Automatic monitoring system for catastrophe of high slope of expressway

Abstract

The invention relates to a mode recognition method or device, and particularly discloses an automatic monitoring system for catastrophe of a high slope of an expressway, and the automatic monitoring system comprises a data collection and feature extraction module, a deep learning training module, a risk assessment model construction module, an early warning notification module, and an automatic early warning and processing suggestion module. A complete risk assessment framework is formed through the risk data collection and basic credibility grading module, the Bayesian network algorithm calculation and weight weighting module and the risk level calculation module, the system is high in systematicness, the basic credibility of each risk factor after adjustment is calculated through the Bayesian network algorithm, and the risk assessment efficiency is improved. And weight weighting is carried out, so that a risk assessment result is more accurate, and high, medium and low risk levels can be effectively distinguished.

Description

Technical Field

[0001] The present invention belongs to a method or device for identifying patterns, and particularly relates to an automatic monitoring system for highway high-slope disasters. Background Art

[0002] Highways are a modern means of road transportation, and their development history can be traced back to the 1950s. In China, the construction of highways began in 1988. After more than 30 years of development, a relatively complete highway network has been formed. The development of highways has had a great impact on economic and social development. It not only improves people's travel efficiency, shortens geographical distances, but also promotes regional economic development and urban construction. At the same time, highways also bring many environmental and social problems, such as noise, air pollution, traffic accidents, etc., requiring the government and relevant departments to take effective measures to solve them. In the future, with the development of technology and the changing needs of people for travel modes, the development of highways will also face new challenges and opportunities. This includes exploring innovations in aspects such as intelligence, digitization, and sustainable development to better meet people's travel needs.

[0003] Highway disease slopes refer to slopes beside highways that have suffered a series of different types of diseases for various reasons. These diseases may include, but are not limited to: landslides, collapses, gullies, cracks, looseness, etc. These diseases pose a threat to the normal use and driving safety of highways, and may even lead to traffic accidents in severe cases. Therefore, timely monitoring and maintenance of highway slope diseases are important links in ensuring the safe and unobstructed operation of highways.

[0004] Currently, for the monitoring of highway slope diseases, methods such as conventional surveying, GPS surveying, and sensor surveying are mainly used. Among them, conventional surveying requires the layout of monitoring points on the deformed body. This method has certain limitations, such as limited number of monitoring points, low measurement efficiency, and inaccurate data caused by factors such as weather. These problems will have a very great impact on the monitoring of highway slope diseases. To solve these problems, more advanced, efficient, and accurate monitoring methods and technologies need to be sought to improve monitoring efficiency and data accuracy.

[0005] A Chinese patent with the publication number CN115797784A discloses a geological monitoring method and monitoring system for highway slopes, specifically discloses: collecting multi-view images of the highway slope and the corresponding feature points of the multi-view images; inputting the multi-view images and the feature points into neural network model one to output a two-dimensional feature map; inputting the multi-view images, the feature points and the two-dimensional feature map into neural network model two to output a three-dimensional image; performing texture mapping processing on the three-dimensional image to obtain a real scene map of the highway slope; identifying and judging the safety condition of the highway slope according to the real scene map of the highway slope, and performing corresponding maintenance. The three-dimensional image of the slope is obtained relatively clearly, but the risk is not predicted and warned, and it is not applicable to scenarios with relatively fast vehicle speeds such as highways.

