Method and device for predicting road fog accident risk, electronic equipment and medium

Abstract

This application provides a method, apparatus, electronic device, and medium for predicting the risk of road fog accidents. The method involves acquiring meteorological forecast data for each segment of any road within a prediction area for a target time period. If the meteorological forecast data for the corresponding segment meets the preset fog occurrence conditions for that segment, the meteorological forecast data is input into a pre-trained fog accident meteorological probability prediction model to obtain the predicted meteorological condition probability value for the corresponding segment. Based on the mapping relationship between the configured meteorological factor level definition range and the road fog accident risk level, the road fog accident risk level of the segment corresponding to the predicted meteorological condition probability value is determined for the future target time period. This method improves the accuracy of traffic accident risk prediction and increases the prediction time.

Description

Technical Field

[0001] This application relates to the field of traffic safety risk assessment technology, and more specifically, to a method, device, electronic device, and medium for predicting the risk of road fog accidents. Background Technology

[0002] Fog is one of the most common hazardous weather phenomena on roads, such as highways. With increasing road network density and changes in climate, the impact of fog on highway traffic accidents is becoming increasingly serious. Patchy fog, due to its localized, sudden, and small-scale characteristics, is more likely to cause serious road traffic accidents and is a key risk factor for road traffic safety.

[0003] Accident risk prediction technology helps reduce or avoid losses from meteorological disasters affecting highway traffic, especially for extreme weather events like fog. Road traffic accidents are related to vehicle, road, and meteorological factors, and from a predictability perspective, the occurrence of fog plays a significant role in accident occurrence. However, existing traffic accident risk prediction models, while incorporating visibility indicators, lack consideration for other key meteorological variables such as temperature, humidity, and wind that influence fog formation, dissipation, and accident occurrence, resulting in incompleteness. Furthermore, traffic flow data has a short time representativeness, typically only predicting accident risk for the next 5-10 minutes. Therefore, these methods are generally used for traffic accident identification, real-time monitoring and guidance, or short-term prediction and warning.

[0004] In other words, existing traffic accident risk prediction models suffer from low prediction accuracy and short prediction time. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, electronic device and medium for predicting the risk of road fog accidents, so as to solve the above-mentioned problems existing in the prior art, improve the accuracy of traffic accident risk prediction and increase the prediction time.

[0006] Firstly, a method for predicting the risk of road fog accidents is provided, which may include:

[0007] Obtain meteorological forecast data for each segment of any road within the prediction area during a target time period; the prediction area includes at least one road.

[0008] If the meteorological forecast data of the corresponding road segment meets the preset fog occurrence conditions corresponding to the corresponding road segment, then the meteorological forecast data is input into the pre-trained fog accident meteorological probability prediction model to obtain the fog accident meteorological condition probability prediction value corresponding to the corresponding road segment.

[0009] Based on the mapping relationship between the configured meteorological factor level definition range and the road fog accident risk level, the road fog accident risk level of the segmented road corresponding to the predicted meteorological condition probability value of the fog accident is determined in the future target time period; the meteorological factor level definition range is determined based on the predicted meteorological condition probability value of the fog accident corresponding to fog event samples, non-fog event samples and corresponding meteorological forecast data in historical time periods.

[0010] In one possible implementation, before acquiring the meteorological forecast data for each segment of any road within the prediction area during the target time period, the method further includes:

[0011] The training set and test set are obtained by using the fog event samples, non-fog event samples and corresponding meteorological forecast data of each road segment in the prediction area during historical time periods as training samples.

[0012] Based on the training set and the test set, the machine learning model to be trained is trained to obtain a trained meteorological probability prediction model for fog accidents; the machine learning model to be trained is a model composed of a random forest model and a support vector machine model.

[0013] The meteorological probability prediction model for fog accidents is expressed as follows:

[0014] Where P is the predicted probability value of fog accident weather conditions output by the fog accident meteorological probability prediction model, P i Let α be the probability prediction value output by the i-th model, where i takes the values ​​1 and 2. i These are the weighting coefficients.

[0015] In one possible implementation, the meteorological factor level definition range includes a boundary value of 0, a first critical threshold, a second critical threshold, a third critical threshold, a fourth critical threshold, and a boundary value of 1, which increase sequentially; the configuration process of the meteorological factor level definition range includes:

[0016] Based on the training samples of each fog event in the training set and the fog accident meteorological probability prediction model, the predicted value of the fog accident meteorological condition for the corresponding historical time period of the fog event training sample is obtained.

[0017] According to the preset probability interval, the obtained probability prediction values ​​of meteorological conditions for fog accidents are divided into intervals, and the frequency of fog events in each interval is obtained.

[0018] If, in ascending order of probability prediction values, a continuous interval is obtained in which the frequency of fog events within a set of adjacent preset value intervals is greater than the target frequency, then the average value of the probability prediction values ​​of the fog accident meteorological conditions corresponding to the consecutive preset value intervals is determined as the first critical threshold.

[0019] Based on the fog accident information of the segmented roads where each fog event training sample is located in the training set, the fog event frequency of the fog event sample corresponding to the fog event training sample in each segmented road is obtained, where the segmented road is a road segment corresponding to a preset number of kilometers in the segmented road; the maximum fog event frequency among the fog event frequencies of each segmented road is determined as the disaster index of the fog event sample.

[0020] Obtain the predicted probability value of meteorological conditions for fog accidents corresponding to the first fog event training sample with a disaster index of 1 and the predicted probability value of meteorological conditions for fog accidents corresponding to the second fog event training sample with a disaster index greater than 1.

[0021] The average value of the predicted probability of fog accident meteorological conditions P corresponding to the first fog event training sample is determined as the second critical threshold.

[0022] Meanwhile, in accordance with the requirements for business level quantification, the disaster index of each fog event sample corresponding to the second fog event training sample is divided into two fog event disaster index intervals.

