CN113257000B - Intelligent detection early warning system and method for road black ice - Google Patents

Intelligent detection early warning system and method for road black ice Download PDF

Info

Publication number
CN113257000B
CN113257000B CN202110189087.1A CN202110189087A CN113257000B CN 113257000 B CN113257000 B CN 113257000B CN 202110189087 A CN202110189087 A CN 202110189087A CN 113257000 B CN113257000 B CN 113257000B
Authority
CN
China
Prior art keywords
road
black ice
decision tree
tree model
intelligent
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110189087.1A
Other languages
Chinese (zh)
Other versions
CN113257000A (en
Inventor
刘洋
胡增
王鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Applied Technology Co Ltd
Original Assignee
China Applied Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Applied Technology Co Ltd filed Critical China Applied Technology Co Ltd
Priority to CN202110189087.1A priority Critical patent/CN113257000B/en
Publication of CN113257000A publication Critical patent/CN113257000A/en
Application granted granted Critical
Publication of CN113257000B publication Critical patent/CN113257000B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/01—Detecting movement of traffic to be counted or controlled
    • G08G1/048—Detecting movement of traffic to be counted or controlled with provision for compensation of environmental or other condition, e.g. snow, vehicle stopped at detector
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D21/00—Measuring or testing not otherwise provided for
    • G01D21/02—Measuring two or more variables by means not covered by a single other subclass
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
    • G06F1/26—Power supply means, e.g. regulation thereof
    • G06F1/32—Means for saving power
    • G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
    • G06F1/3234—Power saving characterised by the action undertaken
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/24323—Tree-organised classifiers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
    • G06F18/256—Fusion techniques of classification results, e.g. of results related to same input data of results relating to different input data, e.g. multimodal recognition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/02—Alarms for ensuring the safety of persons
    • G08B21/10—Alarms for ensuring the safety of persons responsive to calamitous events, e.g. tornados or earthquakes
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
    • G16Y10/00—Economic sectors
    • G16Y10/35—Utilities, e.g. electricity, gas or water
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
    • G16Y20/00—Information sensed or collected by the things
    • G16Y20/10—Information sensed or collected by the things relating to the environment, e.g. temperature; relating to location
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
    • G16Y40/00—IoT characterised by the purpose of the information processing
    • G16Y40/10—Detection; Monitoring
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
    • G16Y40/00—IoT characterised by the purpose of the information processing
    • G16Y40/20—Analytics; Diagnosis

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Business, Economics & Management (AREA)
  • Environmental & Geological Engineering (AREA)
  • Mathematical Physics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Multimedia (AREA)
  • Toxicology (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • Geology (AREA)
  • Emergency Management (AREA)
  • Traffic Control Systems (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an intelligent detection and early warning system and method for black ice on a road, which comprises a front-end system and an Internet of things platform in signal connection with the front-end system, wherein the front-end system comprises: the main body structure comprises a base arranged on an anti-collision fence of a highway, a vertical rod vertically arranged on the base and an electric box, wherein the bottom of the electric box is communicated with the external environment; the power supply system comprises a solar panel arranged at the top end of the upright rod and a storage battery which is arranged in the electric box and is electrically connected with the solar panel; the sensor system comprises a plurality of redundant temperature sensors, a humidity sensor and a condensation ice sensor which are arranged in an electric box; the intelligent gunlock recognition system comprises an AI camera with edge computing capability and an RS-485 communication line; the sensor system and the intelligent bolt system identification system are electrically connected with the storage battery.

