Rock-soil slope displacement monitoring and early warning system

Abstract

The invention relates to the technical field of intelligent monitoring, and particularly discloses a rock-soil side slope displacement monitoring and early warning system, which utilizes an artificial intelligence technology based on deep learning to carry out data analysis on monitoring images and side slope parameter data of a rock-soil side slope in each historical day. The surface deformation characteristics of the rock-soil side slope and the global correlation change characteristics of the side slope parameters are excavated, and the displacement trend of the rock-soil side slope is represented based on the fusion characteristics of the surface deformation characteristics and the global correlation change characteristics, so that whether early warning prompt needs to be carried out or not is intelligently judged. Therefore, real-time monitoring and early warning of displacement of the rock-soil slope can be realized, and the efficiency and accuracy of slope safety monitoring are improved, so that the disaster risk of the slope is reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent monitoring technology, and more specifically, to a rock and soil slope displacement monitoring and early warning system. Background Art

[0002] Geotechnical slopes are a type of landform that is widely found in nature and are also a common structural form in engineering construction. However, due to the influence of natural factors (such as rainfall, earthquakes, etc.) and human activities (such as excavation, filling, etc.), geotechnical slopes often suffer from instability, landslides and other disasters, posing a serious threat to people's lives and property safety.

[0003] Traditional methods for monitoring the displacement of geotechnical slopes mostly rely on manual field measurements, which not only require a large workload and poor timeliness, but are also affected by factors such as weather and personnel skills, making it difficult to guarantee the accuracy and reliability of the measurement results. In addition, since most slope deformation and instability occur in extreme conditions such as rainstorms and earthquakes, organizing on-site measurements at this time will seriously threaten the personal safety of monitoring personnel. At the same time, this method of manual field measurements can often only provide limited point data, making it difficult to fully and dynamically reflect the displacement of geotechnical slopes. Therefore, an optimized geotechnical slope displacement monitoring and early warning system is expected. Summary of the invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a rock slope displacement monitoring and early warning system, which uses artificial intelligence technology based on deep learning to perform data analysis on the monitoring images and slope parameter data of the rock slope on each day of history, and mines out the surface deformation characteristics of the rock slope and the global correlation change characteristics of the slope parameters, and characterizes the displacement trend of the rock slope based on the fusion characteristics of the two, so as to intelligently judge whether an early warning prompt is needed. In this way, real-time monitoring and early warning of rock slope displacement can be achieved, the efficiency and accuracy of slope safety monitoring can be improved, and the risk of slope disasters can be reduced.

[0005] Accordingly, according to one aspect of the present application, a rock slope displacement monitoring and early warning system is provided, which includes:

[0006] The slope monitoring module is used to obtain the slope parameter data and slope monitoring images of the monitored slope on each day of history, wherein the slope parameter data include displacement, inclination, crack width, earth pressure value, rainfall and groundwater level;

[0007] A slope parameter time series association coding module is used to extract parameter time series association features from the slope parameter data of each historical day to obtain a slope parameter global context semantic association feature vector;

[0008] A slope state time series change feature extraction module is used to extract the slope state time series change feature of the slope monitoring images of each historical day to obtain a slope state time series change feature vector;

[0009] A multi-dimensional feature fusion module, used for fusing the slope state temporal variation feature vector and the slope parameter global context semantic association feature vector to obtain a global representation feature vector of the rock and soil slope displacement state;

[0010] The monitoring result generating module is used to determine the monitoring result based on the global characterization feature vector of the displacement state of the rock and soil slope.

[0011] In the above-mentioned rock and soil slope displacement monitoring and early warning system, the slope parameter time series association coding module includes: a slope parameter data regularization unit, which is used to arrange the various slope parameters in the slope parameter data of each historical day according to the time dimension as slope parameter time series input vectors to obtain a sequence of slope parameter time series input vectors; a slope parameter time series feature extraction unit, which is used to extract time series features from each slope parameter time series input vector in the sequence of slope parameter time series input vectors to obtain a sequence of slope parameter time series association feature vectors; and an inter-parameter association coding unit, which is used to extract inter-parameter context association features from the sequence of slope parameter time series association feature vectors to obtain the slope parameter global context semantic association feature vector.