[0023] Based on the two fog event disaster index intervals, the K-means clustering algorithm is used to cluster the disaster index of the fog event samples corresponding to the training samples of the second fog event, so as to obtain the average value of the predicted probability value of the fog accident meteorological conditions corresponding to the two fog event disaster index intervals.

[0024] The average value with the smallest value among the obtained probability prediction values ​​of fog accident meteorological conditions is used as the third critical threshold, and the average value with the largest value is used as the fourth critical threshold.

[0025] In one possible implementation, after determining the road fog accident risk level of the segmented road corresponding to the predicted probability value of the fog accident meteorological conditions in the future target time period, the method further includes:

[0026] Based on the historical road traffic volume and corresponding congestion index of each segment of the predicted area within the historical time period, the traffic factor classification threshold is determined.

[0027] Based on the holiday characteristic data, hourly characteristic data, seasonal characteristic data and trend characteristic data of each road segment within the historical time period, a prophet prediction model is established.

[0028] For any road segment, the prediction algorithm of the prophet prediction model is used to predict the hourly traffic flow of the road segment in the future target time period, and the traffic factor prediction value of the road segment in the future target time period is obtained.

[0029] If the traffic factor prediction value of the segmented road is higher than the traffic factor classification threshold, the risk level of the road fog accident is upgraded according to the preset level value to obtain the first road fog accident risk level of the segmented road.

[0030] In one possible implementation, based on the historical road traffic volume and corresponding congestion index of each segment of road within the predicted area during the historical time period, a traffic factor classification threshold is determined, including:

[0031] After sorting the historical hourly traffic flow of all road segments within the predicted area from smallest to largest by province or region, the hourly traffic flow is divided into multiple percentile intervals and multiple hourly traffic flows within each interval by using a preset percentile as the interval. Each percentile interval is bounded by a dividing percentile, and the difference between two adjacent dividing percentiles is the preset percentile.

[0032] Based on the hourly traffic flow within each interval, the average congestion index corresponding to each interval is obtained.

[0033] For any interval, calculate the linear fitting curve between the average congestion index before any percentile of the interval and the corresponding percentile, and the first goodness of fit of the corresponding fitting curve; at the same time, calculate the index fitting curve between the average congestion index after that percentile of the interval and the corresponding percentile, and the second goodness of fit of the corresponding fitting curve.

[0034] Calculate the average goodness of fit between the first goodness of fit and the second goodness of fit;

[0035] The percentile corresponding to the calculated maximum average goodness of fit is determined as the threshold for traffic factor classification.

[0036] In one possible implementation, after determining the road fog accident risk level of the segmented road corresponding to the predicted probability value of the fog accident meteorological conditions in the future target time period, the method further includes:

[0037] Retrieve the locations of historically frequent fog-prone road sections from the stored database of frequently fog-prone road sections;

[0038] For any road segment, if the location of the road segment is within a preset distance range of a historically frequent fog-prone road section, the risk level of the road fog accident is increased according to a preset level value to obtain the second road fog accident risk level of the road segment.

[0039] In one possible implementation, the method further includes:

[0040] For any road segment, if the predicted traffic factor value of the road segment is higher than the traffic factor classification threshold, or if the location of the road segment is within a preset distance range of the location of historically frequent foggy road sections, then the road foggy accident risk level with the higher risk level between the first road foggy accident risk level and the second road foggy accident risk level is selected as the road foggy accident risk level of the road segment in the future target time period.

[0041] Secondly, a device for predicting the risk of road fog accidents is provided, the device may include:

[0042] The acquisition unit is used to acquire meteorological forecast data of each segment of any road within the prediction area during a target time period; the prediction area includes at least one road.

[0043] Furthermore, if the meteorological forecast data for the corresponding road segment meets the preset fog occurrence conditions for the corresponding road segment, the meteorological forecast data is input into the pre-trained fog accident meteorological probability prediction model to obtain the fog accident meteorological condition probability prediction value for the corresponding road segment.

[0044] The determining unit is used to determine the road fog accident risk level of the segmented road corresponding to the predicted meteorological condition probability value of the fog accident in the future target time period based on the mapping relationship between the configured meteorological factor level definition range and the road fog accident risk level; the meteorological factor level definition range is determined based on the predicted meteorological condition probability value of the fog accident corresponding to fog event samples, non-fog event samples and corresponding meteorological forecast data in historical time periods.

[0045] Thirdly, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0046] Memory, used to store computer programs;

[0047] When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.

[0048] Fourthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of any of the methods described in the first aspect above.

[0049] The method for predicting road fog accident risk provided in this application involves acquiring meteorological forecast data for each segment of any road within a prediction area during a target time period. If the meteorological forecast data for the corresponding segment meets the preset fog occurrence conditions for that segment, the meteorological forecast data is input into a pre-trained fog accident meteorological probability prediction model to obtain the predicted meteorological condition probability value for the corresponding segment. Based on the mapping relationship between the configured meteorological factor level definition range and the road fog accident risk level, the road fog accident risk level for the segment corresponding to the predicted meteorological condition probability value is determined within the future target time period. The meteorological factor level definition range is determined based on historical fog event samples, non-fog event samples, and the predicted meteorological condition probability value for fog accidents corresponding to the meteorological forecast data. This method fully considers the dynamic changes of meteorological factor indicators, improves the accuracy of traffic accident risk prediction, and increases the prediction time by overcoming the problem of short prediction time when using traffic flow data. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 A frequency distribution histogram between the frequency of fog events and the predicted probability of fog accident meteorological conditions is provided in this application embodiment;

[0052] Figure 2 A flowchart illustrating a method for predicting the risk of road fog accidents provided in an embodiment of this application;

[0053] Figure 3 A schematic diagram of a road fog accident risk prediction device provided in an embodiment of this application;

[0054] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0055] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0056] For risk prediction of road fog accidents, existing technologies do not incorporate as many key influencing indicators as possible, especially traffic flow state prediction results; furthermore, they are insufficient in terms of dynamic indicator driving or prediction timeframes. To facilitate operationalization and visualization, threshold definitions are typically used to quantify the levels of indicator factors in accident risk prediction models. However, for meteorological factor level definitions, existing technologies are mostly based on the data's own distribution without considering the differences in the resulting accident severity; for traffic factor level definitions, there is a lack of a grading scheme applicable to any traffic flow data collection method.

