CN109801491B - Intelligent navigation method, device and equipment based on risk assessment and storage medium - Google Patents

Abstract

The invention discloses an intelligent navigation method, an intelligent navigation device, intelligent navigation equipment and a storage medium based on risk assessment, wherein the method comprises the following steps: the method comprises the steps of obtaining a route navigation request sent by a client, obtaining a plurality of original navigation routes based on an initial point and a target point of route navigation, further obtaining traffic data to be evaluated corresponding to each navigation road section, carrying out risk evaluation on the traffic data to be evaluated corresponding to the navigation road sections by adopting a pre-trained road risk recognition model, obtaining road section risk probability corresponding to the navigation road sections, obtaining total risk probability corresponding to the original navigation routes based on the road section risk probability corresponding to at least one navigation road section in each original navigation route, determining the original navigation route with the minimum total risk probability as a recommended navigation route, and sending the recommended navigation route to the client. The method can realize route planning based on safety factors and guarantee driving safety.

Description

Intelligent navigation method, device and equipment based on risk assessment and storage medium

Technical Field

The invention relates to the technical field of intelligent navigation, in particular to an intelligent navigation method, device, equipment and storage medium based on risk assessment.

Background

Planning an optimal navigation route in a navigation road network is a basic function of a navigation system. When planning a route, a current navigation system mainly plans according to whether a road is congested, whether the road is charged or not, whether the road is the shortest distance or not, and the like, namely, factors such as speed, time, cost and the like are emphasized in the route planning process, and route planning is not performed based on safety factors, so that the route planning is not beneficial to ensuring the safety of a journey.

Disclosure of Invention

The embodiment of the invention provides an intelligent navigation method, device, equipment and storage medium based on risk assessment, and aims to solve the problem that the current navigation system cannot plan a route based on safety factors.

An intelligent navigation method based on risk assessment comprises the following steps:

the method comprises the steps of obtaining a route navigation request sent by a client, wherein the route navigation request comprises an initial place, a target place and a navigation type;

acquiring at least one original navigation route based on the starting position and the target position, wherein each original navigation route comprises at least one navigation road section;

if the navigation type is a safety recommendation type, acquiring traffic data to be evaluated corresponding to each navigation road section;

inputting the traffic data to be evaluated corresponding to the navigation road section into a road risk recognition model which is trained by adopting a decision tree algorithm in advance, and performing risk evaluation on the traffic data to be evaluated corresponding to the navigation road section according to the classification logic of the decision tree algorithm to obtain the road section risk probability corresponding to the navigation road section;

acquiring a total risk probability corresponding to each original navigation route based on a road section risk probability corresponding to at least one navigation road section in each original navigation route;

and determining the original navigation route with the minimum total risk probability as a recommended navigation route, and sending the recommended navigation route to the client.

An intelligent navigation device based on risk assessment, comprising:

the route navigation request acquisition module is used for acquiring a route navigation request sent by a client, wherein the route navigation request comprises an initial location, a target location and a navigation type;

the original navigation route acquisition module is used for acquiring at least one original navigation route based on the starting point and the target point, and each original navigation route comprises at least one navigation road section;

the traffic data to be evaluated acquiring module is used for acquiring the traffic data to be evaluated corresponding to each navigation road section if the navigation type is the safety recommendation type;

the road section risk probability obtaining module is used for inputting the traffic data to be evaluated corresponding to the navigation road section into a road risk recognition model which is trained by adopting a decision tree algorithm in advance, carrying out risk evaluation on the traffic data to be evaluated corresponding to the navigation road section according to the classification logic of the decision tree algorithm, and obtaining the road section risk probability corresponding to the navigation road section;

the total risk probability obtaining module is used for obtaining total risk probability corresponding to the original navigation route based on the road section risk probability corresponding to at least one navigation road section in each original navigation route;

and the recommended navigation route module is used for determining the original navigation route with the minimum total risk probability as the recommended navigation route and sending the recommended navigation route to the client.

A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the intelligent navigation method based on risk assessment when executing the computer program.

A computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, carries out the steps of the above-mentioned intelligent navigation method based on risk assessment.

According to the intelligent navigation method, the intelligent navigation device, the intelligent navigation equipment and the intelligent navigation storage medium based on risk assessment, at least one original navigation route can be quickly determined according to the starting location and the target location. When the navigation type is the safety recommendation type, according to the traffic data to be evaluated of each navigation road section in the original navigation route, a road risk identification model is inquired, the road section risk probability corresponding to the navigation road section is obtained, and the obtained road section risk probability is ensured to have objectivity. And acquiring a corresponding total risk probability based on the road section risk probability of at least one navigation road section, so that the total risk probability reflects the driving risk of the original navigation route on the whole, and the objectivity of the result is ensured. And finally, determining the original navigation route with the minimum total risk probability as a recommended navigation route so as to achieve the purpose of intelligent navigation based on safety factors.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments of the present invention will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without inventive labor.

FIG. 1 is a schematic diagram of an application environment of an intelligent navigation method based on risk assessment according to an embodiment of the present invention;

FIG. 2 is a flowchart of an intelligent navigation method based on risk assessment according to an embodiment of the present invention;

FIG. 3 is another flow chart of a method for intelligent navigation based on risk assessment in accordance with an embodiment of the present invention;

FIG. 4 is another flow chart of a risk assessment based intelligent navigation method according to an embodiment of the present invention;

FIG. 5 is another flow chart of a method for intelligent navigation based on risk assessment in an embodiment of the present invention;

FIG. 6 is another flow chart of a method for intelligent navigation based on risk assessment in accordance with an embodiment of the present invention;

FIG. 7 is another flow chart of a method for intelligent navigation based on risk assessment in accordance with an embodiment of the present invention;

FIG. 8 is another flow chart of a method for intelligent navigation based on risk assessment in an embodiment of the present invention;

FIG. 9 is a schematic diagram of an intelligent navigation device based on risk assessment according to an embodiment of the present invention;

FIG. 10 is a schematic diagram of a computer device according to an embodiment of the invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.

The intelligent navigation method based on risk assessment provided by the embodiment of the invention can be applied to the application environment shown in fig. 1. Specifically, the intelligent navigation method based on risk assessment is applied to an intelligent navigation system, the intelligent navigation system comprises a client and a server shown in fig. 1, and the client and the server are communicated through a network and used for achieving risk assessment in a route planning process and ensuring driving safety. The client is also called a user side, and refers to a program corresponding to the server and providing local services for the client. The client may be installed on, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server may be implemented as a stand-alone server or as a server cluster consisting of a plurality of servers.

In an embodiment, as shown in fig. 2, an intelligent navigation method based on risk assessment is provided, which is described by taking the server in fig. 1 as an example, and includes the following steps:

s201: and acquiring a route navigation request sent by a client, wherein the route navigation request comprises a starting place, a target place and a navigation type.

The route navigation request is a request sent by a user to the server through the client for route planning. The starting point is the starting point of the route navigation request, and specifically may be a position autonomously input by the user through the client, or may be a position autonomously located by a GPS module provided in the client. The target location is the end point of the route navigation request. The navigation type is determined by the route planning priority principle selected by the route navigation request. The navigation types include, but are not limited to, common distance recommendation types (types of routing rules based on distance and distance) and cost recommendation types (types of routing rules based on cost), and safety recommendation types (types of routing rules based on risk of safety factor).

S202: based on the starting location and the target location, at least one original navigation route is obtained, and each original navigation route comprises at least one navigation road section.

The original navigation route is a passable navigation route determined by the server according to the starting location and the target location. The server firstly queries a navigation system database based on the starting location and the target location to obtain all original navigation routes connecting the starting location and the target location. For example, there are 5 original navigation routes from Shenzhen (origin site) to Guangzhou (destination site), and item 1 is an original navigation route via Guangzhou branches at Kyoto, Macau High and Shenhai High; the 2 nd navigation route is an original navigation route passing through Longda high speed and Jinggang Australian high speed; item 3 is an original navigation route passing through the high speed of Kyoto hong Australia and the high speed of Guangdong Yangtze river; item 4 is the original navigation route of the great high speed and the water country road. Item 5 is the original navigation route passing through the wide deep highway and the wide garden expressway.