Description

Intelligent detection early warning system and method for road black ice
Technical Field
The invention relates to the field of road traffic, in particular to an intelligent detection and early warning system and method for road black ice.
Background
Black Ice, known under the name Black Ice, refers to a very thin layer of Ice that is difficult to find on roads. Because the thickness of the road is very thin, the asphalt color of the road can be transmitted out of the road, so that the visibility of the ice surface is very low, and the road is integrated with the road and is very difficult to find. Since the temperature of the thin black ice is not particularly low, a layer of melted water is present on the surface of the ice, allowing the ice to slide more than a thick ice having a low temperature. The black ice can cause the vehicle to slip, spin and brake for a significantly longer distance.
Disclosure of Invention
In order to solve the technical problems, the invention provides an intelligent detection and early warning system and method for road black ice.
In order to solve the technical problems, the invention adopts the following technical scheme:
the utility model provides an intelligent detection early warning system of road black ice, includes front end system and the thing networking platform with front end system signal connection, front end system includes:
the main body structure comprises a base arranged on an anti-collision fence of a highway, a vertical rod vertically arranged on the base and an electric box, wherein the bottom of the electric box is communicated with the external environment;
the power supply system comprises a solar panel arranged at the top end of the upright rod and a storage battery which is arranged in the electric box and is electrically connected with the solar panel;
the sensor system comprises a plurality of redundant temperature sensors, a humidity sensor and a condensation ice sensor which are arranged in an electric box;
the intelligent gunlock recognition system comprises an AI camera with edge computing capability and an RS-485 communication line;
the sensor system and the intelligent bolt system identification system are electrically connected with the storage battery.
A detection and early warning method of an intelligent detection and early warning system for road black ice comprises the following steps:
step A: the sensor system periodically detects and judges the occurrence condition of the black ice on the road through a pre-constructed decision tree model;
and B, step B: training a convolutional neural network through a road black ice image with textural features, activating an AI camera and acquiring a road image within the working time period of a sensor system, and inputting the road image into the convolutional neural network to judge whether the road has black ice;
step C: if the sensor system and the intelligent gunlock recognition system both judge that black ice appears on the road, an early warning is sent out through an early warning unit; if only one of the sensor system and the intelligent gunlock recognition system judges that the road is black ice, prompting is carried out through the monitoring terminal;
wherein, step A includes:
step A1: collecting multiple groups of historical road characteristic data and corresponding black ice occurrence conditions to form multiple characteristic data, and calculating the entropy of each characteristic data;
step A2: selecting feature data with the maximum entropy as current splitting features, and constructing a decision tree model by calculating the information gain rate of each feature data;
step A3: and inputting real-time road characteristic data into the decision tree model, and judging the occurrence condition of the black ice.
In particular, the road characteristic data includes an atmospheric temperature, a ground humidity, and an air humidity of the road region.
Specifically, in step A1, three-quarters of the feature data are used as the training data set, one-quarter of the feature data are used as the test data set, and the entropy of each feature data in the training data set is calculated
Figure BDA0002944590740000021
p i Indicating the probability of the occurrence of black ice on the road.
Specifically, in step A1, a C4.5 decision tree algorithm model is introduced, the feature data with the maximum entropy is selected as the current splitting feature, the information gain rate of each feature data is calculated, the feature data with the highest information gain rate is used as the root node of the decision tree model, the feature data with the information gain rate higher than the average level is used as the upper node of the decision tree model, and the feature data with the information gain rate lower than the average level is used as the leaf node of the decision tree model.
Specifically, before road characteristic data are input into a decision tree model, pruning operation and evaluation are carried out on the decision tree model; calculating GINI coefficient when pruning the decision tree model
Figure BDA0002944590740000022
p k Representing the probability value of the black ice on the road; calculating loss
C α (T)＝C(T)+α|T leaf |；
Wherein C (T) = ∑ [ Gini (p) × n L ]，n L Is the number of samples in the leaf node, | T leaf L is the number of leaf nodes, and alpha is a regularization term;
