A method and system for detecting a group fog based on a state of an intelligent stud optical communication link
By using a smart road stud optical communication link status method to detect fog patches, parameters such as signal strength and signal-to-noise ratio are monitored and optimized in real time. Combined with graph neural network analysis of link status changes, accurate prediction and location of fog patches are achieved, which solves the shortcomings of traditional methods in terms of real-time performance and wide-area coverage, and improves the preventive capabilities of traffic safety management.
Patent Information
- Application Number
- CN202610344223.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-20
- Publication Date
- 2026-06-12
Smart Images
Figure CN122194344A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of road meteorological monitoring technology, and in particular to a method and system for detecting fog based on the status of an intelligent road stud optical communication link. Background Technology
[0002] Fog patches are a common meteorological phenomenon on highways and other roads, which can rapidly lead to a significant decrease in visibility, seriously affecting traffic safety. Traditional methods for detecting fog patches mainly rely on fixed visibility meters and video analysis. Visibility meters are typically installed on road sections to measure changes in visibility in real time and determine the occurrence of fog patches by setting link status thresholds. However, visibility meters have limitations in deployment cost and coverage, especially in long road sections or in adverse environments where comprehensive coverage is difficult. In addition, video analysis methods process road images through cameras to analyze changes in visibility, but are greatly affected by ambient light, especially at night or in low-light conditions, making it difficult to guarantee accuracy.
[0003] While traditional visibility meters and video analytics can detect fog patches, they have limitations in terms of real-time performance and wide-area coverage. First, visibility meters typically have limited coverage and require costly equipment installation, failing to provide comprehensive road monitoring and hindering seamless monitoring over long stretches or in adverse weather conditions. Second, video analytics is highly dependent on ambient lighting conditions, especially at night or in low-light conditions, where accuracy and reliability decrease, making it unable to handle all-weather environmental changes. Summary of the Invention
[0004] In view of the aforementioned existing problems, the present invention is proposed.
[0005] Therefore, this invention provides a fog detection method based on the status of the intelligent road spike optical communication link to solve the problem of achieving low-cost, high-density, and real-time fog monitoring using traditional fog monitoring methods.
[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for detecting patchy fog based on the optical communication link status of intelligent road spikes. The method includes: collecting link communication data and performing baseline state learning to generate baseline communication data; monitoring environmental changes and real-time fluctuations in link status based on the baseline communication data, automatically adjusting and optimizing signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay to generate a link status threshold setting; periodically sending probe frames based on the link status threshold setting, calculating real-time communication parameters, and generating link status data; comparing and analyzing the link status data with the baseline communication data to generate abnormal link status; based on the abnormal link status, using a pre-trained graph neural network to analyze state changes between multiple links, predicting the expansion path of the patchy fog, and generating prediction results for the location and propagation trend of the patchy fog; and based on the prediction results for the location and propagation trend of the patchy fog, using the topological relationship of the link status to calculate the core location and influence range of the patchy fog, generating accurate positioning information for the patchy fog, performing dynamic baseline correction, and updating the baseline communication data.
[0007] As a preferred embodiment of the fog detection method based on the optical communication link status of smart road spikes described in this invention, the specific steps for generating reference communication data are as follows: The smart road spikes collect signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay to form link communication data; Perform statistical analysis on the link communication data, calculate the average and standard deviation of signal strength, signal-to-noise ratio, bit error rate and successful handshake delay in each link, and establish the link baseline state; Based on the link baseline status, the normal fluctuation range of signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay is defined, and baseline communication data is generated.
[0008] As a preferred embodiment of the fog detection method based on the optical communication link status of smart road spikes described in this invention, the specific steps for setting the link status threshold are as follows: Based on baseline communication data, the system monitors changes in ambient temperature, humidity, and light intensity in real time, as well as fluctuations in signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay in the link communication data, generating link status fluctuation data. Based on link status fluctuation data, the signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay are automatically optimized and adjusted to generate optimized link communication parameters. Based on the optimized link communication parameters, the effective fluctuation range of signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay is defined, and the link state threshold settings are generated.
[0009] As a preferred embodiment of the fog detection method based on the optical communication link status of smart road spikes described in this invention, the specific steps for generating link status data are as follows: Based on the link status threshold setting, the system periodically sends the sequence number, timestamp, sender ID, and preamble detection information to adjacent road spikes to generate link status detection frames; After receiving the link status probe frame, the system uses statistical analysis methods to calculate and extract real-time communication parameters, and generates real-time communication parameters. The real-time communication parameters are compared with the link status threshold settings one by one to determine whether the signal strength, signal-to-noise ratio, bit error rate and successful handshake delay exceed the effective fluctuation range, and link status data is generated.
[0010] As a preferred embodiment of the fog detection method based on the optical communication link status of smart road spikes described in this invention, the specific steps for generating abnormal link status are as follows: When the difference between the signal strength in the link status data and the reference signal strength in the reference communication data exceeds the effective fluctuation range of the signal strength, it is determined to be an abnormal signal attenuation. When the signal-to-noise ratio in the link status data is lower than the effective fluctuation range of the signal-to-noise ratio, it is determined that the signal quality has degraded. When the bit error rate in the link status data is higher than the effective fluctuation range of the bit error rate, it is determined that the communication quality has deteriorated. When the successful handshake delay in the link status data exceeds the effective fluctuation range of the successful handshake delay, the link is determined to be unstable. The system integrates the identified signal attenuation anomalies, signal quality degradation, communication quality deterioration, and link instability to generate an abnormal link status.
