CN117975732B

CN117975732B - Intelligent traffic control system and method for tunnel

Info

Publication number: CN117975732B
Application number: CN202410362066.9A
Authority: CN
Inventors: 陈大伟; 杨晋文; 周宗华; 王武现; 邵长青; 刘兵; 左学峰; 余朝阳; 宋青波; 张宏伟
Original assignee: China Railway 16th Bureau Group Co Ltd
Current assignee: China Railway 16th Bureau Group Co Ltd
Priority date: 2024-03-28
Filing date: 2024-03-28
Publication date: 2024-05-28
Anticipated expiration: 2044-03-28
Also published as: CN117975732A

Abstract

The application relates to the technical field of vehicle traffic control, and provides a tunnel intelligent traffic control system and method, comprising the following steps: acquiring a tunnel driving data sequence and a detection vector; determining a multi-mode characteristic state diagnosis coefficient based on the change amount of each element in the tunnel driving data sequence between adjacent peak points on a time sequence curve and the change regularity of detection vectors at each acquisition time and the historical acquisition time; determining a data stagnation coefficient based on the discrete degree of elements in the tag stabilization sequence of the state transition probability matrix of each mode data under different time scales; and (3) adaptively determining fading gain based on the data stagnation coefficient and the dragging influence coefficient at two ends, obtaining a state detection fusion vector by adopting a KF algorithm, obtaining a vehicle motion state detection result based on the state detection fusion vector, and performing traffic control on the vehicle based on the motion state detection result. According to the application, the accuracy of vehicle motion state detection is improved by optimizing the KF algorithm, so that the traffic control efficiency is improved.

Description

Intelligent traffic control system and method for tunnel

Technical Field

The application relates to the technical field of vehicle traffic control, in particular to a tunnel intelligent traffic control system and method.

Background

At present, the main working procedures of tunnel construction are blasting excavation, slag discharge, primary support, inverted arch, secondary lining and the like, so that the space in a tunnel is usually narrow due to the influence of tunnel construction progress factors, and the safety risk caused by traffic jam and dislocation of vehicles and personnel exists, so that the method is not beneficial to large-scale vehicle dispatching and transportation.

In addition, the current state detection and traffic control of vehicles in a tunnel in a construction period mainly comprises the steps of vehicle positioning, data acquisition, vehicle scheduling, information display, alarm processing and the like, when the vehicles in the tunnel are subjected to traffic control, the real-time collection of the traffic condition information and related data analysis of the vehicles in the tunnel is difficult to achieve due to the complexity of the environment and the diversification of the information, the remote intelligent scheduling of the traveling vehicles in the tunnel cannot be realized, the emergency is encountered, the traffic order in the tunnel is disturbed, even the traffic is paralyzed, the construction progress is slowed down, the construction of each working face is mutually interfered, and the construction period is delayed.

Disclosure of Invention

The application provides a tunnel intelligent traffic control system and a tunnel intelligent traffic control method, which aim to solve the problem of low traffic control efficiency caused by large real-time detection error of the vehicle motion state in a tunnel, and the adopted technical scheme is as follows:

In a first aspect, an embodiment of the present application provides a tunnel intelligent traffic control method, including the steps of:

(1) The driving data of the vehicles at each collection time is utilized to form a tunnel driving data sequence of each vehicle at each collection time; obtaining a detection vector of a driving state image at each acquisition time based on the driving state image by using a driving state detection model;

(2) Determining a multi-mode characteristic state diagnosis coefficient of each vehicle at each acquisition time based on the consistency of the detection vector and the motion state reflected by the tunnel driving data sequence, wherein the multi-mode characteristic state diagnosis coefficient determining method comprises the following steps:

(2.1) determining abnormal contribution degree of movement behaviors of each element based on the distribution characteristics of extremum variation on a time sequence curve of each element in a tunnel running data sequence of each vehicle at each acquisition time;

(2.2) determining a multi-mode characteristic state diagnosis coefficient of each vehicle at each acquisition time based on the abnormal contribution degree of the motion behavior of each element in the tunnel running data sequence of each vehicle at each acquisition time and the similarity of the detection vector at each acquisition time and the historical time;

(3) Determining an fading gain of each vehicle at each acquisition time based on the influence of the vehicle motion state of the historical time on the vehicle motion state reflected by the two modal data at each time, wherein the determining method of the fading gain comprises the following steps:

(3.1) determining a data stagnation coefficient of each mode data of each vehicle at each acquisition time based on the discrete degree of elements in the tag stabilization sequence of the state transition probability matrix of each mode data under different time scales;

(3.2) determining a two-end drag influence coefficient of each vehicle at each acquisition time based on the abnormality score of each vehicle multimode characteristic state diagnosis coefficient;

(3.3) adaptively determining the fading gain of each vehicle at each acquisition time in the KF algorithm based on the drag influence coefficient at two ends of each vehicle at each acquisition time and the data stagnation coefficients of the two modal data;

(4) Adopting a KF algorithm to determine a state detection fusion vector of each vehicle at each acquisition time based on the detection vector, the running reset vector and the fading gain of each vehicle at each time and the historical time; determining a detection result of the motion state of each vehicle at each moment in the tunnel based on the state detection fusion vector by using an SVM model;

(5) And completing traffic control on vehicles in the tunnel based on the detection result of the motion state of each vehicle at each moment in each work area in the tunnel.

Preferably, the method for determining the abnormal contribution degree of the movement behavior of each element based on the distribution characteristics of the extremum variation on the time sequence curve of each element in the tunnel driving data sequence of each vehicle at each acquisition time comprises the following steps:

Taking values of the same element in all tunnel driving data sequences and corresponding acquisition moments as inputs, taking a fitting curve obtained by a curve fitting algorithm as a time sequence curve of each element, and taking the ratio of the absolute value of the difference value between the ordinate coordinates of any two adjacent peak points to the absolute value of the difference value between the abscissa coordinates of any two adjacent peak points on the time sequence curve as the extremum variation between any two adjacent peak points;

taking the ratio of the maximum value to the minimum value in all extreme value variation amounts from the starting moment to the collecting moment on a time sequence curve of each element in the tunnel driving data sequence of each vehicle as a first variation amount;

calculating the sum of absolute values of differences between the ordinate of two peak points of each acquisition time and the time interval of each acquisition time on the time sequence curve of each element, and taking a calculation result taking a natural constant as a base and the inverse number of the sum of the absolute values as an index as a molecule;

Taking the sum of a Hurst index and 0.01 of a sequence formed by all extreme value variation quantities from the starting moment to the collecting moment on a time sequence curve of each element in a tunnel driving data sequence of each vehicle as a denominator;

And taking the product of the ratio of the numerator and the denominator and the first variation as the abnormal contribution degree of the movement behaviors of each element in the tunnel driving data sequence of each vehicle at each acquisition time.