[0012] In the above-mentioned rock and soil slope displacement monitoring and early warning system, the slope parameter time series feature extraction unit is used to: pass the sequence of the slope parameter time series input vectors through a slope parameter time series feature extractor based on a one-dimensional convolution layer to obtain a sequence of the slope parameter time series associated feature vectors.

[0013] In the above-mentioned rock and soil slope displacement monitoring and early warning system, the inter-parameter association encoding unit is used to: pass the sequence of the slope parameter time series association feature vectors through the inter-parameter context association encoder based on the Transformer model to obtain the slope parameter global context semantic association feature vector.

[0014] In the above-mentioned rock and soil slope displacement monitoring and early warning system, the slope state temporal change feature extraction module includes: an image optimization unit, which is used to perform image preprocessing on the slope monitoring images of each historical day to obtain a sequence of optimized slope monitoring images; an image temporal change feature extraction unit, which is used to use an image temporal change feature encoder to process the sequence of optimized slope monitoring images to obtain the slope state temporal change feature vector.

[0015] In the above-mentioned rock and soil slope displacement monitoring and early warning system, the image optimization unit is used to: perform histogram equalization preprocessing and CLAHE color correction on the slope monitoring images of each historical day to obtain the sequence of optimized slope monitoring images.

[0016] In the above-mentioned rock and soil slope displacement monitoring and early warning system, the image temporal change feature encoder is a convolutional neural network model using a temporal attention mechanism.

[0017] In the above-mentioned rock and soil slope displacement monitoring and early warning system, the monitoring result generation module includes: a feature optimization unit, which is used to perform expected activation in parameter space on the global characterization feature vector of the rock and soil slope displacement state to obtain an optimized global characterization feature vector of the rock and soil slope displacement state; and an early warning decision unit, which is used to pass the optimized global characterization feature vector of the rock and soil slope displacement state through a classifier-based slope displacement monitor to obtain the monitoring result, and the monitoring result is used to indicate whether to send an early warning prompt.

[0018] In the above-mentioned rock and soil slope displacement monitoring and early warning system, the feature optimization unit is used to: perform expected activation of the parameter space of the global characterization feature vector of the rock and soil slope displacement state by the following optimization formula to obtain the optimized global characterization feature vector of the rock and soil slope displacement state, wherein the optimization formula is:

[0019]

[0020] Among them, v i is the eigenvalue of the ith position of the global characteristic vector of the displacement state of the rock soil slope, α and β represent weight hyperparameters, sigmoid represents the activation function, and v i ′ is the eigenvalue of the ith position of the global characterization eigenvector of the optimized rock and soil slope displacement state.

[0021] In the above-mentioned rock and soil slope displacement monitoring and early warning system, the early warning decision unit is used to: use the classifier-based slope displacement monitor to process the optimized rock and soil slope displacement state global characterization feature vector using the following classification formula to obtain the monitoring result, wherein the classification formula is:

[0022] O=softmax{(W n ,B n ):…:(W1,B1)|X}

[0023] Among them, W1 to W n is the weight matrix, B1 to B n is the bias vector, X is the global characterization feature vector of the optimized rock and soil slope displacement state, softmax is the normalized exponential function, and O is the monitoring result.

[0024] Compared with the prior art, the rock and soil slope displacement monitoring and early warning system provided by this application uses artificial intelligence technology based on deep learning to analyze the monitoring images and slope parameter data of the rock and soil slope on each day of history, dig out the surface deformation characteristics of the rock and soil slope and the global correlation change characteristics of the slope parameters, and characterize the displacement trend of the rock and soil slope based on the fusion characteristics of the two, so as to intelligently judge whether an early warning is needed. In this way, real-time monitoring and early warning of rock and soil slope displacement can be achieved, the efficiency and accuracy of slope safety monitoring can be improved, and the risk of slope disasters can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0026] Figure 1 4 is a block diagram of a rock slope displacement monitoring and early warning system according to an embodiment of the present application.

[0027] Figure 2 Schematic diagram of the architecture of a rock slope displacement monitoring and early warning system according to an embodiment of the present application.

[0028] Figure 3 It is a block diagram of a slope parameter time series association coding module in a rock and soil slope displacement monitoring and early warning system according to an embodiment of the present application.

[0029] Figure 4 It is a block diagram of a slope state time series change feature extraction module in a rock and soil slope displacement monitoring and early warning system according to an embodiment of the present application.