[0011] As a preferred embodiment of the fog detection method based on the optical communication link status of smart road spikes described in this invention, the generation of the pre-trained graph neural network involves the following specific steps. Select a dataset containing the connection relationships and link status information between link nodes, design a graph neural network architecture according to task requirements, define the input features of each link node and edge, and construct a training dataset for the graph neural network. Using the training dataset, the graph neural network parameters are adjusted through forward and backward propagation algorithms to generate graph neural network weights; The weights of the graph neural network are validated using a validation dataset. Once the validation achieves the expected results, the pre-trained weights are saved, and the pre-trained graph neural network is generated.
[0012] As a preferred embodiment of the fog detection method based on the optical communication link status of smart road spikes described in this invention, the specific steps for generating the predicted location and propagation trend of the fog are as follows: Abnormal link states are input into a pre-trained graph neural network to analyze the state changes between multiple link nodes, capture the dependencies and information propagation patterns between links, and generate link state change analysis results. Based on the link state change analysis results, a pre-trained graph neural network is used to learn the propagation mode of fog, predict the expansion path of fog, and generate fog expansion path prediction. By combining the predicted path of fog spread, abnormal link status, and historical link status data, the location and propagation trend of fog are predicted, generating prediction results for the location and propagation trend of fog.
[0013] As a preferred embodiment of the fog detection method based on the optical communication link status of smart road spikes described in this invention, the specific steps for generating accurate fog location information are as follows: By combining the location of the fog patch, the prediction results of its propagation trend, and the link status data, and by evaluating changes in the link status, the dense area of the fog patch is estimated, and the core location of the fog patch is generated. Based on the core location of the fog and link status data, the connection status of the link and the degree of signal attenuation are calculated to generate the range of influence of the fog. By combining the core location of the fog patch with its influence range, the overall influence area of the fog patch is calculated, generating accurate location information for the fog patch.
[0014] As a preferred embodiment of the fog detection method based on the optical communication link status of smart road spikes described in this invention, the specific steps for updating the reference communication data are as follows: Based on the accurate positioning information of the fog, the link status change analysis results are monitored in real time, the baseline communication data is dynamically corrected, and dynamic baseline correction data is generated. Based on the dynamic baseline correction data, adjust the real-time communication parameters and update the baseline communication data.
[0015] Secondly, this invention provides a fog detection system based on the optical communication link status of intelligent road spikes, comprising: a data acquisition module for acquiring link communication data and performing baseline state learning to generate baseline communication data; a threshold setting module for automatically adjusting and optimizing signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay based on the baseline communication data, monitoring environmental changes and real-time fluctuations in link status, and generating link status threshold settings; a detection calculation module for periodically sending detection frames based on the link status threshold settings, calculating real-time communication parameters, and generating link status data; a link detection module for comparing and analyzing the link status data with the baseline communication data to generate abnormal link status; a path prediction module for analyzing the status changes between multiple links using a pre-trained graph neural network based on the abnormal link status, predicting the expansion path of the fog, and generating fog location and propagation trend prediction results; and a positioning correction module for calculating the core location and influence range of the fog based on the fog location and propagation trend prediction results, utilizing the topological relationship of the link status, generating accurate fog location information, performing dynamic baseline correction, and updating the baseline communication data.
[0016] The beneficial effects of this invention are as follows: by using a pre-trained graph neural network to analyze the changes in the state of multiple links, it achieves accurate prediction of the expansion path of fog, can identify the core location and influence range of fog in advance, and thus realizes active, real-time and accurate fog detection and positioning, improves the prevention capabilities of road traffic safety management, avoids the lag and inefficiency of traditional methods, and has social value and application potential. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 This is a flowchart of a fog detection method based on the status of the optical communication link of smart road spikes.
[0019] Figure 2 This is a schematic diagram of a fog detection system based on the status of the optical communication link of a smart road spike.
[0020] Figure 3 A flowchart generated for setting link status thresholds.
[0021] Figure 4 A flowchart for generating accurate location information for fog patches. Detailed Implementation
[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.
[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.
[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for detecting fog based on the status of a smart road spike optical communication link, including the following steps: S1. Collect link communication data and perform baseline state learning to generate baseline communication data.
[0026] S1.1. The signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay are collected by the smart road spike to form link communication data.
[0027] Specifically, on the smart road spikes, data such as signal strength, signal-to-noise ratio (SNR), bit error rate (BER), and successful handshake delay are collected. Optical communication is conducted between the smart road spikes, with data collected every second, for example, at 1-second intervals. Signal strength is measured by the strength of the received signal, SNR is assessed by the ratio of signal quality to noise, BER reflects the error rate in data transmission, and successful handshake delay refers to the signal transmission delay during communication establishment. This data is collected in real-time during each transmission, forming link communication data.
[0028] S1.2 Perform statistical analysis on the link communication data, calculate the average and standard deviation of signal strength, signal-to-noise ratio, bit error rate and successful handshake delay in each link, and establish the link baseline state.