In this embodiment, at least one navigation segment is determined in each original navigation route according to the road name and the area where the original navigation route is located. It can be understood that if the distance of the original navigation route is short, it may only contain one navigation segment; if the distance of the original navigation route is longer, determining at least two navigation road sections according to the name of the road passed by the original navigation route or the located area.

S203: and if the navigation type is the safety recommendation type, acquiring the traffic data to be evaluated corresponding to each navigation road section.

The traffic data to be evaluated refers to data which are collected in the route planning process and are needed for evaluating driving risks of the vehicles. The traffic data to be evaluated comprises but is not limited to driver information to be evaluated, vehicle information to be evaluated, road section information to be evaluated, place information to be evaluated and time information to be evaluated. The driver information to be evaluated is information for indicating the identity of the driver who triggers the route navigation request, and may specifically include the name, the identification card number, and the driver license number of the driver. The vehicle information to be evaluated is information indicating a vehicle associated with the client that triggered the route navigation request, and includes, but is not limited to, the model number and the traveled mileage of the vehicle. The road section information to be evaluated is information indicating a road section corresponding to each navigation road section. The location information to be evaluated is information indicating the location where each navigation section is located. The time information to be evaluated is information corresponding to the current time for triggering the route navigation request.

In this embodiment, when the navigation type is the safety recommendation type, it indicates that the user requests the server to perform route planning by using the safety factor as a route planning principle through the client, and therefore traffic data to be evaluated, which affects driving safety, needs to be collected, so as to perform risk identification based on the traffic data to be evaluated in the following. It can be understood that when the navigation type is the distance recommendation type and the cost recommendation type, the traffic data to be evaluated does not need to be acquired, that is, the risk evaluation is not performed based on the traffic data to be evaluated.

S204: and inputting the traffic data to be evaluated corresponding to the navigation road section into a road risk recognition model which is trained by adopting a decision tree algorithm in advance, and performing risk evaluation on the traffic data to be evaluated corresponding to the navigation road section according to the classification logic of the decision tree algorithm to obtain the road section risk probability corresponding to the navigation road section.

The road risk recognition model is a model for recognizing risks, which is formed by training in advance based on model training data formed by historical traffic accident data, and a decision tree algorithm is adopted for model training in the model training process. The historical traffic accident data is data of historical traffic accidents that have occurred (i.e., traffic accidents that have occurred). Corresponding to the traffic data to be evaluated, the historical traffic accident data includes, but is not limited to, historical driver information, historical vehicle information, historical road segment information, historical location information, and historical time information. The historical driver information is information for indicating the identity of a driver in a historical traffic accident, and specifically may include the name, identification number, and driver license number of the driver. Historical vehicle information is information used to indicate vehicles in historical traffic accidents, including, but not limited to, the model number and miles driven of the vehicle. The historical link information is information for indicating a link on which the historical traffic accident is located. The historical location information is information for indicating where the historical traffic accident is located. The historical time information is information indicating a time corresponding to a historical traffic accident. The road risk recognition model is trained on model training data formed on the basis of historical traffic accident data, so that risk recognition can be performed on traffic data to be evaluated corresponding to the model training data, and therefore the road section risk probability corresponding to the navigation road section can be rapidly obtained, and the obtained road section risk probability has objectivity and intuitiveness. The road section risk probability is the risk probability obtained by the road risk identification model identifying the traffic data to be evaluated of a certain navigation road section.

In this embodiment, the road risk recognition model is specifically a model obtained by training model training data by using a decision tree algorithm, that is, the road risk recognition model stores risk probabilities corresponding to different model training data in a tree structure. The classification structure of the road risk identification model comprises a root node, leaf nodes and intermediate nodes for connecting the root node and the leaf nodes, wherein no father node is arranged above the root node, and no child node is arranged below the leaf nodes. The root node and each intermediate node correspond to a classification logic, each leaf node corresponds to a risk probability, and the risk probability is determined according to the number of model training data satisfying all classification logics of the corresponding branch and the number of all model training data, and can be a quotient of the two numbers. Specifically, after acquiring traffic data to be evaluated corresponding to a navigation road section, a server inputs the traffic data to be evaluated into the road risk identification model, and when performing risk evaluation on the traffic data to be evaluated corresponding to the navigation road section according to the classification logic of the decision tree algorithm, the classification logic of a root node is called to classify the traffic data to be evaluated, and an intermediate node to which the root node belongs is determined; and further classifying according to the classification logic of the intermediate node, so as to determine the intermediate node … … of the next level, and so on until determining the corresponding leaf node, and determining the risk probability corresponding to the leaf node as the road segment risk probability corresponding to the navigation road segment.

S205: and acquiring the total risk probability corresponding to the original navigation route based on the road section risk probability corresponding to at least one navigation road section in each original navigation route.

Since each original navigation route includes at least one navigation segment, and each navigation segment corresponds to a segment risk probability, in this embodiment, the server adds the segment risk probabilities corresponding to at least one navigation segment in each original navigation route, so as to obtain the total risk probability corresponding to the original navigation route. As can be appreciated, the total risk probability reflects the driving risk of the original navigation route as a whole. As can be understood, the total risk probability corresponding to the original navigation road segment is obtained by calculating the road segment risk probability of at least one navigation route, so that the obtained total risk probability is more objective by comprehensively considering the difference of objective conditions (such as the road segment information to be evaluated, the time information to be evaluated, and the weather information to be evaluated determined based on the place information to be evaluated and the time information to be evaluated) of each navigation road segment.

S206: and determining the original navigation route with the minimum total risk probability as a recommended navigation route, and sending the recommended navigation route to the client.

When the navigation type in the route navigation request is the safety recommendation type, sequencing the total risk probability corresponding to all the original navigation routes, determining the original navigation route with the minimum total risk probability as the recommended navigation route, and sending the recommended navigation route to the client so that the user drives the vehicle according to the recommended navigation route displayed by the client. It is to be understood that the recommended navigation route is a navigation route from the start location to the target location that is finally recommended to the user.

In the intelligent navigation method based on risk assessment provided by the embodiment, at least one original navigation route can be quickly determined according to the starting location and the target location. When the navigation type is the safety recommendation type, according to the traffic data to be evaluated of each navigation road section in the original navigation route, a road risk identification model is inquired, the road section risk probability corresponding to the navigation road section is obtained, and the obtained road section risk probability is ensured to have objectivity. And acquiring a corresponding total risk probability based on the road section risk probability of at least one navigation road section, so that the total risk probability reflects the driving risk of the original navigation route on the whole, and the objectivity of the result is ensured. And finally, determining the original navigation route with the minimum total risk probability as a recommended navigation route so as to achieve the purpose of intelligent navigation based on safety factors.

In an embodiment, as shown in fig. 3, step S201 is to obtain a route navigation request sent by a client, where the route navigation request includes a start location, a target location, and a navigation type, and specifically includes the following steps:

s301: and acquiring a voice navigation request sent by a client, wherein the voice navigation request comprises voice data to be recognized.

The voice navigation request is a voice-form request sent by a user to the server through the client for route planning. The voice data to be recognized is data which reflects the navigation requirement and the navigation type of the user and is input by the user through the client. For example, a user can enter "i want to go from Guangzhou to Shenzhen, secure up" through a client's voice capture device (e.g., microphone).

S302: and performing voice recognition on the voice data to be recognized by adopting a voice recognition model to obtain text data to be recognized.

The voice recognition model is a pre-trained model for recognizing text content in the voice data. The voice recognition model in the embodiment can adopt a voice static decoding network, and because the static decoding network completely expands the search space, the decoding speed is very high when the text translation is carried out, so that the text data to be recognized can be quickly acquired. The voice static decoding network is obtained by training with training voice data in a specific field, and the training voice data in the specific field can be understood as voice data corresponding to pre-stored historical navigation data. The voice static decoding network is obtained by training based on training voice data in a specific field, so that the pertinence is strong when the voice data to be recognized in the specific field is recognized, and the decoding accuracy is high.

S303: and extracting the text data to be recognized by adopting a keyword extraction algorithm to obtain target keywords.

