Traffic accident subject information anomaly detection method, system and device and storage medium

Abstract

The invention relates to a traffic accident subject information anomaly detection method, system and device and a storage medium. The serious probability that an accident to be detected is a serious traffic accident is obtained through prediction; calculating statistical parameters of the individual characteristic factors relative to the individual injury degrees according to the individual injury degrees and the regression model of the individual characteristic factors in the traffic accident, and determining individual characteristic factor sets corresponding to different individual injury degrees according to the statistical parameters; according to the individual feature factors and the individual feature factor set in the to-be-detected accident, the injury abnormal probability of the individual injury degree in the to-be-detected accident is calculated; and determining an abnormal result of the subject information of the to-be-detected accident according to the severity probability and the injury abnormal probability. According to the method and the device, the abnormal result of the main information in the traffic accident is detected by utilizing the severity prediction model of the traffic accident and the multistage coupling of the individual injury abnormal probability in the prediction accident, and the problem that the main information abnormity of the traffic accident cannot be detected is solved.

Description

Technical Field

[0001] The present application relates to the field of traffic safety technology, and in particular to a method, system, device and storage medium for detecting abnormal information of traffic accident subjects. Background Art

[0002] Traffic accident details are usually an important indicator for evaluating driving safety in a region. By identifying the severity of the accident, the extent of injuries, and other key information through accident records and information about accident-related personnel, relevant departments can obtain real and effective data and take targeted measures to improve traffic safety.

[0003] Due to the complexity of traffic scenarios and the lag in the evolution of accident data, reported accidents may contain false positives in subject information such as the severity of the accident and the degree of injury. However, there is currently a lack of methods for detecting anomalies in subject information of traffic accidents.

[0004] There is currently no effective solution to the problem that related technologies are unable to detect abnormal information about traffic accident subjects. Summary of the Invention

[0005] Based on this, it is necessary to provide a traffic accident subject information anomaly detection method, system, device and storage medium that can detect traffic accident subject information anomalies in response to the above technical problems.

[0006] First, in this embodiment, a method for detecting abnormal information of a traffic accident subject is provided, comprising:

[0007] Based on the well-trained severity prediction model, the severity probability of the accident to be detected being a serious traffic accident is predicted;

[0008] Calculating statistical parameters of the individual characteristic factors relative to the individual injury levels based on a regression model of the individual injury levels and individual characteristic factors in traffic accidents, and determining sets of individual characteristic factors corresponding to different individual injury levels based on the statistical parameters;

[0009] Calculating the abnormal probability of injury of the individual injury degree in the accident to be detected based on the individual characteristic factors and the set of individual characteristic factors in the accident to be detected;

[0010] An abnormal result of the subject information of the accident to be detected is determined according to the severity probability and the injury abnormality probability.

[0011] In some embodiments, the method further comprises:

[0012] Constructing a first data set based on accident information of historical traffic accidents;

[0013] Based on the first data set, the severity prediction model is trained using an extreme gradient boosting algorithm.

[0014] In some embodiments, the method further comprises:

[0015] A second data set is constructed based on the subject information of historical traffic accidents. The injury degree of individuals without injury is used as a reference class. The injury degree and individual characteristic factors of the second data set are dummy-coded.

[0016] The regression model is constructed based on the individual injury degree and individual characteristic factors processed by dummy coding.

[0017] In some embodiments, the statistical parameters include significance level, correlation coefficient and odds ratio;

[0018] The determining of a set of individual characteristic factors corresponding to different degrees of injury of the individuals according to the statistical parameters includes:

[0019] Screening the individual characteristic factors to obtain an individual characteristic factor set according to the significance level and the significance threshold;

[0020] According to the odds ratio, the individual characteristic factor set is divided into a promotion set and an inhibition set.

[0021] In some embodiments, calculating the injury abnormality probability of the individual injury degree in the accident to be detected based on the individual characteristic factors and the individual characteristic factor set in the accident to be detected includes:

[0022] Determining the promoting individual characteristic factors in the accident to be detected based on the promoting sets corresponding to the injury degrees of each individual, and calculating the first importance coefficient of the promoting individual characteristic factors according to the correlation coefficient;

[0023] Determining, based on the suppression set corresponding to each individual injury degree, a suppression individual characteristic factor in the accident to be detected, and calculating a second importance coefficient of the suppression individual characteristic factor according to a correlation coefficient;

[0024] determining a predicted probability of the individual according to a difference between all the first importance coefficients and all the second importance coefficients;

[0025] If the maximum value among the predicted probabilities of the individual injury degrees is greater than the predicted probability corresponding to the individual injury degrees reported in the accident to be detected, the maximum value among the predicted probabilities of the individual injury degrees is used as the injury abnormality probability.