[0057] To address the aforementioned issues, this application provides a method for predicting the risk of road fog accidents. This method requires acquiring meteorological forecast data for each segment of any road within the prediction area during a target time period. The road segments are defined by the location of meteorological stations along the respective roads within the prediction area. If the meteorological forecast data for a corresponding road segment meets preset fog occurrence conditions, the forecast data is input into a pre-trained fog accident meteorological probability prediction model to obtain the predicted fog accident meteorological condition probability value for that road segment. Based on the mapping relationship between the configured meteorological factor level definition range and the road fog accident risk level, the road fog accident risk level for the road segment corresponding to the predicted fog accident meteorological condition probability value is determined within the future target time period. The meteorological factor level definition range is determined based on historical fog accident information and the corresponding predicted fog accident meteorological condition probability values. This method references the meteorological environment. When the meteorological environment meets the preset fog occurrence conditions, the matching relationship between the currently obtained predicted fog accident meteorological condition probability value and the configured meteorological factor level definition range is detected. This fully considers the dynamic changes of meteorological factor indicators, improves the accuracy of the prediction, and allows for earlier disaster prevention deployment.

[0058] Furthermore, this method can also take into account road environment and traffic conditions. By integrating as many accident risk influencing factors as possible, the accuracy of predictions can be further improved.

[0059] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0060] Before describing the method for predicting the risk of road fog accidents provided in this application, it is necessary to construct a meteorological probability prediction model for fog accidents and determine the scope of meteorological factor levels. The scope of meteorological factor levels includes successively increasing boundary values ​​of 0, the first critical threshold, the second critical threshold, the third critical threshold, the fourth critical threshold, and the boundary value of 1; the scope of meteorological factor levels corresponds to five risk levels for road fog accidents: Level 1, Level 2, Level 3, Level 4, and Level 5.

[0061] (1) The specific process of constructing a meteorological probability prediction model for fog accidents includes:

[0062] Step 1: Obtain historical fog accident information for the predicted area from the Traffic Management Science Research Institute of the Ministry of Public Security. The predicted area is a region that may include at least one road. For example, obtain information on 883 fog accidents in Jiangsu Province from 2012 to 2016. Fog accident information may include the time of the accident, the province where the accident occurred, the road where the accident occurred, and the specific marker location. Obtain meteorological forecast data along the road for the corresponding historical time period from the National Meteorological Information Center to extract meteorological forecast information for the surrounding area of ​​the accident.

[0063] Step 2: Based on the location of the meteorological station, define the effective range of any road segment within the forecast area. This means that the meteorological forecast data for any road segment within the forecast area is predicted by combining meteorological observation data collected by that meteorological station. Any road may include at least one road segment obtained by dividing the road according to the location of the meteorological station.

[0064] Step 3: Based on the occurrence time and road marker location of each fog incident, integrate the fog incidents in the same time period and road segment corresponding to the prediction area to obtain fog event samples for each road segment. For example, taking 883 fog incidents from 2012 to 2016 as an example, 465 fog event samples can be obtained.

[0065] Step 4: Use historical fog event samples, non-fog event samples, and corresponding weather forecast data for each road segment within the prediction area as training samples to obtain training and test sets. These sets are then used to train the machine learning model to obtain a trained fog accident meteorological probability prediction model. The training samples include fog event training samples consisting of fog event samples and corresponding weather forecast data, and non-fog event training samples consisting of non-fog event samples and corresponding weather forecast data.

[0066] The ratio of fog event samples to non-fog event samples in the training samples is 1:3. 90% of the training samples are used as the training set, and the remaining 10% are used as the test set.

[0067] To further improve the prediction accuracy of machine learning models, feature selection methods such as recursive feature elimination and principal component analysis can be used to preprocess the independent variables of the training set, and establish a relationship model between the accident occurrence time, geographical location, meteorological factors (such as visibility, relative humidity, wind speed, wind direction, temperature and other basic and derived variables) and the hourly fog accident probability, that is, a fog accident meteorological probability prediction model.

[0068] In some embodiments, the machine learning model to be trained in this application can be a model based on a random forest model and a support vector machine model; therefore, the fog accident meteorological probability prediction model is also a model based on a random forest model and a support vector machine model. The fog accident meteorological probability prediction model can be expressed as:

[0069] Where P is the predicted probability value of fog accident weather conditions output by the fog accident meteorological probability prediction model, and α i P represents the weight coefficients corresponding to the i-th model. i Let P1 be the probability prediction value output by the i-th model, where i takes the value 1 or 2. P1 can be the probability prediction value output by the random forest model, and P2 can be the probability prediction value output by the support vector machine model. α1 and α2 can take the values ​​0.48 and 0.52, respectively.

[0070] It should be noted that α i The recall rates of the random forest model and the support vector machine model can be determined based on the training set mentioned above. For example, if the recall rates of the random forest model and the support vector machine model are 0.754 and 0.816 respectively, the weight coefficients can be allocated according to the proportion of the recall rates.

[0071] (2) The specific process for determining the scope of meteorological factor levels includes:

[0072] Step 11: Based on the fog event training samples and the fog accident meteorological probability prediction model in the training set, obtain the predicted value of the fog accident meteorological condition for the corresponding historical time period of the fog event training samples.

[0073] The training samples of each fog event in the training set are input into the fog accident meteorological probability prediction model to obtain the predicted value P of the fog accident meteorological condition for each historical time period corresponding to the training samples of each fog event output by the fog accident meteorological probability prediction model.

