Rural highway potential safety hazard checking system and method

Abstract

The invention discloses a rural highway potential safety hazard checking system and method, particularly relates to the technical field of highway potential safety hazard checking, and is used for solving the problem of low safety management efficiency in a rural highway potential safety hazard checking process. By deploying the stress sensor, the displacement sensor and the acceleration sensor, different areas of a rural highway are monitored in real time, data acquisition frequency optimization and abnormal value detection are combined, data precision is ensured, and data are transmitted through a wireless network. And carrying out feature extraction on the collected data, constructing roadbed settlement and traffic load features, inputting the features into a nonlinear autoregression model for multi-step prediction, accurately predicting a future roadbed settlement trend, identifying a settlement risk in advance and providing early warning. And in combination with simulation verification information, the prediction precision and robustness of the model are further improved. By analyzing different generated signals and dynamically adjusting strategies, the system is ensured to flexibly cope with different settlement risks, and the safety management efficiency of rural roads is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of highway safety hazard inspection, and more specifically, to a rural highway safety hazard inspection system and method. Background Art

[0002] Rural roads are the foundation of the transportation network in rural areas and play an important role in promoting agricultural production and improving farmers' living conditions. Especially in areas where geological disasters occur frequently, the safety of roads is more easily affected. Common rural road safety hazards include road damage, roadbed settlement, landslides, road cracks, sharp turns and steep slopes, lack of traffic signs and markings, etc. These hazards will not only lead to frequent traffic accidents, but also affect the travel safety and economic development of rural residents. Therefore, it is particularly important to regularly conduct safety hazard inspections and discover and rectify potential problems.

[0003] Deficiencies in existing technologies: Structural instability caused by uneven road settlement is a potential safety hazard on rural roads. It not only affects the service life of roads, but also directly threatens driving safety. Uneven settlement refers to the phenomenon that the degree of settlement of different parts of the roadbed is inconsistent. The roadbeds in different areas will experience different degrees of settlement due to differences in soil properties, load distribution, drainage conditions and other factors, resulting in uneven road surfaces. This settlement phenomenon will gradually develop into structural instability, that is, the bearing capacity and stability of the road will be reduced. Regular inspections of rural roads can temporarily alleviate some problems, but regular inspections often have lags and limitations. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a rural road safety hazard inspection system and method to solve the problem of inaccurate rural road safety hazard inspection in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A method for checking safety hazards on rural roads comprises the following steps: Deploy stress sensors, displacement sensors, and acceleration sensors to monitor different areas of rural roads, optimize the data collection frequency of sensors, detect abnormal values ​​of collected data, and transmit data through wireless networks; Extract features from the collected data to obtain roadbed settlement features and traffic load features, and combine them into feature vectors. Input the feature vectors into a nonlinear autoregressive model to predict roadbed settlement, determine the results of multi-step predicted roadbed settlement, and identify settlement risks. Analyze the multi-step prediction process of roadbed settlement and obtain simulation verification information generated during the multi-step prediction verification process; The simulation verification information is analyzed and different signals are generated, and corresponding strategy adjustments are made according to the different signals generated.

[0006] In a preferred embodiment, stress sensors, displacement sensors and acceleration sensors are deployed to monitor different areas of rural roads, the data acquisition frequency of the sensors is optimized and abnormal value detection of the collected data is performed, and data transmission is performed through a wireless network. The specific process is as follows: Areas where sensors are deployed include low-lying areas, steep slopes, and chronically overloaded sections; Use stress sensors to monitor the changes in the bearing capacity of roadbed soil and detect the pressure distribution of the roadbed under different loads; Use displacement sensors to detect the displacement of roadbed settlement and identify the phenomenon of uneven settlement; Use acceleration sensors to monitor foundation vibration and identify settlement caused by foundation vibration; The equipment data includes the data collected by monitoring the displacement sensor and the data collected by monitoring the laser ranging device through the optical path sensor; After the sensor is installed, the initial sensor readings are recorded as a benchmark for settlement monitoring, and the sensitivity and calibration coefficient of the sensor are adjusted. The sensor collected data is transmitted to the central management platform via the wireless network for storage and processing.