[0026] In some embodiments, determining the abnormal result of the subject information of the accident to be detected based on the severity probability and the abnormal injury probability includes:

[0027] When the severity probability is greater than a preset threshold value of a serious traffic accident, calculating the individual abnormality probability in the accident to be detected based on the severity probability and the injury abnormality probability;

[0028] The maximum value of the individual abnormality probabilities of the individuals is taken as the abnormality result of the subject information.

[0029] In some embodiments, during the training of the severity prediction model and the regression model, the characteristic factors in the data set are randomly masked.

[0030] Secondly, in this embodiment, a traffic accident subject information anomaly detection system is provided, comprising:

[0031] The severity probability prediction module is used to predict the severity probability of the accident to be detected as a serious traffic accident based on the trained severity prediction model;

[0032] An injury degree anomaly detection module is configured to calculate statistical parameters of the individual characteristic factors relative to the individual injury degree based on a regression model of the individual injury degree and individual characteristic factors in a traffic accident, and determine sets of individual characteristic factors corresponding to different individual injury degrees based on the statistical parameters; and calculate the probability of injury anomaly of the individual injury degree in the accident to be detected based on the individual characteristic factors and the set of individual characteristic factors in the accident to be detected;

[0033] The subject information anomaly detection module is used to determine the abnormal result of the subject information of the accident to be detected based on the severity probability and the injury abnormality probability.

[0034] In a third aspect, a computer device is provided in this embodiment, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for detecting abnormal information of a traffic accident subject as described in the first aspect above is implemented.

[0035] In a fourth aspect, a storage medium is provided in this embodiment, on which a computer program is stored. When the program is executed by a processor, the method for detecting abnormal information of a traffic accident subject as described in the first aspect above is implemented.

[0036] Compared to related technologies, the traffic accident subject information anomaly detection method, system, device, and storage medium provided in this embodiment predicts the severity probability of the accident to be detected being a serious traffic accident based on a well-trained severity prediction model; calculates statistical parameters of the individual characteristic factors relative to the individual injury degree based on a regression model of the individual injury degree and individual characteristic factors in the traffic accident, and determines the set of individual characteristic factors corresponding to different individual injury degrees based on the statistical parameters; calculates the injury anomaly probability of the individual injury degree in the accident to be detected based on the individual characteristic factors in the accident to be detected and the set of individual characteristic factors; and determines the anomaly result of the subject information of the accident to be detected based on the severity probability and the injury anomaly probability. Through this embodiment, the anomaly result of the subject information in the traffic accident is detected by utilizing the multi-level coupling of the traffic accident severity prediction model and the predicted individual injury anomaly probability in the accident, thereby solving the problem of being unable to detect anomalies in the subject information of the traffic accident.

[0037] The details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0039] Figure 1 This is a hardware structure block diagram of a terminal of a method for detecting abnormal traffic accident subject information in an embodiment;

[0040] Figure 2 is a flow chart of a method for detecting abnormal information of a traffic accident subject in an embodiment;

[0041] Figure 3 is a flow chart of a method for detecting abnormal traffic accident subject information in another embodiment;

[0042] Figure 4 is a schematic diagram of a traffic accident subject information anomaly detection framework in one embodiment;

[0043] Figure 5 The present invention is a structural block diagram of a traffic accident subject information anomaly detection system in an embodiment.

[0044] In the figure: 102, processor; 104, memory; 106, transmission device; 108, input and output device; 10, severity probability prediction module; 20, injury degree anomaly detection module; 30, subject information anomaly detection module. DETAILED DESCRIPTION

[0045] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0046] Unless otherwise defined, the technical terms or scientific terms involved in this application should have the general meaning understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "an", "a", "the", "these" and the like in this application do not indicate quantitative restrictions, and they can be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "plurality" involved in this application refers to two or more. "And / or" describes the relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, and B exists alone. Generally, the character " / " indicates that the related objects are in an "or" relationship. The terms "first," "second," "third," etc. used in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0047] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. For example, running on a terminal, Figure 1 This is a hardware structure diagram of the terminal of the traffic accident subject information anomaly detection method of this embodiment. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown) a processor 102 and a memory 104 for storing data, wherein the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. The above terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0048] Memory 104 can be used to store computer programs, such as software programs and modules of application software, such as the computer program corresponding to the traffic accident subject information anomaly detection method in this embodiment. Processor 102 executes the computer program stored in memory 104 to execute various functional applications and data processing, thereby implementing the aforementioned method. Memory 104 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, memory 104 may further include memory remotely located relative to processor 102, and such remote memory may be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0049] The transmission device 106 is used to receive or send data via a network. The network may include a wireless network provided by the terminal's telecommunications provider. In one embodiment, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0050] In this embodiment, a method for detecting abnormal information of traffic accident subjects is provided. Figure 2 is a flow chart of the method for detecting abnormal traffic accident subject information in this embodiment. Figure 2 As shown, the method includes the following steps:

[0051] Step S201 : Based on the well-trained severity prediction model, the severity probability of the accident to be detected being a serious traffic accident is predicted.