The keyword extraction algorithm refers to an algorithm for extracting keywords from text data. In this embodiment, the server performs word segmentation on the text data to be recognized by using a word segmentation tool, and then performs word-out-of-use processing on the word segmentation result by using a word-out-of-use algorithm to obtain the target keyword. The target keyword is a keyword that may reflect a navigation requirement and a navigation type.

S304: and matching a preset keyword library based on the target keywords to obtain the starting place, the target place and the navigation type which are successfully matched.

The preset keyword library is a database for storing keywords of navigation requirements and navigation types. Specifically, the server matches a preset keyword library based on the target keywords, and obtains an initial location, a target location and a navigation type which are successfully matched. For example, the target keywords identified in "i want to go from guangzhou to Shenzhen, choosing the safest route" are "me", "want", "from", "Guangzhou", "to", "Shenzhen", "choose", "most", "safe", and "route". In the expression of configuring 'from C to D' in the preset keyword library, C is a starting place and D is a target place, so that Guangzhou is determined as the starting place and Shenzhen is determined as the target place. If the identified target keywords comprise safety keywords or safety-related keywords, determining that the navigation type of the target keywords is a safety recommendation type; if the identified target keywords comprise the cost or keywords related to the cost, the navigation type is determined to be a cost recommendation type; and if the identified target keywords comprise the distance and the keywords related to the distance, determining that the navigation type is the distance recommendation type.

In the intelligent navigation method based on risk assessment provided by the embodiment, a voice navigation request is input in a voice form, and the voice data to be recognized is recognized by adopting a voice recognition model, so that the text data to be recognized can be quickly acquired; and then, extracting keywords from the text data to be recognized by adopting a keyword extraction algorithm to obtain target keywords, and quickly obtaining the corresponding starting point, target point and navigation type by utilizing the target keywords and a preset keyword library, so that the two hands of a user are liberated, the route navigation can be carried out without manual input, and the condition that the user needs to manually input a route navigation request in the driving process is avoided.

In one embodiment, in order to further liberate both hands of the user, the user does not need to manually operate the client to input the voice navigation request during the driving of the vehicle, and therefore, a voice wake-up function needs to be arranged in the intelligent navigation system. As shown in fig. 4, before step S201, that is, before acquiring a voice navigation request sent by a client, the voice navigation request including voice data to be recognized, the intelligent navigation method based on risk assessment further includes the following steps:

s401: and acquiring original voice data acquired by the client in real time.

The original voice data is the voice data actually collected by the client. The original voice data can be understood as voice data collected when the client is in a dormant state, and is used for detecting whether the user intentionally performs voice navigation according to the original voice data. In this embodiment, whether the user intends to perform voice navigation is specifically realized by detecting whether the original voice data includes a wake-up keyword.

S402: and carrying out endpoint detection and feature extraction on the original voice data to obtain target acoustic features.

Voice Activity Detection (VAD), also known as Voice Activity Detection or Voice boundary Detection, refers to a process of detecting the presence or absence of Voice in a noisy environment. In this embodiment, a Voice Activity Detector (Voice Activity Detector) for realizing Voice endpoint detection is built in the client, and is used to eliminate environmental noise and acquire Voice data spoken by a user. The voice feature extraction refers to a process of extracting components with identification in the audio signal to remove background noise or other information irrelevant to the identity identification process. In this embodiment, the server performs endpoint detection and feature extraction on the original voice data to eliminate noise in the original voice data and obtain the target acoustic feature. The target acoustic feature may be a mel-frequency cepstral coefficient or other speech feature.

S403: and recognizing the acoustic characteristics of the target by adopting a voice recognition model to obtain target text data.

The same voice recognition model is acquired in step S403 and step S302, and two additional voice recognition models do not need to be trained, thereby simplifying the processing flow. The implementation process of step S403 is similar to step S302, and is not described herein again.

S404: and if the target text data contains the awakening keyword, controlling the client to enter a voice navigation interface within a preset time period and receiving a voice navigation request.

The awakening keyword is a keyword which is preset by the server and used for awakening the client to enter the voice navigation interface. The preset time period is a preset time period for limiting the time of voice wakeup. In this embodiment, when the server identifies that the target text data includes the wake-up keyword, the server may control the client to perform a voice navigation interface to obtain the voice navigation request. Further, the server controls the client to enter the voice navigation interface within a preset time period when the wake-up keyword is identified in the original voice data, so that the client only obtains the voice navigation request within the preset time period. It can be understood that the client enters the voice navigation interface within a preset time period, and if the voice navigation request sent by the client is obtained within the preset time period, the subsequent step S302 is executed; if the voice navigation request sent by the client is not acquired within the preset time period, the client enters the dormant state again, and the steps S401 to S403 are executed repeatedly, so that the problem of high power consumption caused by the fact that the client is in the voice navigation interface for a long time to wait for receiving the voice navigation request is solved.

The intelligent navigation system is also provided with a voice awakening function and is used for receiving a voice navigation request when the server detects that the target voice data acquired by the client in real time contains the awakening keyword according to the preset awakening keyword. Voice wakeup is sometimes called Keyword detection (Keyword detection), that is, a number of wakeup keywords is generally small (1-2 are more, and a special case can be expanded to more) when the wakeup keywords are detected in continuous voice. Voice wakeup is to process continuous voice stream, such as detecting a target keyword in a microphone recording (i.e., target voice data) continuously for 24 hours by a voice switch; voice wakeup can be combined with a voice recognition technology, and is used for detecting a position where voice starts, replacing a key, for example, "alexa" is used as a wakeup keyword in Amazon Echo, and once the wakeup keyword is detected, the client is controlled to perform a voice navigation interface, and start recording for voice recognition.

In the intelligent navigation method based on risk assessment provided by the embodiment, the target acoustic features are obtained by performing endpoint detection and feature extraction on the original voice data acquired in real time; and then, acquiring a voice recognition model for recognition to obtain target text data, and controlling the client to enter a voice navigation interface within a preset time period when the target text data contains a wake-up keyword, so that a voice navigation request can be acquired, the condition that the client waits for receiving the voice navigation request for a long time is avoided, and the power consumption is saved.

In an embodiment, since the pre-trained road risk identification model is required to perform risk assessment on the traffic data to be assessed corresponding to the navigation road segment in step 204, the road risk identification model capable of performing risk assessment needs to be pre-trained before the route planning is performed based on the route navigation request sent by the client. As shown in fig. 5, before step 201, that is, before obtaining the route navigation request sent by the client, the intelligent navigation method based on risk assessment further includes the following steps:

s501: and acquiring historical traffic accident data from the traffic police platform database through a third-party interface, wherein the historical traffic accident data comprises historical driver information, historical vehicle information, historical road section information, historical place information and historical time information.

The third-party interface is an interface which is preset in the intelligent navigation system and is used for connecting a third-party platform (including but not limited to a traffic police platform database and a weather platform database). The traffic police platform database is a database for storing traffic accident data recorded by the traffic police platform and driving behavior data corresponding to information of each driver. The historical traffic accident data is data stored in the traffic police platform database for recording historical traffic accidents (i.e., traffic accidents that have occurred). In this embodiment, the server may obtain, from the traffic police platform database through the third-party interface, historical traffic accident data that was entered into the traffic police platform database before the current time of the system.

Wherein the historical traffic accident data includes historical driver information, historical vehicle information, historical road segment information, historical location information, and historical time information. The historical driver information is information for indicating the identity of a driver in a historical traffic accident, and may specifically include the name, identification number, and driver license number of the driver. The historical vehicle information is information indicating vehicles in the historical traffic accident, including, but not limited to, the model number and miles driven of the vehicle. The historical link information is information for indicating a link on which the historical traffic accident is located. The historical location information is information for indicating where the historical traffic accident is located. The historical time information is information indicating a time corresponding to a historical traffic accident.

S502: and inquiring a traffic police platform database based on the historical driver information, acquiring historical driving behavior data corresponding to the historical driver information, inquiring a behavior information data table based on the historical driving behavior data, and acquiring a driver evaluation index.