[0029] Specifically, signal strength, signal-to-noise ratio (SNR), bit error rate (BER), and successful handshake delay are calculated for each link. The dataset for each link is recorded at multiple time points, and all collected data are organized chronologically. The arithmetic mean method is used to calculate the average of these parameters for each link, reflecting the average communication performance under normal conditions. The standard deviation formula is used to calculate the standard deviation of each parameter. By measuring the dispersion of signal strength, SNR, BER, and successful handshake delay, parameters with high standard deviations indicate significant fluctuations in communication performance across different time points, suggesting poor link stability. For example, if the standard deviation of signal strength is greater than, say, 15 dB based on actual communication conditions, it indicates significant fluctuations in signal strength across different time periods, resulting in unstable communication quality. Similarly, if the standard deviation of the BER exceeds (e.g., greater than 0.05 based on communication performance standards), it indicates unstable communication quality. Through statistical analysis of signal strength, SNR, BER, and successful handshake delay, a baseline state of the link is established.
[0030] S1.3. Based on the link baseline status, define the normal fluctuation range of signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay, and generate baseline communication data.
[0031] Specifically, based on the link baseline status, the fluctuations of signal strength, signal-to-noise ratio (SNR), bit error rate (BER), and successful handshake delay are analyzed over different time periods. By statistically analyzing the average and standard deviation of each link communication parameter, the fluctuation range of each parameter under normal operating conditions can be determined. Based on the statistical results, the normal fluctuation range of each communication parameter is defined. For example, the normal fluctuation range of signal strength is defined as the baseline average value ± 2 times the standard deviation. If the difference between the upper and lower limits of the normal fluctuation range of signal strength is within 30 dB (i.e., the standard deviation is approximately less than 15 dB), the signal strength of the link is considered to be within the normal range. The preset 15 dB standard deviation threshold references the common fluctuation levels of stable links in the communication environment (e.g., the relevant requirements for the stability of received signal strength indication in the IEEE-802.11 protocol series can be referenced) to ensure communication under relatively stable conditions. The normal fluctuation range of SNR is defined as the baseline average value ± 2 times the standard deviation. If the difference between the upper and lower limits of the normal fluctuation range of SNR is within 6 dB (i.e., the standard deviation is approximately less than 3 dB), the signal quality is considered good. The standard deviation threshold of B references conventional standards in the communications field (e.g., the relevant definitions of the relationship between bit error rate performance and signal-to-noise ratio in standards such as ITU-T G.826), and is suitable for ensuring that the signal is not significantly interfered with. The normal fluctuation range of the bit error rate is defined as the baseline average value ± 2 standard deviations. If the upper limit of the normal fluctuation range of the bit error rate is less than 0.05, the communication quality is considered stable. The standard deviation threshold of 0.05 references the fluctuation range of the bit error rate under normal communication conditions (e.g., the bit error rate performance standard based on G.821 recommendations). Exceeding 0.05 usually indicates a communication problem. The normal fluctuation range of the successful handshake delay is defined as the baseline average value ± 2 standard deviations. If the difference between the upper and lower limits of the normal fluctuation range of the successful handshake delay is within 100 milliseconds (i.e., the standard deviation is about 50 milliseconds), the link is considered stable and responsive. The standard deviation threshold of 50 milliseconds references the response time standard of high-speed communication networks (e.g., the typical round-trip time requirement of the TCP / IP protocol stack), ensuring that the link completes the handshake quickly and stably under normal conditions. Based on the normal fluctuation range, baseline communication data is generated.
[0032] S2. Based on the baseline communication data, monitor environmental changes and real-time fluctuations in link status, automatically adjust and optimize signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay, and generate link status threshold settings.
[0033] S2.1 Based on the reference communication data, monitor in real time the changes in ambient temperature, humidity, and light, as well as the fluctuations in signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay in the link communication data, and generate link status fluctuation data.
[0034] Specifically, based on baseline communication data, the system monitors changes in temperature, humidity, and light intensity in the environment in real time, collecting fluctuations in these parameters as external factors affecting link communication quality. It also monitors fluctuations in signal strength, signal-to-noise ratio (SNR), bit error rate (BER), and successful handshake delay in the link communication data, recording these changes at different points in time. Through analysis of real-time changes in signal strength, SNR, BER, and successful handshake delay, the system can obtain the fluctuation amplitude and trend of each parameter. Combined with fluctuations in environmental data, it generates link status fluctuation data.
[0035] S2.2 Based on the link status fluctuation data, automatically optimize and adjust the signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay to generate optimized link communication parameters.
[0036] Specifically, based on link status fluctuation data, the system analyzes the fluctuation range and trends of signal strength, signal-to-noise ratio (SNR), bit error rate (BER), and successful handshake delay. By comparing the actual fluctuations of these parameters with their normal fluctuation ranges, the system determines the degree of deviation for each parameter. Based on this deviation, the system automatically adjusts the optimized value of each communication parameter. For example, if the standard deviation of signal strength exceeds 15 dB, the system adjusts the signal strength fluctuation range to ensure communication stability; if the standard deviation of BER exceeds 0.05, the system automatically optimizes the BER threshold to ensure data transmission accuracy; if the standard deviation of SNR exceeds 3 dB, the system adjusts the SNR fluctuation range to maintain signal quality; and if the standard deviation of successful handshake delay exceeds 50 milliseconds, the system automatically adjusts the handshake delay fluctuation range to ensure link response speed. These adjustments ensure that communication parameters remain within the normal fluctuation range, maintaining stable communication quality and generating optimized link communication parameters.
[0037] S2.3. Based on the optimized link communication parameters, define the effective fluctuation range of signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay, and generate link state threshold settings.