[0052] Specifically, this embodiment categorizes traffic accidents based on whether or not there are casualties. For example, traffic accidents can be classified into general traffic accidents without casualties and severe traffic accidents with casualties. Accident information from historical traffic accidents is collected to construct a dataset. Sample data in the dataset is classified and labeled according to the historical accident classifications. A severity prediction model is trained to predict the probability of a traffic accident being a severe accident, which can serve as a classifier.

[0053] Obtain the accident to be detected and its accident information, and use the trained severity prediction model to predict the probability that the accident to be detected is a serious traffic accident. Accident information includes, but is not limited to, accident type, collision mode, collision angle, type of roadside protection facilities, physical separation of roads, road surface conditions, road surface conditions, road surface structure, intersection section type, road alignment, road type, traffic signal method, weather, visibility, lighting conditions, season, presence of large vehicles, presence of pedestrians or non-motor vehicles, vehicle driving status, maximum speed, and other accident characteristic factors.

[0054] Step S202 , based on the regression model of individual injury degree and individual characteristic factors in traffic accidents, calculate the statistical parameters of individual characteristic factors relative to individual injury degree, and determine the individual characteristic factor sets corresponding to different individual injury degrees based on the statistical parameters.

[0055] Specifically, for predicting the severity of an individual's injury in a traffic accident, a regression model is established, using the individual's injury severity as the dependent variable and individual characteristic factors as the independent variables. This regression model is used to analyze the relationship between the individual's injury severity and these individual characteristic factors. By calculating the statistical parameters of the individual characteristic factors relative to the individual's injury severity, the degree of their influence on the individual's injury severity is analyzed, and a set of individual characteristic factors corresponding to each injury severity is obtained. Statistical parameters include, but are not limited to, the correlation coefficient B value, standard deviation, Z value to assist in determining significance, significance level P value, and odds ratio (OR) value.

[0056] The set of individual characteristic factors includes a promoting set and an inhibiting set. For example, individual characteristic factors in the promoting set may be factors that promote an individual's injury level, while individual characteristic factors in the inhibiting set may be factors that inhibit an individual's injury level. Individual injury levels can include no injury, minor injury, severe injury, and death; individual characteristic factors may include gender, age, driving experience, mode of transportation, type of accident, injury site, on-site conditions, and the use of seat belts and helmets.

[0057] Step S203 : determining the injury abnormality probability of the individual injury degree in the accident to be detected based on the individual characteristic factors and the individual characteristic factor set in the accident to be detected.

[0058] Specifically, after obtaining the set of individual characteristic factors corresponding to the degree of injury of each individual, the inhibition set and promotion set therein can be used to obtain the predicted degree of injury of the individual in the accident to be detected. Predicting the degree of injury of the individual is specifically to compare the individual characteristic factors reported in the accident to be detected with the inhibition set and promotion set corresponding to different degrees of injury of the individual, and calculate the probability of the accident to be detected at each degree of injury of the individual according to the inhibition individual characteristic factors and promotion individual characteristic factors in the accident to be detected, where the individual injury degree with the highest probability is the individual injury degree predicted for the accident to be detected. In this way, it can be judged whether the individual injury degree reported in the accident to be detected is consistent with the predicted individual injury degree. If they are consistent, it means that there is no injury abnormality. If they are inconsistent, it means that there is an injury degree abnormality in the accident to be detected, and its injury abnormality probability is the probability value of the predicted individual injury degree.

[0059] Step S204: determining the abnormal result of the subject information of the accident to be detected based on the severity probability and the injury abnormality probability.

[0060] Specifically, based on the severity probability of the accident being detected being a serious traffic accident, when the accident is determined to be a serious accident, the injury abnormality probability of the individual injury level in the accident is further obtained. By combining the severity probability and the injury abnormality probability in a multi-level manner, the abnormality probability of each individual in the accident is comprehensively determined, thereby obtaining an abnormal result for the subject information of the accident being detected. The abnormal result can be an abnormal probability value, or it can be further obtained based on the abnormal probability value and a threshold to determine whether the subject information is abnormal.