[0006] A Chinese patent with the publication number CN113723446A discloses a geological disaster monitoring and warning method, device, computer device and storage medium, specifically discloses: obtaining monitoring data of a plurality of different sensors arranged on the same slope; each of the monitoring data includes rainfall monitoring data and deep displacement monitoring data; inputting each of the monitoring data into a monitoring model for geological disaster monitoring to obtain all monitoring results; performing weighted processing on all the monitoring results to obtain a final monitoring result; judging whether there is a landslide risk in the final monitoring result; if there is a landslide risk in the final monitoring result, sending a warning signal to the terminal; wherein, the monitoring model is obtained by training a deep forest algorithm with the monitoring data obtained by a number of sensors with landslide category and non-landslide category labels as a sample set, and the monitoring model is evaluated using the ROC curve. However, this existing technology relies on the monitoring data as a sample set to train the deep forest algorithm, and the monitoring model is evaluated using the ROC curve, with a relatively high degree of dependence on historical records. When the data volume is small or there are historical deviations, the training sample set will deviate, resulting in judgment distortion, and the current data with judgment distortion will return to the sample set, causing a cyclic error.

[0007] The Chinese patent with the publication number CN114812528B discloses an automatic monitoring system applied to the damaged slopes of expressways, specifically including: a data acquisition module: used to scan the expressway slopes through a three-dimensional laser scanner to obtain scanned point cloud data; a model construction module: constructing a slope monitoring model based on the scanned point cloud data, and analyzing the damaged state of the slopes according to the slope monitoring model to obtain an analysis result; an early warning processing module: when the analysis result shows that there is a slope disease, an early warning is processed. Through the point cloud data obtained by the three-dimensional laser scanning technology of the present invention, the three-dimensional spatial position information, reflectivity information, color texture information, etc. of the target object can be quickly, accurately, and comprehensively recorded. Therefore, not only the measurement efficiency can be improved, but also the slope disease conditions can be monitored by constructing a model based on the point cloud data, and the early warning of the emerging disease conditions can be carried out in time, thus avoiding resource losses and personal injuries. However, the implementation of this prior art depends on the mapping results of visible light or near-visible light, has high requirements for the mapping quality, depends on the professional level of the mapping personnel and the optical environment at the mapping site, is greatly affected by personnel and the outside, and has large errors.

[0008] In response to the problems in the related art, no effective solution has been proposed yet. Summary of the Invention

[0009] Object of the Invention: To provide an automatic monitoring system for disasters of high slopes of expressways to overcome the above-mentioned technical problems existing in the related prior art.

[0010] Technical Solution: An automatic monitoring system for disasters of high slopes of expressways, including: a data collection and feature extraction module, a deep learning training module, a risk assessment model construction module, a warning notification module, and an automatic warning and processing suggestion module;

[0011] The data collection and feature extraction module is used to use remote sensing technology to monitor the expressway slopes in real time, collect multi-dimensional data, and extract key features;

[0012] The deep learning training module is used to input the key features into a deep learning model for training to predict the evolution of slope diseases;

[0013] The risk assessment model construction module is used to construct a slope disease risk assessment model based on the prediction results of the deep learning model and combined with the comprehensive factors around the expressway;

[0014] The warning notification module is used to formulate corresponding warning levels for high slope diseases according to the risk assessment results and send them to the PC side or the mobile terminal;

[0015] The automatic early warning and treatment suggestion module is used to set the threshold of the early warning standard. When the occurrence probability and severity reach the threshold of the early warning standard, it automatically triggers an early warning, generates a detection report and formulates treatment measures.

[0016] Furthermore, the deep learning training module includes: a data acquisition module, a prediction model construction and training module, a cross-validation module, a model evaluation and selection module, and a result analysis module;

[0017] The data acquisition module is used to obtain the historical data of slope disease monitoring, and the historical data includes key features and disease evolution situations;

[0018] The prediction model construction and training module is used to select the RBF neural network model as the prediction model for the high slopes of highway diseases, take the key features as the number of nodes in the input layer of the network, and the disease evolution situation as the number of nodes in the output layer of the network, and use the data parallel method to train the number of nodes in the input layer;

[0019] The cross-validation module is used to train the prediction model using the training set, observe the error on the validation set at the same time, and judge whether the model has overfitting phenomenon. For the overfitting situation, by reducing or controlling the training cycle and stopping the training of the network before the inflection point of the number of nodes, the training effect can be achieved;

[0020] The model evaluation and selection module is used to evaluate the fitting degree of the optimal individual in the prediction model to the number of key features, obtain the optimal RBF neural network model, and predict the disease evolution situation;

[0021] The result analysis module is used to compare with the actual disease evolution situation according to the prediction result and analyze the accuracy of the optimal RBF neural network model.