[0007] In a preferred embodiment, the sensor collected data is transmitted to the central management platform via a wireless network for storage and processing. The specific process is as follows: Set the data collection frequency according to the type of sensor and monitoring requirements; The data collected by the sensor is used to calculate the settlement value or stress change. The stress sensor collects the pressure value of the roadbed, the displacement sensor detects the vertical settlement of the roadbed, and the acceleration sensor records the vibration acceleration of the roadbed. The collected data is tested for outliers, and the mean and standard deviation are calculated. A fixed-size sliding window is used to calculate the change in the adjacent data of each data point. If the change exceeds the set threshold, the data is considered abnormal and removed, and linear interpolation is performed in the case of data loss.

[0008] In a preferred embodiment, feature extraction is performed on the collected data to obtain roadbed settlement features and traffic load features, which are combined into feature vectors. The feature vectors are input into a nonlinear autoregressive model to predict roadbed settlement, and the roadbed settlement results of multi-step predictions are determined to identify settlement risks. The specific process is as follows: The data are processed using feature extraction formulas, traffic load features are extracted using the traffic load cumulative effect formula, roadbed settlement features are extracted using the roadbed settlement cumulative rate formula, and the extracted features are preprocessed; The extracted roadbed settlement characteristics and traffic load characteristics are combined into a feature vector and input into a nonlinear autoregressive model; The input layer of the nonlinear autoregressive model is composed of a feature vector sequence consisting of settlement and traffic load in historical data. The hidden layer uses a nonlinear activation function to perform nonlinear transformation on the input features, and the output layer outputs the predicted settlement at the future moment. The data set is divided into training set, validation set and test set, and the mean square error is used as the loss function during training; Use the Adam optimizer to update the model parameters and use the Bayesian optimization algorithm to tune the model's hyperparameters; The nonlinear autoregressive model is used to make single-step and multi-step predictions of sedimentation.

[0009] In a preferred embodiment, the multi-step prediction process of the roadbed settlement is analyzed to obtain simulation verification information generated in the multi-step prediction verification process. The specific process is as follows: Acquire simulation verification information generated during the multi-step prediction verification process, the simulation verification information includes fitting related information and error trend information; Fitting-related information includes the dynamic goodness-of-fit index, and error trend information includes the prediction error accumulation index; The obtained dynamic goodness of fit index and prediction error cumulative index are combined to generate a prediction matching coefficient; The dynamic goodness of fit index is directly proportional to the prediction matching coefficient, and the prediction error accumulation index is inversely proportional to the prediction matching coefficient.

[0010] In a preferred embodiment, the simulation verification information is analyzed and different signals are generated, and corresponding strategies are adjusted according to the generated different signals. The specific steps are as follows: Compare the generated prediction matching coefficient with the set prediction matching threshold; If the prediction matching coefficient is greater than or equal to the prediction matching threshold, a multi-step prediction stability signal is generated, and no additional adjustment is required, and the prediction result can be used as normal data; If the prediction matching coefficient is less than the prediction matching threshold, a multi-step prediction error signal is generated, and the model needs to be optimized in time to determine the source of the prediction error.

[0011] A rural highway safety hazard inspection system, used to implement the rural highway safety hazard inspection method, comprising: The sensor acquisition module is used to deploy stress sensors, displacement sensors and acceleration sensors to monitor different areas of rural roads, optimize the data acquisition frequency of sensors, detect abnormal values ​​of collected data, and transmit data through wireless networks; The settlement prediction module is used to extract features from the collected data, obtain roadbed settlement features and traffic load features, and combine them into feature vectors. The feature vectors are input into a nonlinear autoregressive model to predict roadbed settlement, determine the roadbed settlement results of multi-step predictions, and identify settlement risks. The simulation analysis module is used to analyze the multi-step prediction process of the roadbed settlement and obtain the simulation verification information generated in the multi-step prediction verification process; The strategy adjustment module is used to analyze the simulation verification information and generate different signals, and make corresponding strategy adjustments according to the generated different signals.

[0012] Technical effects and advantages of the present invention: The present invention deploys stress sensors, displacement sensors and acceleration sensors to monitor different areas of rural roads in real time, and realizes efficient monitoring of roadbed settlement. The sensor network optimizes the data acquisition frequency and detects outliers to ensure the accuracy and reliability of the collected data, and realizes real-time transmission through a wireless network. By extracting features from the collected data, constructing roadbed settlement features and traffic load features, and further inputting nonlinear autoregressive models to perform multi-step settlement prediction, the settlement trend at multiple moments in the future can be accurately predicted. Automatically identify settlement risks during the prediction process, provide early warning information, and help take maintenance measures in advance.