[0061] Through the above steps, the severity probability of the accident to be detected can be predicted, and the characteristic factor correlation analysis of the individual injury degree can be performed to identify the abnormal probability of injury of the individual injury degree. Finally, based on the severity probability and the abnormal probability of injury, multi-level coupled subject information anomaly detection is performed from the two levels of anomaly of the accident to be detected and anomaly of the individual injury degree in the accident to be detected, which solves the problem of being unable to detect anomaly of the subject information of traffic accidents.

[0062] In some embodiments, the method further comprises the following steps:

[0063] A first data set is constructed based on accident information of historical traffic accidents; based on the first data set, an extreme gradient boosting algorithm is used to train a severity prediction model.

[0064] Specifically, accident information includes but is not limited to accident types (rear-end collision, rollover, collision with guardrail, multi-vehicle chain collision, etc.), collision mode (frontal collision, side collision, rear-end collision, corner collision, etc.), collision angle (0°, 90°, 180° rear-end collision, etc.), road side protection facility type (no protection, concrete guardrail, corrugated beam guardrail, green belt, etc.), road physical isolation (no isolation, central isolation belt, isolation fence, double yellow line, etc.), road surface condition (flat, potholes, cracks, etc.), road surface condition (dry, waterlogged, icy, snowy, etc.), road surface structure (asphalt pavement, cement concrete pavement, gravel pavement, etc.), intersection section type (crossroads, T-intersection, roundabout, straight line, etc.), accident characteristic factors such as road sections, curved road sections, etc.), road alignment (straight line, gentle curve, sharp curve, continuous curve, etc.), road type (expressway, urban expressway, national highway, provincial highway, rural road, etc.), traffic signal method (traffic light, stop sign, no signal control, traffic police command, etc.), weather (sunny, rainy, snowy, foggy, sandstorm, etc.), visibility (1000 meters, 500 meters, 200 meters, etc.), lighting conditions (natural light during the day, street lights at night, no street lights at night, etc.), season (spring, summer, autumn, winter), whether there are large cars, whether there are pedestrians or non-motor vehicles, vehicle driving status (going straight, changing lanes, reversing, sudden braking, speeding, etc.), maximum driving speed, etc.

[0065] The extreme gradient boosting (XGBoost) algorithm is used to train a severity prediction model. This model is used to classify the severity of traffic accidents. It can obtain the general probability of predicting the accident to be a general traffic accident and the severe probability of predicting it to be a severe traffic accident.

[0066] During the model training process, the sample data in the first dataset, which includes accident information and characteristic factors for both general and serious traffic accidents, is classified and labeled using the aforementioned classification criteria. The sample data is then divided into a training set and a test set according to a certain ratio. The accident characteristic factors are then encoded, standardized, and subjected to feature engineering processing before being input into the model for training. The training process also uses a GridSearch method to determine the optimal values for model parameters such as the column sampling rate and the minimum number of leaf node samples. Finally, the severity prediction model gradually constructs multiple decision trees using an additive model. Each tree is fitted with the residuals of the previous tree, and a weighted voting process is used to output the result, namely, the predicted probability of a serious traffic accident.

[0067] In this embodiment, a severity prediction model is constructed based on accident characteristic factors to provide a prediction of the severity probability of the accident to be detected being a serious traffic accident from the perspective of traffic accidents, providing accident-level information for subsequent analysis of whether the subject information is abnormal.

[0068] In some embodiments, the method further comprises the following steps:

[0069] A second data set was constructed based on the main information of historical traffic accidents. The injury degree of uninjured individuals was used as the reference class, and the individual injury degree and individual characteristic factors in the second data set were dummy coded. A regression model was constructed based on the dummy coded individual injury degree and individual characteristic factors.

[0070] Specifically, the subject information includes the individuals in the accident to be detected, the degree of individual injuries, and individual characteristic factors. Individual characteristic factors include gender, age, driving experience, mode of transportation (small car, motorcycle, electric bicycle, truck, bus), type of accident (speeding, drunk driving, running a red light, failure to maintain a safe distance, fatigue driving, etc.), injury site (head, chest, limbs, spine, etc.), on-site conditions (weather, lighting, etc.), the use of seat belts and helmets, etc.

[0071] The individual injury severity and individual characteristic factors in the second dataset were dummy-coded, and a regression model was established for correlation analysis between the individual injury severity and these individual characteristic factors. Among the individual characteristic factors, age and driving experience were treated as discrete intervals. For example, the age field was used as an example, and the resulting encoding is shown in Table 1. Multi-category features such as gender and seatbelt and helmet use can be encoded using two parameters. The injury severity is also discrete, with 1 representing death, 2 representing severe injury, 3 representing minor injury, and 4 representing no injury. No injury was used as the reference category, and the encoding is shown in Table 2.