and when the decision tree model is evaluated, importing the test data set into the decision tree model, and verifying the prediction precision of the decision tree model on the black ice generation condition.
Specifically, in the step B, a road image with black ice is collected in advance, graying processing, binarization processing and texture extraction are sequentially carried out, and the obtained road black ice image with texture features is led into a convolutional neural network for training; and the AI camera inputs the acquired road image into the convolutional neural network to judge whether black ice appears on the road.
Specifically, when the road image is subjected to the graying processing, the grayscale value Gray (i, j) =0.299R (i, j) +0.578G (i, j) +0.114B (i, j) of any pixel (i, j) is calculated by using the weighted average algorithm; where R (i, j) is the red channel value of the pixel, G (i, j) is the green channel value of the pixel, and B (i, j) is the blue channel value of the pixel.
Specifically, when the road image is subjected to binarization processing, an adaptive threshold segmentation method is adopted.
Specifically, when texture extraction is performed on a road image, a gray level co-occurrence matrix of the road image is extracted, wherein the gray level is 0-1, and the method comprises the following steps of:
selecting any point T (x, y) on the road image and a deviation point T '(x + a, y + b) thereof to form a point pair (T, T'), wherein the corresponding gray values are (i, j), the values of (a, n) are (0,1), (1,0), (1,1) and (-1,1), and four gray level co-occurrence matrixes G1, G2, G3 and G4 are obtained in total;
and respectively calculating the contrast, energy and entropy of the texture features of the road image for the four gray level co-occurrence matrixes.
Compared with the prior art, the invention has the beneficial technical effects that:
the sensor system and the intelligent gunlock recognition system are used for detecting the occurrence condition of black ice on a road, and the sensor system and the intelligent gunlock recognition system have higher power consumption, so that the intelligent gunlock recognition system can work at a certain time interval, and the time interval can be set manually; during the operation of the sensor system and the intelligent bolt face recognition system, the occurrence condition of the black ice on the road is detected and judged respectively and simultaneously, and the occurrence condition of the black ice on the road is comprehensively judged according to the detection results of the two systems, so that the accuracy of black ice detection is improved.
Drawings
FIG. 1 is a diagram of a detection and warning system according to the present invention;
FIG. 2 is a flow chart of a detection and early warning method of the present invention;
FIG. 3 is the code of the adaptive thresholding method API interface of the present invention;
FIG. 4 is a parameter description of the API interface of the adaptive threshold partitioning method of the present invention.
Detailed Description
A preferred embodiment of the present invention will be described in detail below with reference to the accompanying drawings.
The main body structure is used for carrying a power supply system, an intelligent bolt face recognition system and a sensor system; the base is arranged on an anti-collision fence of a road; the electric box is arranged on the base and is 0.4-0.5 m away from the ground; the upright pole is fixed with the electric box; the storage battery is arranged in the electric box. Solar panel installs on the pole setting top, and solar energy supplies power for intelligent rifle bolt identification system, sensor system.
The AI camera is installed in the pole setting, and the AI camera has marginal computing power, can embed black ice, ordinary ice surface, snow surface, surface of water and normal road surface image recognition algorithm, has 4G communication ability, can regularly take a candid photograph analysis, remote control candid photograph, long-range real-time video of watching.
And the RS485 communication line is connected with the AI camera and is used for signal transmission.
The sensor system is arranged in the electric box and is provided with a plurality of paths of redundant temperature sensors, humidity sensors and condensation ice sensors, and the redundant sensors are additionally arranged outside the normal working sensors for improving the reliability of the system.
The bottom of the electrical appliance box is designed into a micro weather station mode, and the electrical appliance box is ventilated, can reach the same weather conditions as the road surface, and can be dustproof and waterproof.
The Internet of things platform is located at a far end and comprises a monitoring server, a monitoring terminal and an early warning unit, wherein the monitoring server, the monitoring terminal and the early warning unit are used for receiving data results of road surface detection and providing road surface conditions for workers.
The detection early warning system transmits a real-time video and an analysis result back to the Internet of things platform through the 4G signal.
A detection and early warning method of an intelligent detection and early warning system for road black ice comprises the following steps:
step A: the sensor system periodically detects and judges the occurrence condition of the black ice on the road through a pre-constructed decision tree model;