[0030] Figure 5 It is a block diagram of a monitoring result generating module in a rock and soil slope displacement monitoring and early warning system according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] Below, the embodiments of the present application will be described in more detail in conjunction with the accompanying drawings, and the above-mentioned and other purposes, features and advantages of the present application will become more obvious. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein. At the same time, the accompanying drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification, and are used together with the embodiments of the present application to explain the present application, and do not constitute a limitation on the present application. In the drawings, the same reference numerals generally represent the same parts or steps.

[0032] Figure 1 4 is a block diagram of a rock slope displacement monitoring and early warning system according to an embodiment of the present application. Figure 2 Schematic diagram of the architecture of the rock slope displacement monitoring and early warning system according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the rock and soil slope displacement monitoring and early warning system 100 includes: a slope monitoring module 110, which is used to obtain the slope parameter data and slope monitoring images of the monitored slope body on each historical day, wherein the slope parameter data includes displacement, inclination, crack width, soil pressure value, rainfall and groundwater level; a slope parameter time series association coding module 120, which is used to extract the parameter time series association features of the slope parameter data of each historical day to obtain the slope parameter global context semantic association feature vector; a slope state time series change feature extraction module 130, which is used to extract the slope state time series change features of the slope monitoring images of each historical day to obtain the slope state time series change feature vector; a multi-dimensional feature fusion module 140, which is used to fuse the slope state time series change feature vector and the slope parameter global context semantic association feature vector to obtain the rock and soil slope displacement state global characterization feature vector; a monitoring result generation module 150, which is used to determine the monitoring result based on the rock and soil slope displacement state global characterization feature vector.

[0033] As mentioned in the above background technology, the traditional method of monitoring the displacement of rock and soil slopes mainly relies on manual field measurement, and the limitations of this method are manifested in many aspects. First of all, manual field measurement is not only labor-intensive, but also requires a large amount of manpower and material resources, and the timeliness of monitoring is poor. In today's era of rapid technological development, this monitoring method cannot meet the needs of high-speed and real-time monitoring of slope displacement. More importantly, most of the deformation and instability of slopes occur under extreme climatic conditions, such as heavy rain, earthquakes, etc. In this case, conducting on-site measurements will greatly endanger the personal safety of monitoring personnel and may cause serious accidents. Therefore, this monitoring method has huge hidden dangers in terms of safety. In addition, the manual field measurement method can only provide limited point data, which is difficult to fully and dynamically display the displacement status of the rock and soil slope, and cannot provide accurate and comprehensive data support for slope stability analysis.

[0034] In response to the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to analyze the monitoring images and slope parameter data of the rock and soil slope on each day of history, to mine the surface deformation characteristics of the rock and soil slope and the global correlation change characteristics of the slope parameters, and to characterize the displacement trend of the rock and soil slope based on the fusion characteristics of the two, so as to intelligently judge whether an early warning is needed. In this way, real-time monitoring and early warning of the displacement of the rock and soil slope can be achieved, the efficiency and accuracy of slope safety monitoring can be improved, and the risk of slope disasters can be reduced.

[0035] In the above-mentioned rock and soil slope displacement monitoring and early warning system 100, the slope monitoring module 110 is used to obtain the slope parameter data and slope monitoring images of the monitored slope body on each historical day, wherein the slope parameter data includes displacement, inclination, crack width, earth pressure value, rainfall and groundwater level. It should be understood that the change of the slope is often a dynamic process that evolves over time. By collecting parameters such as slope monitoring images, slope displacement, inclination, crack width, earth pressure value, rainfall and groundwater level on each historical day, a multi-dimensional slope state database can be constructed to provide comprehensive information on the slope state. Furthermore, by analyzing and processing the slope data on each historical day, the change trend of the slope state over time can be excavated, and the deformation mode and potential sliding trend of the slope can be identified, so as to better understand the dynamic characteristics and potential risks of the slope, and provide rich original materials for subsequent data analysis and slope displacement early warning management.