Preferably, the method for determining the multimode characteristic state diagnosis coefficient of each vehicle at each collection time based on the abnormal contribution degree of the motion behavior of each element in the tunnel running data sequence of each vehicle at each collection time and the similarity of the detection vector at each collection time and the historical time of the detection vector at each collection time comprises the following steps:

taking the accumulated sum of the abnormal contribution degrees of the movement behaviors of all elements in the tunnel running data sequence of each vehicle at each acquisition time as a first product factor;

Taking the accumulated results of pearson correlation coefficients among detection vectors corresponding to each vehicle at each collection time and at one historical time of each collection time at all historical times of each collection time as denominators;

Taking the ratio of the first element to the denominator in the corresponding detection vector of each vehicle at each acquisition time as a second product factor;

the multi-mode characteristic state diagnosis coefficient of each vehicle at each acquisition time consists of a first product factor and a second product factor, wherein the multi-mode characteristic state diagnosis coefficient is in direct proportion to the first product factor and the second product factor respectively.

Preferably, the method for determining the data stagnation coefficient of each mode data of each vehicle at each acquisition time based on the discrete degree of the element in the tag stability sequence of the state transition probability matrix of each mode data at different time scales comprises the following steps:

Taking the value of the same label in the detection vector of the driving state image of each vehicle at all the acquisition moments as input, and taking a fitting curve obtained by using a nonlinear least square method as a time sequence curve of each label;

S1: taking a time sequence curve of each label as input, adopting a differential algorithm to obtain peak points and trough points on the time sequence curve, taking a time interval between any two adjacent peak points corresponding to acquisition moments as a time window, taking the width of each time window with the same size as a time scale, and taking a sequence obtained by arranging all time scales according to descending order as a time scale sequence;

S2: taking the difference value between the number of acquisition moments of each tag and 1 in each time scale as a denominator, taking the ratio of the conversion times between two values of each tag in each time scale and the denominator as the probability of conversion between the two values of each tag in each time scale, and taking a matrix formed by the probabilities of conversion between all values of each tag in each time scale as a state transition matrix of each tag in each time scale;

S3: taking a state transition matrix of each label on each time scale as input, and taking a sequence consisting of all singular values obtained by adopting an SVD algorithm as a label stabilization sequence of the state transition matrix;

S4: taking the accumulated result of the difference value between the variation coefficients of the label stabilizing sequence on the previous element and the label stabilizing sequence on the next element in the time scale sequence of each label in the detection vector corresponding to each vehicle at each acquisition moment as the data stagnation coefficient of the driving state image of each vehicle at each acquisition moment;

S5: and replacing the time sequence curve of each element in the tunnel driving data sequence with the time sequence curve of each label, and repeating S1-S4 to obtain the data stagnation coefficient of the tunnel driving data sequence of each vehicle at each acquisition moment.

Preferably, the method for determining the drag influence coefficient at two ends of each vehicle at each acquisition time based on the abnormality score of the multimode characteristic state diagnosis coefficient of each vehicle comprises the following steps:

Taking the multi-mode characteristic state diagnosis coefficient of each vehicle at all acquisition moments as input, acquiring an abnormal score of each multi-mode characteristic state diagnosis coefficient by adopting an RRCF algorithm, and counting extreme points in all abnormal scores, wherein the acquisition moment corresponding to the nearest extreme point on the left side and the right side of the abnormal score of each multi-mode characteristic state diagnosis coefficient is respectively taken as the left end moment and the right end moment of the acquisition moment corresponding to each multi-mode characteristic state diagnosis coefficient;

Calculating the accumulation results of the multimode characteristic state diagnosis coefficients of the preset number of acquisition moments taken by each vehicle at the left end moment and the right end moment of each acquisition moment, and taking the product of the calculation result taking the accumulation results as true numbers and the multimode characteristic state diagnosis coefficients of each vehicle at each acquisition moment as the two-end dragging influence coefficients of each vehicle at each acquisition moment.

Preferably, the method for adaptively determining the fading gain of each vehicle at each acquisition time in the KF algorithm based on the drag influence coefficient at both ends of each vehicle at each acquisition time and the data stagnation coefficients of the two mode data comprises the following steps:

Taking the product of the sum of the tunnel running data sequence of each vehicle at each acquisition time and the data stagnation coefficient of the driving state image and the dragging influence coefficient at the two ends of each vehicle at each acquisition time as an fading detection correction coefficient of each vehicle at each acquisition time;

Taking the fading detection correction coefficient as a base number, taking the product of the calculation result of taking the time interval between each acquisition time and the initial time of each vehicle starting to move in the tunnel as an index and the Kalman gain of each acquisition time in the KF algorithm as the fading gain of each vehicle at each acquisition time.

Preferably, the method for determining the state detection fusion vector of each vehicle at each acquisition time by adopting the KF algorithm based on the detection vector, the running reset vector and the fading gain of each vehicle at each time and the historical time thereof comprises the following steps:

Determining decision weights of each element in the tunnel running data sequence in each vehicle at each acquisition time based on the tunnel running data sequence of each vehicle at each acquisition time;

Taking the product of the decision weight of each element in each tunnel driving data sequence and each element as a reset value of each element, and taking a vector formed by the reset values of all elements in each tunnel driving data sequence as a driving reset vector corresponding to each tunnel driving data sequence;

And taking the detection vector and the running reset vector of each vehicle at all the acquisition moments from the first acquisition moment to each acquisition moment as input, replacing the Kalman gain of the corresponding time step in the traditional KF algorithm with the fading gain of each acquisition moment, and outputting the state detection fusion vector of each vehicle at each acquisition moment by using the KF algorithm.

Preferably, the method for determining the decision weight of each element in the tunnel driving data sequence in each vehicle at each acquisition time based on the tunnel driving data sequence of each vehicle at each acquisition time comprises the following steps:

Taking tunnel running data sequences of all the acquisition moments of each vehicle as input, and dividing the tunnel running data sequences into a preset number of cluster clusters by adopting a k-shape clustering algorithm; counting probability distribution histograms of abnormal contribution degrees of motion behaviors of the same elements in the tunnel driving data sequence at all acquisition moments of each vehicle in each cluster;

taking the accumulation of the Pasteur distance between each element in the tunnel driving data sequence in each cluster and the probability distribution histogram of the abnormal contribution degree of the motion behavior of any element in the rest as a first accumulated value;

Taking the accumulated result of the Pasteur distance between the probability distribution histograms of abnormal contribution degree of the motion behaviors of each element in the tunnel driving data sequence in each cluster and the same element in any other cluster as a second accumulated value;

Taking the sum of the first accumulated value and the second accumulated value as an intra-class and intra-class distinguishing coefficient of each element in the tunnel driving data sequence in each cluster;

Calculating the ratio of the inside-outside distinguishing coefficient of each element in the tunnel running data sequence in each cluster to the accumulated sum of the inside-outside distinguishing coefficients of each element in the tunnel running data sequences in all the clusters, and taking the product of the ratio and the abnormal contribution degree of the motion behavior of each element in the tunnel running data sequence of each vehicle at each acquisition time as the decision weight of each element in the tunnel running data sequence of each vehicle at each acquisition time.

Preferably, the method for completing traffic control on the vehicles in the tunnel based on the detection result of the motion state of each vehicle at each moment in each work area in the tunnel comprises the following steps:

When the real-time detection model of the vehicle motion state in the tunnel detects that the vehicle motion state is abnormal, sending a detection result of the vehicle motion state at the current moment to an alarm processing module in the intelligent traffic control system, wherein the alarm processing module is configured to process alarm information; after alarm information appears in the alarm module in the intelligent traffic control system, a manager immediately sends an alarm signal to the vehicle according to the vehicle information and the vehicle driver information to inform a prisoner in a work area where the vehicle is located to go to the vehicle for assistance, so that the implementation and adjustment of the motion state of the vehicle are completed, and the control of the traffic of the vehicle in the tunnel work area is realized.