and B: training a convolutional neural network through a road black ice image with textural features, activating an AI camera and acquiring a road image within the working time period of a sensor system, and inputting the road image into the convolutional neural network to judge whether the road has black ice;
and C: and comprehensively judging the occurrence condition of the black ice on the road through the detection results of the sensor system and the gun bolt recognition system.
Specifically, step C is divided into two cases, C1 and C2.
Wherein case C1 is: and if the sensor system and the intelligent bolt face recognition system both judge that black ice appears on the road, an early warning is sent out through the early warning unit.
Case C2 is: in order to prevent the sensor system or the intelligent gun bolt recognition system from generating false alarm, if only one of the sensor system and the intelligent gun bolt recognition system judges that black ice appears on a road, the situation is only prompted through the monitoring terminal, and a worker is required to verify whether the situation is false alarm or not.
Case C2 is divided into two cases, C21 and C22.
Wherein case C21 is: if the sensor system judges that the black ice appears on the road, the worker can call the monitoring picture of the AI camera to continuously observe for a period of time so as to judge whether the black ice appears.
Case C22 is: if the intelligent gunlock recognition system judges that black ice appears on the road, the worker resets the sensor system, the sensor system works again, and numerical value changes of the multiple paths of redundant temperature sensors, the humidity sensor and the condensation ice sensor are continuously observed to judge whether the black ice appears.
Wherein, step A includes:
step A1: and collecting multiple groups of historical road characteristic data and corresponding black ice occurrence conditions to form multiple characteristic data, and calculating the entropy of each characteristic data.
In particular, the road characteristic data includes an atmospheric temperature, a ground humidity, and an air humidity of the road region.
In this embodiment, three-quarters of the feature data are used as the training data set, and one-quarter of the feature data are used as the test data set.
Construction of a sample corpus H = { (M) from a training dataset 1 ,N 1 ),(M 2 ,N 2 ),…(M i ,N i ) In which M is i Measured values representing road characteristic data, N i And i represents the actual condition of the road with black ice, namely whether the road has the black ice or not, and the reference numeral of the training data.
Computing entropy for each feature data in a training dataset
Figure BDA0002944590740000051
p i Indicating the probability of the occurrence of black ice on the road.
Step A2: and selecting the feature data with the maximum entropy as the current splitting feature, and constructing a decision tree model by calculating the information gain rate of each feature data.
Specifically, in the step A2, a C4.5 decision tree algorithm model is introduced, the feature data with the maximum entropy is selected as the current splitting feature, the information gain rate of each feature data is calculated, the feature data with the highest information gain rate is used as the root node of the decision tree model, the feature data with the information gain rate higher than the average level is used as the upper node of the decision tree model, and the feature data with the information gain rate lower than the average level is used as the leaf node of the decision tree model, so that the decision tree model is constructed, and the output value is designed.
Specifically, before road characteristic data are input into a decision tree model, pruning operation and evaluation are carried out on the decision tree model; the criteria for carrying out the pruning operation are: the more leaf nodes, the larger the penalty C function.
When the decision tree model is pruned, the GINI coefficient is calculated firstly
Figure BDA0002944590740000052
p k Representing the probability value of the black ice on the road; recalculating loss
C α (T)＝C(T)+α|T leaf |；
Wherein C (T) = ∑ [ Gini (p) × n L ]，n L Is the number of samples in the leaf node, | T leaf And | is the number of leaf nodes, and alpha is a regularization term.
The parameter alpha controls the degree of influence between the two, a larger alpha facilitates the selection of a simpler decision tree model, and a smaller alpha facilitates the selection of a more complex decision tree model.
And when the decision tree model is evaluated, importing the test data set into the decision tree model, and verifying the prediction precision of the decision tree model on the black ice generation condition.
Step A3: and inputting real-time road characteristic data into the decision tree model, and judging the occurrence condition of the black ice.
The method comprises the steps that an AI camera collects road images, in consideration of the problem of power consumption, a sensor system collects data once every 5-10 minutes, the sensor system is in a sleep working mode at other times, and the AI camera is awakened to capture road images while the sensor system operates; the AI camera may also be set to capture video at intervals, such as capturing a 15-second video stream every 3 minutes after 8 pm to 6 am, where the video stream is processed into a sequence of images by the AI camera to form an image group.