[0036] In the above-mentioned rock and soil slope displacement monitoring and early warning system 100, the slope parameter time series association coding module 120 is used to extract parameter time series association features from the slope parameter data of each historical day to obtain a slope parameter global context semantic association feature vector. Figure 3 FIG. 4 is a block diagram of a slope parameter time series correlation coding module in a rock slope displacement monitoring and early warning system according to an embodiment of the present application. Figure 3 As shown, the slope parameter time series association coding module 120 includes: a slope parameter data regularization unit 121, which is used to arrange the slope parameters in the slope parameter data of each historical day into slope parameter time series input vectors according to the time dimension to obtain a sequence of slope parameter time series input vectors; a slope parameter time series feature extraction unit 122, which is used to extract time series features from each slope parameter time series input vector in the sequence of slope parameter time series input vectors to obtain a sequence of slope parameter time series association feature vectors; and an inter-parameter association coding unit 123, which is used to extract inter-parameter context association features from the sequence of slope parameter time series association feature vectors to obtain the slope parameter global context semantic association feature vector.

[0037] Specifically, the slope parameter data regularization unit 121 is used to arrange the various slope parameters in the slope parameter data of each historical day into slope parameter time series input vectors according to the time dimension to obtain a sequence of slope parameter time series input vectors. It should be understood that the slope parameter data has continuity and time series correlation in the time dimension. Therefore, in order to facilitate the capture of the time series change trend of each slope parameter data, in the technical solution of the present application, the various slope parameters of each historical day are arranged into slope parameter time series input vectors according to the time dimension to retain the time sequence relationship of each slope parameter data, so that in the subsequent data time series analysis process, the evolution law of the slope state over time and the potential risk trend can be better captured, providing an effective data basis for the subsequent data time series analysis task extraction.

[0038] Specifically, the slope parameter time series feature extraction unit 122 is used to extract time series features from each slope parameter time series input vector in the sequence of the slope parameter time series input vectors to obtain a sequence of slope parameter time series associated feature vectors. In a specific example of the present application, the encoding method for extracting time series features from each slope parameter time series input vector in the sequence of the slope parameter time series input vectors is to pass the sequence of the slope parameter time series input vectors through a slope parameter time series feature extractor based on a one-dimensional convolutional layer to obtain a sequence of slope parameter time series associated feature vectors. It should be understood that the one-dimensional convolutional layer is a basic component in a convolutional neural network (CNN), which has a powerful feature extraction capability when processing time series data, and can capture the time series change information and local time series associated features in the data. In the technical solution of the present application, the slope parameter time series feature extractor based on the one-dimensional convolution layer can effectively capture the local time correlation in the time series data of each slope parameter by using a one-dimensional convolution kernel to perform sliding convolution processing on each slope parameter time series input vector, and extract the important time series patterns and trends in the time series data of each slope parameter, such as periodic changes, local mutations, etc., to provide effective data support for subsequent slope status analysis.

[0039] Specifically, the inter-parameter association encoding unit 123 is used to perform inter-parameter context association feature extraction on the sequence of the slope parameter time series association feature vectors to obtain the slope parameter global context semantic association feature vector. In a specific example of the present application, the encoding method for performing inter-parameter context association feature extraction on the sequence of the slope parameter time series association feature vectors is to pass the sequence of the slope parameter time series association feature vectors through an inter-parameter context association encoder based on the Transformer model to obtain the slope parameter global context semantic association feature vector. It should be understood that, considering the complexity and dynamics of the slope state, the correlation analysis between various slope parameters is of great significance for understanding the slope state. Therefore, in the technical solution of the present application, in order to fully capture the dependency relationship between various slope parameters, an inter-parameter context association encoder based on the Transformer model is used to perform inter-parameter context association feature extraction on the sequence of the slope parameter time series association feature vectors. Those skilled in the art should know that the Transformer model models the complex relationship between various slope parameters based on the self-attention mechanism, which can effectively capture the long-distance dependency in the sequence of the temporal correlation feature vectors of the slope parameters, and mine the global contextual semantic correlation features between various slope parameters, so as to better reflect the inherent laws of the slope state, more comprehensively understand the stability and potential risks of the slope, and provide strong support for the subsequent slope displacement warning. In addition, compared with the traditional recurrent neural network (RNN), the Transformer model does not need to calculate step by step in sequence when processing sequence data, and can process the entire sequence at the same time, which improves the efficiency and accuracy of data processing, thereby enhancing the overall performance and reliability of slope displacement warning management.