[0022] Furthermore, the prediction model construction and training module includes: a network structure initialization module, a dataset division module, an RBF neural network model training module, and an RBF neural network model monitoring module;

[0023] The network structure initialization module initializes the RBF neural network model, takes the number of key features as the number of nodes in the input layer of the RBF neural network model, and the disease evolution situation as the number of nodes in the output layer of the network. According to the number of input and output nodes, it initializes the structure of the RBF neural network model and determines the number of nodes in the hidden layer;

[0024] The dataset division module is used to divide the key feature dataset into a training set and a validation set according to a set ratio;

[0025] The RBF neural network model training module is used to determine the centers of the number of hidden layer nodes through a clustering algorithm, calculate the activation values on each number of hidden layer nodes, and update the network connection weights by the least squares method;

[0026] The RBF neural network model monitoring module is used to set thresholds according to the actual requirements and application scenarios of highway disease high slope prediction, calculate the mean square error on the training set, and stop training if the set threshold is reached.

[0027] Further, the risk assessment model construction module includes: an RBF neural network model prediction module, a target importance assignment module, a factor weighted score calculation module, and a slope disease risk assessment value calculation module;

[0028] The RBF neural network model prediction module is used to predict the occurrence probability and severity of slope diseases using the optimal RBF neural network model;

[0029] The target importance assignment module is used to calculate the factor values of slope disease risk types, traffic flow, vehicle types, and vehicle speeds respectively according to the prediction results combined with historical data, and assign their respective target importance;

[0030] The factor weighted score calculation module is used to set a hierarchical protection risk assessment model according to the factor values of slope disease risk types, traffic flow, vehicle types, and vehicle speeds and the corresponding target importance, and calculate the weighted score of each factor;

[0031] The slope disease risk assessment value calculation module is used to add up the weighted scores of all factors to obtain the slope disease risk assessment value.

[0032] Further, the expression of the hierarchical protection risk assessment model is:

[0033] R = {X, W, W * , S * , M}

[0034] W = (w A , w B , w C , w D )

[0035] In the formula, X is the risk type of highway slope diseases;

[0036] W is the risk category weight vector of highway slope diseases;

[0037] W * is the risk factor weight matrix of highway slope diseases;

[0038] S* is the set of fuzzy evaluation comments;

[0039] M is the evaluation matrix for highway slope diseases;

[0040] w A is the weight vector of slope disease risk types;

[0041] w B is the weight vector of traffic flow;

[0042] w C is the weight vector of vehicle types;

[0043] w D is the weight vector of vehicle speeds.

[0044] Furthermore, the target importance assignment module includes: a factor value calculation module, a hazard assessment module, a target hierarchy module, an importance assessment module, and a target importance calculation module;

[0045] The factor value calculation module is used to normalize historical data and calculate the factor values of slope disease risk types, traffic flow, vehicle types, and vehicle speeds;

[0046] The hazard assessment module is used to determine decision-making goals for slope disease risk types, traffic flow, vehicle types, and vehicle speeds according to the occurrence probability and severity of hazards;

[0047] The target hierarchy module is used to split the decision-making goal into several levels according to the decision-making goal to form a target hierarchy;

[0048] The importance assessment module is used to make pairwise comparisons between factors at each level to form a judgment matrix and evaluate the importance degree between factors;

[0049] The target importance calculation module is used to add up the values in each column of the judgment matrix to obtain the sum of the column, and then divide each element by the sum of its corresponding column to obtain the corresponding target importance of each factor.

