CN-121982456-A - Evaluation model construction method for disaster prevention and reduction of highway slope

CN121982456ACN 121982456 ACN121982456 ACN 121982456ACN-121982456-A

Abstract

The invention relates to the field of image data processing, in particular to a method for constructing an evaluation model for disaster prevention and reduction of a highway slope. The method comprises the steps of obtaining sample time sequence images, identifying a slope area moving, merging frame sequences which continuously exist and have consistent moving trend into target time periods, calculating a moving fluctuation index of the slope area, determining a similar time period of each target time period, calculating normal disturbance possibility of the slope area, determining an effective deterioration interval, calculating deterioration degree parameters, determining time duration parameters, determining structure damage degree parameters, calculating early warning priority of each slope area, and training a neural network by taking the sample time sequence images as input and the early warning priority as output to obtain a slope instability evaluation model. The invention can effectively distinguish normal engineering operation disturbance and disastrous slope instability, and realizes accurate evaluation and grading early warning of slope instability risks.

Inventors

LIU WEIMIN
GUO QIYOU
XI HUALIN
ZHANG WENCHENG
WANG PENG
LEI JIE
LI BAOTIAN
LUO TAO
ZHAO DONG
Cai Qinge
WEI XUEYONG
DONG XIAOBO
YANG XIAOMING
ZHU JIAN
LIU SHULIN

Assignees

中交第一公路勘察设计研究院有限公司
西安中交公路岩土工程有限责任公司

Dates

Publication Date: 20260505
Application Date: 20260407

Claims (10)