[0013] On the basis of multi-step prediction, the prediction accuracy of the model is fed back and adjusted in combination with simulation verification information to further improve the robustness of the model. Strategy adjustment is made according to the different signals generated, which can flexibly respond to different settlement risk scenarios and improve the safety management efficiency of rural roads. The advantage is that it realizes intelligent and accurate roadbed settlement prediction and risk warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The present invention is a flow chart of a method for troubleshooting safety hazards on rural roads.

[0015] Figure 2 The present invention is a schematic structural diagram of a rural highway safety hazard inspection system. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Example 1: Figure 1As shown, a method for checking safety hazards on rural roads comprises the following steps: Deploy stress sensors, displacement sensors, and acceleration sensors to monitor different areas of rural roads, optimize the data collection frequency of sensors, detect abnormal values ​​of collected data, and transmit data through wireless networks; Extract features from the collected data to obtain roadbed settlement features and traffic load features, and combine them into feature vectors. Input the feature vectors into a nonlinear autoregressive model to predict roadbed settlement, determine the results of multi-step predicted roadbed settlement, and identify settlement risks. Analyze the multi-step prediction process of roadbed settlement and obtain simulation verification information generated during the multi-step prediction verification process; The simulation verification information is analyzed and different signals are generated, and corresponding strategy adjustments are made according to the different signals generated.

[0018] In the process of checking safety hazards on rural roads, the deployment of sensor networks is the basis of intelligent roadbed settlement prediction and early warning system. The reasonable layout and configuration of sensor networks directly affect the accuracy of settlement monitoring and the effectiveness of early warning. The specific steps are as follows: Due to the diversity of rural road environments, the risk of roadbed settlement varies in different terrains and sections. Therefore, it is necessary to first identify key monitoring areas, which usually include: low-lying areas, which are prone to water accumulation or long-term immersion of the roadbed, leading to high-risk areas for settlement; steep slope sections, where the roadbed is prone to settlement under pressure and geological instability due to large changes in terrain; long-term overloaded sections, where overloaded vehicles frequently pass through, where the roadbed is subject to greater stress and is prone to settlement problems; After determining the monitoring area, it is necessary to select the appropriate sensor type according to the monitoring requirements, that is, the specific sensor selection and configuration for different monitoring needs, including but not limited to the following sensor selections: Use stress sensors to monitor changes in the bearing capacity of roadbed soil and detect the pressure distribution of the roadbed under different loads. Bury stress sensors deep in the roadbed, especially at bridge joints, steep slopes and overloaded sections to monitor pressure changes at different locations. Stress sensors need to be calibrated regularly to ensure data accuracy. The sensor density should be reasonably configured according to the length of the road section and the number of key areas. Displacement sensors detect tiny displacements of roadbed settlement and identify early signs of uneven settlement. Displacement sensors can be placed on the surface and underground, mainly in areas with high settlement risks, such as low-lying sections, near culverts, and under bridges. Accelerometers monitor ground vibrations, especially vibration shocks caused by overloaded vehicles, help identify settlements caused by ground vibrations, and are placed on the roadbed surface and shallow soil to monitor dynamic loads and geological activities caused by traffic. Underground sensors (such as stress sensors and displacement sensors): Dig a trench of a specific depth and place the sensor at the bottom or middle layer of the foundation to directly detect the stress and displacement changes of the soil. The burial depth depends on the thickness of the roadbed, usually between 0.5 meters and 2 meters; Surface sensors (such as accelerometers): installed on the surface of the road or roadbed, fixed on a solid bracket or base to avoid interference when vehicles pass by; After the sensor is installed, the initial sensor readings are recorded as a benchmark for subsequent settlement monitoring. By comparing the measurement data of standard equipment, the sensitivity and calibration factor of the sensor are adjusted to ensure that the sensor can accurately reflect environmental changes. The data collected by the sensor is then transmitted to the central management platform via the wireless network.

