Infrared radiation identification method for damage state of rock slope

Abstract

The invention provides an infrared radiation identification method for a damage state of a rock slope, and relates to the technical field of rock stratum control, and the method comprises the steps: obtaining the infrared radiation time series data of a target rock slope; performing noise removal and enhancement processing on the infrared radiation time sequence data to obtain an enhanced image; correcting the enhanced image according to the data of the environment where the target rock slope is located to obtain a corrected image; performing feature extraction on the corrected image to obtain damage features; and inputting the damage features into a trained damage state recognition model to obtain a recognition result. According to the method, through advanced image processing technologies (noise removal and image enhancement), the image quality is improved, and the subsequent feature extraction is more accurate. And image correction is carried out by using environmental data, and the influence of environmental factors on infrared radiation is considered, so that the interference of external factors is reduced, and the accuracy of damage identification is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rock stratum control, in particular to an infrared radiation identification method for damage state of rock slope. BACKGROUND

[0002] Rock slope is often found in highway, railway, dam and mine engineering. The stability of the slope is crucial for the safe operation of these projects. Traditional slope monitoring methods such as crack measurement and displacement monitoring can provide certain safety evaluation, but they often cannot monitor the overall state of the slope in real time and comprehensively.

[0003] The prior art with application number CN202310170133.2 provides an infrared radiation intelligent identification method for damage state of bearing coal rock. Using coal rock infrared radiation monitoring technology, from the perspective of computer vision, firstly, stress-strain and infrared radiation time series data during the damage process of bearing coal rock are obtained, and based on the crack strain model method, infrared radiation grayscale image classification data set of different progressive damage stages is established. A convolutional neural network (CNN) classifier is built, and the prepared data set is input into the CNN model for training. Finally, the trained model is used for intelligent identification of the damage stage of bearing coal rock. It realizes non-destructive, real-time intelligent monitoring and judgment of the damage stage of bearing coal rock, and has certain practical significance for infrared radiation damage monitoring and early warning in mining engineering and geotechnical engineering. However, the infrared radiation data may be affected by environmental factors such as temperature fluctuations, humidity, and thermal noise of the equipment itself. These factors may affect the quality and accuracy of the infrared image, thereby affecting the damage identification result. SUMMARY

[0004] In order to overcome the shortcomings of the prior art, the purpose of the present application is to provide an infrared radiation identification method for damage state of rock slope.

[0005] To achieve the above purpose, the present application provides the following solutions:

[0006] An infrared radiation identification method for damage state of rock slope, comprising:

[0007] obtaining infrared radiation time series data of the target rock slope;

[0008] performing noise removal and enhancement processing on the infrared radiation time series data to obtain an enhanced image;

[0009] correcting the enhanced image according to environmental data of the target rock slope to obtain a corrected image;

[0010] performing feature extraction on the corrected image to obtain damage features;

[0011] The damage characteristics are input into a trained damage state recognition model to obtain a recognition result.

[0012] Preferably, performing noise removal and enhancement processing on the infrared radiation time series data to obtain an enhanced image includes:

[0013] Taking each pixel of the infrared radiation time series data as the center, a square window is taken, and the change of each pixel under the square window is detected;

[0014] When the change in pixel points is greater than a preset threshold, the infrared radiation time series data in the corresponding square window is smoothed using the median value of the pixel points in the square window to obtain the smoothed pixel points;

[0015] Continuously moving the square window until the smoothing process of the entire infrared radiation time series data is completed to obtain a smoothed image;

[0016] constructing a pixel enhancement function based on the convolution value of the smoothed image;

[0017] The pixel enhancement function is used to perform image enhancement processing on the infrared radiation time series data to obtain the image to be evaluated.

[0018] Preferably, the calculation formula for the change amount of each pixel under the detection square window is: Among them, f x (x i ,y j ) represents the gradient value of the infrared radiation time series data in the x direction, f y (x i ,y j ) represents the gradient value of the infrared radiation time series data in the y direction, f(x i ,y j ) represents the infrared radiation time series data in (x i ,y j ) position, r represents the gray value of the pixel f(x i ,y j ) changes.