1. The method for constructing the evaluation model for disaster prevention and reduction of the highway slope is characterized by comprising the following steps of: Acquiring a sample time sequence image in real time through a chain camera array distributed in an engineering operation area; For each camera, identifying a side slope area where movement occurs based on the inter-frame difference of the sample time sequence images of each camera, and merging frame sequences which continuously exist and have consistent movement trend into a target period of the camera; calculating a movement fluctuation index of each side slope region according to the movement amplitude, the morphological change degree and the depth change degree of each side slope region in each target period; Determining a similar time period of each target time period, and calculating the normal disturbance possibility of each side slope region according to the movement fluctuation index change characteristic of the side slope region in the similar time period in the target time period; For any side slope region, the normal disturbance possibility of the side slope region in each target period is arranged in time sequence, the period in which the side slope region continuously descends is determined as an effective deterioration interval, and the deterioration degree parameter of the effective deterioration interval is calculated; Determining a time duration parameter according to the duration and the interval of a target period of the side slope region, and determining a structural damage degree parameter according to the registration accuracy of the side slope region and the same region in the adjacent camera images; The time persistence parameter, the deterioration degree parameter and the structural damage degree parameter are synthesized, and the early warning priority of each slope area is calculated; And training the neural network by taking the sample time sequence image as input and the early warning priority as output to obtain a slope instability assessment model.
2. The method for constructing the evaluation model for disaster prevention and reduction of the highway slope according to claim 1, wherein the identifying the moving slope region based on the inter-frame difference of the sample time sequence image for each camera specifically comprises: Carrying out semantic segmentation on each frame of sample time sequence image, and marking a side slope region in each frame of sample time sequence image; Performing differential operation on the slope region marked by the current frame and the slope region marked by the previous frame; if the pixel points with the difference result being not zero exist, judging that the current frame has a side slope area with movement, and marking the area where the pixel points with the difference result being not zero are located as the side slope area with movement.
3. The method for constructing the evaluation model for disaster prevention and reduction of the highway slope according to claim 1, wherein the merging the frame sequences with continuous movement and consistent movement trend into the target period of the camera specifically comprises: Calculating the moving direction of each slope region in each moving frame relative to the corresponding region in the previous frame; If two adjacent frames of images are marked as moving frames and the moving direction consistency of each slope area in the two frames meets a first preset condition, combining the two adjacent frames into an initial target period; If the time interval between two adjacent initial target time periods is smaller than a second preset condition and the consistency of the moving directions of the slope areas in the end frame of the former initial target time period and the start frame of the latter initial target time period meets the first preset condition, combining the two adjacent initial target time periods into one target time period.
4. The method for constructing the evaluation model for disaster prevention and reduction of the highway slope according to claim 1, wherein the calculating the movement fluctuation index of the slope region according to the movement amplitude, the morphological change degree and the depth change degree for each slope region in each target period comprises the following steps: calculating the average value of the motion vector amplitude values of the slope region at each moment in the target period as a motion amplitude parameter for each slope region in each target period; Calculating the ratio of the area of the slope region at the ending moment and the area of the starting moment of the target period as a morphological change degree parameter of the slope region in the target period; Acquiring depth information of the slope region at the starting time and the ending time of the target period, and calculating the ratio of the depth mean value of the ending time to the depth mean value of the starting time as a depth change degree parameter; dividing the movement amplitude parameter by the duration of the target period to obtain the movement amplitude of unit time; Multiplying the movement amplitude, the morphological change degree parameter and the depth change degree parameter of unit time to obtain the movement fluctuation index of the slope region.
5. The method for constructing the evaluation model for disaster prevention and reduction of the highway slope according to claim 4, wherein the obtaining the depth information of the slope region at the start time and the end time of the target period specifically comprises: in each target period, edge detection is carried out on the slope area in each frame of image, and edge points are obtained; Counting the times of marking the edge points as the edge points in a target period aiming at each edge point, and calculating the ratio of the times to the total frame number of the target period to be used as the frequency of occurrence parameter of the edge points; Calculating the ratio of the curvature mean value to the curvature variance when the edge point is marked as the edge point, and taking the ratio as the morphological stability parameter of the edge point; multiplying the appearance frequency parameter by the morphological stability parameter to obtain the geometric contribution degree of the edge point; Marking edge points with geometric contribution degree meeting preset conditions as feature points; inputting the feature points into an ORB algorithm, and registering images of adjacent cameras at the starting moment and the ending moment of a target period respectively to obtain homonymous feature points; According to the parallax of the homonymy feature points, calculating the depth information of each feature point; And aiming at each side slope region, obtaining the depth information of the side slope region at the starting moment and the ending moment according to the depth information mean value of the characteristic points contained in the side slope region at the starting moment and the ending moment of the target period.
6. The method for constructing the evaluation model for disaster prevention and reduction of the highway slope according to claim 1, wherein the determining the similar time period of each target time period specifically comprises: For each historical period, calculating the similarity between the current target period and the historical period according to the same number of slope areas in the current target period and the historical period and the movement fluctuation index difference value of the same slope areas, wherein the historical period is the period which is the same as the engineering operation procedure of the current target period in the historical data; And determining the historical time period with the similarity meeting the preset similarity threshold as the similar time period of the current target time period.
7. The method for constructing the evaluation model for disaster prevention and reduction of the highway slope according to claim 1, wherein the calculating the normal disturbance probability of the slope region according to the moving fluctuation index variation characteristic of each slope region in the similar period within the target period comprises: For each side slope region, determining a target period of the side slope region marked as the moving of the region for the last time from the historical data, and acquiring the number of similar periods corresponding to the target period; screening all target time periods of the slope area in the same engineering operation procedure from the historical data, obtaining movement fluctuation indexes of the target time periods, and calculating variances of the movement fluctuation indexes; determining a first target period and a last target period from all target periods of the same engineering operation procedure, acquiring movement fluctuation indexes of the two target periods, and calculating a difference value of the two target periods; The number of similar time periods is multiplied by the ratio of the variance to the difference to obtain the normal disturbance probability of the side slope area.
8. The method for constructing an evaluation model for disaster prevention and reduction of a highway slope according to claim 1, wherein the method for constructing an evaluation model for disaster prevention and reduction of a highway slope according to claim 1 is characterized in that for any one of the slope areas, the normal disturbance possibilities thereof in each of the target periods are arranged in time series, and the period in which the continuous decline is determined as an effective deterioration interval, and the deterioration degree parameter of the effective deterioration interval is calculated, and specifically comprises: For each side slope region, acquiring normal disturbance possibility corresponding to all target time periods of the side slope region, and arranging the normal disturbance possibility according to a time sequence to obtain a normal disturbance possibility time sequence; Performing first-order differential operation on the normal disturbance possibility time sequence, and identifying continuous sequence segments with negative differential values; Determining a period corresponding to each continuous negative change in the continuous sequence segments as an effective deterioration interval; And calculating the lengths of all effective deterioration zones, and taking the maximum zone length as the deterioration degree parameter of the slope area.
9. The method for constructing the evaluation model for disaster prevention and reduction of the highway side slope according to claim 1, wherein the determining the time duration parameter according to the duration and the interval of the target period of the side slope region and determining the structural failure degree parameter according to the registration accuracy of the side slope region and the same region in the adjacent camera images specifically comprises: For each slope region, acquiring the duration of all target time periods of the slope region, and calculating the average value of the duration of all target time periods as the average time period; Acquiring time intervals between adjacent target time periods in all target time periods of the side slope region, and calculating the average value of all the intervals to serve as average interval duration; dividing the average time period duration by the average interval duration to obtain a time duration parameter of the side slope region; and when the slope region is marked as the moving slope region for the last time, calculating the average value of the registration errors in the registration of the adjacent camera images and the registration error of the same slope region in the adjacent camera images, and taking the average value as the structural damage degree parameter of the slope region.
10. An evaluation model construction system for disaster prevention and reduction of a highway slope, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the computer program when executed by the processor realizes the steps of an evaluation model construction method for disaster prevention and reduction of a highway slope according to any one of claims 1-9.