The historical driving behavior data is behavior data recorded by the traffic police platform database whether the driver violates a rule or not, the severity of violation of the rule and the like. Because people are the most important factor of the occurrence of traffic accidents, the historical driving behavior data can reflect the driving technique and the quality of the driver, for example, the driver does not observe the road condition and directly drives into a driving lane, so that the vehicle can easily cause the occurrence of traffic accidents, is accustomed to driving at a high speed, overtaking carelessly, does not pay attention to traffic signs and marking lines and the like. It is to be understood that the historical driving behavior data is behavior data reflecting whether the driver of the historical traffic accident data has a violation prior to the current time of the system or the severity of the violation. The historical driving behavior data can be specifically determined by ticket data, deduction data, other historical traffic accident data and the like.

The behavior information data table is a data table stored in advance in a database for recording a driver evaluation index corresponding to each piece of historical driving behavior data. There are currently several situations for violation fines under the rules of delivery: direct revoking of the driver's license, 12 points of credit (including one and N), 6 points of credit, 3 points of credit, 2 points of credit, 1 point of credit and no penalty. The behavior information data table is several evaluation levels set according to the penalty condition of the violation fine, for example, it can be divided into types of P1, P2, P3 and P4, where P1 is a type for direct revoke license and 12 credits, P2 is a type for 6 credits existing once and 12 credits for 1 year, P3 is a type for 3 credits at most but N number of penalties, and P4 is a type for one penalty not exceeding 3 credits but N number of penalties.

S503: and inquiring a vehicle information data table based on the historical vehicle information to obtain a vehicle evaluation index corresponding to the historical vehicle information.

The vehicle information data table is a data table stored in advance in a database for recording a vehicle evaluation index corresponding to each piece of vehicle information. In the vehicle information data table, several evaluation grades for recording various vehicle evaluation factors affecting the vehicle evaluation index and any combination of two or more of the vehicle evaluation factors are represented by C1, C2, C3, and C4, respectively. The vehicle evaluation factors include, among others, vehicle dynamics parameters, vehicle age (recency), and vehicle maintenance times. For example, the four grades may be divided into 8 years or more, 5-8 years, 2-5 years, or less than 2 years, respectively, according to the age of the vehicle, or the number of vehicle repairs may be divided into 8, 5-8, 2-5, or 2.

S504: and inquiring a road section information data table based on the historical road section information to obtain a road section evaluation index corresponding to the historical road section information.

The link information data table is a data table stored in advance in the database for recording a link evaluation index corresponding to each link. In the link information data table, several evaluation levels corresponding to various link evaluation factors affecting the link evaluation index and any combination of at least one link evaluation factor are recorded, and are respectively represented by L1, L2, L3 and L4. For example, the link evaluation factor includes link attributes including a hybrid road, a vehicle one-way road, a vehicle multi-way road, and an expressway. The expressway includes a fast speed lane (e.g., more than 100 km/h) and a slow speed lane. The mixed road refers to a road without a road center line, namely, a road without a yellow line in the center of the road to distinguish opposite lanes. The general road has no marked line, and is a mixed road of motor vehicles, non-motor vehicles and people and vehicles.

S505: historical weather information corresponding to the historical place information and the historical time information is obtained from a weather platform database through a third-party interface, a weather information data table is inquired based on the historical weather information, and weather evaluation indexes corresponding to the historical weather data are obtained.

The historical weather information refers to weather information when a certain historical traffic accident occurs. The weather information data table is a data table for storing weather evaluation indexes corresponding to each type of weather information. Various weather evaluation factors influencing weather evaluation indexes and several evaluation grades corresponding to at least one weather evaluation factor combination are recorded in the weather information data sheet, wherein the evaluation grades are Q1, Q2, Q3 and Q4 respectively. In the current weather assessment process, the assessment grade can be determined according to an assessment grade classification table which is set by an intelligent navigation system independently, and can also be determined according to early warning grades such as typhoon, rainstorm, high temperature, cold tide, heavy fog, thunderstorm, strong wind, sand storm, hail, snow disaster, road icing and the like. For example, Q1, Q2, Q3, and Q4 correspond to the red, orange, yellow, and blue warnings, respectively, of a typhoon warning.

S506: and inquiring a calendar table based on the historical time information to obtain a time evaluation index corresponding to the historical time information.

The calendar table records data of annual holidays. In this embodiment, the history table is queried based on the time information to determine whether the historical traffic accident occurs on holidays or on peak hours of commuting. For example, its time assessment indicators may be divided into holidays, weekday peak periods, weekday off-peak periods, and weekends, corresponding to T1, T2, T3, and T4, respectively.

S507: and obtaining model training data corresponding to the historical traffic accident data based on the driver evaluation index, the vehicle evaluation index, the road section evaluation index, the weather evaluation index and the time evaluation index corresponding to the historical traffic accident data.

For each historical traffic accident data in the traffic police platform database, the processing is performed according to the steps S502-S506, the corresponding driver assessment index, vehicle assessment index, road section assessment index, weather assessment index and time assessment index are obtained, and the parameters such as P, C, L, Q and T can be respectively adopted for representation, that is, the model training data corresponding to each historical traffic accident data is obtained and can be represented as (P, C, L, Q, T), and the following table lists the specific situation corresponding to each parameter identifier. It can be understood that the factors corresponding to each historical traffic accident data are respectively converted into corresponding evaluation indexes so as to facilitate data analysis.

Parameter description in Table-model training data

S508: and performing model training on the model training data by adopting a decision tree algorithm to obtain a road risk identification model based on a decision tree.

The decision tree algorithm is a common classification algorithm, and a classifier is obtained by learning a plurality of samples containing a group of attributes and a category, so that the classifier can normally classify a newly appeared object. That is, the decision tree algorithm is a method of approximating discrete function values, and essentially the decision tree is a process of classifying data by a series of rules. In this embodiment, a C4.5 decision tree algorithm may be employed.

Specifically, when a decision tree algorithm is adopted to carry out model training on model training data, the tree-shaped classification structure comprises root nodes, leaf nodes and intermediate nodes for connecting the root nodes and the leaf nodes, and the classification logics of the root nodes and the intermediate nodes are determined according to the importance degree of each evaluation index so as to form an original risk identification model; classifying all the model training data according to the original risk recognition model to store all the model training data into a data set corresponding to the leaf node to which the model training data belong in a classified manner, and determining the quotient of the number of the model training data in the data set and the number of all the model training data as the risk probability of the leaf node to form the road risk recognition model.

In the intelligent navigation method based on risk assessment provided by this embodiment, historical driver information, historical vehicle information, historical road section information, historical location information, and historical time information in all historical traffic accident data are converted to obtain corresponding driver assessment indexes, vehicle assessment indexes, road section assessment indexes, weather assessment indexes, and time assessment indexes, so as to determine a classification attribute based on each assessment index, and form model training data. And training the formed model training data by adopting a decision tree algorithm to obtain a corresponding road risk recognition model so as to carry out risk recognition on the road risk recognition model.

In an embodiment, as shown in fig. 6, step S508, namely, performing model training on the model training data by using a decision tree algorithm to obtain a road risk recognition model based on a decision tree, specifically includes the following steps:

s601: and determining the corresponding class label characteristic and at least two training attribute characteristics of each model training data based on the model training data, and storing the model training data in a training data set.

The training attribute features may be understood as respective evaluation indexes of the model training data. The category label features may be understood as different risk levels. Specifically, all model training data may be divided into a plurality of sets according to the intersection of all evaluation indexes of the model training data, the risk probability of each set is determined based on the quotient of the model training data and all model training data in each set, and the grade division may be performed according to the risk probability, such as high risk, medium risk, or low risk grade.

S602: and performing information gain rate calculation on the class label characteristics and the training attribute characteristics of the model training data to obtain the information gain rate corresponding to each training attribute characteristic.

Wherein the information gain ratio calculation is a process for calculating an information gain ratio of the training attribute features. The information gain ratio is a gain ratio, which is a mathematical term, and is specifically a ratio of the information gain to the split information degree. The information gain is the degree of information uncertainty reduction, and the information gain is biased to the characteristic of more values. The split informativeness is the number of branches when a feature is split. The information gain rate is a ratio of the information gain and the splitting information degree, so that the information gain is normalized by the information splitting degree, and the accuracy is higher when classification is performed based on the information gain rate. In this embodiment, the server performs information gain ratio calculation on the class label features and the training attribute features of the model training data to obtain an information gain ratio corresponding to each training attribute feature, so that the information gain ratio is used as a classification basis for model training, the accuracy of the classification process can be higher, and the recognition rate of the trained model can be improved.

