TWI758915B - Driving risk analysis apparatus and method thereof - Google Patents

Abstract

A driving risk analysis apparatus and a corresponding method are provided. In an embodiment, the method includes setting a plurality of risk factors related to driving safety, collecting a plurality of risk events of at least one vehicle; and calculating a risk value according to the risk factors and the risk events. In another embodiment, the method includes setting a plurality of risk factors related to driving safety, collecting an accident history, a plurality of first risk events and a plurality of second risk events of at least one vehicle; and predicting an accident risk probability corresponding to the second risk events according to the accident history and the first risk events.

Description

Driving risk analysis equipment and method

The present invention relates to a driving risk analysis technology, and particularly relates to a driving risk analysis device and method based on driving risk factors and risk events.

With the increasing awareness of driving safety, in the management of commercial vehicles, how to assess driving risks and even predict accident probability has become an important issue. Therefore, for many vehicles and many drivers in the fleet, there is an urgent need for a feasible system or method to quantitatively assess the driving risk of the driver, or estimate the driver's accident risk probability, as an aid for management decision-making.

In order to solve the above problems, the present invention provides a driving risk analysis device, comprising: a risk data collection module for setting multiple risk factors related to driving safety and collecting multiple risk events of at least one vehicle, wherein each risk event includes identifying value, event type and risk level, and the event type and risk level of each risk event correspond to the same risk factor among the corresponding risk factors; the risk coefficient adjustment module is used to set or adjust the various risk factors of each risk factor. The proportion of the risk level coefficients; and a risk calculation module for calculating the risk values corresponding to the identification values according to the risk levels and the proportion coefficients.

The present invention further provides a driving risk analysis device, comprising: a risk data collection module for setting a plurality of risk factors related to driving safety, and collecting the accident history of at least one vehicle, a plurality of first risk events and a plurality of second risk events; A risk factor training module, used for data training based on the accident history and the first risk events, to calculate the accident probability value of each of the risk factors; and an accident prediction module, used for the accident based on each of the risk factors The probability value and the second risk events predict the accident risk probability corresponding to the second risk events.

The present invention further provides a driving risk analysis method, comprising: setting a plurality of risk factors related to driving safety; collecting a plurality of risk events of at least one vehicle, wherein each risk event includes an identification value, an event type and a risk level, and each risk The event type and the risk level of the event all correspond to the same risk factor among the corresponding risk factors; set or adjust the weighting factor of each risk level of each of the risk factors; and calculate the Calculate the risk value corresponding to each identification value.

The present invention further provides a driving risk analysis method, comprising: setting a plurality of risk factors related to driving safety; collecting the accident history, a plurality of first risk events and a plurality of second risk events of at least one vehicle; according to the accident history and the first risk events Perform data training on risk events to calculate the accident probability value of each risk factor; and predict the accident risk probability corresponding to the second risk events according to the accident probability value of each risk factor and the second risk events.

In one embodiment, the driving risk value of the driver is calculated according to the risk factor, and an objective quantitative value is provided, so as to identify drivers with high driving risk and provide guidance for improvement, or to identify drivers with low driving risk. encourage risky drivers. In addition, in another embodiment, a machine learning method is used and data training is performed according to historical data to predict the accident risk probability.

100,300,500: Driving risk analysis equipment

110, 310, 510: Risk Data Collection Module

112,312,512: Risk Factor Setting Module

114,314,514: Risk Data Entry Module

116,316,516:Database Module

120,520: Risk factor adjustment module

130,530: Risk calculation module

320,540: Risk Factor Training Module

330,550: Accident Prediction Module

S210~S240, S410~S440: method steps

FIG. 1 is a block diagram of a driving risk analysis device according to a first embodiment of the present invention.

FIG. 2 is a flowchart of a driving risk analysis method according to the first embodiment of the present invention.

3 is a block diagram of a driving risk analysis device according to a second embodiment of the present invention.

FIG. 4 is a flowchart of a driving risk analysis method according to a second embodiment of the present invention.

FIG. 5 is a block diagram of a driving risk analysis device according to a third embodiment of the present invention.