[0040] In the above-mentioned rock and soil slope displacement monitoring and early warning system 100, the slope state time series change feature extraction module 130 is used to extract the slope state time series change feature of the slope monitoring images of each historical day to obtain the slope state time series change feature vector. Figure 4 FIG. 4 is a block diagram of a slope state temporal variation feature extraction module in a rock slope displacement monitoring and early warning system according to an embodiment of the present application. Figure 4 As shown, the slope state temporal change feature extraction module 130 includes: an image optimization unit 131, which is used to perform image preprocessing on the slope monitoring images of each historical day to obtain a sequence of optimized slope monitoring images; an image temporal change feature extraction unit 132, which is used to use an image temporal change feature encoder to process the sequence of optimized slope monitoring images to obtain the slope state temporal change feature vector.

[0041] Specifically, the image optimization unit 131 is used to perform image preprocessing on the slope monitoring images of each historical day to obtain a sequence of optimized slope monitoring images. It should be understood that considering that the slope monitoring images may be affected by various factors during the acquisition process, such as illumination changes, sensor noise, etc., resulting in poor image quality. Therefore, in order to further improve the visual effect of the image, it is necessary to perform image preprocessing on the slope monitoring images of each historical day, so as to provide a better data basis for subsequent slope state analysis. In a specific example of the present application, the image preprocessing on the slope monitoring images of each historical day to obtain a sequence of optimized slope monitoring images is to perform histogram equalization preprocessing and CLAHE color correction on the slope monitoring images of each historical day to obtain the sequence of optimized slope monitoring images. Among them, histogram equalization and CLAHE (Contrast Limited Adaptive Histogram Equalization, contrast limited adaptive histogram equalization) are both technologies for enhancing image contrast. Specifically, histogram equalization is a method to enhance image contrast by redistributing image pixel values ​​so that the image histogram presents a uniform distribution. CLAHE is a method based on histogram equalization that adaptively adjusts the local area of ​​the image to achieve contrast enhancement and prevent image distortion caused by excessive image enhancement. By combining these two preprocessing methods, the quality and visual effects of slope monitoring images can be effectively improved, the visibility of key features in the image can be enhanced, and it is helpful to better identify and analyze small changes on the slope surface, providing a good data basis for subsequent slope status analysis.

[0042] Specifically, the image temporal change feature extraction unit 132 is used to process the sequence of the optimized slope monitoring images using an image temporal change feature encoder to obtain the slope state temporal change feature vector. In a specific example of the present application, the image temporal change feature encoder is a convolutional neural network model using a time attention mechanism. It should be understood that convolutional neural networks have good performance in the field of image processing. By using a convolutional neural network model to extract image features from the sequence of the optimized slope monitoring images, it is possible to capture subtle changes in the slope state at the image level, thereby providing a strong basis for the temporal change analysis of the slope state. At the same time, since the state change of the slope is a process that progresses over time, in order to fully understand the evolution of the slope state over time, in the technical solution of the present application, a temporal attention mechanism is introduced into the convolutional neural network model to capture the long-distance dependency in the sequence of the optimized slope monitoring images, further dig out the temporal change characteristics of the slope state, and help the model pay attention to the importance of features at different time points in the sequence of the optimized slope monitoring images, so as to more accurately capture the key moments and patterns of slope changes. Specifically, the temporal attention mechanism assigns different weights to the image features at each time point by calculating the similarity between the image features at each time point in the sequence of the optimized slope monitoring images and the overall features, so that in the process of analyzing the temporal changes in the slope state, more attention can be paid to the features of important time points, thereby improving the accuracy of slope state prediction and early warning.

[0043] In a specific embodiment of the present application, a convolutional neural network model using a temporal attention mechanism is used to process the sequence of the optimized slope monitoring images to obtain the slope state temporal change feature vector, including: extracting adjacent first optimized slope monitoring images and second optimized slope monitoring images from the sequence of the optimized slope monitoring images; passing the first optimized slope monitoring image and the second optimized slope monitoring image through the first convolutional layer and the second convolutional layer of the convolutional neural network model respectively to obtain a first convolutional feature map and a second convolutional feature map; calculating the position point multiplication between the first convolutional feature map and the second convolutional feature map to obtain a temporal attention map; inputting the temporal attention map into a Softmax activation function to obtain a temporal attention feature map; calculating the position point multiplication between the second convolutional feature map and the temporal attention feature map to obtain a temporal enhancement feature map corresponding to the second optimized slope monitoring image; looping and iterating the above steps, the temporal enhancement feature map corresponding to the last optimized slope monitoring image in the sequence of the optimized slope monitoring images is the slope state temporal change feature map; performing global mean pooling processing on the slope state temporal change feature map along the channel dimension to obtain the slope state temporal change feature vector.