[0074] Step 12: Divide the obtained probability prediction values ​​of fog accident meteorological conditions into intervals according to the preset probability intervals, and obtain the frequency of fog events in each interval.

[0075] According to the preset probability interval, such as 0.01 probability interval, the obtained probability prediction value of fog accident meteorological conditions is divided into intervals to obtain multiple probability intervals, and the frequency of fog events in each interval is calculated.

[0076] Step 13: Determine the intervals corresponding to the frequency of fog events that meet the preset target frequency conditions from the frequency of fog events in each interval as the target interval, and determine the first critical threshold based on the predicted probability value of fog accident meteorological conditions corresponding to the target interval.

[0077] According to the probability prediction values ​​in ascending order, obtain the continuous intervals in which the frequency of fog events within the first adjacent preset value intervals is not less than the target frequency, and determine the consecutive preset value intervals (e.g., 3 intervals) as the target intervals; calculate the average value of the probability prediction value of fog accident meteorological conditions corresponding to each target interval, and determine it as the first critical threshold.

[0078] While fog accidents are inherently uncertain as a driving behavior, the consecutive occurrence of multiple fog accidents in similar times and locations reflects an increasing inevitability of traffic accidents induced by extreme weather, suggesting a need to raise the level of meteorological disaster risk. Therefore, a remaining critical threshold is established based on the number of traffic accidents per hour within a road segment unit. Here, a road segment unit refers to a road segment corresponding to a preset kilometer length, such as a segment corresponding to every 1 kilometer.

[0079] To measure the severity of each fog event at the same level, road segment units are determined based on the spatial clustering characteristics of fog accidents. Fog-related traffic accidents exhibit a clear "clustering" phenomenon. Statistical analysis of historical disaster information for the predicted area shows that if multiple fog accidents occur within one hour, they are generally within a 4-kilometer radius, with over 65% occurring within a 1-kilometer radius. Therefore, a 1-kilometer road segment unit is defined.

[0080] Step 14: Based on the fog accident information of the road segments where each fog event training sample is located in the training set, obtain the disaster index of the fog event sample corresponding to each fog event training sample; determine the second critical threshold, the third critical threshold and the fourth critical threshold based on the obtained disaster index of the corresponding fog event sample.

[0081] In practice, based on the fog accident information of the segmented roads where each fog event training sample is located in the training set, the fog accident frequency of the corresponding fog event training sample in each segmented road is obtained. The maximum fog accident frequency among the fog accident frequencies of each segmented road is determined as the fog accident frequency of the fog event sample, which can also be called the disaster index of the fog event sample.

[0082] Obtain the predicted meteorological condition probability P of a fog accident corresponding to a first fog event training sample with a disaster index of 1, and the predicted meteorological condition probability P of a fog accident corresponding to a second fog event training sample with a disaster index greater than 1. The first fog event training sample may include at least one fog event training sample with a disaster index of 1, and the second fog event training sample may include at least one fog event training sample with a disaster index greater than 1.

[0083] (1) The average value of the predicted probability of fog accident meteorological conditions P corresponding to the training samples of the first fog event is determined as the second critical threshold.

[0084] (2) According to the business level quantification requirements, the disaster index of each fog event sample corresponding to the second fog event training sample is divided into two fog event disaster index intervals. Among them, the two fog event disaster index intervals are continuous intervals in terms of the disaster index values ​​of the fog event samples, that is, the frequency of the termination boundary of the first fog event disaster index interval and the frequency of the start boundary of the second fog event disaster index interval are continuous in terms of values;

[0085] It is understandable that the frequency of the starting boundary of the first fog event disaster index interval is 2, and the frequency of the ending boundary of the second fog event disaster index interval is the maximum value of the disaster index of each fog event sample in the training set or positive infinity.

[0086] Based on the disaster index intervals of the two fog events, the K-means clustering algorithm is used to cluster the disaster index of the fog event samples corresponding to the training samples of the second fog event, and the average value of the predicted meteorological condition probability of fog accidents corresponding to the disaster index intervals of the two fog events is obtained.

[0087] The average value with the smallest value among the obtained probability prediction values ​​of meteorological conditions for fog accidents is used as the third critical threshold, and the average value with the largest value is used as the fourth critical threshold.

[0088] As can be seen, based on the above steps, the meteorological factor level definition ranges can be obtained as follows: [0, first critical threshold), [first critical threshold, second critical threshold), [second critical threshold, third critical threshold), [third critical threshold, fourth critical threshold) and [fourth critical threshold, 1].

[0089] A mapping relationship was established between the above five meteorological factor level boundaries and different levels of road fog accident risk levels. Specifically, the first level boundary corresponds to level 1 road fog accident risk level, the second level boundary corresponds to level 2 road fog accident risk level, the third level boundary corresponds to level 3 road fog accident risk level, the fourth level boundary corresponds to level 4 road fog accident risk level, and the fifth level boundary corresponds to level 5 road fog accident risk level. As the risk level value increases, the degree of risk increases sequentially.

[0090] For example, the risk levels of road fog accidents corresponding to the five levels in the meteorological factor level definition range are as follows: low risk (level 1) (or "level V"), relatively low risk (level 2) (or "level IV"), medium risk (level 3) (or "level III"), relatively high risk (level 4) (or "level II") and high risk (level 5) (or "level I").