The embodiments of the present invention are described below by means of specific embodiments, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification.

FIG. 1 is a block diagram of a driving risk analysis device 100 according to a first embodiment of the present invention. The driving risk analysis device 100 includes a risk data collection module 110, a risk coefficient adjustment module 120, and a risk calculation module 130, wherein the risk data collection module 110 includes a risk factor setting module 112, a risk data input module 114, And the database module 116. Each module in Figure 1 can be software, hardware or firmware; if it is hardware, it can be a processing unit, processor, computer or server with data processing and computing capabilities; if it is software or firmware , it may include instructions executable by a processing unit, processor, computer or server. In addition, each of the modules in FIG. 1 can be integrated into the same hardware device, or distributed in multiple hardware devices.

FIG. 2 is a flowchart of a driving risk analysis method executed by the driving risk analysis device 100 . The driving risk analysis apparatus 100 and the method executed thereon can be used for the analysis and management of driving risks of a fleet, and the fleet can include at least one vehicle driven by at least one driver.

First, in step S210, the risk factor setting module 112 of the risk data collection module 110 sets a plurality of risk factors related to driving safety, such as the three risk factors F ₁ , F ₂ and F ₃ shown in Table 1 below, wherein , and each such risk factor includes its name and description.

Multiple risk levels can be set for each risk factor according to the speed of vehicles in the fleet. Taking the risk factors F ₁ , F ₂ and F ₃ in Table 1 as an example, the risk factor F ₁ can be set to two risk levels L ₁ and L ₂ , the risk factors F ₂ and F ₃ can each be set to three risk levels L ₁ , L ₂ and L ₃ , as shown in Table 2 below. S and L in Table 2 are the vehicle speed and the speed limit of the road on which the vehicle is located in Table ₁ related to the risk factor F3, respectively.

Although Table 1 and Table 2 list three risk factors, and each risk factor is set with two or three risk levels, the present invention is not limited thereto. In other embodiments, the number of risk factors and risk levels can be set or adjusted according to actual needs.

Next, in step S220, the risk data input module 114 of the risk data collection module 110 collects a plurality of risk events of the vehicles of the fleet, such as the twelve risk events R ₁ to R ₁₂ shown in Table 3 below.

Each vehicle in the fleet can be equipped with an auxiliary system that cooperates with the driving risk analysis device 100 , and the auxiliary system can be hardware, firmware or software. The auxiliary system has stored the settings of risk factors and risk levels in advance (for example, the settings in Table 1 and Table 2), and is used to instantly detect whether the behavior of the vehicle belongs to the description of any risk factor. Whenever the behavior of any vehicle in the fleet meets the description of any risk factor, the auxiliary system of the vehicle will determine the risk level corresponding to the behavior of the vehicle according to the speed of the vehicle, and generate a corresponding risk event. The auxiliary system can instantly upload the risk events of the vehicle to the risk data input module 114 of the risk data collection module 110 through wireless transmission, or the auxiliary system can store the risk events of the vehicle in a data file, and then The data file is imported into the risk data input module 114 of the risk data collection module 110 . The database module 116 can store the risk factors set by the risk factor setting module 112 and the risk events received by the risk data input module 114 for other modules (such as the risk factor adjustment module 120 and the risk calculation module 130 ) )Inquire.

As shown in Table 3, each of the risk events R ₁ to R ₁₂ includes an identification value, an event type, a travel speed, a road speed limit, and a risk level. In each of the risk events, the identification value may be the identification value of the vehicle that generated the risk event, or the identification value of the driver of the vehicle that generated the risk event, or the identification value of the vehicle that generated the risk event and its The identification value of the driver; the event type is the risk factor that the behavior of the vehicle he belongs to when the risk event occurs; the driving speed and the road speed limit are the driving speed of the vehicle he belongs to when the risk event occurs and the road speed limit, respectively The speed limit of the road where the vehicle is located; the risk level is the risk level corresponding to the speed of the vehicle it belongs to when the risk event occurs. The event type and the risk level of each risk event correspond to the same risk factor among the above risk factors.