[0019] Preferably, the output formula of the smoothed pixel point is:

[0020] I denoised (x,y)=median(I original (xh,yh),…,I original (x+h,y+h)); where I denoised (x,y) is

[0021] After smoothing, h is the side length of the square window, Ioriginal (x,y) is the value of a pixel point on the infrared radiation time series data at the (x,y) position.

[0022] Preferably, a pixel enhancement function is constructed based on the convolution value of the smoothed image, including:

[0023] The smoothed image is convolved to obtain a convolved image;

[0024] An enhancement coefficient is determined based on the pixel value of the convolved image;

[0025] A pixel enhancement function is constructed based on the enhancement coefficient; wherein the pixel enhancement function is:

[0026] Wherein S(x,y) represents the enhanced image, E(x,y) represents the enhancement coefficient, G(x,y) represents the pixel value of the convolved image at (x,y), I(x,y) represents the smoothed image, and σ represents the mean square error between the smoothed image and the convolved image.

[0027] Preferably, the enhanced image is corrected according to environmental data of the target rock slope to obtain a corrected image, including:

[0028] The environmental data is obtained by using a collection device; the environmental data includes temperature data, humidity data, solar radiation intensity, wind speed data and wind direction data;

[0029] The environmental data is processed for missing values, detected for outliers and normalized to obtain standard data;

[0030] The correlation between each of the standard data and infrared radiation is analyzed using exploratory data analysis methods, and the mapping relationship between each of the standard data and infrared radiation is displayed through icons;

[0031] An environmental impact analysis model is constructed based on the mapping relationship;

[0032] An ideal infrared image without environmental impact is simulated according to the environmental impact analysis model;

[0033] The gray value of the enhanced image is corrected according to the ideal infrared image to compensate for changes caused by environmental factors, to obtain the corrected image.

[0034] Preferably, the environmental impact analysis model includes any one of a linear regression model, a multiple regression model and a machine learning model.

[0035] Preferably, the damage state recognition model comprises a CNN neural network and an LSTM network; the CNN neural network model comprises two stacked convolutional layers, a ReLU activation function and a pooling layer arranged in sequence; the CNN neural network model is used to output damage feature classified data; the LSTM network is used to receive the damage feature classified data, and through an LSTM unit updating and transmission process and a linear layer conversion process, a predicted recognition result is obtained; the recognition result comprises a damage category, a damage probability and a damage degree.

[0036] According to the specific embodiments provided by the present application, the following technical effects are disclosed.

[0037] The present application provides an infrared radiation recognition method for damage state of rock slope, comprising: acquiring infrared radiation time series data of a target rock slope; performing noise removal and enhancement processing on the infrared radiation time series data to obtain an enhanced image; correcting the enhanced image according to environmental data of the target rock slope to obtain a corrected image; extracting features from the corrected image to obtain damage features; inputting the damage features into a trained damage state recognition model to obtain a recognition result. The present application improves the image quality through advanced image processing technology (noise removal and image enhancement), so that the subsequent feature extraction is more accurate. And the image is corrected using environmental data, which considers the influence of environmental factors on infrared radiation, thereby reducing the interference of external factors and improving the accuracy of damage recognition. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0039] Figure 1 The method flowchart provided by the embodiments of the present application. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0041] The purpose of the present application is to provide an infrared radiation recognition method for damage state of rock slope, which can improve the accuracy of damage recognition.

[0042] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0043] Figure 1 The method flowchart provided by the embodiment of the present application is shown in Figure 1 The present application provides an infrared radiation identification method for rock slope damage state, which comprises:

[0044] Step 100: acquiring infrared radiation time series data of a target rock slope;

[0045] Step 200: performing noise removal and enhancement processing on the infrared radiation time series data to obtain an enhanced image;

[0046] Step 300: correcting the enhanced image according to environmental data of the target rock slope to obtain a corrected image;

[0047] Step 400: performing feature extraction on the corrected image to obtain damage features;

[0048] Step 500: inputting the damage features into a trained damage state identification model to obtain an identification result.