[0072] Table 1

[0073]

[0074] Table 2

[0075]

[0076] By constructing a regression model that associates individual injury levels with individual characteristic factors in this embodiment, the causes of individual injury levels can be analyzed and subsequently used to predict individual injury levels.

[0077] In some embodiments, during the training of the severity prediction model and the regression model, the characteristic factors in the data set are randomly masked.

[0078] Specifically, since the construction of the above-mentioned model involves a large number of factors participating in the analysis, and the data used for modeling are all high-quality data after review, the trained model may be overfitting and unable to adapt to the quality of the data to be reviewed, resulting in low accuracy of the constructed model when actually used.

[0079] In order to prevent the impact of local abnormal data in the accident to be detected on the final results, a robust design with random masking of sample factors is added during model training. The specific design method is as follows:

[0080] The sample data is randomly split into a training sample set and a test sample set in a ratio of 3:1; 5% of the sample data is randomly selected from the training sample data, and the values corresponding to 1-2 feature factors randomly selected from the first data set and the second data set are masked, where the masking process can set the values corresponding to the feature factors to missing values.

[0081] In some embodiments, determining a set of individual characteristic factors corresponding to different injury levels of individuals based on statistical parameters includes the following steps:

[0082] According to the significance level and significance threshold, the individual characteristic factor set is screened from the individual characteristic factors; according to the odds ratio, the individual characteristic factor set is divided into a promotion set and an inhibition set.

[0083] Specifically, the established regression model was subjected to a Hosmer-Lemeshow goodness-of-fit test. After passing the model fit test, the statistical parameters of individual characteristic factors relative to individual injury severity were calculated to analyze the degree of influence of individual characteristic factors on individual injury severity, thereby obtaining a set of individual characteristic factors corresponding to each injury severity. Statistical parameters include, but are not limited to, the correlation coefficient B value, standard deviation, Z value for auxiliary significance judgment, significance level P value, and odds ratio (OR) value. Table 3 shows the statistical parameter analysis table, with no injury as the reference class.

[0084] Table 3

[0085]

[0086] Based on the significance level and significance threshold, individual characteristic factors with a significance level below the significance threshold are considered to have a significant influence relationship with the different degrees of injury of the individual who is relatively harmless, while individual characteristic factors with a significance level above the significance threshold are considered to have no significant influence relationship with the different degrees of injury of the individual who is relatively harmless. In this way, individual characteristic factors with a significance level below the significance threshold are selected from all individual characteristic factors to form the individual characteristic factor set. For example, the significance threshold can be 0.05.

[0087] Furthermore, an individual injury degree analysis is performed based on the individual characteristic factors in the individual characteristic factor set, and the individual characteristic factor set is divided according to the odds ratio OR value. The individual characteristic factor set is divided into a promotion set with an OR value greater than 1 and an inhibition set with an OR value less than 1. For example, the individual characteristic factors in the promotion set may be factors that promote an individual to a certain injury degree, and the individual characteristic factors in the inhibition set may be factors that inhibit an individual to a certain injury degree.

[0088] After obtaining the individual characteristic factor set corresponding to the degree of injury of each individual in this embodiment, the inhibition set and promotion set therein can be used to predict the degree of injury of the individual in the accident to be detected, so as to better identify abnormalities in the reported degree of injury of the individual.

[0089] In some embodiments, calculating the injury abnormality probability of an individual injury degree in an accident to be detected based on the individual characteristic factors and the set of individual characteristic factors in the accident to be detected includes the following steps:

[0090] Based on the promotion set corresponding to each individual injury degree, the promoting individual characteristic factors in the accident to be detected are determined, and the first importance coefficient of the promoting individual characteristic factors is calculated based on the correlation coefficient; based on the inhibition set corresponding to each individual injury degree, the inhibitory individual characteristic factors in the accident to be detected are determined, and the second importance coefficient of the inhibitory individual characteristic factors is calculated based on the correlation coefficient; based on the difference between all the first importance coefficients and all the second importance coefficients, the predicted probability of the individual injury degree is determined; if the maximum value among the predicted probabilities of the individual injury degree is greater than the predicted probability corresponding to the reported individual injury degree in the accident to be detected, then the maximum value among the predicted probabilities of the individual injury degree is taken as the injury abnormality probability.

[0091] Specifically, the prediction of individual injury degree is to compare the individual characteristic factors reported in the accident to be detected with the inhibition set and promotion set corresponding to different individual injury degrees, and calculate the predicted probability of the accident to be detected at each individual injury degree based on the inhibition individual characteristic factors and promotion individual characteristic factors in the accident to be detected. The individual injury degree with the highest probability is the individual injury degree predicted for the accident to be detected.