[0091] In one example, based on the historical time period corresponding to the fog event training samples in the training set obtained in steps 11-13 above, the predicted meteorological conditions for fog accidents can be divided into 0.01 probability intervals. After obtaining the frequency of fog events in each interval, a histogram is plotted between the frequency of fog events and the predicted meteorological conditions for fog accidents, as shown below. Figure 1 As shown, the horizontal axis (X-axis) represents the predicted probability of fog accidents under meteorological conditions; the vertical axis (Y-axis) represents the frequency of fog events. Following the left-to-right order of the horizontal axis, the frequency of fog events in each interval is iterated. When the frequency of fog events in interval A and the subsequent two consecutive intervals B and C are both not less than 2, these three intervals are designated as target intervals. The mean of the predicted probability of fog accidents under meteorological conditions corresponding to these three target intervals is calculated to be 0.19, and this is determined as the first critical threshold, i.e., the first critical threshold is 0.19. At this point, the risk level of a Level 1 road fog accident is determined based on the boundary value of 0 and the range of the first critical threshold of 0.19.

[0092] The prediction area is divided into sections with 1-kilometer intervals. The frequency of fog incidents (I) is then counted for each road segment within the segment containing the fog event sample corresponding to each fog event training sample in the training set. i (i∈n, n is the number of road segment units), and with max(I1, I2, I3, ..., I n The disaster index I is used as a sample of the fog event.

[0093] Based on the accident locations of training samples of fog events in the training set, and after obtaining the disaster index of each fog event sample in the prediction area, the occurrence of a single fog accident is considered the most random. Therefore, a critical threshold is defined as I=1, i.e., the disaster index of the fog event sample is 1. Specifically, the average value of the predicted meteorological probability of fog accidents corresponding to the fog event training samples with a disaster index of 1, which is 0.62, is determined as the second critical threshold. At this point, the risk level of a level 2 road fog accident is determined based on the range between the second critical threshold of 0.62 and the first critical threshold of 0.19.

[0094] For I > 1, i.e., when the disaster index of the fog event sample is greater than 1, K-means clustering is used to determine the risk levels of road fog accidents at levels 3, 4 and 5. Specifically, K-means clustering is used to divide the disaster index corresponding to the fog event training sample with a disaster index greater than 1 into two fog event disaster index intervals [2,5] and [6,∞). The average value of the predicted probability value of the fog accident meteorological conditions in the corresponding fog event disaster index interval is calculated to obtain the third critical threshold of 0.77 and the fourth critical threshold of 0.88.

[0095] The scope of meteorological factor levels and the corresponding risk level of road fog accidents can be expressed as follows:

[0096] ①0≤P<0.19: Level 1

[0097] ②0.19≤P<0.62: Level 2

[0098] ③ 0.62 ≤ P < 0.77: Level 3

[0099] ④ 0.77≤P<0.88: Level 4

[0100] ⑤ 0.88≤P≤1: Level 5

[0101] Based on the established critical threshold, the effectiveness was validated using training samples of fog events in the test set. Validation results show that over 95.7% of the fog event samples correspond to risk levels of 2 or higher; as the risk level increases, the average number of fog accidents within the affected area rises, and approximately 79% of multiple fog events occurring at multiple locations under the background of road fog accident risk levels of 3 or higher.

[0102] Figure 2 This is a flowchart illustrating a method for predicting the risk of road fog accidents, provided in an embodiment of this application. Figure 2 As shown, the method may include:

[0103] Step S210: Obtain meteorological forecast data for each segment of any road within the prediction area during the target time period.

[0104] The predicted area includes at least one road.

[0105] In practice, meteorological forecast data for each segment of any road within the prediction area is obtained for the target time period. Meteorological forecast data may include forecasts for fog / visibility, relative humidity, wind, and temperature.

[0106] Step S220: If the meteorological forecast data of the corresponding road segment meets the preset fog occurrence conditions corresponding to the road segment, then input the meteorological forecast data into the pre-trained fog accident meteorological probability prediction model to obtain the fog accident meteorological condition probability prediction value corresponding to the corresponding road segment.

[0107] In practice, the collected meteorological forecast data will be preprocessed to form meteorological forecast indicators for fog, including meteorological background conditions, relative humidity, daily temperature drop, and wind force.

[0108] To determine whether the meteorological forecast indicators for fog patches in each segment of any road meet the corresponding fog patch occurrence conditions for that segment within the target time period. Since different geographical environments result in different meteorological environments, corresponding fog patch occurrence conditions can be preset for different regions. For example, the fog patch occurrence conditions for Jiangsu Province and Anhui Province are shown in Table 1.

[0109]

[0110] If the fog weather forecast indicators for any segment of road within the target time period meet the preset fog occurrence conditions for the forecast area, then the weather forecast data is input into the pre-trained fog accident weather probability prediction model to obtain the fog accident weather condition probability prediction value corresponding to the weather forecast data.

[0111] Step S230: Based on the mapping relationship between the configured meteorological factor level definition range and the road fog accident risk level, determine the road fog accident risk level of the segmented road corresponding to the predicted meteorological condition probability value of fog accident in the future target time period.

[0112] Furthermore, to improve the accuracy of predictions, traffic factors that characterize road traffic flow and / or road factors that indicate whether a road is in a fog-prone area can be used to adjust and correct the risk level of road fog accidents in the future target time period.

[0113] (1) Adjusting and correcting the risk level of road fog accidents based on traffic factors:

[0114] Traffic risk in foggy weather is basically proportional to traffic volume. Based on the hourly traffic flow changes, traffic factors are divided into two levels, that is, a traffic factor classification threshold is determined to characterize the off-peak (normal) and peak (risk) traffic flow status respectively.

[0115] The system obtains the historical traffic flow and corresponding congestion index of each road segment within the prediction area over a historical period, and determines the traffic factor classification threshold based on the historical traffic flow and corresponding congestion index. Among these, different provinces or regions have different control standards for roads within their jurisdiction, such as highways, and the traffic factor classification threshold can be determined separately for each province or region.

[0116] In some embodiments, the method for determining the traffic factor classification threshold may include:

[0117] The historical hourly traffic flow of all road segments within the predicted area is sorted from smallest to largest by province or region. Then, the sorted hourly traffic flow is divided into cumulative percentiles at preset percentile intervals to obtain multiple percentile intervals and multiple hourly traffic flows within the corresponding intervals.