[0044] In the above-mentioned rock and soil slope displacement monitoring and early warning system 100, the multi-dimensional feature fusion module 140 is used to fuse the slope state time series change feature vector and the slope parameter global context semantic association feature vector to obtain the rock and soil slope displacement state global characterization feature vector. It should be understood that the slope state time series change feature vector represents the intuitive changes on the slope surface, while the slope parameter global context semantic association feature vector reveals the physical changes inside the slope. By fusing the slope state time series change feature vector and the slope parameter global context semantic association feature vector, the complementary information in the slope parameter data and the slope monitoring image can be fully utilized to provide more comprehensive slope state information, improve the characterization accuracy of the rock and soil slope displacement state, and thus more accurately understand and predict the displacement trend of the slope. In addition, in some cases, a single data source may be affected by noise or outliers. Here, by fusing multiple data sources, the dependence on a single data source can be reduced, and the stability and robustness of the system can be improved. In a specific embodiment of the present application, the slope state temporal variation feature vector and the slope parameter global context semantic association feature vector are fused in a weighted summation manner to obtain a global representation feature vector of the displacement state of the rock and soil slope.

[0045] In the above-mentioned rock and soil slope displacement monitoring and early warning system 100, the monitoring result generation module 150 is used to determine the monitoring result based on the global characterization feature vector of the rock and soil slope displacement state. Figure 5 FIG. 1 is a block diagram of a monitoring result generation module in a rock slope displacement monitoring and early warning system according to an embodiment of the present application. Figure 5 As shown, the monitoring result generation module 150 includes: a feature optimization unit 151, which is used to perform expected activation in parameter space on the global characterization feature vector of the rock and soil slope displacement state to obtain an optimized global characterization feature vector of the rock and soil slope displacement state; an early warning decision unit 152, which is used to pass the optimized global characterization feature vector of the rock and soil slope displacement state through a classifier-based slope displacement monitor to obtain the monitoring result, and the monitoring result is used to indicate whether to send an early warning prompt.

[0046] Specifically, the feature optimization unit 151 is used to perform expected activation of the parameter space of the global characterization feature vector of the displacement state of the rock and soil slope to obtain the optimized global characterization feature vector of the displacement state of the rock and soil slope. In particular, in the technical solution of the present application, considering that in practical applications, the acquired slope parameter data and slope monitoring images may be affected by various factors, including sensor errors, inaccurate data collection, etc., which may cause abnormal values ​​or noise in some data, thereby affecting the generation process of the global characterization feature vector of the displacement state of the rock and soil slope. At the same time, in the feature extraction process, the slope parameter time series feature extractor based on the one-dimensional convolution layer and the parameter context association encoder based on the Transformer model may be affected by uneven data distribution or abnormal values, resulting in local abnormal distribution in the global characterization feature vector of the displacement state of the rock and soil slope generated in the end. This local abnormal distribution may cause the slope displacement monitor to misjudge when identifying the displacement state of the rock and soil slope, thereby causing false alarm or missed alarm warning. That is, if the abnormal distribution in the global characterization feature vector of the displacement state of the rock and soil slope affects the judgment logic of the monitor, the system may send an erroneous early warning prompt, affecting the accuracy of emergency response and risk management. Moreover, this abnormal distribution interferes with the accuracy and consistency of the global characterization feature vector of the displacement state of the rock and soil slope, thereby reducing the overall accuracy of the monitoring system, which may lead to instability and unreliability of the monitoring results, making the monitoring and early warning of the displacement state of the rock and soil slope unreliable. Therefore, in order to solve the problem of local abnormal distribution in the global characterization feature vector of the displacement state of the rock and soil slope, in the technical solution of the present application, the global characterization feature vector of the displacement state of the rock and soil slope is activated in the expected activation of the parameter space.

[0047] In a specific example of the present application, the feature optimization unit 151 is used to: perform desired activation of the parameter space of the global characterization feature vector of the displacement state of the rock and soil slope using the following optimization formula to obtain the optimized global characterization feature vector of the displacement state of the rock and soil slope, wherein the optimization formula is:

[0048]

[0049] Among them, v i is the eigenvalue of the ith position of the global characteristic vector of the displacement state of the rock soil slope, α and β represent weight hyperparameters, sigmoid represents the activation function, and v i ′ is the eigenvalue of the ith position of the global characterization eigenvector of the optimized rock and soil slope displacement state.