In a second aspect, an embodiment of the present application further provides a tunnel intelligent traffic control system, including a memory, a processor, and a computer program stored in the memory and running on the processor, where the processor implements the steps of any one of the tunnel intelligent traffic control methods described above when executing the computer program.

The beneficial effects of the application are as follows: according to the application, the multimode characteristic state diagnosis coefficient is constructed by analyzing the consistency of the driving state image and the tunnel driving data sequence in the vehicle motion state characteristics, and the multimode characteristic state diagnosis coefficient considers the consistency of the characteristics of the vehicle and the driver reflected by two modes when the motion state is abnormal, so that the error probability of single-mode data judgment can be reduced; and determining an fading detection correction coefficient of each vehicle at each acquisition time according to the influence degree of the motion state of the vehicle at the historical moment on the motion state of the vehicle at the current moment under different time scales, wherein the fading detection correction coefficient considers the short-term persistence characteristic that each mode data is influenced by the motion state of the historical vehicle under different time scales. And based on the fading detection correction coefficient, the fading gain when estimating the vehicle motion state is adaptively determined, so that the optimization of the KF algorithm is realized, the accuracy of vehicle motion state detection is improved, and the effect of vehicle traffic control in the tunnel is improved.

Drawings

In order to more clearly illustrate the embodiments of the application or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the application, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.

FIG. 1 is a flow chart of a method for intelligent traffic control in a tunnel according to an embodiment of the present application;

FIG. 2 is a flowchart illustrating a method for controlling intelligent traffic in a tunnel according to an embodiment of the present application;

FIG. 3 is a schematic view showing a change of steering wheel angle during abnormal movement of a vehicle according to an embodiment of the present application;

FIG. 4 is a schematic diagram of a state transition probability matrix of a 1 st type tag at a minimum time scale according to an embodiment of the present application;

FIG. 5 is a schematic diagram showing the physical dimensions of a positioning chip according to an embodiment of the present application;

FIG. 6 is a schematic diagram of a hardware topology relationship in a tunnel intelligent traffic control system according to an embodiment of the present application;

FIG. 7 is a block diagram of a software system in a tunnel intelligent traffic control system according to an embodiment of the present application;

FIG. 8 is a system topology of an opening vehicle management and control system according to one embodiment of the present application;

Fig. 9 is a schematic structural diagram of a computer device according to an embodiment of the present application.

Detailed Description

The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.

Referring to fig. 1, a flowchart of a tunnel intelligent traffic control method according to an embodiment of the application is shown, the method includes the following steps:

and S001, carrying out data acquisition processing on vehicles running in the tunnel to respectively obtain a tunnel running data sequence and a detection vector.

The tunnel for vehicle traffic control comprises a plurality of construction work areas; the tunnel positive holes are formed by face areas at two ends, and intersections which are in cross connection with the inclined shafts are formed in the face areas; a vehicle velocimeter is arranged at the position of the front 100 meters of each intersection, and traffic prompt traffic lights and RFID base station sensors are arranged in all driving directions of each intersection; the vehicles enter the tunnel construction work area from a specified hole to carry out construction operation, the vehicles which go out are arranged at each intersection in the work area to pass preferentially, the vehicles which go in wait for passing pass rules, and the vehicles are driven out from the hole after the construction operation is finished.

The vehicle velocimeter includes: the system comprises a video detection assembly, a Doppler radar (speed measuring radar), a snapshot host, a sensor, a license plate recognition system and a system management platform, wherein the speed monitoring and the video image monitoring are carried out on a running vehicle in a tunnel, and the video detection assembly further comprises a camera arranged above a driving position.

Further, in order to realize effective control of vehicle traffic in the tunnel, real-time accurate detection of the motion state of vehicles in each work area is required, the running state of the vehicles can be accurately mastered through detection of the state information of the vehicles, statistics of traffic flow, lane change, abnormal running and other conditions of the vehicles are further judged, safety running of drivers is prompted through various guidance approaches, and further guarantee is provided for safety running of road vehicles. The anti-interference capability of detecting the vehicle motion state by only data of a single mode is poor, and the accuracy of the detection result is low. Therefore, the application improves the detection effect of the vehicle motion state through analyzing two modes of image data and driving data and data fusion, and further improves the efficiency of traffic control in the tunnel, and the implementation flow of the whole scheme is shown in figure 2.

Specifically, the application utilizes various sensors in a vehicle velocimeter to collect running data of vehicles in a tunnel at each collecting time, the time interval between two adjacent collecting times is 0.5s, the sensors comprise a direction sensor, a speed sensor, a temperature sensor, a pressure sensor and a rotating speed sensor, and the running data comprises 7 running data of steering wheel angle, speed, acceleration, side wheel rotating speed, torque converter temperature and torque converter pressure. In order to avoid the phenomenon of data loss in the data transmission process, firstly, carrying out k neighbor filling processing on each running data by using a mean filling method, secondly, carrying out Z-score normalization on the result of each running data after filling, taking a sequence formed by processing results of all running data of a vehicle under each acquisition realization as a tunnel running data sequence of the vehicle at each acquisition moment, and recording the tunnel running data sequence of the a vehicle at the ith acquisition moment as a tunnel running data sequence of the a vehicleK-nearest neighbor filling and Z-score normalization are well known techniques, and detailed procedures are not repeated.

Further, in order not to obstruct the sight of the driver and not to influence the normal driving of the driver, a camera is installed above the driving position to collect the face image of the driver, each obtained face image is used as input, the two-sided filtering denoising algorithm is used for processing, the denoised image is used as a driving state image, the two-sided filtering denoising algorithm is a known technology, and the specific process is not repeated. Secondly, the disclosed taiwan Qinghua university-DDD data set is used as a training set to train a driving state detection model, the driving state detection model is MTCNN (Multi-TASK CASCADED Neural Networks), a Multi-classification cross entropy function is used as a loss function, an Adam algorithm is used as an optimization algorithm, the labels of the training set comprise three detection results of 0, 1 and 2, namely normal detection results, fatigue detection results and yawning detection results, driving state images corresponding to each acquisition moment are input into the driving state detection model, the driving state detection model is used for outputting detection vectors corresponding to each acquisition moment to obtain the driving state images, the detection vectors are formed by three labels according to a confidence descending order, training of a neural network is a known technology, and detailed processes are omitted. For example, if the confidence of the labels 0, 1, and 2 in the detection result of the driving state image corresponding to the ith acquisition time is 0.1, 0.7, and 0.2, respectively, the detection vector corresponding to the ith acquisition time is。

So far, the tunnel driving data sequence and the detection vector of the vehicle at each acquisition time are obtained and are used for detecting the motion state of the vehicle in the tunnel subsequently.

Step S002 determines a multi-mode feature state diagnostic coefficient for each vehicle at each acquisition time based on the consistency of the detection vector with the motion state reflected by the tunnel travel data sequence.