[0092] The importance coefficient I of individual characteristic factor i d,i The calculation is as follows:

[0093]

[0094] Where d represents the individual injury degree, N represents the total number of individual injury degrees, and B represents the correlation coefficient of individual characteristic factor i under individual injury degree d.

[0095] According to the above formula, the first importance coefficient of the promoting individual characteristic factor in the accident to be detected and the second importance coefficient of the inhibiting individual characteristic factor in the accident to be detected are calculated respectively. According to the difference between all the first importance coefficients and all the second importance coefficients, the predicted probability of the individual injury degree is determined. The specific calculation method is as follows:

[0096] p n =sum(set_gt(I d,i ))–sum(set_lt(I d,i ));

[0097] Among them, p n Indicates the predicted probability of individual injury level, set_gt(I d,i ) represents the first importance coefficient of promoting individual characteristic factors, set_lt(I d,i ) represents the second importance coefficient of suppressing individual characteristic factors, and sum() represents cumulative calculation.

[0098] This can be used to determine whether the reported individual injury level in the accident to be detected is consistent with the predicted individual injury level. If they are consistent, it means that there is no injury abnormality. If they are inconsistent, it means that there is an injury abnormality in the accident to be detected, and the probability of the injury abnormality is the probability value of the predicted individual injury level.

[0099] By analyzing and predicting the probability of individual injury levels using the individual injury level correlation factor obtained through regression model analysis in this embodiment, it is possible to further identify whether there are any abnormalities in the individual injury levels reported in the accident to be detected, as well as the probability of abnormal injuries.

[0100] In some embodiments, determining abnormal results of subject information of an accident to be detected based on the severity probability and the injury abnormality probability includes the following steps:

[0101] When the severity probability is greater than the preset threshold of a serious traffic accident, the individual abnormal probability in the accident to be detected is calculated based on the severity probability and the abnormal probability of injury; the maximum value of the individual abnormal probability of each individual is taken as the abnormal result of the subject information.

[0102] Specifically, the probability of predicting that the accident to be detected is a serious traffic accident is obtained. Combined with the preset threshold value P for serious traffic accidents A , if the probability of serious traffic accidents is greater than the preset threshold This indicates that there are individual casualties in the predicted accident to be detected, and the degree of injury to the individual in the accident to be detected is reported for abnormal identification. For example, the preset threshold for serious traffic accidents can be set to 0.8.

[0103] When the accident to be detected is predicted to be a serious traffic accident, the abnormal probability p of the injury degree of the individual in the accident to be detected is further obtained. abn Based on the severity probability and injury abnormality probability, the individual abnormality probability in the accident to be detected is calculated. The specific calculation is as follows:

[0104]

[0105] in, is the probability of abnormal information of the jth individual accident subject in the i-th traffic accident; p abn is the injury abnormality probability of the individual injury level in the accident to be detected; A serious probability.

[0106] The abnormal probability p of the subject information in the accident to be detected is further obtained as follows:

[0107] p=max(P i,1 d ,P i,1 2 ,…,P i,n d );

[0108] The abnormal result can be an abnormal probability value, or it can be further used to determine whether the subject information is abnormal based on the abnormal probability value and threshold. For all incidents to be detected, the abnormal probability p of all subject information predictions can be obtained according to the above steps. The abnormal result is obtained by sorting the probability from highest to lowest as a recommended order.

[0109] Furthermore, the abnormal results can provide an explanatory basis for data anomalies. The individual characteristic factors and the set of individual characteristic factors of each individual in the accident to be detected are analyzed one by one, and the promoting individual characteristic factors and the inhibiting individual characteristic factors are analyzed as the explanatory basis for the abnormality of the subject in the accident to be detected.

[0110] In this embodiment, multi-level coupled subject information anomaly detection can be performed from the two levels of anomaly of the accident to be detected and anomaly of the degree of individual injury in the accident to be detected, and abnormal information such as the degree of individual injury that is mistakenly reported in the accident to be detected can be further repaired.

[0111] The present embodiment is described and illustrated below through preferred embodiments.

[0112] Figure 3 is a flow chart of the traffic accident subject information anomaly detection method of this embodiment, as shown in FIG. Figure 3 As shown, the method includes the following steps:

[0113] Step S301: construct a first data set based on accident information of historical traffic accidents; and based on the first data set, use the extreme gradient boosting algorithm to train and obtain a severity prediction model.

[0114] Step S302 : construct a second data set based on the main information of historical traffic accidents, take the injury degree of uninjured individuals as the reference class, and construct a regression model based on the individual injury degree and individual characteristic factors processed by dummy coding in the second data set.