[0050] Furthermore, the slope disease risk assessment value calculation module includes: a risk category and factor definition module, a risk data collection and analysis module, and a comprehensive assessment module;

[0051] The risk category and factor definition module is used to define the weight vectors of slope disease risk types, traffic flow, vehicle types, and vehicle speeds, and set corresponding weight vectors for each risk category;

[0052] The risk data collection and analysis module is used to collect data related to each risk category, assign a basic credibility to each factor according to the risk level, and calculate the adjusted basic credibility of each risk factor;

[0053] The comprehensive evaluation module is used to calculate the basic credibility of each risk category at the risk level, calculate the belief function and the likelihood function, and obtain the evaluation result.

[0054] Further, the risk data collection and analysis module includes: a risk data collection and basic credibility grading module, a Bayesian network algorithm calculation and weight weighting module, and a risk level calculation module;

[0055] The risk data collection and basic credibility grading module is used to collect data related to each risk category and divide the basic credibility of each risk into three levels: high, medium, and low according to the risk level.

[0056] The Bayesian network algorithm calculation and weight weighting module is used to calculate the adjusted basic credibility of each risk factor using the Bayesian network algorithm and perform weight weighting.

[0057] The risk level calculation module is used to calculate the final risk level of each risk according to the weight and the divided level of the basic credibility.

[0058] Further, the early warning notification module includes: a level division module, an information push module, an information interaction module, and an information feedback module;

[0059] The level division module is used to divide the high-risk area into warning levels of red, orange, and yellow according to the slope disease risk assessment result, and sort out the warning level, geographical location, and time information;

[0060] The information push module is used to design the message format for the PC side and the mobile terminal, simplify the message content, and automatically notify and push;

[0061] The information interaction module is used to develop a program or function module for receiving warning messages in the PC side or the mobile terminal, and receive, parse, and display the warning information;

[0062] The information feedback module is used to set up a user feedback function to collect the opinions and suggestions of users on the warning information.

[0063] Further, the automatic early warning and treatment suggestion module includes: an early warning system setting module, a detection report production module, and a treatment measure formulation module;

[0064] The early warning system setting module is used to determine the monitoring indicators and thresholds of the occurrence probability and severity, and set up the early warning system;

[0065] The detection report generation module is used to generate a detection report by a computer program after the early warning system automatically triggers an early warning. The detection report includes the monitoring data of the structure, the reason for the early warning, and the early warning suggestions.

[0066] The treatment measure formulation module is used to formulate corresponding treatment measures according to the information in the detection report.

[0067] Beneficial effects:

[0068] 1. The present invention uses an RBF neural network model for prediction, which has strong non-linear fitting ability and can capture the complex relationship between key features and disease evolution, thereby improving the prediction accuracy. By observing the errors on the training set and the validation set through the cross-validation module, for the overfitting phenomenon, the model structure can be adjusted in time or the training cycle can be controlled to avoid overfitting. The model evaluation and selection module evaluates the optimal RBF neural network model, ensuring that the selected model has good fitting degree and reliability. The entire model includes multiple modules such as data acquisition, model construction and training, cross-validation, model evaluation and selection, and result analysis, forming a complete prediction system.

[0069] 2. The present invention forms a complete risk assessment framework through the risk data collection and basic credibility grading module, the Bayesian network algorithm calculation and weight weighting module, and the risk level calculation module, which has strong systematicness. By calculating the adjusted basic credibility of each risk factor through the Bayesian network algorithm and performing weight weighting, the risk assessment result is more accurate, and it can effectively distinguish different risk levels of high, medium, and low. The information push module can automatically notify and push early warning information for the PC side and the mobile terminal, improving the real-time nature of information transmission. The level division module visually divides the high-risk area into warning levels of red, orange, and yellow, which is easy for users to understand and take corresponding measures. The information interaction module and the information feedback module allow users to receive early warning messages on the PC side or the mobile terminal and provide a feedback function, enhancing the interactivity between the system and the users.