The single-mode vehicle driving related data, whether traveling data or image data, has false detection of the vehicle state when detecting the vehicle motion state in the tunnel. For example, when the image data tracks the vehicle position to judge the vehicle motion state, large-area change in adjacent video frames may occur in the background of the tunnel, so as to influence the tracking and positioning result and control the vehicle in the tunnel; if only the motion related data such as acceleration and speed are analyzed, the actual situation of uneven part of the road in the tunnel is easily ignored. Therefore, the application considers the verification and analysis of the data of different modes at the same time, and improves the detection accuracy of the vehicle motion state at each acquisition time.

Specifically, the driving environment in the tunnel is complex, in order to provide powerful guarantee for construction safety in the tunnel as far as possible, the traffic accident rate in the tunnel is reduced, when a driver drives a vehicle to run, the driver needs to make timely and accurate judgment on the surrounding environment and the change of the road in order to ensure that the vehicle runs normally, so that the steering wheel is correctly controlled, and the motion state of the vehicle is maintained to be stable for a long time. When the vehicle motion state is affected by the influence of the driver or the rough road or the like in an abnormal state, for example, when the driver is in a fatigue state, the driver's ability to perceive the tunnel environment, judgment ability, and steering ability to the steering wheel are all degraded. For example, when the movement state of the vehicle in the tunnel is abnormal due to fatigue of the driver, the value of the steering wheel angle appears to fluctuate continuously in a small range while traveling on a straight road, and the value of the steering wheel angle appears to change drastically while traveling on a turning road; and the value of the steering wheel angle remains unchanged for a period of time, the steering wheel is kept still, and the steering wheel is greatly and rapidly corrected when the driver is alert from the fatigue state, and the value of the steering wheel angle is greatly and rapidly changed in a very short time, as shown in fig. 3.

Further, for any element in the tunnel driving data sequence, taking the value of each element as an ordinate and the acquisition time of each element as an abscissa, taking the same element in all tunnel driving data sequences at all acquisition times as input, taking a fitting curve obtained by a nonlinear least square method as a time sequence curve of each element, marking all peak points on the time sequence curve, wherein the nonlinear least square method is a known technology, and specific processes are not repeated. And for the time sequence curve of any element, taking the ratio of the absolute value of the difference value between the ordinate of any two adjacent peak points to the absolute value of the difference value between the abscissa of any two adjacent peak points on the time sequence curve as the extreme value variation between any two adjacent peak points.

In another embodiment, for any element in the tunnel driving data sequence, the value of each element may be taken as an ordinate, the acquisition time of each element may be taken as an abscissa, the same element in all the tunnel driving data sequences at all the acquisition times may be taken as input, the fitting curve obtained by using the piecewise fitting algorithm is taken as a time sequence curve of each element, and the piecewise fitting algorithm is a known technology, and a specific process is not repeated.

Further, there are areas with uneven pits on the ground in the tunnel, and when a vehicle passes through the areas, irregular changes of elements in the tunnel driving data sequence at corresponding acquisition time can be caused, but the changes are different from the data changes in the abnormal movement state of the vehicle, so that the influence on the traffic of the vehicle in the tunnel can be avoided in advance. However, with respect to the image data, the facial expression of the driver hardly changes when the vehicle passes through the area of the dishing, or the difference between the facial variation due to the vehicle body shake and the facial expression in the fatigue state is very large, and the difference between the corresponding detection vectors is large. Therefore, in order to reduce the error of detecting the vehicle motion state by single data, the application considers the probability of evaluating the vehicle motion state abnormality at each acquisition time based on the consistency of the detection vector and the motion state reflected by the tunnel driving data sequence.

Based on the analysis, a multi-mode characteristic state diagnosis coefficient is constructed here and used for representing the abnormal condition of each vehicle motion state in the tunnel at each acquisition moment. Calculating a multi-mode characteristic state diagnosis coefficient of the a vehicle at the ith acquisition time:

In the method, in the process of the invention, Is the abnormal contribution degree of the motion behavior of the b element in the tunnel running data sequence of the a-th vehicle at the ith acquisition time,/>Is a sequence formed by all extreme value variation quantities between the starting moment and the i acquisition moment on the time sequence curve of the b element in the tunnel running data sequence of the a-th vehicle,/>、/>Respectively, is a function of taking maximum value and minimum value,/>、/>Sequences/>, respectivelyMaximum value, minimum value,/>Is an exponential function based on natural constants,/>Is the sum of absolute values of differences between the ith acquisition time and the ordinate of the two peak points with the smallest time interval on the time sequence curve of the b element,/>Is the sequence/>Hurst index of all elements in the interior,/>Is a parameter-adjusting factor for preventing denominator from being 0,/>The magnitude of (1) is checked to be 0.1, the hurst index is a known technology, and the detailed process is not repeated;

is the multi-mode characteristic state diagnosis coefficient of the a-th vehicle at the ith acquisition time, n is the number of elements in the tunnel driving data sequence,/> Is the first element in the detection vector corresponding to the ith acquisition time of the (a) th vehicle, j is the jth acquisition time before the ith acquisition time,/>、/>The detection vectors corresponding to the ith vehicle at the ith acquisition time and the jth acquisition time are respectively/(respectively)The pearson correlation coefficients between vectors are known techniques, and detailed description thereof is omitted.

Wherein the greater the probability that the motion state of the a-th vehicle in the tunnel at the ith acquisition moment is in an abnormal condition is, the more irregular the extreme value variation of each element in the tunnel running data sequence of the a-th vehicle at the ith acquisition moment is,、/>The greater the difference between the first variation/>The larger the value of (a) is, the larger the probability of the ith acquisition time being in a short-time abnormal change on the time sequence curve of the b-th element of the a-th vehicle is, the smaller the ordinate between the ith acquisition time and the two peak points with the smallest time interval is on the time sequence curve of the b-th element of the a-th vehicle is/>The smaller the value of/>The larger the value of/>The poorer the stability of all elements in the interior, the more abnormal the extremum change amount at different moments,/>, theThe smaller the value of/>The larger the value of/>The larger the value of (a) is, the first product factorThe greater the value of (2); the greater the possibility that the a-th vehicle is in an abnormal motion state in the tunnel at the i-th acquisition time, the greater the difference between detection vectors corresponding to the a-th vehicle at the i-th acquisition time and the j-th acquisition time is, and the greater the difference between detection vectors is/>The smaller the value of (1) is, the greater the probability that the driver of the a-th vehicle in the tunnel is in a fatigue state at the ith acquisition moment is, the lower the confidence that the 1 st element in the detection vector is 0 is, and the greater the probability that the driver is in a fatigue state isThe larger the value of (2), the second product factor/>The greater the value of (2); i.e./>The greater the value of (a) is, the higher the probability of abnormality in the motion state of the a-th vehicle in the tunnel at the i-th acquisition time is. The multimode characteristic state diagnosis coefficient considers the consistency of the characteristics of the vehicle and the driver reflected by two modes when the motion state is abnormal, and can reduce the error probability of single-mode data judgment.

So far, the multimode characteristic state diagnosis coefficient of each vehicle at each acquisition time in the tunnel is obtained and is used for constructing a vehicle motion fusion detection vector of each subsequent acquisition time of each vehicle.