[0050] Here, in view of the local abnormal distribution in the global characterization feature vector of the displacement state of the rock and soil slope, in the technical solution of the present application, the global characterization feature vector of the displacement state of the rock and soil slope is activated in the parameter space of the expected activation, which activates the feature vector elements that have an important impact on the classification decision in an expected manner by dynamically adjusting the parameter space. When implementing the expected activation, not only the importance of a single feature vector element is considered, but also the interaction and synergy between the feature vector elements are considered. Specifically, by performing the expected activation of the parameter space of the global characterization feature vector of the displacement state of the rock and soil slope, the feature vector elements that contribute more to the model performance can be automatically identified and enhanced, and at the same time, the parameter space can be dynamically adjusted according to the learning state of the current model and the activation of the feature vector elements to maintain the activity and diversity of the feature representation. In this way, the problem of reduced classification performance due to local abnormal distribution can be effectively alleviated, and the sensitivity and adaptability of the model to key features in the data can also be improved, so that a high classification accuracy and robustness can be maintained in various complex data environments.

[0051] Specifically, the early warning decision unit 152 is used to obtain the monitoring result of the optimized geotechnical slope displacement state global characterization feature vector through a slope displacement monitor based on a classifier, and the monitoring result is used to indicate whether to send an early warning prompt. It should be understood that in the field of machine learning, classifiers are mainly used to classify and predict input data. In the technical solution of the present application, a classifier model is used to construct a slope displacement monitor. Specifically, a large number of labeled training data sets are used to train the classifier so that it can learn the characteristic differences between the normal slope displacement state and the abnormal slope displacement state, and establish a decision boundary on whether to send an early warning prompt, so that the classifier can accurately identify the abnormal slope displacement state in practical applications and issue an early warning prompt. Furthermore, the trained classifier is used as a slope displacement monitor in actual early warning decisions. The slope displacement monitor based on the classifier performs feature learning on the global characterization feature vector of the optimized rock and soil slope displacement state, and classifies and maps the global characterization feature vector of the optimized rock and soil slope displacement state according to the decision boundary learned during the training process, thereby outputting a monitoring result indicating whether to send an early warning prompt. In this way, an automated slope displacement early warning function is realized. Once the monitoring result indicates that there is a risk of displacement of the slope, an early warning prompt can be automatically sent to notify relevant personnel in time to take action, thereby reducing potential safety risks.

[0052] In a specific example of the present application, the early warning decision unit 152 is used to: use the classifier-based slope displacement monitor to process the optimized geotechnical slope displacement state global characterization feature vector using the following classification formula to obtain the monitoring result, wherein the classification formula is:

[0053] O=softmax{(W n ,B n ):…:(W1,B1)|X}

[0054] Among them, W1 to W n is the weight matrix, B1 to B n is the bias vector, X is the global characterization feature vector of the optimized rock and soil slope displacement state, softmax is the normalized exponential function, and O is the monitoring result.

[0055] In summary, the rock and soil slope displacement monitoring and early warning system according to the embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to analyze the monitoring images and slope parameter data of the rock and soil slope on each day of history, and mines out the surface deformation characteristics of the rock and soil slope and the global correlation change characteristics of the slope parameters, and characterizes the displacement trend of the rock and soil slope based on the fusion characteristics of the two, so as to intelligently judge whether an early warning is needed. In this way, real-time monitoring and early warning of rock and soil slope displacement can be achieved, the efficiency and accuracy of slope safety monitoring can be improved, and the risk of slope disasters can be reduced.

[0056] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0057] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0058] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0059] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A rock and soil slope displacement monitoring and early warning system, characterized in that: include: The slope monitoring module is used to obtain the slope parameter data and slope monitoring images of the monitored slope on each day of history, wherein the slope parameter data include displacement, inclination, crack width, earth pressure value, rainfall and groundwater level; A slope parameter time series association coding module is used to extract parameter time series association features from the slope parameter data of each historical day to obtain a slope parameter global context semantic association feature vector; A slope state time series change feature extraction module is used to extract the slope state time series change feature of the slope monitoring images of each historical day to obtain a slope state time series change feature vector; A multi-dimensional feature fusion module, used for fusing the slope state temporal variation feature vector and the slope parameter global context semantic association feature vector to obtain a global representation feature vector of the rock and soil slope displacement state; The monitoring result generating module is used to determine the monitoring result based on the global characterization feature vector of the displacement state of the rock and soil slope.