Before recognition, image processing needs to be performed on the acquired single image or image group.
Specifically, in the step B, a road image with black ice is collected in advance, graying processing, binarization processing and texture extraction are sequentially carried out, and the obtained road black ice image with texture features is led into a convolutional neural network for training; and the AI camera inputs the acquired road image into the convolutional neural network to judge whether the black ice appears on the road.
Specifically, when the road image is subjected to the graying processing, the grayscale value Gray (i, j) =0.299R (i, j) +0.578G (i, j) +0.114B (i, j) of any pixel (i, j) is calculated by using the weighted average algorithm; where R (i, j) is the red channel value of the pixel, G (i, j) is the green channel value of the pixel, and B (i, j) is the blue channel value of the pixel.
Specifically, when the road image is subjected to binarization processing, a self-adaptive threshold segmentation method is adopted; specifically, a corresponding API interface in the Opencv software library is called, as shown in fig. 3 to 4, where:
InputArray src denotes an input image;
OutputArray dst represents an output image, which is in accordance with the input image size;
int ADAPTIVE method represents an algorithm used for calculating a threshold in a neighborhood, and has two values, namely ADAPTIVE _ THRESH _ MEAN _ C and ADAPTIVE _ THRESH _ GAUSSIAN _ C;
the calculation method of ADAPTIVE _ THRESH _ MEAN _ C is to calculate the average value of the neighborhood and then subtract the value of the parameter double C;
the calculation method of ADAPTIVE _ THRESH _ GAUSSIAN _ C is to calculate the GAUSSIAN mean value of the field and then subtract the value of the parameter double C;
int threshold type is a threshold type, and only two values are provided, namely threshold _ BINARY and threshold _ BINARY _ INV;
the calculation unit of adaptiveThreshold is a neighborhood block of image pixels, int blockSize is the size of a local neighborhood, 3, 5, 7 and the like;
double C is the offset adjustment amount;
the final threshold value is obtained by subtracting the offset value adjustment amount from the mean value and Gaussian calculation threshold value, so that the analysis of the black ice condition of the road is facilitated, and the influence of uneven illumination distribution on image processing is reduced.
Specifically, when texture extraction is performed on a road image, a gray level co-occurrence matrix of the road image is extracted, wherein the gray level is 0-1, and the method comprises the following steps:
selecting any point T (x, y) on the road image and a deviation point T '(x + a, y + b) thereof to form a point pair (T, T'), wherein the corresponding gray values are (i, j), the values of (a, b) are (0,1), (1,0), (1,1) and (-1,1), and four gray level co-occurrence matrixes G1, G2, G3 and G4 are obtained in total;
and respectively calculating the contrast, energy and entropy of the texture features of the road image for the four gray level co-occurrence matrixes.
Contrast con = ∑ Σ (i-j) 2 p (x, y); if the contrast is high, the characteristic texture is obvious, and if the contrast is low, the characteristic texture is fuzzy;
energy Asm =sigma-p (x, y) 2 (ii) a The energy is small, which indicates that the characteristic texture is fine; large energy meterThe apparent characteristic texture is thick;
entropy Ent = -Sigma p (i, j) log 2 p (i, j), wherein (i, j) represents the gray value of a pair of points (T, T') in the image; p (i, j) is the probability of the occurrence of the point pair (T, T').
And then, introducing the road black ice image with the texture features into a convolutional neural network for training, wherein the convolutional neural network in the embodiment is a VGG-16 network.
And the AI camera compares the acquired single image or image group with the road black ice image with the texture characteristics to judge whether the road generates black ice.
The detection results of the two systems are used for comprehensively judging the occurrence condition of the black ice on the road, so that the accuracy of black ice detection is improved; the intelligent gunlock recognition system, the sensor system feedback and the historical experience can be simultaneously integrated, for example, information such as the time interval of road black ice generation, the time period of black ice generation, the road section prone to black ice generation and the like in the past year is obtained, a comprehensive early warning result is obtained, then related data are returned to the Internet of things platform through the 4G communication system, the monitoring server responds, the monitoring terminal displays the road condition, and the early warning unit starts a corresponding warning process.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein, and any reference signs in the claims are not intended to be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment may contain only a single embodiment, and such description is for clarity only, and those skilled in the art should integrate the description, and the embodiments may be combined as appropriate to form other embodiments understood by those skilled in the art.