[0049] Preferably, the noise removal and enhancement processing on the infrared radiation time series data to obtain an enhanced image comprises:

[0050] taking a square window centered on each pixel point of the infrared radiation time series data, and detecting the change amount of each pixel point under the square window;

[0051] when the change amount of a pixel point is greater than a preset threshold, using the median value of the pixel points in the square window to perform smoothing processing on the infrared radiation time series data in the corresponding square window to obtain a smoothed pixel point;

[0052] continuously moving the square window until the smoothing process of the entire infrared radiation time series data is completed to obtain a smoothed image;

[0053] constructing a pixel enhancement function based on the convolution value of the smoothed image;

[0054] performing image enhancement processing on the infrared radiation time series data using the pixel enhancement function to obtain the to-be-evaluated image.

[0055] Preferably, the calculation formula for detecting the change amount of each pixel point under the square window is: wherein, f x (x i ,yj represents a gradient value of the infrared radiation time series data in the x direction, f y (x i ,y j ) represents a gradient value of the infrared radiation time series data in the y direction, f(x i ,y j ) represents a gray value of the infrared radiation time series data at the (x i ,y j ) position, and r represents a change amount of the pixel point f(x i ,y j ).

[0056] Preferably, the output formula of the smoothed pixel point is:

[0057] I denoised (x,y) = median(I original (x-h,y-h),…,I original (x+h,y+h)); wherein I denoised (x,y) is

[0058] a smoothed pixel point, h is the side length of the square window, and I original (x,y) is the value of the pixel point on the infrared radiation time series data at the (x,y) position.

[0059] Specifically, in actual application, since noise points cannot be avoided in the image acquisition process, and the noise points generally have a large difference from the surrounding pixels, the embodiment can find the noise points on the image by detecting the change amount of each pixel point under the square window, and can better maintain the detail information while removing the noise by using the median of the pixel sequence under the square window as the smoothed pixel point.

[0060] Preferably, a pixel enhancement function is constructed based on the convolution value of the smoothed image, including:

[0061] performing convolution processing on the smoothed image to obtain a convolved image;

[0062] determining an enhancement coefficient based on the pixel value of the convolved image;

[0063] constructing a pixel enhancement function based on the enhancement coefficient; wherein the pixel enhancement function is:

[0064] wherein S(x,y) represents the enhanced image, E(x,y) represents the enhancement coefficient, G(x,y) represents the pixel value of the convolved image at (x,y), I(x,y) represents the smoothed image, and σ represents the mean square error between the smoothed image and the convolved image.

[0065] Specifically, the application can quantify the difference between the original image and the convolved image by calculating the mean square error between the two, when the mean square error is large, it means that the original image and the convolved image are different, and need to be enhanced; when the mean square error is small, it means that the original image and the convolved image are small, and can be considered not to be enhanced. In image enhancement, according to the size of the mean square error, it can be determined whether the image needs to be enhanced, so that the image can achieve better visual effect.

[0066] Preferably, the enhanced image is corrected according to environmental data of the target rock slope to obtain a corrected image, comprising:

[0067] The environmental data is obtained by a collection device; the environmental data includes temperature data, humidity data, solar radiation intensity, wind speed data and wind direction data;

[0068] The environmental data is processed for missing values, detected for outliers and normalized to obtain standard data;

[0069] The correlation between each standard data and infrared radiation is analyzed by exploratory data analysis method, and the mapping relationship between each standard data and infrared radiation is displayed by icon;

[0070] An environmental impact analysis model is constructed based on the mapping relationship;

[0071] An ideal infrared image without environmental impact is simulated according to the environmental impact analysis model;

[0072] The gray value of the enhanced image is corrected according to the ideal infrared image to compensate for the changes caused by environmental factors, and the corrected image is obtained.