Next, in step S230, the risk coefficient adjustment module 120 sets or adjusts the proportion coefficient of each risk level of each risk factor, such as shown in Table 4 below.

Generally speaking, the higher the driving speed of the vehicle, the higher the driving risk. Therefore, in the same risk factor, the proportion coefficient of the risk level will form an increasing function of the driving speed of the vehicle. In other words, the higher the driving speed, the higher the corresponding specific gravity coefficient. For example, the specific gravity coefficient of the risk factor F2 in Table ₄ forms an arithmetic sequence that increases with the travel speed, and the specific gravity coefficient of the risk factor F3 in Table ₄ forms an arithmetic sequence that increases with the travel speed.

After the setting of the proportion coefficient of each of the risk levels is completed, if necessary, the risk coefficient adjustment module 120 can also adjust the proportion coefficient of each of the risk levels. For example, the risk factor adjustment module 120 can increase the proportion factor of the risk factor or the risk level corresponding to the vehicle behavior prone to accidents according to the recent accident statistics. Alternatively, the management personnel can manually adjust the proportion coefficient according to the application requirements through the risk coefficient adjustment module 120 .

Next, in step S240, the risk calculation module 130 calculates the risk value corresponding to each identification value in the risk event according to the risk level of the risk factor and its proportion coefficient, wherein, for each identification value, the identification value corresponds to The risk value is the sum of the products of the number of occurrences of each risk level and the corresponding proportion coefficient in the risk events including the identification value.

For example, according to the risk events R ₁ to R ₄ shown in Table 3, the risk value F ( D ₁ ) of the identification value D ₁ can be calculated. Specifically, among the risk events R ₁ to R ₄ including the identification value D ₁ , the risk levels with occurrence times greater than zero are L ₁ and L ₂ of the risk factor F ₁ , L ₂ of the risk factor F ₂ , and risk factor F respectively For L ₁ of ₃ , the number of occurrences of these risk levels is 1, and the proportion coefficients corresponding to the corresponding risk levels are 1.2, 1, 1.2, and 1.3, respectively, so the risk value F ( D ₁ ) of the identification value D ₁ is

F ( D ₁ )=1×1.2+1×1+1×1.2+1×1.3=4.7.

For another example, according to the risk events R ₅ to R ₁₂ shown in Table 3, the risk value F ( D ₂ ) of the identification value D ₂ can be calculated. In detail, among the risk events R ₅ to R ₁₂ including the identification value D ₂ , the risk levels with occurrence times greater than zero are L ₂ of the risk factor F ₁ , L ₁ and L ₂ of the risk factor F ₂ , and risk factors, respectively. For L ₁ and L ₂ of F ₃ , the occurrence times of these risk levels are 2, 1, 1, 3, and 1, respectively, and the proportion coefficients corresponding to these risk levels are 1, 1.44, 1.2, 1.3, and 1.2, respectively. Therefore, The risk value F ( D ₂ ) of the identification value D ₂ is

F ( D ₂ )=2×1+1×1.44+1×1.2+3×1.3+1×1.2=10.94.

The value at risk calculated using the above method can be used to evaluate driver performance and to assist fleet management.

FIG. 3 is a block diagram of a driving risk analysis device 300 according to a second embodiment of the present invention. The driving risk analysis device 300 includes a risk data collection module 310, a risk factor training module 320, and an accident prediction module 330, wherein the risk data collection module 310 includes a risk factor setting module 312, a risk data input module 314, And the database module 316. The risk factor setting module 312, the risk data input module 314, and the database module 316 respectively include the functions of the risk factor setting module 112, the risk data input module 114, and the database module 116 of the first embodiment, And it can still include the functions required by the second embodiment. Each module in Figure 3 can be software, hardware or firmware; In the case of hardware, it can be a processing unit, processor, computer or server with data processing and computing capabilities; in the case of software or firmware, it can include instructions executable by the processing unit, processor, computer or server . In addition, each of the modules in FIG. 3 can be integrated into the same hardware device, or distributed in multiple hardware devices.