[0070] 3. The entire early warning system in the present invention includes multiple modules, such as a risk category and factor definition module, a risk data collection and analysis module, a comprehensive evaluation module, etc., forming a complete risk assessment and early warning framework. By using the Bayesian network algorithm to calculate the adjusted basic credibility of each risk factor and performing weight weighting, the risk assessment result is more accurate, and it can effectively distinguish different risk levels of high, medium, and low. The information push module can automatically notify and push early warning information for the PC side and mobile terminals, improving the real-time nature of information transmission. According to the risk assessment result of the slope disease, corresponding early warning levels are formulated, and high-risk areas are divided into red, orange, and yellow early warning levels. The early warning levels, geographical locations, and time information are sorted out, which helps to improve the accuracy and reliability of early warning. The information interaction module is used to develop a program or functional module for receiving early warning messages in the PC side or mobile terminal, and receive, parse, and display early warning information, which enables users to conveniently obtain early warning information. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0072] Figure 1 It is a schematic block diagram of an automatic monitoring system for highway high slope disasters according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0073] In the following description, a large number of specific details are given to provide a more thorough understanding of the present invention. However, it is obvious to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present invention, some well-known technical features in the art are not described.

[0074] According to an embodiment of the present invention, an automatic monitoring system for highway high slope disasters is provided.

[0075] As Figure 1 shown, the automatic monitoring system for highway high slope diseases according to an embodiment of the present invention includes: a data collection and feature extraction module 1, a deep learning training module 2, a risk assessment model construction module 3, an early warning notification module 4, and an automatic early warning and treatment suggestion module 5;

[0076] The data collection and feature extraction module 1 is used to use remote sensing technology to monitor the highway slope in real time, collect multi-dimensional data, and extract key features.

[0077] It should be noted that remote sensing technology can provide fast, large-scale, and continuous observation information, which is of great significance for the stability assessment and disaster warning of highway slopes. The following are the steps to use remote sensing technology to conduct real-time monitoring of highway slopes, collect multi-dimensional data, and extract key features: According to the monitoring objectives and accuracy requirements, select appropriate remote sensing sensors (such as optical sensors, radar sensors, laser scanners, etc.) and platforms (such as satellites, unmanned aerial vehicles, ground vehicles, etc.). Use the selected remote sensing equipment to conduct real-time observation of the highway slope to obtain the original remote sensing data. These data may include: digital orthophoto images, elevation data, multi-spectral data, synthetic aperture radar data, etc. Conduct preprocessing operations on the original remote sensing data, such as noise removal, radiometric calibration, geometric correction, image registration, etc., to eliminate the influence of systematic errors and environmental factors on data quality. Extract key features of the slope from the preprocessed remote sensing data, such as elevation, slope, aspect, landform, crack distribution, vegetation cover, etc. Integrate and comprehensively analyze the extracted multi-dimensional feature information to evaluate the slope stability and potential disaster risks. Real-time monitor the stability changes of the slope, and issue an alarm in a timely manner when abnormal conditions are found, providing a scientific basis for disaster prevention and mitigation.

[0078] The deep learning training module 2 is used to input the key features into the deep learning model for training to predict the evolution of slope diseases.

[0079] Preferably, the deep learning training module 2 includes: a data acquisition module, a prediction model construction and training module, a cross-validation module, a model evaluation and selection module, and a result analysis module;

[0080] The data acquisition module is used to acquire historical data on slope disease monitoring, and this historical data includes key features (such as geological conditions, climate conditions, etc.) and disease evolution situations (such as cracks, landslides, etc.);

[0081] The prediction model construction and training module is used to select the RBF neural network model as the prediction model for high slopes of highway diseases, use the key features as the number of nodes in the input layer of the network, and the disease evolution situation as the number of nodes in the output layer of the network, and adopt the data parallel method to train the number of nodes in the input layer;

[0082] The cross-validation module is used to train the prediction model using the training set, and at the same time observe the error on the validation set and judge whether the model has an overfitting phenomenon. For the case of overfitting, stop training the network before the inflection point of the number of nodes by reducing or controlling the training cycle to achieve the training effect;

[0083] The model evaluation and selection module is used to evaluate the fitting degree of the optimal individual in the prediction model to the number of key features, obtain the optimal RBF neural network model, and predict the disease evolution situation.