[0019] The sensor network will continuously collect the settlement data of the roadbed and transmit it to the central management platform through the wireless network for storage and processing. The specific steps are as follows: According to the types of different sensors and monitoring requirements, set the appropriate data collection frequency. Stress sensors collect data every 10-15 minutes, especially on roads where overloaded vehicles frequently pass by, so as to capture the impact of instantaneous loads on the roadbed. Displacement sensors are set with a longer collection cycle (such as 30 minutes or 1 hour) to detect long-term settlement changes. Acceleration sensors need to monitor instantaneous foundation vibration changes, and it is recommended to set them to collect data once a minute. The data collected by each sensor node is used to calculate the settlement value or stress change through the formula. The pressure value collected by the stress sensor can be expressed as: ,in, is stress, F is the force collected by the sensor, and A is the contact area. Stress data can help analyze the settlement pressure distribution, especially when overloaded vehicles are driving, to monitor the instantaneous load on the roadbed; The displacement sensor detects the vertical settlement of the roadbed, and the formula is: , where S is the sedimentation, is the initial height, is the roadbed height at time t, that is, the settlement difference at different times is calculated through multiple collected data, so as to infer the trend of roadbed settlement; The accelerometer records the vibration acceleration, and the formula is: , where a is the acceleration, is the speed at time t, is the initial velocity (i.e., the velocity when vibration starts), and t is the time interval (from the initial time to time t). The acceleration value can help identify the intensity of vibration of the foundation, especially for analyzing the impact of traffic flow on the roadbed; Each sensor node is equipped with a wireless communication module, which is responsible for the collection and transmission of local data. Each node sends data to a relay station or gateway via wireless signals, and the transmission rate and energy efficiency of the network are calculated based on the size and transmission time of the data packet transmitted by the data collection node; Since the sensor may be affected by environmental noise or data collection errors, it is necessary to detect outliers on the data, calculate the mean and standard deviation of the data, and use a fixed-size sliding window to calculate the change in the adjacent data of each data point. If the change exceeds the set threshold, it is considered abnormal and the outlier is removed; During the sensor data transmission process, data loss may occur and interpolation processing is required. The commonly used interpolation method is linear interpolation, that is, assuming that the changes of the two data points before and after the missing point are linear, the values ​​of the two data points before and after are used to estimate the missing value. The formula is as follows: ,in, is a missing value, , is the value of two known data points before and after, The time point corresponding to the missing value, , is the time point corresponding to the two known data points before and after; By reasonably setting the data collection frequency, applying the transmission protocol and performing data cleaning preprocessing, the efficient and reliable operation of the sensor network can be ensured.