FIG. 4 is a flowchart of a driving risk analysis method executed by the driving risk analysis device 300 . The driving risk analysis apparatus 300 and the method executed by the driving risk analysis apparatus 300 can be used for the analysis and management of the driving risk of a fleet, and the fleet may include at least one vehicle driven by at least one driver.

First, in step S410 , the risk factor setting module 312 of the risk data collection module 310 sets a plurality of risk factors related to driving safety, such as the three risk factors F ₁ , F ₂ and F ₃ shown in Table 1 . In this embodiment, the risk factor does not need to set a risk level.

Next, in step S420, the risk data input module 314 of the risk data collection module 310 collects the accident history and a plurality of first risk events of the vehicles of the fleet. The incident history can be provided by an external server or imported directly. The first risk events can be uploaded or directly imported from the auxiliary system of the vehicle as described in the first embodiment.

After collecting a plurality of first risk events, the risk data input module 314 can count the accident history and the first risk events to generate a plurality of first samples, such as the eight samples S ₁ to S shown in Table 5 below ₈ .

Each _of the samples S1 to _S8 is generated by the risk data input module 314 counting a plurality of first risk events generated within a time period, each of the samples S1 to _S8 corresponds to _a different time period, and each The first risk events corresponding to the samples S ₁ to S ₈ may include the same identification value or a plurality of different identification values. Each _of the samples S1 to _S8 includes accident information A. The accident information A indicates whether the vehicle generating the first risk event corresponding to the sample has an accident, such as a car accident, within the time period corresponding to the sample. The risk data input module 314 may query the accident history to generate accident information A according to the time period of each sample. In addition, each of the samples S ₁ to S ₈ further includes the occurrence times C ₁ , C ₂ and C ₃ of the risk factors F ₁ , F ₂ and F ₃ . Specifically, the occurrence times C ₁ , C ₂ and C ₃ are the occurrence times of the risk factors F ₁ , F ₂ and F ₃ in the first risk events in the time period corresponding to each of the samples, respectively. For example, the occurrence times C ₁ , C ₂ and C ₃ in the sample S ₁ are the occurrence times of the first risk events of the risk factors F ₁ , F ₂ and F ₃ in the time period corresponding to the sample S ₁ , respectively ,So on and so forth. The risk data input module 314 can store the generated first sample in the database module 316 for query by the risk factor training module 320 and the accident prediction module 330 .

In another embodiment, the accident history and the first risk events can be collected by the database module 316 to generate and store the first samples.

In yet another embodiment, the risk factor training module 320 can count the accident history and the first risk events to generate the first samples.

Next, in step S430, the risk factor training module 320 performs data training according to the first samples to calculate the accident probability value of each risk factor, the overall probability value and the composite probability value of the risk factors. In one embodiment, the risk factor training module 320 can execute a risk factor training method using the first samples and applying a multinomial Naive Bayes classifier in a machine learning method. classifier) training risk factor probability value, and its training formula is described as follows.

The formula for the accident probability value P ( C _i |A) of each risk factor is as follows, where A represents an accident, C _i is the number of occurrences of the above risk factors, i is an integer and 1<=i<=N, N is The number of risk factors. In this embodiment, N is equal to three.

The formula for the overall probability value P (A) of these risk factors is as follows.

The composite probability value of these risk factors P ( C ₁ , C ₂ , C ₃ ) = P ( C ₁ ) × P ( C ₂ ) × P ( C ₃ ), where P ( C ₁ ), P ( C ₂ ) and P ( C ₃ ) are the sum of occurrences of each risk factor in all samples divided by the sum of occurrences of all risk factors in all samples. For example, P ( C1 )=sum of occurrences C1 _{of risk factor} F1 in all samples divided by the sum of occurrences of all risk factors in all samples, and so on.

If the first samples S ₁ to S ₈ in Table 5 are used as input data, the calculation of the accident probability value P ( C _i |A) of each risk factor and the overall probability value P (A) of these risk factors The results are as follows.