[0084] The result analysis module is used to compare with the actual disease evolution situation according to the prediction result and analyze the accuracy of the optimal RBF neural network model.

[0085] Preferably, the prediction model construction and training module includes: a network structure initialization module, a data set division module, an RBF neural network model training module, and an RBF neural network model monitoring module.

[0086] The network structure initialization module initializes the RBF neural network model, takes the number of key features as the number of nodes in the input layer of the RBF neural network model, takes the disease evolution situation as the number of nodes in the network output layer, initializes the structure of the RBF neural network model according to the number of input and output nodes, and determines the number of nodes in the hidden layer.

[0087] The data set division module is used to divide the key feature data set into a training set and a validation set according to a set ratio.

[0088] The RBF neural network model training module is used to determine the centers of the number of nodes in the hidden layer through a clustering algorithm, calculate the activation values on each number of nodes in the hidden layer, and update the network connection weights by the least squares method.

[0089] The RBF neural network model monitoring module is used to set a threshold according to the actual requirements and application scenarios of highway disease high slope prediction, calculate the mean square error on the training set, and stop training if the set threshold is reached.

[0090] The risk assessment model construction module 3 is used to construct a slope disease risk assessment model based on the prediction result of the deep learning model and combined with the comprehensive factors around the highway.

[0091] Preferably, the risk assessment model construction module 3 includes: an RBF neural network model prediction module, a target importance assignment module, a factor weighted score calculation module, and a slope disease risk assessment value calculation module.

[0092] The RBF neural network model prediction module is used to predict the occurrence probability and severity of slope diseases using the optimal RBF neural network model.

[0093] The target importance assignment module is used to calculate the factor values of slope disease risk types, traffic flow, vehicle types, and vehicle speeds respectively according to the prediction result combined with historical data, and assign their respective target importance.

[0094] The factor weighted score calculation module is used to calculate the weighted score of each factor according to the factor values of slope disease risk type, traffic flow, vehicle type, and vehicle speed and the corresponding target importance setting level protection risk assessment model;

[0095] The slope disease risk assessment value calculation module is used to add up the weighted scores of all factors to obtain the slope disease risk assessment value.

[0096] Preferably, the expression of the level protection risk assessment model is:

[0097] R = {X, W, W * , S * , M}

[0098] W = (w A , w B , w C , w D )

[0099] Wherein, X is the risk type of highway slope disease;

[0100] W is the risk category weight vector of highway slope disease;

[0101] W * is the risk factor weight matrix of highway slope disease;

[0102] S * is the fuzzy comment set;

[0103] M is the evaluation matrix of highway slope disease;

[0104] w A is the weight vector of slope disease risk type;

[0105] w B is the weight vector of traffic flow;

[0106] w C is the weight vector of vehicle type;

[0107] w D is the weight vector of vehicle speed.

[0108] Specifically, by establishing a hierarchical evaluation system based on level protection and using a fuzzy evaluation method based on evidence theory to process the fuzzy values in the evaluation, the evaluation results are finally quantified.

[0109] Preferably, the target importance assignment module includes: a factor value calculation module, a hazard assessment module, a target hierarchy module, an importance assessment module, and a target importance calculation module;

[0110] The factor value calculation module is used to normalize historical data and calculate the factor values of slope disease risk types, traffic flow, vehicle types, and vehicle speeds;

[0111] The hazard assessment module is used to determine decision-making objectives for slope disease risk types, traffic flow, vehicle types, and vehicle speeds according to the occurrence probability and severity of hazards;

[0112] The target hierarchy module is used to split the decision-making objective into several levels according to the decision-making objective to form a target hierarchy;

[0113] The importance assessment module is used to make pairwise comparisons between factors at each level to form a judgment matrix and evaluate the importance degree between factors;

[0114] The target importance degree calculation module is used to add up the values in each column of the judgment matrix to obtain the sum of the column, and then divide each element by the sum of its corresponding column to obtain the corresponding target importance degree of each factor.