[0118] Each percentile interval is defined by a dividing percentile as the interval boundary, and the difference between two adjacent dividing percentiles is the preset percentile. For example, if the preset percentile for the interval is 5%, the percentile intervals include 0%-5%, 5%-10%, ..., 95%-100%. In the 5%-10% percentile interval, 5% and 10% are the dividing percentiles of the interval, i.e., the interval boundary.

[0119] The average congestion index is calculated based on the hourly traffic flow within each interval. As road utilization increases, interference between vehicles intensifies, and the congestion index's characteristics with traffic volume growth change significantly, typically shifting from near-linear growth to near-exponential growth.

[0120] For any interval, calculate the linear fitting curve between the average congestion index before any percentile of the interval and the corresponding percentile, and the first goodness of fit of the corresponding fitting curve; at the same time, calculate the exponential fitting curve between the average congestion index after the percentile of the interval and the corresponding percentile, and the second goodness of fit of the corresponding fitting curve.

[0121] Calculate the average goodness of fit of the first and second goodness of fits, and determine the percentile corresponding to the maximum average goodness of fit as the threshold for traffic factor classification.

[0122] In one example, hourly traffic flow data and hourly congestion index for the predicted area are extracted by province for recent years (at least one year, such as 2019-2020). The traffic flow percentile intervals are divided into intervals of 5 percentiles, and the average congestion index is calculated for the intervals of 0%-5%, 5%-10%, ..., 95%-100%.

[0123] The linear fitting formulas for the average congestion index (Y) and percentile (X) before the percentile division (Formula 1) and the exponential fitting formulas for Y and X after the division (Formula 2) can be calculated separately. The first and second goodness-of-fit values ​​of the two fitting curves are recorded. The percentile corresponding to the maximum average goodness-of-fit value is taken as the abrupt change point in the linear change of the congestion index. Taking roads within Jiangsu Province as an example, part of the calculation process is shown in Table 2.

[0124] Table 2

[0125]

[0126] As shown in Table 2, when the 55th percentile of the historical hourly traffic flow data set within the historical time period is used as the dividing point, the average goodness of linear fit before the dividing percentile (0%-55% interval) and the goodness of exponential fit after the dividing percentile (55%-100% interval) reach the highest. This indicates that the characteristics and abrupt changes of the congestion index with traffic volume growth are optimally characterized, and this can be considered the dividing line between a free and stable traffic flow state and a saturated conflict state. Therefore, the 55th percentile of the aforementioned historical hourly traffic flow data set is used as the traffic factor classification threshold for roads within Jiangsu Province. If the hourly traffic flow of any segment of road exceeds the classification threshold in the future target time period, it is considered a peak (risk) traffic state; if it is below the classification threshold, it is considered an off-peak (normal) traffic state.

[0127] Furthermore, after determining the threshold for traffic factor classification, a prophet prediction model can be established based on the holiday characteristic data, hourly characteristic data, seasonal characteristic data, and trend characteristic data of each road segment in historical time periods.

[0128] For any road segment, the prediction algorithm of the Prophet prediction model is used to predict the hourly traffic flow of the road segment in the future target time period, and the traffic factor prediction value of the road segment in the future target time period is obtained.

[0129] If the predicted traffic factor value for a road segment is higher than the traffic factor classification threshold, it indicates that the road segment will be in a peak (risk) traffic condition during the future target time period. This makes traffic accidents more likely. Therefore, the risk level of road fog accidents is increased according to a preset level value, resulting in the first road fog accident risk level for that road segment. For example, if the road segment is in a peak (risk) traffic condition during the future target time period, and the corresponding fog accident risk level is level 3, with a preset level value of 1, the first road fog accident risk level for that road segment is level 4.

[0130] If the predicted traffic factor value of the road segment is not higher than the traffic factor classification threshold, it indicates that the road segment is in off-peak (normal) traffic conditions in the future target time period, and the road fog accident risk level of the road segment is maintained.

[0131] It should be noted that, to improve prediction accuracy, real-time traffic flow data for each road segment within the prediction area can be obtained. Specifically, a historical database of hourly traffic volume and congestion index is constructed and updated in real time; then, based on data from the past N years (N≥1) at the current moment, a traffic factor classification threshold is determined (i.e., dynamic updating of the traffic factor classification threshold is achieved), thereby obtaining the predicted traffic factor values ​​and traffic factor prediction levels for each road segment within the prediction area in the future target time period; finally, the risk level of road fog accidents is dynamically corrected.

[0132] (2) Adjusting and correcting the risk level of road fog accidents based on road factors:

[0133] In practice, information on road sections prone to fog in the past, published by the Ministry of Public Security, is obtained from the Internet or other means to form a database of road sections with potential fog hazards.

[0134] For any road segment within the prediction area, determine whether the location of the road segment is within a preset distance range (e.g., within 500 meters) of the location of historically frequent fog-prone road segments; if yes, mark the road segment as a special road segment; otherwise, mark the road segment as a normal road segment.

[0135] If the road segment is a special section, the risk level of fog accidents needs to be upgraded according to the preset level value to obtain the second risk level of fog accidents for that road segment. For example, if the road segment is a special section and the corresponding fog accident risk level is level 3, and the preset level value is level 1, then the second risk level of fog accidents for that road segment is level 4.

[0136] If the road segment is a regular road segment, the road fog accident risk level for that road segment will remain unchanged.

[0137] (3) Adjust and correct the risk level of road fog accidents based on traffic factors and road factors:

[0138] For any segment of road within the prediction area, if the predicted traffic factor value of the segment of road is higher than the traffic factor classification threshold, or if the location of the segment of road is within a preset distance range of the location of historically frequent foggy road sections, then the road foggy accident risk level with the higher risk level between the first road foggy accident risk level and the second road foggy accident risk level is selected as the road foggy accident risk level of the segment of road in the future target time period.