And step S003, determining a comprehensive fading detection correction coefficient of each vehicle at each acquisition time based on the influence degree of the vehicle motion state at the historical time on the vehicle motion state at the current time under different time scales, and adaptively determining the fading gain based on the comprehensive fading detection correction coefficient.

Further, when the vehicle runs in the tunnel, both the driving state image of the vehicle and the tunnel running data sequence of the vehicle are dynamically changed and are influenced by the same mode data at the historical moment. For example, if the driver in the a-th vehicle is in a tired state in a yawning state at the i-th acquisition time, there is a certain potential performance feature, such as eyelid closure, in the driving state image of the driver in the a-th vehicle at the i-1 st acquisition time; if the driver in the a-th vehicle at the i-th acquisition time is in a state of rapid steering wheel adjustment from fatigue state, there is also a certain potential performance feature such as a large jitter of the face in the driving state image of the driver in the a-th vehicle at the i-1 st acquisition time. For the tunnel running data sequence of the a-th vehicle at the i-th acquisition time, if the motion state of the a-th vehicle at the i-th acquisition time is abnormal, one or more elements in the tunnel running data sequence of the a-th vehicle at the i-1-th acquisition time also can fluctuate in a short time compared with the normal motion state. On the other hand, if an abnormality occurs at a certain time, whether it is a driving state image of the vehicle or a tunnel running data sequence of the vehicle, the driving state image or the tunnel running data sequence of the vehicle at the subsequent acquisition time is affected to different extents. The influence of the same type of modal data at adjacent moments on different modal data at each acquisition moment is different, the influence of historical data is eliminated aiming at the data at each mode when data fusion is carried out, and the detection precision of the vehicle motion state at the current moment is improved.

Specifically, for any one mode data, taking a driving state image as an example, taking the value of each label in the detection vector of the driving state image of the a-th vehicle at all acquisition moments as input, taking a fitting curve obtained by using a nonlinear least square method as a time sequence curve of each label, wherein the abscissa of the time sequence curve is the acquisition moment corresponding to the detection vector of the label and the value of the label, and the nonlinear least square method is a known technology, and the specific process is not repeated. Taking the 1 st label as an example, taking a time sequence curve of the 1 st label as an input, and adopting a differential algorithm to obtain peak points and trough points on the time sequence curve, wherein the differential algorithm is a known technology, and the specific process is not repeated. And secondly, taking the time interval between any two adjacent peak points corresponding to the acquisition time as a time window.

Further, the widths of all time windows on the time sequence curve of the 1 st label and the occurrence probability of each width are counted, each width is taken as a time scale, and the sequence obtained by arranging all time scales in descending order is taken as a time scale sequence. Next, determining a state transition probability matrix of the 1 st label under each time scale based on the change probability of the detection vector in the same scale on the time sequence curve of the 1 st label, wherein the state transition probability matrix of the 1 st label under the minimum time scale is shown as figure 4, and the elements in the first row and the second columnFor the probability of transition of tag 1 from 0 to 1,/>The calculation formula of (2) is as follows:

In the method, in the process of the invention, Is the probability that the 1 st tag is converted from 0 to 1 at the minimum time scale, n is the number of all acquisition moments of the 1 st tag at the minimum time scale,/>Is the number of times the 1 st tag transitions from 0 to 1 at two adjacent acquisition times.

Secondly, according to the steps, the state transition probability matrixes of the 1 st label under all time scales are respectively obtained, each state transition probability matrix is taken as input, a sequence formed by all singular values of each state transition probability matrix obtained by using a SVD (Singular Value Decomposition) algorithm is taken as a label stabilizing sequence of each state transition probability matrix, the SVD algorithm is a known technology, and the specific process is not repeated. If the motion state of the vehicle at the current moment is greatly affected by the motion state at the historical moment, the transition probability of the same tag value of the same tag is more unstable with the gradual increase of the time scale.

Further, taking the a-th vehicle as an example, taking the multi-mode feature state diagnosis coefficients of the a-th vehicle at all the acquisition moments as input, acquiring the abnormality score of each multi-mode feature state diagnosis coefficient by adopting RRCF (Robust Random Cut Forst) algorithm, secondly counting the extreme points in all the abnormality scores, and taking the acquisition moment corresponding to the nearest extreme point on the left side and the right side of the abnormality score of each multi-mode feature state diagnosis coefficient as the left end moment and the right end moment of the acquisition moment corresponding to each multi-mode feature state diagnosis coefficient respectively, wherein the RRCF algorithm is a known technology, and the specific process is not repeated.

Based on the analysis, an evanescence detection correction coefficient is constructed here for representing the influence degree of the motion state of the vehicle at each acquisition time on the motion state at the historical time. Calculating an fading detection correction coefficient of the a-th vehicle at the ith acquisition time:

In the method, in the process of the invention, Is the data stagnation coefficient of the driving state image of the a-th vehicle at the ith acquisition time,/>Is the kind of tag in the detection vector, q is the q-th tag,/>Is the category of time scales on the timing curve, c is the c-th time scale,/>、/>The q-th tag is a tag stabilizing sequence corresponding to the c-th time scale, the q-th tag is a tag stabilizing sequence corresponding to the c-1-th time scale, and the number of the tag stabilizing sequences is/are as follows、/>Are respectively/>、/>The variation coefficient of the internal elements is a known technology, and the specific process is not repeated;

Is the two-end dragging influence coefficient of the a-th vehicle at the ith acquisition time,/> Is the multi-mode characteristic state diagnosis coefficient of the a-th vehicle at the ith acquisition time,/>Is the total number of acquisition time with the smallest time interval between the left end time and the right end time of the ith acquisition time,/>The size of (2) is taken to be 10, namely 5 acquisition times are taken to the left of the left end time and the right of the right end time of the ith acquisition time, and p is the taken p-th acquisition time,/>Is the multi-mode characteristic state diagnosis coefficient of the a-th vehicle at the p-th acquisition time,/>Is a logarithmic function with natural constants as bases;

is the fading detection correction coefficient of the a-th vehicle at the ith acquisition time,/> Is the data stagnation coefficient of the tunnel driving data sequence of the a-th vehicle at the ith acquisition time,/>And/>The calculation modes of the two are consistent, and repeated description is omitted.

Wherein, the motion state of the a-th vehicle at the i-th acquisition moment is greatly influenced by the motion state of the history moment, so that the transition probability of the same label value of the same label is more unstable along with the gradual increase of the time scale, the discrete degree of the element in the label stabilization sequence corresponding to the larger time scale is higher,The larger the value of (c) is,The greater the value of (2); because the abnormal state does not appear instantaneously and disappear immediately, namely, each abnormal movement moment has abnormal moments with different degrees on the left side and the right side in the time dimension, the greater the probability of abnormal motion state of the a-th vehicle at the i-th acquisition time, the greater the probability of abnormal motion state of the a-th vehicle at the i-th acquisition timeThe larger the value of/>The larger the value of/>The greater the value of (2); i.e./>The greater the value of the (a) th acquisition time is, the greater the influence of the motion state of the vehicle at the historical time is on the motion state of the vehicle at the ith acquisition time, so that the influence of the historical motion state is reduced when the motion state of the vehicle is estimated, and the accuracy of implementation detection is improved. The fading detection correction coefficient considers the short-term persistence characteristic that each mode data is influenced by the historical vehicle motion state under different time scales, and has the beneficial effects that the influence of the transient abnormality on the motion state detection result caused by the interference factors in the tunnel on the acquired data can be avoided.