Claims (3)

1. The detection and early warning method of the road black ice intelligent detection and early warning system is characterized in that the detection and early warning system comprises a front-end system and an Internet of things platform in signal connection with the front-end system, and the front-end system comprises:
the main body structure comprises a base arranged on an anti-collision fence of a highway, a vertical rod vertically arranged on the base and an electric box, wherein the bottom of the electric box is communicated with the external environment;
the power supply system comprises a solar panel arranged at the top end of the upright rod and a storage battery which is arranged in the electric box and electrically connected with the solar panel;
the sensor system comprises a plurality of redundant temperature sensors, a humidity sensor and a condensation ice sensor which are arranged in an electric box;
the intelligent gunlock recognition system comprises an AI camera with edge computing capability and an RS-485 communication line;
the sensor system and the intelligent bolt face recognition system are electrically connected with the storage battery;
the Internet of things platform is positioned at a far end and comprises a monitoring server, a monitoring terminal and an early warning unit;
the detection early warning method comprises the following steps:
step A: the sensor system periodically detects and judges the occurrence condition of the black ice on the road through a pre-constructed decision tree model;
and B: training a convolutional neural network through a road black ice image with textural features, activating an AI camera and acquiring a road image within the working time period of a sensor system, inputting the road image into the convolutional neural network for comparison, and judging whether the road has black ice or not;
and C: if the sensor system and the intelligent gunlock recognition system both judge that black ice appears on the road, an early warning is sent out through an early warning unit; if only one of the sensor system and the intelligent gunlock recognition system judges that the road is black ice, prompting is carried out through the monitoring terminal;
wherein, step A includes:
step A1: collecting multiple groups of historical road characteristic data and corresponding black ice occurrence conditions to form multiple characteristic data, and calculating the entropy of each characteristic data;
step A2: selecting the feature data with the maximum entropy as the current splitting feature, and constructing a decision tree model by calculating the information gain rate of each feature data;
step A3: inputting real-time road characteristic data into a decision tree model, and judging the occurrence condition of black ice;
in step A1, three-quarters of the feature data are used as a training data set, one-quarter of the feature data are used as a test data set, and the entropy of each feature data in the training data set is calculated
Figure 715308DEST_PATH_IMAGE001
Figure 177513DEST_PATH_IMAGE002
Representing a probability value of black ice appearing on the road;
in the step A2, introducing a C4.5 decision tree algorithm model, selecting feature data with the maximum entropy as current splitting features, calculating the information gain rate of each feature data, taking the feature data with the highest information gain rate as a root node of the decision tree model, taking the feature data with the information gain rate higher than the average level as an upper node of the decision tree model, and taking the feature data with the information gain rate lower than the average level as leaf nodes of the decision tree model;
before inputting the road characteristic data into the decision tree model, pruning and evaluating the decision tree model; calculating GINI coefficient when pruning the decision tree model
Figure 197422DEST_PATH_IMAGE003
Figure 527909DEST_PATH_IMAGE004
Representing the probability value of the occurrence of black ice on the road; calculating loss
Figure 254557DEST_PATH_IMAGE005
Wherein the content of the first and second substances,
Figure 277876DEST_PATH_IMAGE006
，
Figure 785081DEST_PATH_IMAGE007
is the number of samples in the leaf node,
Figure 529046DEST_PATH_IMAGE008
is the leaf node number, alpha is the regularization term;
when the decision tree model is evaluated, importing a test data set into the decision tree model, and verifying the prediction precision of the decision tree model on the black ice generation condition;
b, acquiring a road image with black ice in advance, sequentially carrying out graying processing, binarization processing and texture extraction, and introducing the road image with the texture characteristics into a convolutional neural network for training; the AI camera inputs the acquired road image into a convolutional neural network to judge whether black ice appears on the road;
when the road image is grayed, the weighted average algorithm is adopted, and the gray value of any pixel (i, j)
Figure 500413DEST_PATH_IMAGE009
(ii) a Wherein
Figure 304421DEST_PATH_IMAGE010
Is the red channel value of the pixel,
Figure 298922DEST_PATH_IMAGE011
is the green channel value for the pixel,
Figure 971212DEST_PATH_IMAGE012
is the blue channel value of the pixel;
when texture extraction is carried out on a road image, a gray level co-occurrence matrix of the road image is extracted, the gray level is 0~1, and the method comprises the following steps:
selecting any point on road image
Figure 672451DEST_PATH_IMAGE013
And its deviation point
Figure 37574DEST_PATH_IMAGE014
Form a pair of dots
Figure 253791DEST_PATH_IMAGE015
Corresponding to a gray value of
Figure 605138DEST_PATH_IMAGE016
Wherein the values of (a, b) are (0,1), (1,0), (1,1) and (-1,1) respectively, so as to obtain four gray level co-occurrence matrixes G1, G2, G3 and G4;
and respectively calculating the contrast, energy and entropy of the texture features of the road image for the four gray level co-occurrence matrixes.
2. The detection and early-warning method of the intelligent detection and early-warning system for the black ice on the road as claimed in claim 1, wherein the road characteristic data comprises the atmospheric temperature, the ground humidity and the air humidity of the road area.
3. The detection and early-warning method of the intelligent detection and early-warning system for the black ice on the road as claimed in claim 1, wherein an adaptive threshold segmentation method is adopted when the road image is subjected to binarization processing.
CN202110189087.1A 2021-02-19 2021-02-19 Intelligent detection early warning system and method for road black ice Active CN113257000B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110189087.1A CN113257000B (en) 2021-02-19 2021-02-19 Intelligent detection early warning system and method for road black ice