[0020] The real-time data collected in the previous steps is used for in-depth analysis to establish a prediction model for highway subsidence and provide early warning and subsequent repair suggestions. The data is processed through complex algorithms and models to achieve prediction and dynamic adjustment of subsidence trends. The specific steps are as follows: The historical data and real-time data obtained from the data collection and transmission steps include the roadbed stress, settlement displacement, vibration frequency, etc. within a certain time range in the past. By analyzing these historical data, key features are extracted for the establishment of the prediction model. The main features include: Historical roadbed settlement data, including historical settlement recorded by stress sensors and displacement sensors at different time periods; Historical traffic load data, including vehicle passing frequency, load pressure, and traffic flow recorded by acceleration sensors; Apply feature extraction formulas to process data and construct more representative features, such as the cumulative effect of traffic load The formula is: ,in, is the load of the i-th vehicle, is the displacement of the roadbed caused by the i-th vehicle, and n is the total number of vehicles passing through in a specific time period. The formula for the cumulative rate of roadbed settlement is: ,in, is the cumulative rate of roadbed settlement, is the amount of settlement at time t, is the initial settlement amount, is the time interval (days); In the actual sedimentation process, the sedimentation rate often does not change linearly. We can also use nonlinear analysis based on historical data to extract the nonlinear characteristics of the sedimentation rate. The formula is: ,in, is the sedimentation rate change rate at time t, indicating the speed of change of sedimentation rate, is the sedimentation rate at time t-1; The roadbed settlement accumulation rate and traffic load accumulation effect are used as roadbed settlement characteristics and traffic load characteristics respectively; the collected real-time data and the features extracted from historical data are preprocessed, including normalization processing, which makes the data of different dimensions (such as settlement rate and load) have different dimensions and need to be normalized to have the same dimensions; the missing data are interpolated, and the bidirectional interpolation method combining forward interpolation and reverse interpolation is used to handle the missing sensor data to ensure the continuity of the time series; The extracted roadbed settlement characteristics and traffic load characteristics are combined into a feature vector as the input of the subsequent model. The time series of the combined multidimensional feature vector is expressed as: ,in, is the settlement at time t, is the traffic load characteristic at time t; The nonlinear autoregressive model (NARX) processes multidimensional inputs and captures the nonlinear relationship between input features. The form of the NARX model is: ,in, is the predicted settlement value at time t, is the amount of sedimentation in the past p moments, is the traffic load characteristics of the past q moments, is the error term, f is the nonlinear regression function, which is usually learned and fitted by a neural network (such as a multilayer perceptron, MLP) or a recurrent neural network (such as LSTM); The input layer of the nonlinear autoregressive model is composed of a sequence of feature vectors consisting of settlement and traffic load in historical data, with an input dimension of p+q. The hidden layer uses a nonlinear activation function (such as ReLU or Sigmoid) to perform nonlinear transformation on the input features to capture the complex relationship between various factors. The output layer outputs the predicted settlement at the future time t+k. , where k is the predicted time step; The data set is divided into training set, validation set and test set. The mean square error (MSE) is used as the loss function during training. The formula is: , where is the actual sedimentation value of the ith sample, is the settlement value predicted by the model, n is the number of samples; Use Adam optimizer to update model parameters. Adam optimizer combines momentum and adaptive learning rate, which is suitable for dealing with non-stationary and multi-dimensional input features in time series prediction. At the same time, it performs hyperparameter tuning. It uses Bayesian optimization algorithm to tune the model's hyperparameters, such as learning rate, number of hidden units, time step (values ​​of p, q, etc.), construct a proxy function to approximate the objective function, and gradually find the optimal solution through iteration. The nonlinear autoregressive model can perform single-step or multi-step predictions. Single-step prediction is to predict the amount of settlement at a future moment, and multi-step prediction is used to predict the amount of settlement at multiple moments in the future. Each predicted value will depend on the predicted value at the previous moment, forming a recursive relationship. Multi-step prediction is to predict the values ​​of multiple future moments at one time. In the process of roadbed settlement prediction, multi-step prediction can provide long-term trend assessment, which helps to discover potential risks in advance. If the model produces errors in the initial prediction, then the subsequent predictions will often deviate further from the true value. This cumulative effect will cause the prediction accuracy to decrease with the increase of time step. Compared with short-term prediction, long-term prediction is often more unstable, especially when the system has complex nonlinear characteristics; Analyze the multi-step prediction process of roadbed settlement, obtain simulation verification information generated in the multi-step prediction verification process, and determine the accuracy of the multi-step prediction. The simulation verification information includes fitting related information and error trend information; Fitting-related information includes the dynamic goodness-of-fit index and is calibrated as DFGI, and error trend information includes the prediction error accumulation index and is calibrated as PEAI; The dynamic goodness of fit index is used to evaluate the fitting performance of the model in multi-step forecasting. By making the fitting effects of short-term and long-term forecasts dynamic, it can simultaneously reflect the forecasting ability of the model at different time steps. By considering different time steps, it reflects the performance differences of the forecasting model in short-term and long-term forecasting. The specific functions are as follows: The dynamic goodness of fit index can identify the prediction performance of the model at different time steps in multi-step prediction, helping the model to better balance short-term and long-term predictions, thereby improving the overall accuracy of the prediction. In multi-step predictions, errors tend to accumulate gradually, affecting the prediction effect. Through the analysis of the dynamic goodness of fit index, it can be found which time steps are more sensitive to the prediction error, so as to take measures to reduce the error accumulation; In application scenarios that require multi-step predictions (such as roadbed settlement prediction), the dynamic goodness of fit index can provide decision makers with a powerful model performance evaluation, ensure the reliability of the prediction results, and help optimize the decision-making process, thereby dynamically evaluating the performance of the model in multi-step predictions, which can reflect the accuracy of short-term predictions and evaluate the trend stability of long-term predictions; The logic for obtaining the dynamic goodness of fit index is as follows: Get the actual value at time step t+i in unit time With the predicted value , calculate the error value: , where n is the total number of time steps, and the square of the error for each time step i is calculated: , get the mean of the actual values ​​obtained at each time step per unit time , calculate the actual value of each time step i The mean of the actual value The square difference calculation expression between them is: , calculate the dynamic goodness-of-fit index: .

[0021] It should be noted that the unit time is set according to the time range of actual needs.