[0115] Preferably, the slope disease risk assessment value calculation module includes: a risk category and factor definition module, a risk data collection and analysis module, and a comprehensive assessment module;

[0116] The risk category and factor definition module is used to define the weight vectors of slope disease risk types, traffic flow, vehicle types, and vehicle speeds, and set corresponding weight vectors for each risk category;

[0117] The risk data collection and analysis module is used to collect data related to each risk category, assign basic credibility to each factor according to the risk level, and calculate the adjusted basic credibility of each risk factor;

[0118] The comprehensive assessment module is used to calculate the basic credibility of each risk category at the risk level, and calculate the belief function and the likelihood function to obtain the assessment result.

[0119] Preferably, the risk data collection and analysis module includes: a risk data collection and basic credibility grading module, a Bayesian network algorithm calculation and weight weighting module, and a risk level calculation module;

[0120] The risk data collection and basic credibility grading module is used to collect data related to each risk category, and divide the basic credibility of each risk into three levels: high, medium, and low according to the risk level.

[0121] The Bayesian network algorithm calculation and weight weighting module is used to calculate the adjusted basic credibility of each risk factor using the Bayesian network algorithm and perform weight weighting.

[0122] The risk level calculation module is used to calculate the final risk level of each risk according to the divided levels of weights and basic credibility.

[0123] The early warning notification module 4 is used to formulate corresponding early warning levels for high slope diseases according to the risk assessment results, and send them to the PC side or mobile terminal.

[0124] Preferably, the early warning notification module 4 includes: a level division module, an information push module, an information interaction module, and an information feedback module;

[0125] The level division module is used to divide high-risk areas into red, orange, and yellow early warning levels according to the risk assessment results of slope diseases, and sort out early warning level, geographical location, and time information;

[0126] The information push module is used to design message formats for the PC side and mobile terminal, simplify the message content, and automatically notify and push;

[0127] The information interaction module is used to develop a program or functional module for receiving early warning messages in the PC side or mobile terminal, and receive, parse, and display early warning information;

[0128] The information feedback module is used to set up a user feedback function to collect users' opinions and suggestions on early warning information.

[0129] It should be noted that the risk data collection and analysis module forms a complete risk assessment framework through the risk data collection and basic credibility grading module, the Bayesian network algorithm calculation and weight weighting module, and the risk level calculation module, which has strong systematicness. By calculating the adjusted basic credibility of each risk factor through the Bayesian network algorithm and performing weight weighting, the risk assessment results are more accurate, and it can effectively distinguish different risk levels of high, medium, and low. The information push module can automatically notify and push early warning information for the PC side and mobile terminal, improving the timeliness of information transmission.

[0130] The automatic early warning and treatment suggestion module 5 is used to set the threshold of the early warning standard. When the occurrence probability and severity reach the threshold of the early warning standard, it automatically triggers an early warning, generates a detection report, and formulates treatment measures.

[0131] Preferably, the automatic early warning and treatment suggestion module 5 includes: an early warning system setting module, a detection report production module, and a treatment measure formulation module;

[0132] The early warning system setting module is used to determine the monitoring indicators and thresholds of the occurrence probability and severity, and set up the early warning system;

[0133] The detection report generation module is used to generate a detection report by a computer program after the early warning system automatically triggers an early warning. The detection report includes the monitoring data of the structure, the reason for the early warning, and the early warning suggestions.

[0134] The treatment measure formulation module is used to formulate corresponding treatment measures according to the information in the detection report.