[0139] Corresponding to the above method, this application also provides a device for predicting the risk of road fog accidents, such as... Figure 3 As shown, the device includes:

[0140] The acquisition unit 310 is used to acquire meteorological forecast data of each segment of any road in the prediction area within the target time period.

[0141] Furthermore, if the meteorological forecast data for the corresponding road segment meets the preset fog occurrence conditions for the corresponding road segment, the meteorological forecast data is input into the pre-trained fog accident meteorological probability prediction model to obtain the fog accident meteorological condition probability prediction value for the corresponding road segment.

[0142] The determining unit 320 is used to determine the road fog accident risk level of the segmented road corresponding to the predicted meteorological condition of the fog accident in the future target time period based on the mapping relationship between the configured meteorological factor level definition range and the road fog accident risk level; the meteorological factor level definition range is determined based on the predicted meteorological condition of the fog accident corresponding to the fog event samples, non-fog event samples and corresponding meteorological forecast data in the historical time period.

[0143] The functions of each functional unit of the road fog accident risk prediction device provided in the above embodiments of this application can be implemented through the above-described method steps. Therefore, the specific working process and beneficial effects of each unit in the device provided in the embodiments of this application will not be repeated here.

[0144] This application also provides an electronic device, such as... Figure 4 As shown, it includes a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440.

[0145] Memory 430 is used to store computer programs;

[0146] When the processor 410 executes the program stored in the memory 430, it performs the following steps:

[0147] Obtain meteorological forecast data for each segment of any road within the prediction area during a target time period; the prediction area includes at least one road.

[0148] If the meteorological forecast data of the corresponding road segment meets the preset fog occurrence conditions corresponding to the corresponding road segment, then the meteorological forecast data is input into the pre-trained fog accident meteorological probability prediction model to obtain the fog accident meteorological condition probability prediction value corresponding to the corresponding road segment.

[0149] Based on the mapping relationship between the configured meteorological factor level definition range and the road fog accident risk level, the road fog accident risk level of the segmented road corresponding to the predicted meteorological condition probability value of the fog accident is determined in the future target time period; the meteorological factor level definition range is determined based on the predicted meteorological condition probability value of the fog accident corresponding to fog event samples, non-fog event samples and corresponding meteorological forecast data in historical time periods.

[0150] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0151] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0152] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0153] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0154] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 2 The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.

[0155] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform any of the methods for predicting the risk of road fog accidents described in the above embodiments.

[0156] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the methods for predicting the risk of road fog accidents described in the above embodiments.

[0157] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, the embodiments in this application can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0158] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0159] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0160] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0161] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.

[0162] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims in this application and their equivalents, then this application also intends to include these modifications and variations.

Claims

1. A method of predicting the risk of a road fogging incident, characterized in that, The method comprises: obtaining meteorological forecast data of each segmented road in any road in a prediction area within a target time period; the prediction area comprises at least one road; if the meteorological forecast data of the corresponding segmented road meets a preset fog occurrence condition corresponding to the corresponding segmented road, inputting the meteorological forecast data into a pre-trained fog accident meteorological probability prediction model to obtain a fog accident meteorological condition probability prediction value corresponding to the corresponding segmented road; determining a road fog accident risk level of the segmented road corresponding to the fog accident meteorological condition probability prediction value within a future target time period based on a mapping relationship between a configured meteorological factor level definition range and a road fog accident risk level; the meteorological factor level definition range is determined based on fog event samples, non-fog event samples and fog accident meteorological condition probability prediction values corresponding to the meteorological forecast data in a historical time period; wherein, before obtaining the meteorological forecast data of each segmented road in any road in the prediction area within the target time period, the method further comprises: obtaining fog event samples, non-fog event samples and corresponding meteorological forecast data of each segmented road in the prediction area in the historical time period as training samples to obtain a training set and a test set; training a machine learning model to be trained based on the training set and the test set to obtain a trained fog accident meteorological probability prediction model; the machine learning model to be trained is a model based on a random forest model and a support vector machine model; The fog accident weather probability prediction model is expressed as: ; wherein P is a fog-accident weather condition probability prediction value output by the fog-accident weather probability prediction model, is a probability prediction value output by the i-th model, i takes values of 1 and 2, is a weight coefficient; wherein, after determining the road fog accident risk level of the segmented road corresponding to the fog accident meteorological condition probability prediction value within the future target time period, the method further comprises: determining a traffic factor classification threshold based on the obtained historical road traffic volume and corresponding congestion index of each segmented road in the prediction area within the historical time period; establishing a prophet prediction model according to holiday feature data, hour feature data, seasonal feature data and trend feature data of each segmented road within the historical time period; for any segmented road, using a prediction algorithm of the prophet prediction model to predict the hourly traffic volume of the segmented road within the future target time period to obtain a traffic factor prediction value of the segmented road within the future target time period; if the traffic factor prediction value of the segmented road is higher than the traffic factor classification threshold, the road fog accident risk level is upgraded according to a preset level value to obtain a first road fog accident risk level of the segmented road.

