a preset pruning condition.

10. The method of claim 9, wherein the loss value is calculated according to the number of leaf nodes of the decision tree, the number of training sample input data of the leaf nodes corresponding to each traffic signal phase information, and the number of training sample input data of the leaf nodes corresponding.

11. A traffic information processing method, comprising:

inputting the intersection traffic flow information and intersection road facility information into a prediction model to obtain phase recommendation information, wherein the prediction model is a trained prediction model according to any one of claims 1-10.

12. A model training apparatus comprising:

the training module is used for training a prediction model to be trained according to the training sample input data to obtain a trained prediction model, and the prediction model is used for outputting signal lamp phase recommendation information according to intersection traffic flow information and intersection road facility information;

the preset information data conversion mode comprises a preset numbering mode and a preset calculation mode; the conversion module includes:

the numbering unit is used for converting the traffic signal lamp phase information and the intersection road facility information into numbered data according to the preset numbering mode;

the index unit is used for calculating traffic flow index data according to the intersection traffic flow information according to the preset calculation mode;

the data processing unit is used for obtaining the acquired sample input data according to the number data and the traffic flow index data;

wherein the numbering unit is further configured to:

forming the number into a facility vector;

13. The apparatus of claim 12, wherein the index unit is further configured to:

14. The apparatus of claim 12, wherein the intersection asset information comprises: intersection type information, left-turning lane widening information, right-turning lane channelizing information, left-turning waiting area information, pedestrian crossing facility information, left-turning motor vehicle signal lamp group type, right-turning motor vehicle signal lamp group type information and non-motor vehicle signal lamp group type.

15. The apparatus of claim 12, wherein the training sample input data module comprises:

The first vector unit is used for performing dimension expansion processing on the numbered data to obtain a first vector;

the second vector unit is used for carrying out linear rectification calculation on the first vector to obtain a second vector;

and the second vector processing unit is used for generating training sample input data according to the second vector.

16. The apparatus of claim 15, wherein the second vector processing unit is further to:

17. The apparatus of any of claims 12-16, wherein the predictive model is a decision tree model; the apparatus further comprises:

the sample subset module is used for obtaining a sample subset according to the training sample input data, and the sample subset comprises at least one of the training sample input data;

the node module is used for constructing nodes of the decision tree model according to the sample subset;

And the model module is used for obtaining the prediction model to be trained according to the nodes of the decision tree model.

18. The apparatus of claim 17, wherein the sample subset module comprises:

the characteristic unit is used for obtaining a plurality of characteristics according to the training sample input data; each feature corresponds to a traffic signal lamp phase information or an intersection road facility information;

an empirical entropy unit for determining an information gain value of each of the plurality of features for the training sample input data; the information gain value is determined according to conditional experience entropy of the characteristic on the training sample input data;

and the segmentation unit is used for segmenting the training sample input data according to the information gain value to obtain the sample subset.

19. The apparatus of claim 18, wherein the segmentation unit is further configured to:

determining a target feature according to the information gain value;

20. The apparatus of claim 17, wherein the training module comprises:

The loss value unit is used for calculating a loss value corresponding to a leaf node of the decision tree according to the training sample input data;

and the pruning unit is used for deleting the leaf nodes under the condition that the loss value accords with a preset pruning condition.

21. The apparatus of claim 20, wherein the loss value is calculated from a number of leaf nodes of the decision tree, a number of training sample input data of the leaf nodes corresponding to respective traffic signal phase information, and a number of training sample input data of the leaf nodes corresponding.

22. A traffic information processing apparatus comprising:

the phase prediction module is used for inputting the intersection traffic flow information and the intersection road facility information into a prediction model to obtain phase recommendation information, wherein the prediction model is a trained prediction model according to any one of claims 12-21.

23. An electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-11.

24. A non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the method of any one of claims 1-11.

25. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the steps of the method of any of claims 1-11.