Based on the steps, the fading detection correction coefficient of each vehicle at each acquisition time is obtained respectively, and the fading gain of each vehicle at each acquisition time is determined in a self-adaptive manner by combining the Kalman gain at each acquisition time. The calculation formula of the fading gain of the a-th vehicle at the i-th acquisition time is as follows:

In the method, in the process of the invention, Is the fading gain of the a-th vehicle at the i-th acquisition time,/>Is the fading detection correction coefficient of the a-th vehicle at the ith acquisition time,/>Is the initial moment of the a-th vehicle starting to move in the tunnel,/>Is the Kalman gain of the ith acquisition time of the traditional KF (Kalman Filter) algorithm.

Wherein the longer the total time length of the normal motion state at the historical moment, the smaller the influence of the motion state at the current moment, the smaller the observation noise or error covariance when the KF algorithm estimates the vehicle state, the smaller the influence of the observation data at the current moment on the state estimation,The smaller the value of/>The smaller the value of (2), the more the state estimate depends on the predicted value; conversely, the shorter the total duration of the normal motion state at the historical moment, the greater the influence on the motion state at the current moment, and the greater the influence on the motion state at the current momentIf the process noise is larger or the error covariance is larger, the kalman gain is larger, the state estimation is more dependent on the observed value,The greater the value of (2).

So far, the fading gain under each moment in the KF algorithm is obtained and is used for determining the state detection fusion vector at each moment in the follow-up.

Step S004, a state detection fusion vector of each vehicle at each acquisition time is obtained based on the detection vector and the running reset vector of each vehicle by using a KF algorithm, and the state detection fusion vector is input into a detection model to obtain a detection result of the motion state of each vehicle at each time.

Due to the specificity of the structure in the tunnel, the environmental conditions of different positions of the tunnel have larger differences, such as different brightness of different positions, and different influence on a driver, so that different vehicle motion states are caused; the vehicle motion data collected on the flat road at the turning position is also greatly different from the vehicle motion data collected on the straight flat road, namely the contribution rate of the single data to the vehicle motion state detection result is dynamically changed due to the tunnel environment change. Based on the analysis, the method and the system dynamically adjust the weight of each element in the tunnel driving data sequence by considering the influence of the tunnel environment.

Specifically, the tunnel driving data sequences of all the acquisition moments of the a-th vehicle are taken as input, the tunnel driving data sequences are divided into k clustering clusters by adopting a k-shape clustering algorithm, the k-shape clustering algorithm is a known technology, and the specific process is not repeated. Taking a kth cluster as an example for any cluster, counting probability distribution histograms of abnormal contribution degrees of the motion behaviors of the same element in a tunnel driving data sequence of all the vehicles at all acquisition moments in the kth cluster, wherein the abscissa of the probability distribution histograms is the abnormal contribution degrees of the motion behaviors, the ordinate is a probability value corresponding to the abnormal contribution degrees of the motion behaviors, and the probability distribution histograms of different elements are more similar, so that the influence of all the elements in the kth cluster on the states of the drivers of the vehicles reflected by the kth cluster is more consistent; on the other hand, in different clusters, the more similar probability values corresponding to the abnormal contribution degrees of the motion behaviors in the probability distribution histogram of the abnormal contribution degrees of the motion behaviors of a certain element are, the smaller the influence of the element on the detection result of the motion state of the vehicle is.

Based on the analysis, a decision weight for each element within the tunnel travel data sequence at each acquisition time is determined based on the tunnel travel data sequence for each vehicle at each acquisition time. Calculating a tunnel travel data sequenceDecision weights for the inner b-th element:

In the method, in the process of the invention, Is the inside-outside distinguishing coefficient of the b element in the tunnel running data sequence in the kth cluster, n is the number of elements in the tunnel running data sequence, c is the c element in the tunnel running data sequence,/>、/>The probability distribution histograms of the b-th element and the c-th element in the kth cluster are respectively, m is the number of clusters corresponding to the tunnel driving data sequences at all the acquisition moments of the a-th vehicle, g is the g-th cluster, and the m is/>Is the probability distribution histogram of the b-th element in the g-th cluster,/>、/>Are respectively/>And/>、/>And/>The pasteurization distance between the two layers is a known technology, and the specific process is not repeated;

Is the tunnel driving data sequence/> Decision weights of the b-th element in the interior,/>Is the abnormal contribution degree of the movement behavior of the b-th element in the tunnel driving data sequence of the a-th vehicle at the ith acquisition time.

Wherein in the running process of the vehicle, the smaller the fluctuation of the value range of one element in the tunnel running data sequence under different motion states is, the more inconsistent the influence degree of the different elements on the vehicle motion state reflected by the tunnel running data sequence in the kth cluster is, the lower the similarity of probability distribution histograms corresponding to the different elements in the kth cluster is,The larger the value of (1)/>, the first accumulated valueThe larger the value of the b-th element is, the larger the influence of the vehicle motion state reflected by different clusters is, the distribution of the b-th element changes along with the change of the vehicle motion state, and the larger the difference of the probability distribution histogram of the b-th element on different clusters is, namely/>The larger the value of (2) the second accumulated valueThe larger the value of/>The greater the value of (2); /(I)The larger the value of (a) is, the larger the probability that the motion state of the (a) vehicle in the tunnel at the ith acquisition moment is in abnormal conditions is, and the tunnel driving data sequence/>The higher the likelihood that an internal element can characterize an abnormal motion state, i.e./>The larger the value of (2), the tunnel travel data sequence/>The more sensitive the inner element b is to anomalies in the state of motion of the vehicle.

And respectively acquiring decision weights of all elements in any one tunnel driving data sequence according to the steps. And determining a running reset vector corresponding to each tunnel running data sequence based on the decision weights of all elements in each tunnel running data sequence. Data sequence driven by tunnelFor example, the tunnel driving data sequence/>Decision weight of inner b-th element/>Taking the product of the element b and the element b as the reset value of the element b, and taking the reset values of all elements according to the tunnel driving data sequence/>Vector composed of inner element sequences as tunnel driving data sequence/>Corresponding driving reset vector/>。

Further, the detection vectors and the running reset vectors of all the a-th vehicles at the previous i acquisition moments are used as inputs, the fading gain of the i-th acquisition moment replaces the Kalman gain of the traditional KF algorithm corresponding to the time step, the KF algorithm is used for outputting the state detection fusion vector of the a-th vehicle at the i-th acquisition moment, and the KF algorithm is a known technology and a specific process is not repeated.

Secondly, according to the above-mentioned procedure, obtain the state detection fusion vector of multiple motion states when the vehicle is in various unusual motion states in the normal running in the tunnel respectively, multiple motion states include straight-line running motion state, turning motion state, climbing motion state, downhill motion state to the vehicle motion state at every moment marks the state detection fusion vector, mark the content altogether including: the method comprises the steps of performing normal straight running motion state, normal turning motion state, normal climbing motion state, normal downhill motion state, abnormal straight running motion state, abnormal turning motion state, abnormal climbing motion state and abnormal downhill motion state, performing training on a support vector machine SVM (Support Vector Machine) model by using a marked state detection fusion vector, and performing training on the support vector machine by using the trained SVM model as a real-time detection model of the vehicle motion state in a tunnel, wherein the training on the support vector machine is a known technology, and the specific process is not repeated.