S603: and selecting the training attribute feature with the maximum information gain rate as a target classification feature of the training data set, and dividing the training data set into at least two training subsets by adopting the target classification feature.

Wherein the target classification features are features that divide the training data set. In this embodiment, the training attribute feature with the largest information gain rate is selected as the target classification feature of the training data set. The training subset is a set that partitions the training data set based on the target classification features. For example, in a model training data, the information gain ratio of training attribute features corresponding to a driver evaluation index, a vehicle evaluation index, a road section evaluation index, a weather evaluation index and a time evaluation index is calculated, and the training attribute feature with the largest information gain ratio is selected as a target classification feature so as to perform classification based on the target classification feature. And if the information gain rate corresponding to the driver evaluation index is the maximum, classifying all model training data by taking the maximum information gain rate as a target classification characteristic so as to divide the model training data into at least two training subsets.

S604: and judging whether training attribute features which are not determined as target classification features exist.

In this embodiment, it is determined whether training attribute features that are not determined as target classification features exist in training attribute features corresponding to model training data, so as to determine whether the training subset can be further divided, that is, if training attribute features that are not determined as target classification features exist, it is determined that the training subset cannot be further divided; if the training attribute features which are not determined as the target classification features do not exist in the training subset, the training subset can be further divided.

S605: and if the training attribute features which are not determined as the target classification features exist, updating the training subset into a training data set, and repeatedly executing the step of calculating the information gain rate of the class label features and the training attribute features of the model training data to obtain the information gain rate corresponding to each training attribute feature.

In this embodiment, after the training data set is divided into at least two training subsets by using the target classification feature, if there is a training attribute feature that is not determined as the target classification feature, it is indicated that the training subset can be further divided, at this time, the training subset is updated to the training data set, and the steps S602 to S603 are repeatedly executed until there is no training attribute feature that is not determined as the target classification feature, so as to complete the training process of the road risk recognition model based on the decision tree algorithm.

S606: and if the training attribute characteristics which are not determined as the target classification characteristics do not exist, forming an original risk identification model based on a decision tree algorithm, acquiring data volume corresponding to each leaf node in the original risk identification model, cutting the original risk identification model based on the data volume, and acquiring a road risk identification model based on the decision tree.

In this embodiment, after the training data set is divided into at least two training subsets by using the target classification feature, if there is no training attribute feature that is not determined as the target classification feature, it indicates that neither of the at least two training subsets can be further divided, and the original risk recognition model based on the decision tree algorithm can be obtained.

Specifically, after the original risk recognition model based on the decision tree is formed, all model training data in the training data set may be classified by using the original risk recognition model to determine the data amount of the model training data corresponding to each leaf node. Then, comparing the data volume of the model training data corresponding to each leaf node with a preset threshold value; if the data volume is larger than a preset threshold value, the leaf node is reserved; if the data volume is not greater than the preset threshold, the data volume of the model training data corresponding to the corresponding superior node (namely, the parent node and the grandparent node … … according to which the leaf node is obtained) is sequentially searched upwards, until the data volume of the superior node is greater than the preset threshold, the superior node is determined to be a truncation node, and the original risk recognition model is cut based on the truncation node, so that the road risk recognition model based on the decision tree is obtained. The preset threshold is a preset minimum threshold for performing clipping processing. For example, the preset threshold is set to 5, and if the data size of the model training data corresponding to any leaf node is greater than 5, the leaf node is reserved; if the data volume of the model training data corresponding to any leaf node is not more than 5, the data volume of the model training data corresponding to the upper node of the leaf node is searched upwards, if the data volume of the upper node is more than 5, the upper node is determined to be a cut node, the original risk identification model is cut according to the cut node, so that a simplified road risk identification model is obtained, and the road risk identification model is prevented from being redundant. Furthermore, in the road risk identification model, the risk probability corresponding to each cutoff node can be set as a low-probability risk; the leaf nodes retained by the leaf nodes are determined according to the class label characteristics or the ratio of the data volume of the leaf nodes to the data volume of the training data set.

In the intelligent navigation method based on risk assessment provided by this embodiment, training attribute features in a training data set are selected as a classification basis, and an information gain rate of each training attribute feature is calculated; selecting the training attribute feature with the largest information gain rate as a target classification feature of the training data set so as to divide the training data set into at least two training subsets by using the target classification feature; when training attribute features which are not determined as target classification features do not exist, an original risk identification model based on a decision tree algorithm is formed, data volume corresponding to each leaf node in the original risk identification model is obtained, the original risk identification model is cut based on the data volume, a road risk identification model based on the decision tree is obtained, a simplified road risk identification model is obtained, and model redundancy is avoided. The road risk identification model obtained by the embodiment has high classification accuracy and definite classification rule, so that the identification efficiency is high when the road risk identification model is subsequently utilized for identification.

In one embodiment, the route navigation request further includes a user account, which is an identification for uniquely identifying the user who triggered the route navigation request. As shown in fig. 7, after step S203, that is, after the traffic data to be evaluated corresponding to each navigation road segment is acquired if the navigation type is the safety recommendation type, the intelligent navigation method based on risk evaluation further includes the following steps:

s701: and querying a navigation system database based on the user account to acquire historical navigation data corresponding to the user account, wherein each historical navigation data corresponds to a historical recommendation type.

The historical navigation data refers to navigation data recorded by a route navigation request triggered by a user before the current time of the system. Since the historical navigation data is navigation data formed based on route navigation requests before the current time of the system, the historical navigation data can also record navigation types corresponding to the route navigation requests, and the navigation types are determined as historical recommendation types corresponding to the historical navigation data. It is to be appreciated that the historical recommendation types include, but are not limited to, a cost recommendation type, a distance recommendation type, and a security recommendation type.

S702: and counting historical recommendation frequency corresponding to the safety recommendation type in the historical recommendation types of the historical navigation data.

Because each historical navigation data only corresponds to one historical recommendation type, and the historical recommendation types include but are not limited to a cost recommendation type, a distance recommendation type, a safety recommendation type and the like, the driving habits of the user can be known by counting the historical recommendation frequency corresponding to each historical recommendation type. In this embodiment, the history recommendation frequency corresponding to the safety recommendation type in the history recommendation types of the history navigation data is counted, and specifically, a quotient of the number of the history navigation data corresponding to the safety recommendation type and the number of all history navigation data in the history recommendation types of the history navigation data is calculated.

S703: and if the historical recommendation frequency corresponding to the safety recommendation type is smaller than a preset frequency threshold, acquiring the travel risk information and sending the travel risk information to the client.

The preset frequency threshold is a preset probability threshold for evaluating whether the travel risk recommendation needs to be carried out. The travel insurance information refers to information on insurance relating to a travel. Specifically, when the historical recommendation frequency corresponding to the safety recommendation type is smaller than a preset frequency threshold, it is indicated that the driving habits of the user are less in selection of the safety recommendation type, and the selection of the safety recommendation type in the route navigation request indicates that the traveling is different from the past driving habits, and special situations such as children or old people may exist in the same trip. In order to further guarantee the safety of the user, insurance recommendation related to the journey can be carried out at the moment, namely journey insurance information is pushed to the client. It is understood that the travel insurance information is travel-related insurance information provided by an insurance company cooperating with the intelligent navigation system. Specifically, the travel insurance information includes insurance information related to travel, such as travel personal accident insurance and vehicle accident injury insurance, provided by an insurance company having a cooperative relationship with the intelligent navigation system, for example, if the travel insurance is self-driving travel insurance provided on a safe way, the travel insurance information may include information such as premium and settlement amount corresponding to the self-driving travel insurance. The travel insurance information includes a purchase entry link for the corresponding travel insurance to facilitate the travel insurance purchase by the user.

In the intelligent navigation method based on risk assessment provided by this embodiment, the frequency of navigation based on a safety recommendation type in the driving habits of a user is known by counting the history recommendation frequency of the safety recommendation type in history navigation data corresponding to a user account; and when the historical recommendation frequency is smaller than the preset frequency threshold, the current driving habit is different from the past driving habit, and the travel insurance recommendation can be carried out, so that the travel insurance recommendation is more pertinent.