[0038] Specifically, based on the optimized link communication parameters, the actual fluctuation of each communication parameter is evaluated to determine the effective fluctuation range of signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay. By analyzing the optimized values of each link communication parameter and comparing them with the normal fluctuation range, it is determined whether the change of each parameter is within the normal fluctuation range. Based on the evaluation results, a new threshold range is set for each communication parameter. For example, by adjusting the signal strength fluctuation range, the effective fluctuation range of signal strength is determined to be ±15dB. Based on statistical analysis of actual conditions in the actual communication environment and historical link status data, a signal strength fluctuation exceeding ±15dB usually indicates a chain... The link communication exhibits instability. The effective fluctuation range of the signal-to-noise ratio (SNR) is set to ±3dB. According to standards in the communication field, SNR fluctuations within ±3dB are considered stable, while fluctuations exceeding ±3dB typically indicate a deterioration in signal quality. The effective fluctuation range of the bit error rate (BER) is set to 0.05. Based on actual BER fluctuations, a BER exceeding 0.05 usually indicates a serious transmission problem in the link. The effective fluctuation range of the successful handshake delay is set to ±50 milliseconds. According to the response time standards for high-speed communication networks, a BER fluctuation greater than 50 milliseconds typically indicates a slow link response or link instability. This generates the link status threshold settings.
[0039] S3. Based on the link status threshold setting, periodically send probe frames, calculate real-time communication parameters, and generate link status data.
[0040] S3.1. Based on the link status threshold setting, periodically send the sequence number, timestamp, sender ID and preamble detection information to adjacent road spikes to generate link status detection frames.
[0041] Specifically, based on the link state threshold setting, for example, a 1-second period, probe information containing a sequence number, timestamp, sender ID, and preamble is sent to adjacent spikes. Each probe information is transmitted through the optical communication link. The sequence number ensures that each probe frame has a unique identifier, the timestamp records the transmission time of the probe information, the sender ID indicates the source of the information, and the preamble helps the receiver identify the start position of the probe frame. The receiver obtains the actual value of each communication parameter by measuring the signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay in the probe frame. For example, the signal strength is obtained by the strength of the received signal, the signal-to-noise ratio is calculated by measuring the ratio of the received signal to the background noise, the bit error rate is evaluated by detecting error bits in the transmission, and the successful handshake delay is determined by calculating the time difference between the transmission and reception of the probe frame, thus generating a link state probe frame.
[0042] S3.2 After receiving the link status probe frame, use statistical analysis methods to calculate and extract real-time communication parameters and generate real-time communication parameters.
[0043] Specifically, upon receiving a link status probe frame, the signal strength, signal-to-noise ratio (SNR), bit error rate (BER), and successful handshake delay are extracted from the probe frame. Using a moving average method, the same parameter value is calculated on multiple consecutively received probe frames (e.g., the most recent 10 frames) to obtain smoothed real-time parameter values. For example, by performing a moving average calculation on the signal strength of multiple received probe frames, the real-time signal strength at each time point is calculated. The SNR is evaluated by performing a moving average of the ratio of the received signal to the background noise. The BER is evaluated by comparing the number of correct and erroneous bits during transmission. The successful handshake delay is obtained by calculating the time difference between sending and receiving the probe frame. Based on the results of statistical analysis, real-time communication parameters are generated.
[0044] S3.3. Compare the real-time communication parameters with the link status threshold settings item by item to determine whether the signal strength, signal-to-noise ratio, bit error rate and successful handshake delay exceed the effective fluctuation range, and generate link status data.
[0045] Specifically, each real-time communication parameter is compared with its effective fluctuation range. For example, the actual value of signal strength is compared with its effective fluctuation range to determine whether the signal strength exceeds the effective fluctuation range. Similarly, the actual values of signal-to-noise ratio, bit error rate, and successful handshake delay are compared with their respective effective fluctuation ranges to check for any cases exceeding the effective range. If the actual values of signal strength, signal-to-noise ratio, bit error rate, or successful handshake delay exceed the effective fluctuation range, they are marked as abnormal; otherwise, the parameters are considered to be within the normal range. By comparing each parameter, link status data is generated.
[0046] S4. Compare and analyze the link status data with the baseline communication data to generate abnormal link status.
[0047] S4.1 When the difference between the signal strength in the link status data and the reference signal strength in the reference communication data exceeds the effective fluctuation range of the signal strength, it is determined to be an abnormal signal attenuation.
[0048] Specifically, when the difference between the signal strength in the link status data and the reference signal strength in the reference communication data exceeds the effective fluctuation range of the signal strength, the difference method is used to calculate the difference in signal strength, that is, to compare the signal strength in the link status data with the reference signal strength in the reference communication data; if the calculation result exceeds the preset effective fluctuation range of the signal strength, it is determined that the signal strength is abnormal, indicating abnormal signal attenuation.
[0049] S4.2 When the signal-to-noise ratio in the link status data is lower than the effective fluctuation range of the signal-to-noise ratio, it is determined that the signal quality has deteriorated.
[0050] Specifically, when the signal-to-noise ratio (SNR) in the link status data is lower than the effective fluctuation range of the SNR, the SNR value in the link status data is obtained and compared with the effective fluctuation range of the SNR; if the SNR in the link status data is lower than the effective fluctuation range of the SNR, it indicates that the signal quality has deteriorated.
[0051] S4.3 When the bit error rate in the link status data is higher than the effective fluctuation range of the bit error rate, it is determined as a degradation of communication quality.