If the first samples S ₁ to S ₈ in Table 5 are used as input data, the composite probability value of these risk factors P ( C ₁ , C ₂ , C ₃ ) = P ( C ₁ ) × P ( C ₂ ) × P ( C ₃ ) calculation results are shown in Table 6 below.

In addition to collecting the above-mentioned first risk events, the risk data input module 314 can also collect a plurality of second risk events generated by the vehicles of the fleet within a period of time, and the risk data input module 314 and the database module 316 Or the risk factor training module 320 can count the second risk events to generate a second sample, such as the second sample S ₉ shown in Table 7 below. _The second sample is similar to the above -mentioned first sample, for example, the second sample S9 includes the occurrence times C1, C2 and C3 _{of each of} the risk _{factors F1} , F2 and F3 in the _second risk events .

Next, in step S440, the accident prediction module 330 predicts the first accident probability value according to the above training results (including the accident probability value of each risk factor, the overall probability value of the risk factors, and the composite probability value) and the second sample The accident risk probability P (A| C ₁ , C ₂ , C ₃ ) corresponding to the two samples is calculated as follows.

For example, if the accident risk probability corresponding to the second sample _S9 is predicted, the calculation result is as follows.

After the above training of the risk factors is completed, the accident prediction module 330 can predict the corresponding accident risk probability for the second sample generated subsequently according to the above calculation method. Each of the accident risk probabilities is a relative value, which can represent the difference in driving quality among the vehicle drivers corresponding to each of the second samples.

In another embodiment, the prediction of accident risk probability may add an age factor. Specifically, each of the second samples may include the age of the vehicle driver corresponding to the corresponding second risk event, and the accident risk probability generated by the above prediction may be grouped according to the age, so as to present the vehicle driving in each age group. Differences in driving quality between drivers. In other embodiments, the prediction of accident risk probability may also increase Additional factors are added to further group the differences in driving quality among vehicle drivers within each group.

The above-mentioned second embodiment takes three risk factors as an example, but the present invention is not limited to this. In other embodiments, the number of risk factors can be adjusted according to application requirements, and the above calculation formula can be adjusted according to the number of risk factors Adjust accordingly.

FIG. 5 is a block diagram of a driving risk analysis device 500 according to a third embodiment of the present invention. The driving risk analysis device 500 includes a risk data collection module 510, a risk coefficient adjustment module 520, a risk calculation module 530, a risk factor training module 540, and an accident prediction module 550, wherein the risk data collection module 510 includes risk Factor setting module 512 , risk data input module 514 , and database module 516 . Each module in Figure 5 can be software, hardware or firmware; if it is hardware, it can be a processing unit, processor, computer or server with data processing and computing capabilities; if it is software or firmware , it may include instructions executable by a processing unit, processor, computer or server. In addition, each of the modules in FIG. 5 can be integrated into the same hardware device, or distributed in multiple hardware devices.

In this embodiment, the risk factor setting module 512, the risk data input module 514, and the database module 516 respectively include the risk factor setting module 112, the risk data input module 114, and the database of the first embodiment The functions of the module 116 respectively include the functions of the risk factor setting module 312, the risk data input module 314, and the database module 316 of the second embodiment. Therefore, the risk data collection module 510, the risk coefficient adjustment module 520, and the risk calculation module 530 can execute the driving risk analysis method of the first embodiment shown in FIG. 2, and the risk data collection module 510, the risk factor The training module 540 and the accident prediction module 550 can execute the second method shown in FIG. 4 . The driving risk analysis method of the embodiment is used to calculate the driving risk value of the driver and predict the accident risk probability.

To sum up, the driving risk analysis device and method of the present invention calculates the driving risk value of the driver according to the risk factor, and provides objective quantitative values, so as to discriminate the driver with high driving risk and provide guidance for improvement, or discriminate the driver with low driving risk. And give encouragement. In addition, the driving risk analysis device and method of the present invention can also use a machine learning method and perform data training according to historical data, so as to predict the accident risk probability.

The above-mentioned embodiments are only used to illustrate the principle and effect of the present invention, but are not intended to limit the present invention. Anyone with ordinary knowledge in the technical field can modify and change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention should be as listed in the patent application scope described later.