[0022] The forecast error accumulation index is used to measure the trend of the forecast error gradually accumulating with the increase of time step in the multi-step forecast of the model. By quantifying the error accumulation of the model in the multi-step forecast, it reflects the trend of how the forecast error gradually increases with the time step. Measures the speed of error accumulation in multi-step forecasts. In multi-step forecasts, as the time step increases, the error tends to accumulate gradually. The forecast error accumulation index quantifies the rate of error change over time by considering the time step weight. The higher the forecast error accumulation index, the faster the error accumulation speed, and the more obvious the decrease in the accuracy of the model prediction over time; Identify deficiencies in model long-term forecasts: The forecast error accumulation index is particularly useful for analyzing the performance of a model in long-term forecasts. Short-term forecasts are usually relatively accurate, but the errors in long-term forecasts may increase significantly. The forecast error accumulation index can reveal the deficiencies of the model in long-term forecasts and help identify the reasons for the decline in forecast accuracy. The logic for obtaining the prediction error cumulative index is as follows: Get the actual value at time step t+j The predicted value at time step t+j , calculate the relative error between the predicted value and the actual value: , calculate the error fluctuation range: , where is a constant to prevent the denominator from being zero. The error fluctuation amplitudes of all time steps j are averaged to obtain the historical error fluctuation mean: , m is the total number of time steps, and the prediction error cumulative index is calculated. The calculation expression is: .

[0023] The fitting related information and error trend information are combined to generate the prediction matching coefficient, that is, the obtained dynamic goodness of fit index DFGI and prediction error accumulation index PEAI are combined to generate the prediction matching coefficient. The expression is: , where , is the preset proportional coefficient of the dynamic goodness of fit index and the prediction error accumulation index, and , Both are greater than 0.

[0024] It should be noted that the size of the preset proportional coefficient is a specific value obtained by quantifying each parameter. In order to facilitate subsequent comparison, the size of the coefficient depends on the amount of sample data and the preliminary setting of the corresponding preset proportional coefficient for each group of sample data by technical personnel in this field. It is not unique, as long as it does not affect the proportional relationship between the parameter and the quantized value. For example, the dynamic goodness of fit index is proportional to the predicted matching coefficient. The dynamic goodness of fit index and the predicted error accumulation index are normalized to have the same dimension and range. This can be achieved by subtracting the mean from the original data and dividing it by the standard deviation, or mapping the data to the range of [0, 1].

[0025] The larger the dynamic goodness of fit index is and the smaller the prediction error accumulation index is, the larger the prediction matching coefficient generated by the joint calculation is. A large dynamic goodness of fit index means that the model can capture the trend and fluctuation of the data well, and the matching degree between the predicted value and the actual value is high. A small prediction error accumulation index indicates that the model has better error control in a longer time step, and the error does not accumulate significantly. The prediction is more stable. A high prediction matching coefficient reflects that the model performs well in both short-term and long-term predictions, which means that the model can maintain a high degree of fit in the short term, and there is no significant error accumulation in long time step predictions, and there is a balance between the prediction performances at different time steps. In multi-step prediction, a higher prediction matching coefficient also indicates that the model has a strong generalization ability, which means that the model not only performs well on the training data, but also maintains a high accuracy on the new prediction data. It can usually adapt to different types of data and maintain good prediction performance when facing changes in different time steps. A higher prediction matching coefficient indicates that the model performs well in capturing the long-term trend of settlement, which means that the model can not only identify short-term changes, but also predict settlement trends over long periods of time, helping to identify potential hidden dangers such as roadbed subsidence and structural instability.

[0026] The generated prediction matching coefficient is compared with a preset prediction matching threshold to generate a multi-step prediction stability signal and a multi-step prediction error signal; After obtaining the predicted matching coefficient, the predicted matching coefficient is compared with the predicted matching threshold; If the prediction matching coefficient is greater than or equal to the prediction matching threshold, a multi-step prediction stable signal is generated, which means that the model has a high reliability in the prediction results for a long period of time in the future. This provides strong support for the long-term decision-making of the road maintenance department, and can arrange long-term maintenance plans more confidently. Since the model can maintain a small error accumulation in multiple time steps in the future, it means that the deviation of the prediction results is small. The prediction results of the model can be used in key decision-making links, such as monitoring and maintenance scheduling of rural road settlement. Decision-makers can rely on these results to formulate safe long-term maintenance strategies without additional adjustments. The prediction results can be used as normal data; If the prediction matching coefficient is less than the prediction matching threshold, a multi-step prediction error signal is generated, which indicates that the decisions made based on these prediction results may lead to potential increased risks. For example, failure to detect road settlement problems in a timely manner may lead to structural damage to the road and safety accidents, which may lead to inaccurate maintenance plans, missing the best repair time or underestimating the settlement amount, increasing the difficulty and cost of subsequent maintenance. It is necessary to optimize the model in a timely manner and determine the source of the prediction error in order to improve the prediction performance and reduce error accumulation.