[0052] Specifically, when the bit error rate in the link status data is higher than the effective fluctuation range of the bit error rate, the bit error rate value is extracted from the link status data and compared with the effective fluctuation range of the bit error rate. If the bit error rate in the link status data exceeds the effective fluctuation range of the bit error rate, it indicates that the communication quality has deteriorated.
[0053] S4.4 When the successful handshake delay in the link status data exceeds the effective fluctuation range of the successful handshake delay, the link is determined to be unstable.
[0054] Specifically, when the successful handshake delay in the link state data exceeds the effective fluctuation range of the successful handshake delay, the value of the successful handshake delay is extracted from the link state data and compared with the effective fluctuation range of the successful handshake delay. If the successful handshake delay in the link state data exceeds the effective fluctuation range of the successful handshake delay, it indicates that the link is unstable.
[0055] S4.5 Integrate the determined abnormal signal attenuation, signal quality degradation, communication quality deterioration, and link instability to generate an abnormal link status.
[0056] Specifically, when integrating the judgments of abnormal signal attenuation, degraded signal quality, deteriorated communication quality, and unstable link, the judgment results of each abnormal link state are evaluated one by one to check whether each parameter exceeds the effective fluctuation range. If any one or more of the abnormal conditions of abnormal signal attenuation, degraded signal quality, deteriorated communication quality, or unstable link meet the abnormal criteria, it is marked as an abnormal state. Combined with the judgment results, an abnormal link state is generated, which means that there is a problem with the communication quality of the link at the current moment.
[0057] S5. Based on the abnormal link status, use a pre-trained graph neural network to analyze the state changes between multiple links, predict the expansion path of the fog, and generate prediction results of the fog location and propagation trend.
[0058] S5.1 Select a dataset containing the connection relationships and link status information between link nodes, design a graph neural network architecture according to task requirements, define the input features of each link node and edge, and construct the training dataset for the graph neural network.
[0059] Specifically, the dataset of connection relationships and link status information between link nodes is selected based on the deployment location of smart road studs along the road. The connection relationships between each link node are analyzed to determine the communication dependencies between each link node and its neighboring link nodes, and the connection relationships are transformed into a graph structure. The basic structure of the graph neural network architecture is determined according to the task requirements, including the input features of link nodes and edges. For each link node, the input feature is defined as the link status information of the link node, such as communication parameters such as signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay. For each edge, the input feature is defined as the link status information connecting two link nodes. Considering the changes in communication quality between link nodes, a training dataset for the graph neural network is constructed based on the defined features of link nodes and edges.
[0060] It should also be noted that the task requirements refer to defining the input features and network structure required for the graph neural network architecture based on specific application scenarios and objectives. The task requirements mainly refer to determining how to use the connection relationships and link status information between link nodes to predict the expansion path and location of the fog. The graph neural network architecture is specifically set as two graph convolutional layers with a hidden layer dimension of 128 and using the ReLU activation function.
[0061] S5.2 Using the training dataset, adjust the graph neural network parameters and generate graph neural network weights through forward and backward propagation algorithms.
[0062] Specifically, in a graph neural network, the forward propagation process inputs the training dataset into the input layer, with the features of each link node (such as signal strength and signal-to-noise ratio) as input. The input data is processed through graph convolutional layers, which aggregate information from neighboring nodes and update the representation of each node. Activation layers apply non-linear activation functions (such as ReLU) to the output of the graph convolutional layers to increase the non-linearity of the graph neural network. The output of the graph convolutional layers can be passed to fully connected layers, which further integrate node features and generate higher-level representations. The output layer generates predicted values or classification results. Through layer-by-layer propagation and computation, the graph neural network completes the processing of the input data. The process involves processing the input data and outputting the prediction results of the graph neural network. During this process, the parameters of the graph neural network (i.e., the weights of the links and edges) are updated based on the features of the input training dataset until the error between the prediction result and the expected target result reaches, for example, a mean squared error of less than 0.01. Using the backpropagation algorithm, the gradient is calculated based on the error between the graph neural network output and the actual target, and backpropagated to each parameter of the graph neural network for updating to minimize the error. The weight values in the graph neural network are updated using the gradient descent method, so that the error gradually decreases. This process is repeated until the parameters of the graph neural network converge, generating the graph neural network weights.
[0063] It should also be noted that when judging whether the parameters of a graph neural network have converged, the change in the loss function is usually set to be less than, for example, 0.001. The basis is that during multiple iterations, when the change in the loss function is less than 0.001, it indicates that the learning of the graph neural network has become stable, the parameters of the graph neural network no longer change, and the graph neural network has converged. The expected target result refers to the ideal or target value that the graph neural network model output should achieve based on the actual application requirements. It usually reflects the performance requirements of the graph neural network in a specific task. Each parameter refers to the weights of the links in the graph neural network, the weights of the edges, and the learnable parameters related to the graph neural network structure. These parameters are used to adjust the output of the graph neural network in order to gradually reduce the error between the predicted value and the target value.
[0064] S5.3. Using the validation dataset, validate the weights of the graph neural network. After the validation achieves the expected results, save the pre-trained weights and generate the pre-trained graph neural network.
[0065] Specifically, using the validation dataset, the trained graph neural network weights are input into the graph neural network, forward propagation is performed, and the output results are calculated. The output results are compared with the actual results in the validation dataset to evaluate the prediction effect of the network. If the error between the prediction result and the actual result of the graph neural network is less than the preset allowable range, it means that the graph neural network weights have achieved the expected effect. The pre-trained weights generated during the training process are saved, and the pre-trained graph neural network is generated using the validated weights.