2. The method of claim 1, wherein, The meteorological factor level definition range comprises boundary values 0, a first critical threshold, a second critical threshold, a third critical threshold, a fourth critical threshold and a boundary value 1 in turn; the configuration process of the meteorological factor level definition range comprises: obtaining fog accident meteorological condition probability prediction values of the historical time period corresponding to each fog event training sample in the training set based on the fog event training sample and the fog accident meteorological probability prediction model; after obtaining the fog accident meteorological condition probability prediction values, obtaining the fog event frequency of each interval after dividing the obtained fog accident meteorological condition probability prediction values into intervals according to a preset probability interval; If the fog event frequency in the first adjacent preset number of intervals is not less than the target frequency in order of the probability prediction value from small to large, the average of the fog accident meteorological condition probability prediction values corresponding to the continuous intervals is determined as the first critical threshold value; Based on the fog accident information of each fog event training sample in the training set, the fog event frequency of each road segment unit of the corresponding fog event sample in the segmented road is obtained, and the road segment unit is the road segment corresponding to a preset kilometer; the maximum fog accident frequency in each road segment unit is determined as the disaster index of the fog event sample; The fog accident meteorological condition probability prediction value corresponding to the first fog event training sample with a disaster index of 1 in the training set and the fog accident meteorological condition probability prediction value corresponding to the second fog event training sample with a disaster index greater than 1 in the training set are obtained; The average of the fog accident meteorological condition probability prediction value P corresponding to the first fog event training sample is determined as the second critical threshold value; At the same time, according to the quantitative demand of business level, the disaster indexes of each fog event sample corresponding to the second fog event training sample are divided into two fog event disaster index intervals; According to the two fog event disaster index intervals, the K-means clustering algorithm is used to cluster the disaster indexes of the fog event samples corresponding to the second fog event training sample, and the average of the fog accident meteorological condition probability prediction values corresponding to the two fog event disaster index intervals is obtained; The average value with a smaller value among the obtained average values of the fog accident meteorological condition probability prediction value is taken as the third critical threshold value, and the average value with a larger value is taken as the fourth critical threshold value.

3. The method of claim 1, wherein, Based on the obtained historical road traffic volume and corresponding congestion index of each segmented road in the prediction area in the historical time period, the traffic factor classification threshold is determined, including: The historical hourly traffic volume of all segmented roads in the prediction area is sorted from small to large, and the sorted hourly traffic volume is divided into multiple percentile intervals at a preset percentile interval, to obtain multiple hourly traffic volumes in the corresponding interval; each percentile interval is bounded by the division percentile, and the difference between adjacent two division percentiles is the preset percentile; Based on the multiple hourly traffic volumes in each interval, the average of the congestion index corresponding to each interval is obtained; For any interval, the linear fitting curve between the average of the congestion index before any division percentile of the interval and the corresponding percentile and the first fitting goodness of the corresponding fitting curve are calculated; at the same time, the exponential fitting curve between the average of the congestion index after the division percentile of the interval and the corresponding percentile and the second fitting goodness of the corresponding fitting curve are calculated; The average fitting goodness of the first fitting goodness and the second fitting goodness is calculated; The division percentile corresponding to the maximum average fitting goodness is determined as the traffic factor classification threshold.

4. The method of claim 1, wherein, After determining the road group fog accident risk level of the segmented road corresponding to the fog accident weather condition probability prediction value in the future target time period, the method further comprises: obtaining the historical group fog-prone road segment positions in the stored group fog-prone road segment database; For any segmented road, if the position of the segmented road is located within the preset distance range of the historical group fog-prone road segment position, the road group fog accident risk level is upgraded according to the preset level value, and the second road group fog accident risk level of the segmented road is obtained.

5. The method of claim 4, wherein, The method further comprises: For any segmented road, if the traffic factor prediction value of the segmented road is higher than the traffic factor grading threshold, or the position of the segmented road is located within the preset distance range of the historical group fog-prone road segment position, the road group fog accident risk level with the greater risk level is selected from the first road group fog accident risk level and the second road group fog accident risk level as the road group fog accident risk level of the segmented road in the future target time period.

6. A device for predicting the risk of a road fogging accident, characterized in that, The device comprises: an acquisition unit configured to acquire meteorological forecast data of each segmented road in any road in a prediction area in a target time period; and if the meteorological forecast data of the corresponding segmented road meets the preset group fog occurrence condition corresponding to the corresponding segmented road, the meteorological forecast data is input into a pre-trained fog accident weather probability prediction model to obtain a fog accident weather condition probability prediction value corresponding to the corresponding segmented road; a determination unit configured to determine a road group fog accident risk level of the segmented road corresponding to the fog accident weather condition probability prediction value in a future target time period based on a mapping relationship between a configured meteorological factor level definition range and a road group fog accident risk level, wherein the meteorological factor level definition range is determined based on fog event samples, non-fog event samples and corresponding fog accident weather condition probability prediction values of meteorological forecast data in a historical time period; a training unit configured to obtain a training set and a test set by taking fog event samples, non-fog event samples and corresponding meteorological forecast data of each segmented road in a prediction area in a historical time period as training samples, and train a machine learning model to be trained based on the training set and the test set to obtain a trained fog accident weather probability prediction model, wherein the machine learning model to be trained is a model based on a random forest model and a support vector machine model; The fog accident weather probability prediction model is expressed as: ; wherein P is a fog-accident weather condition probability prediction value output by the fog-accident weather probability prediction model, is a probability prediction value output by the i-th model, i takes values of 1 and 2, is a weight coefficient; The determination unit is further configured to determine a traffic factor grading threshold based on the historical road traffic and the corresponding congestion index of each segmented road in the prediction area in the historical time period after determining the road group fog accident risk level of the segmented road corresponding to the fog accident weather condition probability prediction value in the future target time period. an establishment unit configured to establish a prophet prediction model according to holiday feature data, hour feature data, seasonal feature data and trend feature data of each segmented road in the historical time period. The prediction unit is configured to, for any segmented road, adopt a prediction algorithm of the prophet prediction model to predict the hourly traffic volume of the segmented road in the future target time period, and obtain a traffic factor prediction value of the segmented road in the future target time period. The acquisition unit is further configured to, if the traffic factor prediction value of the segmented road is higher than a traffic factor grading threshold, upgrade the road group fog accident risk level according to a preset level value to obtain a first road group fog accident risk level of the segmented road.

7. An electronic device, comprising: The electronic device includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete mutual communication through the communication bus; The memory is used for storing a computer program; The processor is used for executing the program stored on the memory to realize the method steps in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to realize the method steps in any one of claims 1-5.

Images