2. The rock and soil slope displacement monitoring and early warning system according to claim 1 is characterized in that: The slope parameter time series association coding module includes: A slope parameter data regularization unit is used to arrange the various slope parameters in the slope parameter data of each historical day into slope parameter time series input vectors according to the time dimension to obtain a sequence of slope parameter time series input vectors; A slope parameter time series feature extraction unit is used to extract time series features from each slope parameter time series input vector in the sequence of slope parameter time series input vectors to obtain a sequence of slope parameter time series associated feature vectors; The parameter correlation encoding unit is used to extract the parameter context correlation features from the sequence of the slope parameter temporal correlation feature vectors to obtain the slope parameter global context semantic correlation feature vector.

3. The rock and soil slope displacement monitoring and early warning system according to claim 2 is characterized in that: The slope parameter time series feature extraction unit is used to: The sequence of the slope parameter time series input vectors is passed through a slope parameter time series feature extractor based on a one-dimensional convolutional layer to obtain a sequence of the slope parameter time series associated feature vectors.

4. The rock and soil slope displacement monitoring and early warning system according to claim 3 is characterized in that: The parameter association encoding unit is used to: The sequence of the slope parameter temporal association feature vectors is passed through an inter-parameter context association encoder based on a Transformer model to obtain the slope parameter global context semantic association feature vector.

5. The rock and soil slope displacement monitoring and early warning system according to claim 4 is characterized in that: The slope state temporal change feature extraction module includes: An image optimization unit, used for performing image preprocessing on the slope monitoring images of each historical day to obtain a sequence of optimized slope monitoring images; The image temporal variation feature extraction unit is used to process the sequence of the optimized slope monitoring images using the image temporal variation feature encoder to obtain the slope state temporal variation feature vector.

6. The rock and soil slope displacement monitoring and early warning system according to claim 5 is characterized in that: The image optimization unit is used for: The slope monitoring images of each historical day are respectively subjected to histogram equalization preprocessing and CLAHE color correction to obtain a sequence of optimized slope monitoring images.

7. The rock and soil slope displacement monitoring and early warning system according to claim 6 is characterized in that: The image temporal change feature encoder is a convolutional neural network model using a temporal attention mechanism.

8. The rock and soil slope displacement monitoring and early warning system according to claim 7 is characterized in that: The monitoring result generating module comprises: A feature optimization unit, used for performing expected activation in parameter space on the global characterization feature vector of the displacement state of the rock and soil slope to obtain an optimized global characterization feature vector of the displacement state of the rock and soil slope; The early warning decision unit is used to pass the optimized global characterization feature vector of the rock and soil slope displacement state through the classifier-based slope displacement monitor to obtain the monitoring result, and the monitoring result is used to indicate whether to send an early warning prompt.

9. The rock and soil slope displacement monitoring and early warning system according to claim 8 is characterized in that: The feature optimization unit is used to: The following optimization formula is used to perform expected activation in parameter space on the global characterization feature vector of the displacement state of the rock and soil slope to obtain the optimized global characterization feature vector of the displacement state of the rock and soil slope, wherein the optimization formula is: Among them, v i is the eigenvalue of the ith position of the global characteristic vector of the displacement state of the rock soil slope, α and β represent weight hyperparameters, sigmoid represents the activation function, and v i ′ is the eigenvalue of the ith position of the global characterization eigenvector of the optimized rock and soil slope displacement state.

10. The rock and soil slope displacement monitoring and early warning system according to claim 9 is characterized in that: The early warning decision unit is used to: The classifier-based slope displacement monitor is used to process the optimized geotechnical slope displacement state global characterization feature vector using the following classification formula to obtain the monitoring result, wherein the classification formula is: O=softmax{(W n ,B n ):…:(W1,B1)|X} Among them, W1 to W n is the weight matrix, B1 to B n is the bias vector, X is the global characterization feature vector of the optimized rock and soil slope displacement state, softmax is the normalized exponential function, and O is the monitoring result.

Images