[0066] It should also be noted that the validation dataset refers to the dataset used to evaluate the performance of the graph neural network during the training process. Unlike the training dataset, the validation dataset does not participate in the updating of the graph neural network parameters, but is used to periodically check the performance of the graph neural network during the training process. By inputting the validation dataset into the trained graph neural network, the generalization ability of the graph neural network on the unseen validation dataset can be detected, which helps to determine whether the graph neural network is overfitting. The preset tolerance range is based on the error tolerance setting during training, and is usually determined based on domain standards. For graph neural networks, the preset tolerance range is set between 0.001 and 0.01.
[0067] S5.4 Input the abnormal link state into the pre-trained graph neural network, analyze the state changes between multiple link nodes, capture the dependencies and information propagation patterns between links, and generate link state change analysis results.
[0068] Specifically, abnormal link states are input into a pre-trained graph neural network. The pre-trained graph neural network processes the information of the input link nodes and edges to analyze the state changes between multiple link nodes, capturing the dependencies between links and information propagation patterns. The graph neural network uses the communication parameters of each link node (such as signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay) and link connectivity to learn about the changes in link states. Through the propagation mechanism of the pre-trained graph neural network, the captured dependencies between link nodes are transmitted to adjacent link nodes, thus reflecting the overall trend of link state changes and generating link state change analysis results. The link state change analysis results include the trends of signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay, as well as the dependencies between link nodes and information propagation patterns.
[0069] S5.5 Based on the link state change analysis results, the pre-trained graph neural network is used to learn the propagation mode of the fog, predict the expansion path of the fog, and generate fog expansion path prediction.
[0070] Specifically, the link state change analysis results are input into a pre-trained graph neural network for learning. The pre-trained graph neural network learns the propagation pattern of the fog by processing the state changes between link nodes. That is, by analyzing the trend of link state changes and the dependencies between link nodes, it captures the expansion pattern of the fog in different regions. The pre-trained graph neural network models the changes of multiple link nodes and propagates them to adjacent link nodes to form a prediction of the fog expansion path. The pre-trained graph neural network generates fog expansion path predictions through information propagation and state change modeling between link nodes.
[0071] S5.6 Combine the predicted expansion path of the fog, abnormal link status, and historical link status data to predict the location and propagation trend of the fog, and generate the prediction results of the location and propagation trend of the fog.
[0072] Specifically, by combining fog spread path prediction, abnormal link status, and historical link status data, and considering factors such as signal attenuation, signal-to-noise ratio (SNR) decrease, and bit error rate (BER) increase, the prediction of fog spread path and the abnormal link status are combined. If there is signal strength attenuation, SNR decrease, or BER increase in the link status, it usually reflects environmental deterioration or fog spread. By comparing historical link status and fog spread path prediction, and combining the current link status change trend, it is possible to infer whether it is related to the fog spread path, thereby inferring the link status changes that may affect fog spread. Using historical link status data, by comparing the link status changes of past fog events, the current fog spread path and spread trend are determined, the location and spread trend of the fog are inferred, and the fog location and spread trend prediction results are generated.
[0073] It should also be noted that historical link status data refers to link communication parameter data collected over a period of time, including signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay, reflecting the changes in link status under different time points and environmental conditions throughout history.
[0074] S6. Based on the predicted location and propagation trend of the fog, the core location and influence range of the fog are calculated using the topological relationship of the link status, accurate location information of the fog is generated, dynamic baseline correction is performed, and the reference communication data is updated.
[0075] S6.1 Combine the location of the fog patch, the propagation trend prediction results, and the link status data. By evaluating the changes in the link status, the dense area of the fog patch is calculated, and the core location of the fog patch is generated.
[0076] Specifically, by combining the predicted location and propagation trend of the fog patch with the link status data, and based on the predicted location and propagation trend of the fog patch, the main areas of the fog patch are determined by monitoring the location of the core area of the fog patch and the propagation trend of its surrounding influences, combined with the signal quality changes of the link nodes. Furthermore, by using the signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay in the link status data, areas with abnormal fluctuations in signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay are identified as dense fog patch areas. Finally, by combining the predicted location and propagation trend of the fog patch with the communication quality changes in the link status data, the core location of the fog patch is generated.
[0077] S6.2 Based on the core location of the fog and the link status data, calculate the link connection status and signal attenuation level to generate the fog's influence range.
[0078] Specifically, based on the signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay in the link status data, the difference method is used to calculate the signal attenuation degree within the link area. By comparing the changes in signal strength, the attenuation amplitude of each link is determined. If the signal strength is lower than the effective fluctuation range, it indicates that the signal attenuation degree is significant. Combining the degree of signal attenuation with the connection status between links, the range of the affected links is determined. By synthesizing the status of multiple adjacent links, the affected area of the fog is determined, and the fog influence range is generated.
[0079] For example, by processing the fluctuations in signal strength, signal-to-noise ratio, and bit error rate of each link in the link status data, link areas with signal attenuation exceeding 15dB are identified, indicating significant signal attenuation and potentially within the influence range of fog. By calculating the connectivity between links, areas where multiple links are affected by fog are identified, and these affected areas are integrated to obtain the influence range of fog. If the signal strength change of some links exceeds 15dB and multiple adjacent links are affected, the influence range of fog is determined.