Further, a state detection fusion vector of the vehicle in the tunnel at each moment is obtained in real time, and the state detection fusion vector is input into the SVM model to obtain a detection result of the vehicle motion state at each moment.

And step S005, adjusting the motion state of each vehicle based on the detection result of the motion state of the vehicle in each work area in the tunnel, and completing traffic control of the vehicles in the tunnel.

And determining a detection result of the motion state of each vehicle in each area in the tunnel in real time according to the steps, and realizing control of the vehicles in the tunnel based on the detection result. Specifically, when the real-time detection model of the vehicle motion state in the tunnel detects that the vehicle motion state is abnormal, sending a detection result of the vehicle motion state at the current moment to an alarm processing module in the intelligent traffic control system, wherein the alarm processing module is configured to process alarm information; after alarm information appears in the alarm module in the intelligent traffic control system, a manager immediately sends an alarm signal to the vehicle according to the vehicle information and the vehicle driver information to inform a prisoner in a work area where the vehicle is located to go to the vehicle for assistance, so that the implementation and adjustment of the motion state of the vehicle are completed, and the control of the traffic of the vehicle in the tunnel work area is realized. Taking the a-th vehicle as an example, according to the vehicle information of the a-th vehicle and the identity information of the vehicle driver, the vehicle information visualization interface comprises a license plate number, a work area position, a vehicle state and the like, and the vehicle driver identity information visualization interface comprises a staff number, a contact way, a current state and the like. Meanwhile, a manager immediately enters a visualization module in the intelligent traffic control system, and according to the work area where the vehicle is located, a corresponding work area identifier is selected on a tunnel layout diagram in the visualization module in the intelligent traffic control system, so that the manager can jump to a man-vehicle control visual interface diagram where the work area is located. Further, after a specific working point position is selected, an actual interface diagram of the tunnel layout diagram can be popped up, wherein the diagram comprises a main tunnel and a branch tunnel in the tunnel.

Secondly, for any vehicle, transmitting the detection result of the vehicle motion state of each vehicle at each moment to a data storage module for storage and recording, ensuring the safety, traceability and durability of data, providing data support for subsequent data analysis and business decision, and specifically storing the detection result of the vehicle motion state of all vehicles in each work area into a folder.

Based on the same inventive concept as the method, the embodiment of the application also provides a tunnel intelligent traffic control system, which comprises a memory, a processor and a computer program stored in the memory and running on the processor, wherein the processor realizes the steps of any one of the tunnel intelligent traffic control methods when executing the computer program.

It should be noted that, the tunnel intelligent traffic control system provided in this embodiment executes the statistical scheduling of the entrance and exit of the tunnel vehicle, including: a hardware system and a software system, wherein the hardware system comprises:

A tachometer sensor configured to acquire image data and radar data of a vehicle within a tunnel, the tachometer sensor comprising: a video detection assembly, a Doppler radar;

A positioning chip configured to obtain accurate position information of vehicles through positioning sensors installed in tunnel traffic vehicles to be statistically scheduled, and transmit the vehicle position information to a background scheduling server in real time, providing a real-time data basis for operation management; the physical dimensions of the positioning chip in this embodiment are shown in fig. 5, and the hardware topology relationship in the tunnel intelligent traffic control system is shown in fig. 6.

The data positioning base station acquisition equipment is configured to be installed at a set position in the tunnel, is matched with the positioning chip, receives signals sent by the vehicle positioning chip, obtains position information of the vehicle, and transmits the position information to the background scheduling server through the scheduling controller;

the dispatching controller is configured to dispatch the vehicle dispatching operation of the control center, monitor and control the position and state information of the vehicle in real time, and exchange data with the background dispatching server to realize the dispatching, tracking and management of the vehicle;

A switch, router, firewall network device configured to establish a data communication network of the vehicle dispatch system such that data transmission between the devices is secure, fast, stable;

the scheduling server is configured as background core equipment of the vehicle scheduling system, and is used for receiving, storing, processing and analyzing data generated by the vehicle scheduling system and providing real-time support for vehicle scheduling, management and traffic supervision;

and the storage device is configured to store the data information in the scheduling system and ensure the data reliability, traceability and safety of the scheduling system. In this embodiment, the data information in the scheduling system stored in the storage device includes:

vehicle position information, scheduling records, operation data, alarm information and vehicle motion state detection results.

A database configured to store a large amount of data in the scheduling system, providing a data basis for real-time analysis and comprehensive evaluation of vehicle scheduling;

In this embodiment, the large amount of data in the scheduling system stored in the database includes:

vehicle and driver information, history scheduling, and operating conditions.

The Web server is configured to manage a Web interface of the scheduling system, provides a friendly graphical interface and is convenient for a user to operate and manage;

in this embodiment, the Web interface of the scheduling system includes: the system comprises a vehicle dispatching system background management interface for a dispatcher, a vehicle monitoring platform for a driver, and a data query and report analysis system for a superior management department.

In this embodiment, a block diagram of a software system in a tunnel intelligent traffic control system is shown in fig. 7, where the software system includes:

The data acquisition module is configured to acquire data uploaded by the vehicle-mounted terminal equipment in real time and provide real-time data support for vehicle dispatching; the data uploaded by the vehicle-mounted terminal equipment include, but are not limited to: vehicle position, speed, direction, status information;

A data processing module configured to perform data analysis, computation, filtering, processing of the collected data, including but not limited to calculation of tunnel travel data sequences, detection vectors; calculating abnormal contribution degree of motion behaviors, multimode characteristic state diagnosis coefficients, fading gain and state detection fusion vectors; and the accuracy and the safety of the subsequent data are ensured so as to carry out vehicle dispatching and supervision work.

In this embodiment, the type of the collected data processed by the data processing module includes: judging the motion state of the vehicle, detecting and processing abnormal events, and cleaning and archiving data;

The scheduling management module is configured to execute vehicle scheduling, instruction issuing and operation supervision, including task scheduling, route planning, real-time monitoring and exception handling, and helps to improve scheduling efficiency and management level and ensure high-efficiency and safe operation of the vehicle;

The information display module is configured to display the processed data in the form of charts and reports, provides monitoring, analyzing and early warning functions for managers and decision makers, and helps to realize data visualization and intellectualization of vehicle dispatching;

The alarm processing module is configured to process alarm information, including the condition processing of overspeed abnormal events, timely find and solve the problems existing in the vehicle, reduce the safety risk and the management cost, and ensure the operation safety and the sustainable development;

the data storage module is configured to store data, ensure the safety, traceability and durability of the data and provide data support for subsequent data analysis and business decision;

in this embodiment, the data stored in the data storage module includes:

history, vehicle information, driver information, operation data, detection vector of vehicle, and travel reset vector.

And the permission management module is configured to manage and authorize the system user, ensure personal privacy and system safety and prevent illegal access and malicious attack.