In one embodiment, each recommended navigation route includes at least one frequent accident location, which is a location where traffic accidents frequently occur. As shown in fig. 8, after step S206, that is, after the recommended navigation route is sent to the client, the intelligent navigation method based on risk assessment further includes the following steps:

s801: and in the process that the vehicle runs according to the recommended navigation route, acquiring the current traffic data and the current driving position sent by the client in real time.

The current traffic data refers to data which are acquired in real time and are needed for evaluating driving risks of the vehicles during driving according to the recommended navigation route. The current traffic data includes, but is not limited to, current driver information, current vehicle information, current road segment information, current location information, and current time information. The current driver information is information for indicating the driver identity associated with the client, and may specifically include a name, an identification card number, and a driver license number of the driver. The current vehicle information is information indicating the vehicle with which the client is associated, including but not limited to the model number and miles driven of the vehicle. The current link information is information indicating a link corresponding to the currently located navigation link. The current location information is information indicating a location where the currently located navigation section is located. The current time information is information indicating that the current time corresponds to. The current driving position is the current position of the vehicle acquired by a GPS module carried by the client. Specifically, in the process that the vehicle runs according to the recommended navigation route, the server acquires current traffic data sent by the client in real time so as to carry out road risk assessment; and actually acquires the current driving position for the purposes of route tracking and risk assessment.

S802: and performing risk identification on the current traffic data by adopting a road risk identification model, acquiring current risk probability, and acquiring first risk data if the current risk probability is greater than a preset risk threshold.

The preset risk threshold is a preset threshold for evaluating whether risk reminding is needed. Specifically, when the server performs risk identification on the current traffic data by using the road risk identification model, the current traffic data may be input to the road risk identification model for identification, a leaf node where the current traffic data is located in the tree-shaped classification structure is determined, and a risk probability corresponding to the leaf node is determined as the current risk probability. In this embodiment, if the current risk probability corresponding to the current traffic data is greater than the preset risk threshold, it is indicated that the risk is greater in the process of driving the vehicle according to the recommended navigation route by the user, and at this time, the first risk data needs to be generated so as to perform risk reminding based on the first risk data. It will be appreciated that the first risk data may be a risk level determined from its current risk probability, such as high risk, medium risk and low risk.

S803: and according to the driving direction of the recommended navigation route, if the accident frequent place or the sudden accident information exists in the preset distance range of the current driving position, acquiring second risk data.

The preset distance range refers to a preset distance range. The emergency information refers to accident information obtained by the intelligent navigation system when the user drives a vehicle to run according to the recommended navigation route. Specifically, when the user drives the vehicle to run according to the recommended navigation route, according to the driving direction of the recommended navigation route, in a preset range (for example, 1km) of the current driving position of the vehicle, there are frequent accident points and sudden accident information (for example, there are frequent accident points and sudden accident information in the front of 1km of the driving direction of the vehicle), and second risk data needs to be generated so as to perform risk reminding based on the second risk data.

S804: and filling the first risk data and the second risk data into a preset risk reminding template to obtain risk reminding text data.

The preset risk reminding template is a preset template for risk reminding. For example, the preset risk reminding template is "… … meters ahead of the road, frequent accident sites exist, careful driving", "… … meters ahead of the road, outstanding accidents occur, … … lane blockage, careful driving" or "the road section where the vehicle is currently driven is a high-risk road section, careful driving" or the like. And the server fills the first risk data and the second risk data into the preset risk reminding template to form risk reminding text data.

S805: and converting the risk reminding text data into risk reminding voice data by adopting a TTS technology, and broadcasting the risk reminding voice data through the client.

Among them, TTS (abbreviation of Text To Speech, i.e., "from Text To Speech") is a part of man-machine conversation, enabling a machine To speak. Specifically, after the server obtains the risk reminding text data, the risk reminding text data is converted into risk reminding voice data by adopting a TTS technology, and the risk reminding voice data is played through the client, so that a user can hear the risk reminding voice data, and the purposes of risk reminding and reminding the user of careful driving are achieved.

In the intelligent navigation method based on risk assessment provided by the embodiment, the road risk identification model is adopted to identify the current traffic data, so that the first risk data can be quickly acquired, and the acquisition of the first risk data has objectivity. And then according to the driving mode of the recommended navigation route, when accident frequent places or outstanding accident information exist in the preset range of the current driving position, generating second risk data so as to carry out corresponding risk reminding. And filling the first risk data and the second risk data into a preset risk reminding template, and converting by adopting a TTS technology to broadcast the formed risk reminding voice data so that the user can hear the risk reminding voice data in real time in the driving process to achieve the purpose of reminding the user of careful driving.

It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

In an embodiment, an intelligent navigation device based on risk assessment is provided, and the intelligent navigation device based on risk assessment corresponds to the intelligent navigation method based on risk assessment in the above embodiments one to one. As shown in fig. 9, the intelligent navigation device based on risk assessment includes a route navigation request obtaining module 901, an original navigation route obtaining module 902, a traffic data to be assessed obtaining module 903, a road segment risk probability obtaining module 904, a total risk probability obtaining module 905, and a recommended navigation route module 906. The functional modules are explained in detail as follows:

a route navigation request obtaining module 901, configured to obtain a route navigation request sent by a client, where the route navigation request includes a start location, a target location, and a navigation type.

An original navigation route obtaining module 902 is configured to obtain at least one original navigation route based on the start location and the target location, where each original navigation route includes at least one navigation segment.

And the traffic data to be evaluated acquisition module 903 is configured to acquire traffic data to be evaluated corresponding to each navigation road segment if the navigation type is the safety recommendation type.

And a road section risk probability obtaining module 904, configured to input traffic data to be evaluated corresponding to the navigation road section into a road risk recognition model trained by using a decision tree algorithm in advance, perform risk evaluation on the traffic data to be evaluated corresponding to the navigation road section according to a classification logic of the decision tree algorithm, and obtain a road section risk probability corresponding to the navigation road section.

The total risk probability obtaining module 905 is configured to obtain a total risk probability corresponding to each original navigation route based on a road segment risk probability corresponding to at least one navigation road segment in each original navigation route.

And a recommended navigation route module 906, configured to determine the original navigation route with the smallest total risk probability as a recommended navigation route, and send the recommended navigation route to the client.

Preferably, the route navigation request acquisition module 901 includes a voice navigation request acquisition unit, a voice recognition processing unit, a keyword extraction unit, and a keyword matching unit.

And the voice navigation request acquisition unit is used for acquiring the voice navigation request sent by the client, and the voice navigation request comprises voice data to be recognized.

And the voice recognition processing unit is used for performing voice recognition on the voice data to be recognized by adopting the voice recognition model to acquire the text data to be recognized.

And the keyword extraction unit is used for extracting the text data to be recognized by adopting a keyword extraction algorithm to obtain a target keyword.

And the keyword matching unit is used for matching a preset keyword library based on the target keywords to obtain the starting place, the target place and the navigation type which are successfully matched.

Preferably, before the voice navigation request obtaining unit, the intelligent navigation device based on risk assessment further includes an original voice data obtaining unit, a target acoustic feature obtaining unit, a target text data obtaining unit and a text data acquisition obtaining unit.

And the original voice data acquisition unit is used for acquiring the original voice data acquired by the client in real time.

And the target acoustic feature acquisition unit is used for carrying out endpoint detection and feature extraction on the original voice data to acquire target acoustic features.

And the target text data acquisition unit is used for identifying the target acoustic characteristics by adopting a voice recognition model to acquire target text data.

And the text data acquisition and acquisition unit is used for controlling the client to enter a voice navigation interface within a preset time period and receiving a voice navigation request if the target text data contains the awakening keyword.

Preferably, before the route guidance request obtaining module 901, the intelligent navigation device based on risk assessment further includes a historical traffic accident data obtaining unit, a historical driving behavior data obtaining unit, a vehicle assessment index obtaining unit, a road segment assessment index obtaining unit, a weather assessment index obtaining unit, a time assessment index obtaining unit, a model training data obtaining unit and a road risk identification model obtaining unit.