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110189087.1A CN113257000B (en) 2021-02-19 2021-02-19 Intelligent detection early warning system and method for road black ice

Publications (2)

Publication Number Publication Date
CN113257000A CN113257000A (en) 2021-08-13
CN113257000B true CN113257000B (en) 2022-10-25

Family

ID=77181352

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110189087.1A Active CN113257000B (en) 2021-02-19 2021-02-19 Intelligent detection early warning system and method for road black ice

Country Status (1)

Country Link
CN (1) CN113257000B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104504366A (en) * 2014-11-24 2015-04-08 上海闻泰电子科技有限公司 System and method for smiling face recognition based on optical flow features
CN105788288A (en) * 2014-12-17 2016-07-20 王益 Intelligent system for monitoring icing on pavement and monitoring method thereof
CN106126328A (en) * 2016-06-24 2016-11-16 同济大学 A kind of traffic metadata management method based on event classification and system
CN106951692A (en) * 2017-03-06 2017-07-14 北京师范大学 Exemplary resource elements recognition method and device
CN107229910A (en) * 2017-05-18 2017-10-03 北京环境特性研究所 A kind of remote sensing images icing lake detection method and its system
CN108090507A (en) * 2017-10-19 2018-05-29 电子科技大学 A kind of medical imaging textural characteristics processing method based on integrated approach
CN109919184A (en) * 2019-01-28 2019-06-21 中国石油大学(北京) A kind of more well complex lithology intelligent identification Methods and system based on log data