In this embodiment, the types of management and authorization of the system user by the rights management module include:

User registration, login rights, role rights, and access control.

In the embodiment, an opening vehicle control system is constructed by the hardware system and the software system based on an application environment of opening vehicle control;

The tunnel portal vehicle management and control system comprises: the cloud Internet of things induction positioning base station (data positioning base station acquisition equipment, data acquisition module), RFID tag cards (positioning chips) of hole entering vehicles and personnel equipment, cloud Internet of things server (scheduling server, data processing module, scheduling management module, data storage module, permission management module), hole computer (information display module, alarm processing module), system platform on the cloud Internet of things server is visited by the hole computer through the network, and vehicle personnel business turn over record, on-site vehicle personnel, historical data viewing and scheduling are carried out, and display is carried out on a hole large screen display. The system topology diagram of the vehicle control system for the tunnel portal is shown in fig. 8, and fig. 8 includes two transmission lines, namely a network cable and an HDMI high-definition video signal line, wherein the HDMI high-definition video signal line is used for synchronizing the picture of a computer for the tunnel portal to a display screen for the tunnel portal;

The cloud Internet of things induction positioning base station is connected with the RFID tag card of the hole entering vehicle and the personnel equipment in a wireless signal manner; the cloud Internet of things induction positioning base station is in signal connection with the cloud Internet of things server; and the cloud Internet of things server is in signal connection with the hole computer.

When a vehicle runs to a cloud Internet of things induction positioning base station, the cloud Internet of things induction positioning base station senses chip information of an RFID tag card of a corresponding vehicle, chip data are transmitted to a system platform on a cloud Internet of things server for analysis and judgment, the system platform analyzes collected data, categorizes and counts vehicles and personnel in a hole and synchronously displays the classified and counted data on a large-screen display screen of the hole, and automatic statistics of entering and exiting of the vehicles and the personnel and real-time control of the number of the vehicles and the personnel in the hole are achieved.

The embodiment of the application also provides a computer device, and fig. 9 is a schematic structural diagram of the computer device provided by the embodiment of the application; as shown in fig. 9, the computer device includes: input means 23, output means 24, memory 22 and processor 21; the memory 22 is configured to store one or more programs; when the one or more programs are executed by the one or more processors 21, the one or more processors 21 are caused to implement the tunnel intelligent traffic control method as provided in the above-described embodiments; wherein the input device 23, the output device 24, the memory 22 and the processor 21 may be connected by a bus or otherwise, for example in fig. 9.

The memory 22 is used as a readable storage medium of a computing device and can be used for storing a software program and a computer executable program, and the program instructions corresponding to the tunnel intelligent traffic control method according to the embodiment of the application; the memory 22 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, at least one application program required for functions; the storage data area may store data created according to the use of the device, etc.; in addition, memory 22 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid-state storage device; in some examples, memory 22 may further comprise memory located remotely from processor 21, which may be connected to the device via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The input means 23 is operable to receive input numeric or character information and to generate key signal inputs relating to user settings and function control of the device; the output device 24 may include a display device such as a display screen.

The processor 21 executes various functional applications of the device and data processing by running software programs, instructions and modules stored in the memory 22, i.e. implements the tunnel intelligent traffic control method described above.

The computer equipment provided by the embodiment can be used for executing the tunnel intelligent traffic control system provided by the embodiment, and has corresponding functions and beneficial effects.

Embodiments of the present application also provide a storage medium containing computer-executable instructions, which when executed by a computer processor, are for performing the tunnel intelligent traffic control system as provided by the above embodiments, the storage medium being any of various types of memory devices or storage devices, the storage medium comprising: mounting media such as CD-ROM, floppy disk or tape devices; computer system memory or random access memory, such as DRAM, DDRRAM, SRAM, EDORAM, rambus (Rambus) RAM, or the like; nonvolatile memory such as flash memory, magnetic media (e.g., hard disk or optical storage); registers or other similar types of memory elements, etc.; the storage medium may also include other types of memory or combinations thereof; in addition, the storage medium may be located in a first computer system in which the program is executed, or may be located in a second, different computer system, the second computer system being connected to the first computer system through a network (such as the internet); the second computer system may provide program instructions to the first computer for execution. Storage media includes two or more storage media that may reside in different locations (e.g., in different computer systems connected by a network). The storage medium may store program instructions (e.g., embodied as a computer program) executable by one or more processors.

It should be noted that, the storage medium containing the computer executable instructions provided by the embodiments of the present application is not limited to the tunnel intelligent traffic control system described in the foregoing embodiments, and may also perform the related operations in the tunnel intelligent traffic control method provided by any embodiment of the present application.

In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. The foregoing description of the preferred embodiments of the present application is not intended to be limiting, but rather, any modifications, equivalents, improvements, etc. that fall within the principles of the present application are intended to be included within the scope of the present application.

Claims

1. The intelligent tunnel traffic control method is characterized by comprising the following steps of:

2. The method for controlling intelligent traffic in a tunnel according to claim 1, wherein the method for determining the abnormal contribution degree of the movement behavior of each element based on the distribution characteristics of the extremum variation on the time sequence curve of each element in the tunnel driving data sequence of each vehicle at each acquisition time is as follows:

3. The method for managing and controlling intelligent traffic in tunnels according to claim 1, wherein the method for determining the multi-mode feature state diagnosis coefficient of each vehicle at each collection time based on the abnormal contribution degree of the movement behavior of each element in the tunnel driving data sequence of each vehicle at each collection time and the similarity of the detection vector at each collection time and the historical time thereof is as follows:

4. The tunnel intelligent traffic control method according to claim 1, wherein the method for determining the data stagnation coefficient of each mode data of each vehicle at each acquisition time based on the degree of dispersion of elements in the tag stabilization sequence of the state transition probability matrix of each mode data at different time scales is as follows:

5. The tunnel intelligent traffic control method according to claim 1, wherein the method for determining the two-end dragging influence coefficient of each vehicle at each collection time based on the abnormality score of the multi-mode feature state diagnosis coefficient of each vehicle is as follows:

6. The tunnel intelligent traffic control method according to claim 1, wherein the method for adaptively determining the fading gain of each vehicle at each collection time in the KF algorithm based on the drag influence coefficient at both ends of each vehicle at each collection time and the data stagnation coefficients of two mode data is as follows:

7. The tunnel intelligent traffic control method according to claim 1, wherein the method for determining the state detection fusion vector of each vehicle at each acquisition time based on the detection vector, the driving reset vector, and the fading gain of each vehicle at each time and its history time by KF algorithm is as follows:

8. The method for intelligent traffic control according to claim 7, wherein the method for determining the decision weight of each element in the tunnel driving data sequence in each vehicle at each acquisition time based on the tunnel driving data sequence in each vehicle at each acquisition time is as follows:

9. The intelligent traffic control method according to claim 1, wherein the method for completing traffic control of vehicles in the tunnel based on the detection result of the motion state of each vehicle at each time in each work area in the tunnel comprises the steps of:

10. Tunnel intelligent traffic control system comprising a memory, a processor and a computer program stored in said memory and running on said processor, characterized in that the steps of the tunnel intelligent traffic control method according to any one of claims 1-9 are realized when said computer program is executed by said processor.