And the historical traffic accident data acquisition unit is used for acquiring historical traffic accident data from the traffic police platform database through a third-party interface, wherein the historical traffic accident data comprises historical driver information, historical vehicle information, historical road section information, historical place information and historical time information.

And the historical driving behavior data acquisition unit is used for inquiring the traffic police platform database based on the historical driver information, acquiring historical driving behavior data corresponding to the historical driver information, inquiring the behavior information data table based on the historical driving behavior data and acquiring driver evaluation indexes.

And a vehicle evaluation index acquisition unit for querying the vehicle information data table based on the historical vehicle information and acquiring a vehicle evaluation index corresponding to the historical vehicle information.

And the road section evaluation index acquisition unit is used for inquiring the road section information data table based on the historical road section information and acquiring the road section evaluation index corresponding to the historical road section information.

And the weather evaluation index acquisition unit is used for acquiring historical weather information corresponding to the historical place information and the historical time information from the weather platform database through the third party interface, inquiring a weather information data table based on the historical weather information and acquiring weather evaluation indexes corresponding to the historical weather data.

And the time evaluation index acquisition unit is used for inquiring the calendar table based on the historical time information and acquiring the time evaluation index corresponding to the historical time information.

The model training data acquisition unit acquires model training data corresponding to the historical traffic accident data based on a driver assessment index, a vehicle assessment index, a road section assessment index, a weather assessment index and a time assessment index corresponding to the historical traffic accident data.

And the road risk recognition model acquisition unit is used for carrying out model training on the model training data by adopting a decision tree algorithm to acquire a road risk recognition model.

Preferably, the road risk identification model obtaining unit comprises a determination subunit, a gain rate obtaining subunit, a training subset dividing subunit, an attribute feature judging subunit, a repeated calculation subunit and an identification model obtaining subunit.

And the determining and determining subunit is used for determining the corresponding class label characteristic and at least two training attribute characteristics of each model training data based on the model training data, and storing the model training data in the training data set.

And the gain rate acquisition subunit is used for performing information gain rate calculation on the class label characteristics and the training attribute characteristics of the model training data to acquire the information gain rate corresponding to each training attribute characteristic.

And the training subset dividing subunit is used for selecting the training attribute feature with the largest information gain rate as a target classification feature of the training data set, and dividing the training data set into at least two training subsets by adopting the target classification feature.

And the attribute characteristic judging subunit is used for judging whether training attribute characteristics which are not determined as target classification characteristics exist or not.

And the repeated calculation subunit is used for updating the training subset into a training data set if the training attribute features which are not determined as the target classification features exist, and repeatedly executing the step of calculating the information gain rate of the class label features and the training attribute features of the model training data to obtain the information gain rate corresponding to each training attribute feature.

And the identification model obtaining subunit is used for forming an original risk identification model based on a decision tree algorithm if the training attribute features which are not determined as the target classification features do not exist, obtaining a data volume corresponding to each leaf node in the original risk identification model, cutting the original risk identification model based on the data volume, and obtaining a road risk identification model based on the decision tree.

Preferably, the route navigation request further comprises a user account; after the to-be-evaluated traffic data acquisition module 903, the intelligent navigation device based on risk evaluation further comprises a historical navigation data unit, a historical recommendation frequency unit and a travel risk information acquisition unit.

And the historical navigation data unit is used for inquiring the navigation system database based on the user account to acquire historical navigation data corresponding to the user account, and each historical navigation data corresponds to a historical recommendation type.

And the historical recommendation frequency unit is used for counting the historical recommendation frequency corresponding to the safety recommendation type in the historical recommendation types of the historical navigation data.

And the travel risk information acquisition unit is used for acquiring the travel risk information and sending the travel risk information to the client if the historical recommendation frequency corresponding to the safety recommendation type is smaller than a preset frequency threshold.

Preferably, the recommended navigation route module 906 includes at least one frequent location of an accident. After the recommend navigation route module 906, the intelligent navigation device based on risk assessment further includes: the system comprises a current data acquisition unit, a first risk data acquisition unit, a second risk data acquisition unit, a reminding text data acquisition unit and a reminding voice data broadcast unit.

And the current data acquisition unit is used for acquiring the current traffic data and the current driving position sent by the client in real time in the driving process of the vehicle according to the recommended navigation route.

The first risk data acquisition unit is used for carrying out risk identification on the current traffic data by adopting the road risk identification model, acquiring the current risk probability, and acquiring the first risk data if the current risk probability is greater than a preset risk threshold.

And the second risk data acquisition unit is used for acquiring second risk data if an accident frequent place or sudden accident information exists in a preset distance range of the current driving position according to the driving direction of the recommended navigation route.

And the reminding text data acquisition unit is used for filling the first risk data and the second risk data into a preset risk reminding template to acquire risk reminding text data.

And the reminding voice data broadcasting unit is used for converting the risk reminding text data into risk reminding voice data by adopting a TTS technology and broadcasting the risk reminding voice data through the client.

For the specific definition of the intelligent navigation device based on risk assessment, the above definition of the intelligent navigation method based on risk assessment can be referred to, and details are not repeated here. The modules in the intelligent navigation device based on risk assessment may be wholly or partially implemented by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.

In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 10. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used for storing data adopted or generated in the process of executing the intelligent navigation method based on risk assessment, such as traffic data to be assessed. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement an intelligent navigation method based on risk assessment.

In an embodiment, a computer device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the intelligent navigation method based on risk assessment in the foregoing embodiments are implemented, for example, steps S201 to S206 shown in fig. 2 or steps shown in fig. 3 to fig. 8, which are not described herein again to avoid repetition. Alternatively, when the processor executes the computer program, the functions of each module/unit in the embodiment of the intelligent navigation device based on risk assessment are implemented, for example, the functions of the route navigation request obtaining module 901, the original navigation route obtaining module 902, the traffic data to be assessed obtaining module 903, the road segment risk probability obtaining module 904, the total risk probability obtaining module 905, and the recommended navigation route module 906 shown in fig. 9 are not described herein again for avoiding repetition.

In an embodiment, a computer-readable storage medium is provided, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the intelligent navigation method based on risk assessment in the foregoing embodiments are implemented, for example, steps S201 to S206 shown in fig. 2 or steps shown in fig. 3 to fig. 8, and are not described herein again to avoid repetition. Alternatively, when being executed by the processor, the computer program implements functions of each module/unit in the embodiment of the intelligent navigation device based on risk assessment, for example, functions of the route navigation request obtaining module 901, the original navigation route obtaining module 902, the to-be-assessed traffic data obtaining module 903, the road segment risk probability obtaining module 904, the total risk probability obtaining module 905, and the recommended navigation route module 906 shown in fig. 9, which are not described herein again in order to avoid repetition.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

It should be clear to those skilled in the art that, for convenience and simplicity of description, the foregoing division of the functional units and modules is only used for illustration, and in practical applications, the above function distribution may be performed by different functional units and modules as needed, that is, the internal structure of the device is divided into different functional units or modules, so as to perform all or part of the above described functions.

The above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.