[0115] Step S303 : Based on the fully trained severity prediction model, the severity probability of the accident to be detected being a serious traffic accident is predicted.

[0116] Step S304: Calculate statistical parameters of individual characteristic factors relative to individual injury levels based on the regression model, and determine individual characteristic factor sets corresponding to different individual injury levels based on the statistical parameters; the individual characteristic factor sets include a promotion set and an inhibition set.

[0117] Step S305: Based on the promotion set corresponding to each individual injury degree, determine the promoting individual characteristic factors in the accident to be detected and the first importance coefficient of the promoting individual characteristic factors; based on the inhibition set corresponding to each individual injury degree, determine the inhibition individual characteristic factors in the accident to be detected and the second importance coefficient of the inhibition individual characteristic factors.

[0118] Step S306: Determine the predicted probability of the individual injury degree based on the difference between all the first importance coefficients and all the second importance coefficients; if the maximum value among the predicted probabilities of the individual injury degree is greater than the predicted probability corresponding to the individual injury degree reported in the accident to be detected, then the maximum value among the predicted probabilities of the individual injury degree is used as the injury abnormality probability.

[0119] Step S307: When the severity probability is greater than the preset threshold of a serious traffic accident, the individual abnormal probability in the accident to be detected is calculated based on the severity probability and the injury abnormal probability; the maximum value of the individual abnormal probability of each individual is used as the abnormal result of the subject information.

[0120] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0121] Figure 4 This is a schematic diagram of the traffic accident subject information anomaly detection framework in this embodiment. Figure 4As shown, historical traffic accidents are accessed to predict accident severity and analyze individual injury factors. A severity prediction model is constructed to predict the probability of a serious accident. A regression model is also constructed to analyze individual characteristic factors associated with individual injury severity. This model, combining the probability of traffic accident severity and the probability of individual injury anomalies, is used to develop an accident subject information anomaly recognition model. Reported accidents to be detected are then fed into the accident subject information anomaly recognition model to generate algorithm-recommended anomaly results and explanatory evidence.

[0122] This embodiment also provides a traffic accident subject information anomaly detection system, which is used to implement the above-mentioned embodiments and preferred embodiments. Details already described will not be repeated here. The terms "module," "unit," "subunit," etc. used below refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0123] Figure 5 This is a structural block diagram of the traffic accident subject information anomaly detection system in this embodiment. Figure 5 As shown, the system includes:

[0124] The severity probability prediction module 10 is used to predict the severity probability of the accident to be detected being a serious traffic accident based on the well-trained severity prediction model;

[0125] The injury degree anomaly detection module 20 is used to calculate the statistical parameters of the individual characteristic factors relative to the individual injury degree based on the regression model of the individual injury degree and individual characteristic factors in the traffic accident, and determine the individual characteristic factor sets corresponding to different individual injury degrees based on the statistical parameters; and calculate the injury anomaly probability of the individual injury degree in the accident to be detected based on the individual characteristic factors and the individual characteristic factor sets in the accident to be detected;

[0126] The subject information anomaly detection module 30 is used to determine the abnormal result of the subject information of the accident to be detected based on the severity probability and the injury abnormality probability.

[0127] Through the system provided in this embodiment, it is possible to predict the probability of severity of the accident to be detected, perform characteristic factor correlation analysis on the individual injury degree, identify the abnormal probability of injury of the individual injury degree, and finally, based on the severity probability and the abnormal probability of injury, perform multi-level coupled subject information anomaly detection from two levels: the anomaly of the accident to be detected and the anomaly of the individual injury degree in the accident to be detected, thereby solving the problem of being unable to detect anomalies in the subject information of traffic accidents.

[0128] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0129] This embodiment further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0130] Optionally, the computer device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0131] It should be noted that, for specific examples in this embodiment, reference may be made to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0132] In addition, in conjunction with the traffic accident subject information anomaly detection method provided in the above embodiments, this embodiment may also provide a storage medium for implementation. The storage medium stores a computer program; when the computer program is executed by a processor, it implements any of the traffic accident subject information anomaly detection methods provided in the above embodiments.

[0133] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0134] It should be understood that the specific embodiments described herein are only used to explain this application and are not used to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0135] Obviously, the accompanying drawings are merely examples or embodiments of the present application. A person skilled in the art can also apply the present application to other similar situations based on these drawings without inventive effort. Furthermore, it is understandable that, although the work involved in this development process may be complex and lengthy, certain design, manufacturing, or production changes based on the technical content disclosed in this application are merely routine technical means for a person skilled in the art and should not be considered to constitute a deficiency in the disclosure of the present application.