[0027] After generating a multi-step prediction error signal, you need to respond quickly and adjust the correction measures. You can use Monte Carlo simulation to evaluate the model's prediction uncertainty, simulate multiple prediction results under different input conditions, calculate the prediction confidence interval, evaluate the model's performance under different conditions, determine the source of the model error, and make corresponding improvements and model adjustments.

[0028] It should be noted that the thresholds involved in the embodiments can be determined according to specific scenarios and requirements.

[0029] The present invention deploys stress sensors, displacement sensors and acceleration sensors to monitor different areas of rural roads in real time, and realizes efficient monitoring of roadbed settlement. The sensor network optimizes the data acquisition frequency and detects outliers to ensure the accuracy and reliability of the collected data, and realizes real-time transmission through a wireless network. By extracting features from the collected data, constructing roadbed settlement features and traffic load features, and further inputting nonlinear autoregressive models to perform multi-step settlement prediction, the settlement trend at multiple moments in the future can be accurately predicted. Automatically identify settlement risks during the prediction process, provide early warning information, and help take maintenance measures in advance.

[0030] On the basis of multi-step prediction, the prediction accuracy of the model is fed back and adjusted in combination with simulation verification information to further improve the robustness of the model. Strategy adjustment is made according to the different signals generated, which can flexibly respond to different settlement risk scenarios and improve the safety management efficiency of rural roads. The advantage is that it realizes intelligent and accurate roadbed settlement prediction and risk warning.

[0031] Embodiment 2: A rural highway safety hazard detection system, such as Figure 2 As shown, specifically including: The sensor acquisition module is used to deploy stress sensors, displacement sensors and acceleration sensors to monitor different areas of rural roads, optimize the data acquisition frequency of sensors, detect abnormal values ​​of collected data, and transmit data through wireless networks; The settlement prediction module is used to extract features from the collected data, obtain roadbed settlement features and traffic load features, and combine them into feature vectors. The feature vectors are input into a nonlinear autoregressive model to predict roadbed settlement, determine the roadbed settlement results of multi-step predictions, and identify settlement risks. The simulation analysis module is used to analyze the multi-step prediction process of the roadbed settlement and obtain the simulation verification information generated in the multi-step prediction verification process; The strategy adjustment module is used to analyze the simulation verification information and generate different signals, and make corresponding strategy adjustments according to the generated different signals.

[0032] The above formulas are all dimensionless and calculated numerically. Specific dimension removal can be achieved by various means such as standardization, which will not be elaborated here. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0033] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, an ATA hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium may be a solid-state ATA hard disk.

[0034] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0035] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0036] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0037] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0038] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0039] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for checking safety hazards on rural roads, characterized in that: The steps include: Deploy stress sensors, displacement sensors, and acceleration sensors to monitor different areas of rural roads, optimize the data collection frequency of sensors, detect abnormal values ​​of collected data, and transmit data through wireless networks; Extract features from the collected data to obtain roadbed settlement features and traffic load features, and combine them into feature vectors. Input the feature vectors into a nonlinear autoregressive model to predict roadbed settlement, determine the results of multi-step predicted roadbed settlement, and identify settlement risks. Analyze the multi-step prediction process of roadbed settlement and obtain simulation verification information generated during the multi-step prediction verification process; The simulation verification information is analyzed and different signals are generated, and corresponding strategy adjustments are made according to the different signals generated.

2. A method for troubleshooting rural road safety hazards according to claim 1, characterized in that: Deploy stress sensors, displacement sensors, and acceleration sensors to monitor different areas of rural roads, optimize the data collection frequency of the sensors, detect abnormal values ​​of the collected data, and transmit data through wireless networks. The specific process is as follows: Areas where sensors are deployed include low-lying areas, steep slopes, and chronically overloaded sections; Use stress sensors to monitor the changes in the bearing capacity of roadbed soil and detect the pressure distribution of the roadbed under different loads; Use displacement sensors to detect the displacement of roadbed settlement and identify the phenomenon of uneven settlement; Use acceleration sensors to monitor foundation vibration and identify settlement caused by foundation vibration; The equipment data includes the data collected by monitoring the displacement sensor and the data collected by monitoring the laser ranging device through the optical path sensor; After the sensor is installed, the initial sensor readings are recorded as a benchmark for settlement monitoring, and the sensitivity and calibration coefficient of the sensor are adjusted. The sensor collected data is transmitted to the central management platform via the wireless network for storage and processing.