[0073] Specifically, the embodiment uses a temperature sensor to collect temperature data of the environment where the slope is located; a humidity sensor is used to record the environmental humidity, because humidity can affect the absorption and scattering of infrared radiation; and the solar radiation intensity is recorded, especially when monitoring during the day, solar radiation may significantly affect the infrared image; and the wind speed data and wind direction data are obtained by a wind speed and direction measuring device, the wind speed and direction will affect the cooling rate of the slope surface, thereby affecting the infrared image.

[0074] Further, the embodiment performs exploratory data analysis (EDA) to understand the basic characteristics and distribution of the data, including:

[0075] 1. Correlation analysis: analyze the correlation between temperature, humidity, solar radiation and infrared radiation.

[0076] 2. Visualization: The relationship between environmental factors and infrared radiation is visually displayed through charts (such as scatter plots, line graphs, etc.).

[0077] Preferably, the environmental impact analysis model includes any one of a linear regression model, a multiple regression model, and a machine learning model.

[0078] Optionally, if the data shows a linear relationship, the embodiment can use a linear regression model. If the influence of multiple environmental variables on infrared radiation is considered, the embodiment can use a multiple regression model. If a nonlinear relationship and a large-scale data set need to be processed, the embodiment can use a machine learning model.

[0079] In addition, the embodiment also uses historical data to train the model and adjust parameters to optimize the performance of the model. Cross-validation techniques are used to evaluate the stability and accuracy of the model, and appropriate evaluation indicators (such as R 2 , mean square error, etc.) are used to evaluate the performance of the model.

[0080] Preferably, the damage state recognition model includes a CNN neural network and an LSTM network; the CNN neural network model includes two stacked convolutional layers, a ReLU activation function, and a pooling layer arranged in sequence; the CNN neural network model is used to output data after damage feature classification; the LSTM network is used to receive the data after damage feature classification, and through an LSTM unit update and transmission process and a linear layer conversion process, a predicted recognition result is obtained; the recognition result includes: damage category, damage probability, and damage degree.

[0081] Further, the recognition result of the embodiment is as follows:

[0082] 1. If the damage state is a predefined category (such as no damage, slight damage, moderate damage, and severe damage), the recognition result is one of these categories. In this case, the result is usually a category label.

[0083] 2. Damage probability distribution: The model outputs a probability vector, and each element represents the probability of the corresponding damage category. Users can select the most likely damage category according to the highest probability.

[0084] 3. Continuous numerical value of damage degree: If the damage degree is continuous (such as a score or percentage of damage severity), the result is a real number representing the severity of the damage.

[0085] The beneficial effects of the present application are as follows:

[0086] (1) The present application improves the image quality through advanced image processing techniques (noise removal and image enhancement), making the subsequent feature extraction more accurate. And using environmental data for image correction, considering the influence of environmental factors on infrared radiation, thus reducing the interference of external factors, improving the accuracy of damage identification.

[0087] (2) The present application can realize the continuous monitoring of rock slope by using infrared radiation time series data, and timely discover and warn potential damage changes. And combining with LSTM network processing time series data, the model can capture the dynamic process of damage development, and provide more accurate time-dependent analysis.

[0088] (3) The automatic data processing and feature extraction of the present application reduces manual intervention, improves processing speed and efficiency. And using the combination of CNN and LSTM, the damage feature deep learning and intelligent identification are realized, reducing the analysis burden of professionals.

[0089] (4) The present application is not limited by specific environment or geographical location, and is widely applicable to various rock slope damage state monitoring. And the present application allows flexible adjustment and optimization, which can adjust the model parameters and processing flow according to the specific application scene.

[0090] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts of each embodiment can be referred to each other.

[0091] The principles and implementation modes of the present application are described by applying specific examples in this paper. The above description of the embodiments is only used to help understand the method of the present application and its core idea; at the same time, for the general technical personnel in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In summary, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method for infrared radiation identification of rock slope damage status, characterized in that: include: Obtain infrared radiation time series data of the target rock slope; performing noise removal and enhancement processing on the infrared radiation time series data to obtain an enhanced image; Correcting the enhanced image according to the environmental data of the target rock slope to obtain a corrected image; performing feature extraction on the corrected image to obtain damage features; The damage characteristics are input into a trained damage state recognition model to obtain a recognition result.