Claims (8)

1. An intelligent navigation method based on risk assessment is characterized by comprising the following steps:

the method comprises the steps of obtaining a route navigation request sent by a client, wherein the route navigation request comprises an initial place, a target place, a user account and a navigation type;

acquiring at least one original navigation route based on the starting position and the target position, wherein each original navigation route comprises at least one navigation road section;

if the navigation type is a safety recommendation type, acquiring traffic data to be evaluated corresponding to each navigation road section;

the traffic data to be evaluated comprises driver information to be evaluated, vehicle information to be evaluated, time information to be evaluated, road section information to be evaluated and place information to be evaluated;

querying a navigation system database based on the user account to acquire historical navigation data corresponding to the user account, wherein each historical navigation data corresponds to a historical recommendation type;

calculating the quotient of the quantity of the historical navigation data corresponding to the safety recommendation type and the quantity of all historical navigation data in the historical recommendation types of the historical navigation data, and determining the calculation result as the historical recommendation frequency corresponding to the safety recommendation type;

if the historical recommendation frequency corresponding to the safety recommendation type is smaller than a preset frequency threshold, acquiring travel risk information and sending the travel risk information to the client;

inputting the traffic data to be evaluated corresponding to the navigation road section into a road risk recognition model which is trained by adopting a decision tree algorithm in advance, and performing risk evaluation on the traffic data to be evaluated corresponding to the navigation road section according to the classification logic of the decision tree algorithm to obtain the road section risk probability corresponding to the navigation road section;

acquiring a total risk probability corresponding to each original navigation route based on a road section risk probability corresponding to at least one navigation road section in each original navigation route;

determining an original navigation route with the minimum total risk probability as a recommended navigation route, and sending the recommended navigation route to the client;

acquiring current traffic data and a current driving position sent by a client in real time in the process that the vehicle drives according to the recommended navigation route;

the current traffic data comprises current driver information, current vehicle information, current road section information, current location information and current time information;

performing risk identification on the current traffic data by adopting the road risk identification model to obtain a current risk probability, and if the current risk probability is greater than a preset risk threshold, obtaining first risk data;

according to the driving direction of the recommended navigation route, if accident frequent places or sudden accident information exists in the preset distance range of the current driving position, second risk data are obtained;

filling the first risk data and the second risk data into a preset risk reminding template to obtain risk reminding text data;

and converting the risk reminding text data into risk reminding voice data by adopting a TTS technology, and broadcasting the risk reminding voice data through the client.

2. The intelligent navigation method based on risk assessment according to claim 1, wherein the obtaining of the route navigation request sent by the client, the route navigation request including a start location, a target location and a navigation type, comprises:

acquiring a voice navigation request sent by a client, wherein the voice navigation request comprises voice data to be recognized;

performing voice recognition on the voice data to be recognized by adopting a voice recognition model to obtain text data to be recognized;

extracting the text data to be recognized by adopting a keyword extraction algorithm to obtain target keywords;

and matching a preset keyword library based on the target keywords to obtain an initial place, a target place and a navigation type which are successfully matched.

3. The intelligent navigation method based on risk assessment according to claim 2, wherein before the voice navigation request sent by the obtaining client, the voice navigation request including the voice data to be recognized, the intelligent navigation method based on risk assessment further comprises:

acquiring original voice data acquired by a client in real time;

carrying out end point detection and feature extraction on the original voice data to obtain target acoustic features;

recognizing the target acoustic characteristics by adopting the voice recognition model to obtain target text data;

and if the target text data contains the awakening keyword, controlling the client to enter a voice navigation interface within a preset time period, and receiving the voice navigation request.

4. The intelligent navigation method based on risk assessment according to claim 1, wherein before the obtaining of the route navigation request sent by the client, the intelligent navigation method based on risk assessment further comprises:

acquiring historical traffic accident data from a traffic police platform database through a third-party interface, wherein the historical traffic accident data comprises historical driver information, historical vehicle information, historical road section information, historical place information and historical time information;

inquiring the traffic police platform database based on the historical driver information, acquiring historical driving behavior data corresponding to the historical driver information, inquiring a behavior information data table based on the historical driving behavior data, and acquiring a driver evaluation index;

inquiring a vehicle information data table based on the historical vehicle information to obtain a vehicle evaluation index corresponding to the historical vehicle information;

inquiring a road section information data table based on the historical road section information to obtain a road section evaluation index corresponding to the historical road section information;

acquiring historical meteorological information corresponding to the historical place information and the historical time information from a meteorological platform database through a third-party interface, inquiring a meteorological information data table based on the historical meteorological information, and acquiring meteorological evaluation indexes corresponding to the historical meteorological data;

inquiring a calendar table based on the historical time information to obtain a time evaluation index corresponding to the historical time information;

acquiring model training data corresponding to the historical traffic accident data based on the driver assessment index, the vehicle assessment index, the road section assessment index, the weather assessment index and the time assessment index corresponding to the historical traffic accident data;

and performing model training on the model training data by adopting a decision tree algorithm to obtain a road risk identification model based on a decision tree.

5. The intelligent navigation method based on risk assessment according to claim 4, wherein model training is performed on the model training data by adopting a decision tree algorithm to obtain a road risk recognition model based on a decision tree, and the method comprises the following steps:

determining a class label characteristic and at least two training attribute characteristics corresponding to each model training data based on each model training data, and storing the model training data in a training data set;

calculating information gain ratios of the class label characteristics and the training attribute characteristics of the model training data to obtain the information gain ratio corresponding to each training attribute characteristic;

selecting the training attribute feature with the maximum information gain rate as a target classification feature of the training data set, and dividing the training data set into at least two training subsets by adopting the target classification feature;

judging whether training attribute features which are not determined as target classification features exist or not;

if the training attribute features which are not determined as the target classification features exist, updating the training subset into the training data set, repeatedly executing the step of calculating the information gain ratio of the class label features and the training attribute features of the model training data, and acquiring the information gain ratio corresponding to each training attribute feature;

if the training attribute characteristics which are not determined as the target classification characteristics do not exist, an original risk identification model based on a decision tree algorithm is formed, data volume corresponding to each leaf node in the original risk identification model is obtained, the original risk identification model is cut based on the data volume, and a road risk identification model based on the decision tree is obtained.

6. An intelligent navigation device based on risk assessment, comprising:

the route navigation request acquisition module is used for acquiring a route navigation request sent by a client, wherein the route navigation request comprises an initial place, a target place, a user account and a navigation type;

the original navigation route acquisition module is used for acquiring at least one original navigation route based on the starting point and the target point, and each original navigation route comprises at least one navigation road section;

the traffic data to be evaluated acquiring module is used for acquiring the traffic data to be evaluated corresponding to each navigation road section if the navigation type is the safety recommendation type;

the traffic data to be evaluated comprises driver information to be evaluated, vehicle information to be evaluated, time information to be evaluated, road section information to be evaluated and place information to be evaluated;

the road section risk probability obtaining module is used for inputting the traffic data to be evaluated corresponding to the navigation road section into a road risk recognition model which is trained by adopting a decision tree algorithm in advance, carrying out risk evaluation on the traffic data to be evaluated corresponding to the navigation road section according to the classification logic of the decision tree algorithm, and obtaining the road section risk probability corresponding to the navigation road section;

the total risk probability obtaining module is used for obtaining total risk probability corresponding to the original navigation route based on the road section risk probability corresponding to at least one navigation road section in each original navigation route;

the recommended navigation route module is used for determining the original navigation route with the minimum total risk probability as a recommended navigation route and sending the recommended navigation route to the client;

the current data acquisition module is used for acquiring current traffic data and a current driving position sent by the client in real time in the process that the vehicle drives according to the recommended navigation route;

the current traffic data comprises current driver information, current vehicle information, current time information, current road section information and current location information;

the historical data query module is used for querying a navigation system database based on the user account to acquire historical navigation data corresponding to the user account, and each historical navigation data corresponds to a historical recommendation type;

the history frequency calculation module is used for calculating the quotient of the quantity of the history navigation data corresponding to the safety recommendation type and the quantity of all history navigation data in the history recommendation types of the history navigation data, and determining the calculation result as the history recommendation frequency corresponding to the safety recommendation type;

the journey risk recommendation module is used for acquiring journey risk information and sending the journey risk information to the client if the historical recommendation frequency corresponding to the safety recommendation type is smaller than a preset frequency threshold;

the first risk data acquisition module is used for carrying out risk identification on the current traffic data by adopting a road risk identification model, acquiring current risk probability, and acquiring first risk data if the current risk probability is greater than a preset risk threshold;

the second risk data acquisition module is used for acquiring second risk data if an accident frequent place or sudden accident information exists in a preset distance range of the current driving position according to the driving direction of the recommended navigation route;

the reminding text data acquisition module is used for filling the first risk data and the second risk data into a preset risk reminding template to acquire risk reminding text data;

and the reminding voice data broadcasting module is used for converting the risk reminding text data into risk reminding voice data by adopting a TTS technology and broadcasting the risk reminding voice data through the client.

7. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor when executing the computer program implements the steps of the risk assessment based intelligent navigation method according to any one of claims 1 to 5.

8. A computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, carries out the steps of the risk assessment based intelligent navigation method according to any one of claims 1 to 5.

Images