3. A method for troubleshooting rural road safety hazards according to claim 2, characterized in that: The sensor collected data is transmitted to the central management platform through the wireless network for storage and processing. The specific process is as follows: Set the data collection frequency according to the type of sensor and monitoring requirements; The data collected by the sensor is used to calculate the settlement value or stress change. The stress sensor collects the pressure value of the roadbed, the displacement sensor detects the vertical settlement of the roadbed, and the acceleration sensor records the vibration acceleration of the roadbed. The collected data is tested for outliers, and the mean and standard deviation are calculated. A fixed-size sliding window is used to calculate the change in the adjacent data of each data point. If the change exceeds the set threshold, the data is considered abnormal and removed, and linear interpolation is performed in the case of data loss.

4. A method for checking safety hazards of rural roads according to claim 3, characterized in that: The collected data is subjected to feature extraction to obtain the roadbed settlement characteristics and traffic load characteristics, which are combined into feature vectors. The feature vectors are input into the nonlinear autoregressive model to predict the roadbed settlement, and the roadbed settlement results of the multi-step prediction are determined to identify the settlement risk. The specific process is as follows: The data are processed using feature extraction formulas, traffic load features are extracted using the traffic load cumulative effect formula, roadbed settlement features are extracted using the roadbed settlement cumulative rate formula, and the extracted features are preprocessed; The extracted roadbed settlement characteristics and traffic load characteristics are combined into a feature vector and input into a nonlinear autoregressive model; The input layer of the nonlinear autoregressive model is composed of a feature vector sequence consisting of settlement and traffic load in historical data. The hidden layer uses a nonlinear activation function to perform nonlinear transformation on the input features, and the output layer outputs the predicted settlement at the future moment. The data set is divided into training set, validation set and test set, and the mean square error is used as the loss function during training; Use the Adam optimizer to update the model parameters and use the Bayesian optimization algorithm to tune the model's hyperparameters; The nonlinear autoregressive model is used to make single-step and multi-step predictions of sedimentation.

5. A method for checking safety hazards of rural roads according to claim 4, characterized in that: The multi-step prediction process of roadbed settlement is analyzed to obtain the simulation verification information generated in the multi-step prediction verification process. The specific process is as follows: Acquire simulation verification information generated during the multi-step prediction verification process, the simulation verification information includes fitting related information and error trend information; Fitting-related information includes the dynamic goodness-of-fit index, and error trend information includes the prediction error accumulation index; The obtained dynamic goodness of fit index and prediction error cumulative index are combined to generate a prediction matching coefficient; The dynamic goodness of fit index is directly proportional to the prediction matching coefficient, and the prediction error accumulation index is inversely proportional to the prediction matching coefficient.

6. A method for checking safety hazards on rural roads according to claim 5, characterized in that: Analyze the simulation verification information and generate different signals. Adjust the corresponding strategies according to the generated different signals. The specific steps are as follows: Compare the generated prediction matching coefficient with the set prediction matching threshold; If the prediction match coefficient is greater than or equal to the prediction match threshold, a multi-step prediction stability signal is generated, and no additional adjustment is required, and the prediction result is used as normal data; If the prediction matching coefficient is less than the prediction matching threshold, a multi-step prediction error signal is generated, and the model needs to be optimized in time to determine the source of the prediction error.

7. A rural road safety hazard inspection system, used to implement a rural road safety hazard inspection method according to any one of claims 1 to 6, characterized in that: include: The sensor acquisition module is used to deploy stress sensors, displacement sensors and acceleration sensors to monitor different areas of rural roads, optimize the data acquisition frequency of sensors, detect abnormal values ​​of collected data, and transmit data through wireless networks; The settlement prediction module is used to extract features from the collected data, obtain roadbed settlement features and traffic load features, and combine them into feature vectors. The feature vectors are input into a nonlinear autoregressive model to predict roadbed settlement, determine the roadbed settlement results of multi-step predictions, and identify settlement risks. The simulation analysis module is used to analyze the multi-step prediction process of the roadbed settlement and obtain the simulation verification information generated in the multi-step prediction verification process; The strategy adjustment module is used to analyze the simulation verification information and generate different signals, and make corresponding strategy adjustments according to the generated different signals.

Images