[0080] S6.3 Combine the core location of the fog patch with the range of its influence to calculate the overall area of influence of the fog patch and generate accurate location information for the fog patch.
[0081] Specifically, by combining the core location and the affected area of the fog patch, the central region of the fog patch is determined based on its core location. Combined with the link areas within the affected area where signal attenuation is significant, the impact of the fog patch on the surrounding area is identified. By calculating the distances and signal quality changes between the core location of the fog patch and multiple link nodes within its affected area, the overall impact range of the fog patch on the links is determined. Finally, by combining the link topology and the affected area, the overall affected area of the fog patch is calculated, expressed as: ; Where A represents the overall area affected by the fog, and △S i For the signal attenuation of the i-th link node, d i is the weighted distance between the i-th link node and the core location of the fog, n is the number of link nodes involved in the calculation, and i is the index of the link node; Accurate positioning information for the fog is generated by weighted summation of signal attenuation and distance at multiple link nodes.
[0082] S6.4 Based on the accurate positioning information of the fog, monitor the link status change analysis results in real time, dynamically correct the reference communication data, and generate dynamic baseline correction data.
[0083] Specifically, based on the accurate location information of the fog, the link status change analysis results are monitored in real time to identify changes in signal strength, signal-to-noise ratio (SNR), bit error rate (BER), and successful handshake delay in the link status. These changes are correlated with the accurate location information of the fog to determine whether there are new link status changes or communication quality changes caused by fog expansion. By comparing the link status change analysis results with the accurate location information of the fog, if the signal strength, SNR, BER, or successful handshake delay of the link status exceeds the effective fluctuation range, and the effective fluctuation range is correlated with the changing trend of the core location of the fog, it can be determined that the communication quality change is caused by a new link status change or fog expansion. Based on the link status change analysis results, the effective fluctuation range of signal strength, SNR, BER, and successful handshake delay in the baseline communication data is dynamically adjusted to generate dynamic baseline correction data.
[0084] S6.5. Adjust the real-time communication parameters and update the baseline communication data based on the dynamic baseline correction data.
[0085] Specifically, based on the dynamic baseline correction data, by comparing the fluctuation range of signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay in the real-time communication parameters and the dynamic baseline correction data, it is determined whether the real-time communication parameters need to be adjusted. If the real-time communication parameters exceed the effective fluctuation range, the real-time communication parameters are adjusted according to the dynamic baseline correction data to bring them back to a reasonable range. The adjusted real-time communication parameters are compared with the baseline communication data. If a deviation is found, the baseline communication data is updated according to the new adjustment results to ensure that the baseline communication data more accurately reflects the current link status.
[0086] This embodiment also provides a fog detection system based on the optical communication link status of intelligent road spikes, including: a data acquisition module for acquiring link communication data and performing baseline state learning to generate baseline communication data; a threshold setting module for automatically adjusting and optimizing signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay based on the baseline communication data, monitoring environmental changes and real-time fluctuations in link status, and generating link status threshold settings; a detection calculation module for periodically sending detection frames based on the link status threshold settings, calculating real-time communication parameters, and generating link status data; a link detection module for comparing and analyzing the link status data with the baseline communication data to generate abnormal link status; a path prediction module for analyzing the status changes between multiple links using a pre-trained graph neural network based on the abnormal link status, predicting the expansion path of the fog, and generating fog location and propagation trend prediction results; and a positioning correction module for calculating the core location and influence range of the fog based on the fog location and propagation trend prediction results, utilizing the topological relationship of the link status, generating accurate fog location information, performing dynamic baseline correction, and updating the baseline communication data.
[0087] In summary, this invention achieves accurate prediction of fog expansion paths by utilizing a pre-trained graph neural network to analyze changes in the states of multiple links. It can identify the core location and impact range of fog in advance, thereby realizing proactive, real-time, and accurate fog detection and positioning. This improves the preventive capabilities of road traffic safety management and avoids the lag and inefficiency of traditional methods, demonstrating social value and application potential.
[0088] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A method for detecting patchy fog based on the status of an intelligent road spike optical communication link, characterized in that: include, Collect link communication data and perform baseline state learning to generate baseline communication data; Based on baseline communication data, monitor environmental changes and real-time fluctuations in link status, automatically adjust and optimize signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay, and generate link status threshold settings. Based on the link status threshold setting, probe frames are periodically sent to calculate real-time communication parameters and generate link status data. The link status data is compared and analyzed with the baseline communication data to generate abnormal link status. Based on the abnormal link status, a pre-trained graph neural network is used to analyze the state changes between multiple links, predict the expansion path of the fog, and generate prediction results of the fog location and propagation trend. Based on the predicted location and propagation trend of the fog, the core location and influence range of the fog are calculated using the topological relationship of the link status, accurate location information of the fog is generated, dynamic baseline correction is performed, and the reference communication data is updated.
2. The fog detection method based on the optical communication link status of smart road spikes as described in claim 1, characterized in that: The specific steps for generating the reference communication data are as follows: The smart road spikes collect signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay to form link communication data; Perform statistical analysis on the link communication data, calculate the average and standard deviation of signal strength, signal-to-noise ratio, bit error rate and successful handshake delay in each link, and establish the link baseline state; Based on the link baseline status, the normal fluctuation range of signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay is defined, and baseline communication data is generated.