Description

Evaluation model construction method for disaster prevention and reduction of highway slope Technical Field The invention relates to the field of image data processing, in particular to a method for constructing an evaluation model for disaster prevention and reduction of a highway slope. Background In the road engineering operation process, the common engineering safety risk is slope instability, and the engineering safety risk is mainly represented by slope internal stress unbalance caused by road excavation or filling, so that disasters such as landslide or collapse and the like are caused. Therefore, the slope instability needs to be early warned in time so as to adjust the engineering operation flow and ensure the engineering operation safety. In the prior art, cameras are installed at fixed positions in an engineering operation area, and moving objects in the engineering operation range are detected in an inter-frame difference mode and the like, so that early warning is carried out on slope instability. However, the actual engineering work scene is complex. For example, during engineering operation, conditions such as blasting, mechanical operation and the like may exist, so that the side slope is normally disturbed. The prior art is difficult to effectively distinguish the disturbance of normal engineering operation from the actual slope instability. In addition, the presence of vehicles, personnel or temporary stacked items within the engineering work area may also interfere with the accurate capture of side slope movement information. The factors cause that the prior art is easy to miss judgment or misjudgment of the instability of the side slope, and the accuracy and the reliability of early warning are affected. Disclosure of Invention The invention provides a construction method of an evaluation model for disaster prevention and reduction of a highway slope, which aims to solve the existing problems. The invention relates to a construction method of an evaluation model for highway slope disaster prevention and reduction, which adopts the following technical scheme: The embodiment of the invention provides a method for constructing an evaluation model for disaster prevention and reduction of a highway slope, which comprises the following steps: Acquiring a sample time sequence image in real time through a chain camera array distributed in an engineering operation area; For each camera, identifying a side slope area where movement occurs based on the inter-frame difference of the sample time sequence images of each camera, and merging frame sequences which continuously exist and have consistent movement trend into a target period of the camera; calculating a movement fluctuation index of each side slope region according to the movement amplitude, the morphological change degree and the depth change degree of each side slope region in each target period; Determining a similar time period of each target time period, and calculating the normal disturbance possibility of each side slope region according to the movement fluctuation index change characteristic of the side slope region in the similar time period in the target time period; For any side slope region, the normal disturbance possibility of the side slope region in each target period is arranged in time sequence, the period in which the side slope region continuously descends is determined as an effective deterioration interval, and the deterioration degree parameter of the effective deterioration interval is calculated; Determining a time duration parameter according to the duration and the interval of a target period of the side slope region, and determining a structural damage degree parameter according to the registration accuracy of the side slope region and the same region in the adjacent camera images; The time persistence parameter, the deterioration degree parameter and the structural damage degree parameter are synthesized, and the early warning priority of each slope area is calculated; And training the neural network by taking the sample time sequence image as input and the early warning priority as output to obtain a slope instability assessment model. Optionally, for each camera, identifying a side slope area where movement occurs based on an inter-frame difference of its sample time sequence image specifically includes: Carrying out semantic segmentation on each frame of sample time sequence image, and marking a side slope region in each frame of sample time sequence image; Performing differential operation on the slope region marked by the current frame and the slope region marked by the previous frame; if the pixel points with the difference result being not zero exist, judging that the current frame has a side slope area with movement, and marking the area where the pixel points with the difference result being not zero are located as the side slope area with movement. Optionally, merging the frame sequences with continuous