2. The infrared radiation identification method for rock slope damage status according to claim 1 is characterized in that: The infrared radiation time series data is subjected to noise removal and enhancement processing to obtain an enhanced image, including: Taking each pixel of the infrared radiation time series data as the center, a square window is taken, and the change of each pixel under the square window is detected; When the change in pixel points is greater than a preset threshold, the infrared radiation time series data in the corresponding square window is smoothed using the median value of the pixel points in the square window to obtain the smoothed pixel points; Continuously moving the square window until the smoothing process of the entire infrared radiation time series data is completed to obtain a smoothed image; constructing a pixel enhancement function based on the convolution value of the smoothed image; The pixel enhancement function is used to perform image enhancement processing on the infrared radiation time series data to obtain the image to be evaluated.

3. The infrared radiation identification method for rock slope damage status according to claim 2 is characterized in that: The calculation formula for detecting the change of each pixel under the square window is: Among them, f x (x i ,y j ) represents the gradient value of the infrared radiation time series data in the x direction, f y (x i ,y j ) represents the gradient value of the infrared radiation time series data in the y direction, f(x i ,y j ) represents the infrared radiation time series data in (x i ,y j ) position, r represents the gray value of the pixel f(x i ,y j ) changes.

4. The infrared radiation identification method for rock slope damage status according to claim 3 is characterized in that: The output formula of the smoothed pixel point is: I denoised (x,y) = median(I original (x - h,y - h),…,I original (x + h,y + h)); where, I denoised (x,y) is After smoothing, h is the side length of the square window, I original (x, y) is the value of the pixel point at the (x, y) position on the infrared radiation time series data.

5. The infrared radiation identification method for rock slope damage status according to claim 4 is characterized in that: Constructing a pixel enhancement function based on the convolution value of the smoothed image, comprising: Performing convolution processing on the smoothed image to obtain a convolved image; Determine the enhancement coefficient based on the pixel value of the convolved image; A pixel enhancement function is constructed based on the enhancement coefficient; wherein the pixel enhancement function is: Wherein, S(x,y) represents the enhanced image, E(x,y) represents the enhancement coefficient, G(x,y) represents the pixel value of the convolved image at (x,y), I(x,y) represents the smoothed image, and σ represents the mean square error between the smoothed image and the convolved image.

6. The infrared radiation identification method for rock slope damage status according to claim 1 is characterized in that: Correcting the enhanced image according to the environmental data of the target rock slope to obtain a corrected image includes: Acquiring the environmental data using a collection device; the environmental data includes temperature data, humidity data, solar radiation intensity, wind speed data, and wind direction data; Performing missing value processing, outlier detection, and data normalization on the environmental data to obtain standard data; Analyzing the correlation between each of the standard data and infrared radiation using an exploratory data analysis method, and displaying the mapping relationship between each of the standard data and infrared radiation through icons; Constructing an environmental impact analysis model based on the mapping relationship; simulating an ideal infrared image without environmental impact according to the environmental impact analysis model; The grayscale value of the enhanced image is corrected according to the ideal infrared image to compensate for changes caused by environmental factors, thereby obtaining the corrected image.

7. The infrared radiation identification method for rock slope damage status according to claim 6 is characterized in that: The environmental impact analysis model includes any one of a linear regression model, a multivariate regression model and a machine learning model.

8. The infrared radiation identification method for rock slope damage status according to claim 1 is characterized in that: The damage status recognition model includes a CNN neural network and an LSTM network; the CNN neural network model includes two stacked convolutional layers, a ReLU activation function, and a pooling layer arranged in sequence; the CNN neural network model is used to output data after damage feature classification; the LSTM network is used to receive the data after damage feature classification, and obtain a predicted recognition result through the LSTM unit update and transmission process and the linear layer conversion process; the recognition result includes: damage category, damage probability, and damage degree.

Images