[0135] Specifically, for better understanding by those skilled in the art, in the related embodiments of the present application, technical terms or some nouns that the present application may involve are now explained: The RBF neural network model prediction module (Radial Basis Function) is a special feedforward artificial neural network. The RBF neural network has only one hidden layer, and the role of this layer is to perform a non-linear mapping on the input signal. Each neuron in the hidden layer has a corresponding radial basis function, such as a Gaussian kernel function, a polynomial kernel function, etc. The number of neurons in the hidden layer determines the complexity and fitting ability of the network. As the activation function of the hidden layer neurons, the radial basis function has a local response characteristic, that is, its response decreases as the distance from the center point increases. Common radial basis functions include Gaussian radial basis functions, polynomial radial basis functions, etc.

[0136] In summary, by means of the above technical solutions of the present invention, the present invention uses an RBF neural network model for prediction, which has strong non-linear fitting ability and can capture the complex relationship between key features and disease evolution, thereby improving the prediction accuracy. The cross-validation module observes the errors on the training set and the validation set. For the overfitting phenomenon, the model structure can be adjusted in time or the training cycle can be controlled to avoid overfitting. The model evaluation and selection module evaluates the optimal RBF neural network model, ensuring that the selected model has good fitting degree and reliability. The entire model includes multiple modules such as data acquisition, model construction and training, cross-validation, model evaluation and selection, and result analysis, forming a complete prediction system. The present invention forms a complete risk assessment framework through the risk data collection and basic credibility grading module, the Bayesian network algorithm calculation and weight weighting module, and the risk level calculation module, which has strong systematicness. By calculating the adjusted basic credibility of each risk factor through the Bayesian network algorithm and performing weight weighting, the risk assessment result is more accurate, and it can effectively distinguish different risk levels of high, medium, and low. The information push module can automatically notify and push warning information for the PC side and the mobile terminal, improving the timeliness of information transmission. The level division module intuitively divides the warning levels of high-risk areas into red, orange, and yellow, which is easy for users to understand and take corresponding measures. The information interaction module and the information feedback module allow users to receive warning messages on the PC side or the mobile terminal and provide a feedback function, enhancing the interactivity between the system and the users. The entire warning system in the present invention includes multiple modules such as the risk category and factor definition module, the risk data collection and analysis module, the comprehensive evaluation module, etc., forming a complete risk assessment and warning framework. By calculating the adjusted basic credibility of each risk factor through the Bayesian network algorithm and performing weight weighting, the risk assessment result is more accurate, and it can effectively distinguish different risk levels of high, medium, and low. The information push module can automatically notify and push warning information for the PC side and the mobile terminal, improving the timeliness of information transmission. According to the slope disease risk assessment result, the corresponding warning level is formulated, and the high-risk area is divided into red, orange, and yellow warning levels, and the warning level, geographical location, and time information are sorted out, which helps to improve the accuracy and reliability of the warning. The information interaction module is used to develop a program or functional module for receiving warning messages in the PC side or the mobile terminal, and receive, parse, and display the warning information, which enables users to conveniently obtain the warning information.

[0137] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An automatic monitoring system for high slope disasters on highways, characterized in that: The automatic monitoring system includes: data collection and feature extraction module, deep learning training module, risk assessment model construction module, warning notification module and automatic warning and processing suggestion module; The data collection and feature extraction module is used to monitor the highway slope in real time using remote sensing technology, collect multi-dimensional data, and extract key features; The deep learning training module is used to input key features into the deep learning model for training to predict the evolution of slope diseases; The risk assessment model building module is used to build a slope disease risk assessment model based on the prediction results of the deep learning model and combined with comprehensive factors around the expressway; The warning notification module is used to formulate corresponding warning levels for high slope diseases according to the risk assessment results, and send them to the PC or mobile terminal; The automatic warning and treatment suggestion module is used to set the threshold of the warning standard. When the probability and severity of occurrence reach the threshold of the warning standard, the warning is automatically triggered, a detection report is generated, and treatment measures are formulated.

Images