500: Driving Risk Analysis Equipment

510: Risk Data Collection Module

512: Risk Factor Setting Module

514: Risk Data Input Module

516: Database Module

520: Risk Factor Adjustment Module

530: Risk Computing Module

540: Risk Factor Training Module

550: Accident Prediction Module

Claims (9)

A driving risk analysis device, comprising: a risk data collection module for setting multiple risk factors and collecting multiple risk events of at least one vehicle, wherein each risk event includes identification value, event type and risk level, and each risk event includes The event type and the risk level of the event correspond to the same risk factor among the corresponding risk factors; the risk coefficient adjustment module is used to set or adjust the proportion coefficient of each risk level of each of the risk factors; and a risk calculation module , which is used to calculate the risk value corresponding to each identification value according to the risk levels and the proportion coefficients, wherein the risk value corresponding to each identification value is the risk event including the identification value. The sum of the product of the number of occurrences of the risk level and the corresponding proportion coefficient. The driving risk analysis device according to claim 1, wherein the risk levels are set according to the driving speed of the at least one vehicle. The driving risk analysis device according to claim 2, wherein the proportion coefficients of the risk levels of each of the risk factors form an increasing function of the driving speed. A driving risk analysis device, comprising: a risk data collection module for setting a plurality of risk factors, and for collecting accident history, a plurality of first risk events and a plurality of second risk events of at least one vehicle; a risk factor training module for Execute a risk factor training method, wherein the risk factor training method uses the accident history and the first risk events, and uses a number of naive shell classifiers in the machine learning method for data training to calculate the risk factors The accident probability value of the factor; and an accident prediction module, used for predicting the accident risk probability corresponding to the second risk event by statistical means according to the accident probability value of each of the risk factors and the second risk events. The driving risk analysis device according to claim 4, wherein the risk data collection module or the risk factor training module is reused to count the accident history and the first risk events, so as to generate plural numbers with corresponding time. The first samples of the segment, each of the first samples includes whether the at least one vehicle has had an accident during the corresponding time period and the number of occurrences of each of the risk factors in the first risk events within the corresponding time period , and the risk factor training module performs the data training according to the first samples, and calculates the accident probability value of each risk factor. The driving risk analysis device according to claim 5, wherein the risk factor training module is reused to perform the data training according to the first samples, to calculate the overall probability value and the composite probability value of the risk factors, and The accident prediction module predicts the accident risk probability according to the accident probability value of each of the risk factors, the overall probability value and the composite probability value of the risk factors, and the second risk events. The driving risk analysis device according to claim 4, wherein the risk data collection module or the accident prediction module is multiplexed to count the second risk events to generate the second risk events including each of the risk factors and the accident prediction module predicts the accident risk probability according to the accident probability value of each of the risk factors and the second sample. A driving risk analysis method, comprising: setting a plurality of risk factors by a risk data collection module; collecting a plurality of risk events of at least one vehicle by the risk data collection module, wherein each risk event includes an identification value, an event type and a risk level , and the event type and the risk level of each of the risk events correspond to the same risk factor among the corresponding risk factors; the risk factor adjustment module sets or adjusts the proportion coefficient of each of the risk levels of each of the risk factors; and The risk calculation module calculates the risk value corresponding to each identification value according to the risk levels and the proportion coefficients, wherein the risk value corresponding to each identification value is the risk event including the identification value , the sum of the product of the occurrence times of each risk level and the corresponding proportion coefficient. A driving risk analysis method, comprising: setting a plurality of risk factors by a risk data collection module; collecting the accident history of at least one vehicle, a plurality of first risk events and a plurality of second risk events by the risk data collection module; training by the risk factors The module implements a risk factor training method, wherein the risk factor training method uses the accident history and the first risk events, and uses a number of naive shell classifiers in the machine learning method for data training to calculate each The accident probability value of the risk factor; and according to the accident probability value of each risk factor and the second risk events, the accident prediction module uses a statistical method to predict the accident risk probability corresponding to the second risk events.

Images