3. The fog detection method based on the optical communication link status of smart road spikes as described in claim 1, characterized in that: The specific steps for setting the generated link state threshold are as follows. Based on baseline communication data, the system monitors changes in ambient temperature, humidity, and light intensity in real time, as well as fluctuations in signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay in the link communication data, generating link status fluctuation data. Based on link status fluctuation data, the signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay are automatically optimized and adjusted to generate optimized link communication parameters. Based on the optimized link communication parameters, the effective fluctuation range of signal strength, signal-to-noise ratio, bit error rate, and successful handshake delay is defined, and the link state threshold settings are generated.
4. The fog detection method based on the optical communication link status of intelligent road spikes as described in claim 1, characterized in that: The specific steps for generating link status data are as follows: Based on the link status threshold setting, the system periodically sends the sequence number, timestamp, sender ID, and preamble detection information to adjacent road spikes to generate link status detection frames; Upon receiving a link status probe frame, statistical analysis methods are used to calculate and extract real-time communication parameters. The real-time communication parameters are compared with the link status threshold settings one by one to determine whether the signal strength, signal-to-noise ratio, bit error rate and successful handshake delay exceed the effective fluctuation range, and link status data is generated.
5. The fog detection method based on the optical communication link status of smart road spikes as described in claim 1, characterized in that: The specific steps for generating abnormal link states are as follows: When the difference between the signal strength in the link status data and the reference signal strength in the reference communication data exceeds the effective fluctuation range of the signal strength, it is determined to be an abnormal signal attenuation. When the signal-to-noise ratio in the link status data is lower than the effective fluctuation range of the signal-to-noise ratio, it is determined that the signal quality has degraded. When the bit error rate in the link status data is higher than the effective fluctuation range of the bit error rate, it is determined that the communication quality has deteriorated. When the successful handshake delay in the link status data exceeds the effective fluctuation range of the successful handshake delay, the link is determined to be unstable. The system integrates the identified signal attenuation anomalies, signal quality degradation, communication quality deterioration, and link instability to generate an abnormal link status.
6. The fog detection method based on the optical communication link status of intelligent road spikes as described in claim 1, characterized in that: The specific steps for generating the pre-trained graph neural network are as follows: Select a dataset containing the connection relationships and link status information between link nodes, design a graph neural network architecture according to task requirements, define the input features of each link node and edge, and construct a training dataset for the graph neural network. Using the training dataset, the graph neural network parameters are adjusted through forward and backward propagation algorithms to generate graph neural network weights; The weights of the graph neural network are validated using a validation dataset. Once the validation achieves the expected results, the pre-trained weights are saved, and the pre-trained graph neural network is generated.
7. The fog detection method based on the optical communication link status of intelligent road spikes as described in claim 1, characterized in that: The specific steps for predicting the location and propagation trend of the generated fog are as follows. Abnormal link states are input into a pre-trained graph neural network to analyze the state changes between multiple link nodes, capture the dependencies and information propagation patterns between links, and generate link state change analysis results. Based on the link state change analysis results, a pre-trained graph neural network is used to learn the propagation mode of fog, predict the expansion path of fog, and generate fog expansion path prediction. By combining the predicted path of fog spread, abnormal link status, and historical link status data, the location and propagation trend of fog are predicted, generating prediction results for the location and propagation trend of fog.
8. The fog detection method based on the optical communication link status of smart road spikes as described in claim 1, characterized in that: The specific steps for generating accurate positioning information for the fog are as follows. By combining the location of the fog patch, the prediction results of its propagation trend, and the link status data, and by evaluating changes in the link status, the dense area of the fog patch is estimated, and the core location of the fog patch is generated. Based on the core location of the fog and link status data, the connection status of the link and the degree of signal attenuation are calculated to generate the range of influence of the fog. By combining the core location of the fog patch with its influence range, the overall influence area of the fog patch is calculated, generating accurate location information for the fog patch.
9. The fog detection method based on the optical communication link status of intelligent road spikes as described in claim 1, characterized in that: The specific steps for updating the baseline communication data are as follows: Based on the accurate positioning information of the fog, the link status change analysis results are monitored in real time, the baseline communication data is dynamically corrected, and dynamic baseline correction data is generated. Based on the dynamic baseline correction data, adjust the real-time communication parameters and update the baseline communication data.
10. A fog detection system based on the optical communication link status of a smart road spike, based on the fog detection method based on the optical communication link status of a smart road spike as described in any one of claims 1 to 9, characterized in that: include, The data acquisition module is used to collect link communication data, perform baseline state learning, and generate baseline communication data. The threshold setting module is used to monitor environmental changes and real-time fluctuations in link status based on baseline communication data, automatically adjust and optimize signal strength, signal-to-noise ratio, bit error rate and successful handshake delay, and generate link status threshold settings. The detection calculation module is used to periodically send detection frames based on the link status threshold settings, calculate real-time communication parameters, and generate link status data. The link detection module is used to compare and analyze link status data with baseline communication data to generate abnormal link status. The path prediction module is used to analyze the state changes between multiple links based on the abnormal link status, using a pre-trained graph neural network, to predict the expansion path of the fog and generate prediction results of the fog location and propagation trend. The positioning correction module is used to calculate the core location and influence range of the fog based on the predicted location and propagation trend of the fog and the topological relationship of the link status, generate accurate positioning information of the fog, perform dynamic baseline correction, and update the reference communication data.