[0136] The term "embodiment" as used in this application refers to specific features, structures, or characteristics described in conjunction with the embodiment that can be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is understood, either explicitly or implicitly, by those skilled in the art that the embodiments described in this application can be combined with other embodiments when there is no conflict.

[0137] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of patent protection. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for detecting abnormal information of traffic accident subjects, characterized in that: include: Based on the well-trained severity prediction model, the severity probability of the accident to be detected being a serious traffic accident is predicted; Calculating statistical parameters of the individual characteristic factors relative to the individual injury levels based on a regression model of the individual injury levels and individual characteristic factors in traffic accidents, and determining sets of individual characteristic factors corresponding to different individual injury levels based on the statistical parameters; Calculating the abnormal probability of injury of the individual injury degree in the accident to be detected based on the individual characteristic factors and the set of individual characteristic factors in the accident to be detected; An abnormal result of the subject information of the accident to be detected is determined according to the severity probability and the injury abnormality probability.

2. The method for detecting abnormal traffic accident subject information according to claim 1, characterized in that: The method further comprises: Constructing a first data set based on accident information of historical traffic accidents; Based on the first data set, the severity prediction model is trained using an extreme gradient boosting algorithm.

3. The method for detecting abnormal traffic accident subject information according to claim 1, characterized in that: The method further comprises: A second data set is constructed based on the subject information of historical traffic accidents. The injury degree of individuals without injury is used as a reference class. The injury degree and individual characteristic factors of the second data set are dummy-coded. The regression model is constructed based on the individual injury degree and individual characteristic factors processed by dummy coding.

4. The method for detecting abnormal traffic accident subject information according to claim 1, characterized in that: The statistical parameters include significance level, correlation coefficient and odds ratio; The determining of a set of individual characteristic factors corresponding to different degrees of injury of the individuals according to the statistical parameters includes: Screening the individual characteristic factors to obtain an individual characteristic factor set according to the significance level and the significance threshold; According to the odds ratio, the individual characteristic factor set is divided into a promotion set and an inhibition set.

5. The method for detecting abnormal traffic accident subject information according to claim 1, characterized in that: The calculating, based on the individual characteristic factors and the set of individual characteristic factors in the accident to be detected, the injury abnormality probability of the individual injury degree in the accident to be detected includes: Determining the promoting individual characteristic factors in the accident to be detected based on the promoting sets corresponding to the injury degrees of each individual, and calculating the first importance coefficient of the promoting individual characteristic factors according to the correlation coefficient; Determining, based on the suppression set corresponding to each individual injury degree, a suppression individual characteristic factor in the accident to be detected, and calculating a second importance coefficient of the suppression individual characteristic factor according to a correlation coefficient; determining a predicted probability of the individual's injury degree based on a difference between all of the first importance coefficients and all of the second importance coefficients; If the maximum value among the predicted probabilities of the individual injury degrees is greater than the predicted probability corresponding to the individual injury degrees reported in the accident to be detected, the maximum value among the predicted probabilities of the individual injury degrees is used as the injury abnormality probability.

6. The method for detecting abnormal traffic accident subject information according to claim 1, characterized in that: The determining of the abnormal result of the subject information of the accident to be detected according to the severity probability and the abnormal injury probability includes: When the severity probability is greater than a preset threshold value of a serious traffic accident, calculating the individual abnormality probability in the accident to be detected based on the severity probability and the injury abnormality probability; The maximum value of the individual abnormality probabilities of the individuals is taken as the abnormality result of the subject information.

7. The method for detecting abnormal traffic accident subject information according to any one of claims 1 to 3, characterized in that: During the training of the severity prediction model and the regression model, the characteristic factors in the data set are randomly masked.

8. A traffic accident subject information anomaly detection system, characterized in that: include: The severity probability prediction module is used to predict the severity probability of the accident to be detected as a serious traffic accident based on the trained severity prediction model; An injury degree anomaly detection module is configured to calculate statistical parameters of the individual characteristic factors relative to the individual injury degree based on a regression model of the individual injury degree and individual characteristic factors in a traffic accident, and determine sets of individual characteristic factors corresponding to different individual injury degrees based on the statistical parameters; and calculate the probability of injury anomaly of the individual injury degree in the accident to be detected based on the individual characteristic factors and the set of individual characteristic factors in the accident to be detected; The subject information anomaly detection module is used to determine the abnormal result of the subject information of the accident to be detected based on the severity probability and the injury abnormality probability.

9. A computer device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the traffic accident subject information anomaly detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for detecting abnormality in traffic accident subject information according to any one of claims 1 to 7 are implemented.

Images