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104504366A (en) * 2014-11-24 2015-04-08 上海闻泰电子科技有限公司 System and method for smiling face recognition based on optical flow features
CN105788288A (en) * 2014-12-17 2016-07-20 王益 Intelligent system for monitoring icing on pavement and monitoring method thereof
CN106126328A (en) * 2016-06-24 2016-11-16 同济大学 A kind of traffic metadata management method based on event classification and system
CN106951692A (en) * 2017-03-06 2017-07-14 北京师范大学 Exemplary resource elements recognition method and device
CN107229910A (en) * 2017-05-18 2017-10-03 北京环境特性研究所 A kind of remote sensing images icing lake detection method and its system
CN108090507A (en) * 2017-10-19 2018-05-29 电子科技大学 A kind of medical imaging textural characteristics processing method based on integrated approach
CN109919184A (en) * 2019-01-28 2019-06-21 中国石油大学(北京) A kind of more well complex lithology intelligent identification Methods and system based on log data

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于OpenCV+Python的道路结冰检测系统的设计;李颀等;《陕西科技大学学报(自然科学版)》;20170425(第02期);全文 *

Also Published As

Publication number Publication date
CN113257000A (en) 2021-08-13

Similar Documents

Publication Publication Date Title
CN108446630B (en) Intelligent monitoring method for airport runway, application server and computer storage medium
CN108109385B (en) System and method for identifying and judging dangerous behaviors of power transmission line anti-external damage vehicle
CN106845890B (en) Storage monitoring method and device based on video monitoring
CN111079663A (en) High-altitude parabolic monitoring method and device, electronic equipment and storage medium
CN101430195B (en) Method for computing electric power line ice-covering thickness by using video image processing technology
CN108108772B (en) Insulator pollution flashover state detection method based on aerial image of distribution line
CN114972843A (en) Agricultural pest diagnosis and early warning system based on big data
CN112396635B (en) Multi-target detection method based on multiple devices in complex environment
CN110852164A (en) YOLOv 3-based method and system for automatically detecting illegal building
CN113887412A (en) Detection method, detection terminal, monitoring system and storage medium for pollution emission
CN113947570B (en) Crack identification method based on machine learning algorithm and computer vision
CN113139489A (en) Crowd counting method and system based on background extraction and multi-scale fusion network
CN113887324A (en) Fire point detection method based on satellite remote sensing data
CN115311623A (en) Equipment oil leakage detection method and system based on infrared thermal imaging
CN111767881A (en) Self-adaptive crowd density estimation device based on AI technology
CN115410114A (en) Urban rail flood prevention early warning method and system based on multiple characteristics
CN112883969B (en) Rainfall intensity detection method based on convolutional neural network
CN107977531A (en) A kind of method that ground resistance hard measurement is carried out based on image procossing and field mathematical model
CN111461076A (en) Smoke detection method and smoke detection system combining frame difference method and neural network
CN113257000B (en) Intelligent detection early warning system and method for road black ice
CN117033935B (en) Prediction method of rainfall characteristic under statistics and monitoring based on Bayesian fusion
CN106228192B (en) A kind of transmission line forest fire domain identification method analyzed based on double-threshold cluster
CN117593499A (en) Hydropower station electromechanical equipment fault identification method based on distributed routing inspection strategy
CN111539356A (en) Pavement gap accurate and efficient detection method based on linear array image
CN108802845A (en) A kind of indoor occupant occupation rate method of estimation based on infrared sensor array

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant