CN109801491A - Intelligent navigation method, device, equipment and storage medium based on risk assessment - Google Patents

Abstract

The invention discloses the intelligent navigation methods based on risk assessment, device, equipment and storage medium, this method comprises: obtaining the route guidance request that client is sent, origin and target location based on route guidance, obtain a plurality of original navigation route, further obtain the corresponding traffic data to be assessed in each navigation section, risk assessment is carried out to the corresponding traffic data to be assessed in navigation section using road risk identification model trained in advance, obtain the corresponding section risk probability in navigation section, based on the corresponding section risk probability at least one of each original navigation route navigation section, obtain the corresponding overall risk probability of original navigation route, it is to recommend navigation routine by the smallest original navigation route determination of overall risk probability, navigation routine will be recommended to be sent to client.This method, which can be realized, carries out route planning, guarantee driving safety based on safety factor.

Description

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

Technical field

The present invention relates to intelligent navigation technology field more particularly to a kind of intelligent navigation methods based on risk assessment, dress It sets, equipment and storage medium.

Background technique

Plan that optimal navigation routine is a basic function of navigation system in navigation road network.Present navigation system exists When route planning, it is main according to road whether congestion, pass by one's way whether charge, whether be that shortest distance etc. is planned, the road Ji Qi The factors such as speed, time and expense are laid particular emphasis in line planning process, are not based on safety factor and are carried out route planning, are unfavorable for protecting Hinder safety travel.

Summary of the invention

The embodiment of the present invention provides a kind of intelligent navigation method based on risk assessment, device, equipment and storage medium, with Solve the problems, such as that Present navigation system can not carry out route planning based on safety factor.

A kind of intelligent navigation method based on risk assessment, comprising:

Obtain the route guidance request that client is sent, route guidance request include origin, target location and Navigation type；

Based on the origin and the target location, at least one original navigation route is obtained, it is each described original Navigation routine includes at least one navigation section；

If the navigation type is safe type of recommendation, the corresponding traffic number to be assessed in each navigation section is obtained According to；

By the corresponding traffic data input to be assessed in the navigation section in advance using the road transportation work style of decision Tree algorithms training In dangerous identification model, the sorted logic traffic data to be assessed corresponding to the navigation section according to decision Tree algorithms carries out wind Danger assessment obtains the corresponding section risk probability in the navigation section；

Based on the corresponding section risk probability in navigation section described at least one of each described original navigation route, obtain Take the corresponding overall risk probability of the original navigation route；

It is to recommend navigation routine by the smallest original navigation route determination of overall risk probability, the recommendation navigation routine is sent out Give the client.

A kind of intelligent navigation device based on risk assessment, comprising:

Route guidance request module, for obtaining the route guidance request of client transmission, route guidance request packet Include origin, target location and navigation type；

Original navigation route acquiring module obtains at least one original navigation for being based on origin and target location Route, each original navigation route include at least one navigation section；

Traffic data to be assessed obtains module, if being safe type of recommendation for navigation type, obtains each navigation road The corresponding traffic data to be assessed of section；

Section risk probability obtains module, for adopting the corresponding traffic data input to be assessed in the navigation section in advance In road risk identification model with decision Tree algorithms training, the sorted logic according to decision Tree algorithms is to the navigation section pair The traffic data to be assessed answered carries out risk assessment, obtains the corresponding section risk probability in the navigation section；

Overall risk probability obtains module, for corresponding based at least one of each original navigation route navigation section Section risk probability obtains the corresponding overall risk probability of original navigation route；

Recommend navigation routine module, for being to recommend navigation road by the smallest original navigation route determination of overall risk probability Line will recommend navigation routine to be sent to client.

A kind of computer equipment, including memory, processor and storage are in the memory and can be in the processing The computer program run on device, the processor realize the above-mentioned intelligence based on risk assessment when executing the computer program The step of air navigation aid.

A kind of computer readable storage medium, the computer-readable recording medium storage have computer program, the meter The step of calculation machine program realizes the above-mentioned intelligent navigation method based on risk assessment when being executed by processor.

The above-mentioned intelligent navigation method based on risk assessment, device, equipment and storage medium, according to origin and target Place can quickly determine at least one original navigation route.When navigation type is safe type of recommendation, according to original navigation road The traffic data to be assessed in each navigation section in line, inquiry roads risk identification model obtain the corresponding road in navigation section Section risk probability guarantees that the section risk probability got has objectivity.Section risk based at least one navigation section Probability obtains corresponding overall risk probability, so that the overall risk probability reflects the driving risk of original navigation route on the whole, Guarantee the objectivity of result.Finally, being to recommend navigation routine by the smallest original navigation route determination of overall risk probability, to reach The purpose of intelligent navigation is carried out based on safety factor.

Detailed description of the invention

In order to illustrate the technical solution of the embodiments of the present invention more clearly, below by institute in the description to the embodiment of the present invention Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings Obtain other attached drawings.

Fig. 1 is an application environment schematic diagram of the intelligent navigation method in one embodiment of the invention based on risk assessment；

Fig. 2 is a flow chart of the intelligent navigation method in one embodiment of the invention based on risk assessment；

Fig. 3 is another flow chart of the intelligent navigation method in one embodiment of the invention based on risk assessment；

Fig. 4 is another flow chart of the intelligent navigation method in one embodiment of the invention based on risk assessment；

Fig. 5 is another flow chart of the intelligent navigation method in one embodiment of the invention based on risk assessment；

Fig. 6 is another flow chart of the intelligent navigation method in one embodiment of the invention based on risk assessment；

Fig. 7 is another flow chart of the intelligent navigation method in one embodiment of the invention based on risk assessment；

Fig. 8 is another flow chart of the intelligent navigation method in one embodiment of the invention based on risk assessment；

Fig. 9 is a schematic diagram of the intelligent navigation device in one embodiment of the invention based on risk assessment；

Figure 10 is a schematic diagram of computer equipment in one embodiment of the invention.

Specific embodiment

Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete Site preparation description, it is clear that described embodiments are some of the embodiments of the present invention, instead of all the embodiments.Based on this hair Embodiment in bright, every other implementation obtained by those of ordinary skill in the art without making creative efforts Example, shall fall within the protection scope of the present invention.

Intelligent navigation method provided in an embodiment of the present invention based on risk assessment, should the intelligent navigation based on risk assessment Method can be using in application environment as shown in Figure 1.Specifically, the intelligent navigation method based on risk assessment is somebody's turn to do to apply in intelligence In navigation system, which includes client and server as shown in Figure 1, and client and server pass through network It is communicated, for realizing risk assessment is carried out during route planning, guarantees traffic safety.Wherein, client is also known as User terminal, refers to corresponding with server, provides the program of local service for client.Client it is mountable but be not limited to various On personal computer, laptop, smart phone, tablet computer and portable wearable device.Server can be with independently The server cluster of server either multiple servers composition realize.

In one embodiment, it as shown in Fig. 2, providing a kind of intelligent navigation method based on risk assessment, answers in this way It is illustrated, includes the following steps: for the server in Fig. 1

S201: obtain client send route guidance request, route guidance request include origin, target location and Navigation type.

Wherein, route guidance request is that user is used to carry out asking for route planning by what user end to server was sent It asks.Origin is the starting point of this route guidance request, is specifically as follows user by position of the client from primary input, also It can be the position of the included GPS module autonomous positioning of client.Target location is the terminal of this route guidance request.Navigation Type refers to that this route guidance requests type determined by selected progress route planning priority principle.The navigation type packet It includes but is not limited to common apart from type of recommendation (using the distance of distance as the type of route planning principle) and expense type of recommendation (using the number of expense as the type of route planning principle), further includes safe type of recommendation (using the risk of safety factor as route The type of planning principles).

S202: it is based on origin and target location, obtains at least one original navigation route, each original navigation route Including at least one section of navigating.

Wherein, original navigation route is the navigation road for the P Passable that server is determined according to origin and target location Line.Server is first based on origin and target location enquiry navigation system database, with obtaining connection origin and target All original navigation routes between point.For example, from Shenzhen (origin) to Guangzhou (target location) there are 5 originals Beginning navigation routine, the 1st article is by way of Beijing-Hongkong Australia high speed and this original navigation route of Shen Hai high speed Guangzhou branch line；2nd article is way Through imperial big high speed and this original navigation route of Beijing-Hongkong Australia high speed；3rd article be by way of Beijing-Hongkong Australia high speed and wide deep riverine expressway this One original navigation route；4th article is by way of imperial big high speed and this original navigation route of region of rivers and lakes main road.5th article is by way of wide depth Highway and this original navigation route of the through street Guang Yuan.

In the present embodiment, determine that at least one is led according to its road name and its locating region in each original navigation route Air route section.It is to be appreciated that may only include a navigation section if the distance of original navigation route is shorter；If original The distance of navigation routine is longer, then the road name passed through according to the original navigation route or locating region determine at least two A navigation section.

S203: if navigation type is safe type of recommendation, the corresponding traffic data to be assessed in each navigation section is obtained.

Wherein, traffic data to be assessed refers to that is acquired during route planning drives a vehicle needed for risk for assessing it Data.The traffic data to be assessed includes but is not limited to driver information to be assessed, information of vehicles to be assessed, section to be assessed letter Breath, location information to be assessed and temporal information to be assessed.Wherein, driver information to be assessed, which is used to indicate that, triggers this route The information of the driver identity of navigation requests can specifically include name, ID card No. and the drivers license number of driver.To Assessment information of vehicles is used to indicate that the information of vehicle associated by the client for triggering the request of this route guidance, including but not It is limited to the model and mileage travelled number of vehicle.Road section information to be assessed is used to indicate that the corresponding section in each navigation section Information.Location information to be assessed is used to indicate that the information in place locating for each navigation section.Temporal information to be assessed is to use In the corresponding information of current time for showing to trigger the request of this route guidance.

In the present embodiment, when navigation type is safe type of recommendation, then illustrate that user passes through client request server Route planning is carried out by route planning principle of safety factor, therefore the traffic data to be assessed for influencing traffic safety need to be acquired, Risk identification is carried out based on the traffic data to be assessed so as to subsequent.It is to be appreciated that being apart from type of recommendation in navigation type When with expense type of recommendation, without obtaining traffic data to be assessed, that is, it is not necessarily based on traffic data progress risk to be assessed and comments Estimate.

S204: by the corresponding traffic data input to be assessed in section of navigating in advance using the road transportation work style of decision Tree algorithms training In dangerous identification model, the sorted logic according to decision Tree algorithms carries out risk to the corresponding traffic data to be assessed in navigation section and comments Estimate, obtains the corresponding section risk probability in navigation section.

Wherein, road risk identification model be in advance based on the model training data that are formed by historical traffic casualty data into Row training is formed by the model of risk for identification, carries out model training using decision Tree algorithms during model training. The historical traffic casualty data is the data for the historical traffic accident (traffic accident having occurred and that) having occurred and that.With it is to be evaluated It is corresponding to estimate traffic data, which includes but is not limited to history driver information, history information of vehicles, goes through History road section information, history location information and historical time information.Wherein, history driver information is used to indicate that historical traffic thing Therefore the information of middle driver identity, it can specifically include name, ID card No. and the drivers license number of driver.History vehicle Information is used to indicate that the information of vehicle in historical traffic accident, the including but not limited to model of vehicle and mileage travelled number. History road section information is used to indicate that the information in section locating for historical traffic accident.History location information is used to indicate that history is handed over The information of interpreter's event Location.Historical time information is used to indicate that the information of historical traffic accident corresponding time.The road Transportation work style danger identification model is to be formed by model training data based on historical traffic casualty data and be trained, make its can for The corresponding traffic data to be assessed of model training data carries out risk identification, and so as to quick obtaining, the navigation section is corresponding Section risk probability, so that the section risk probability obtained has objectivity and intuitive.The section risk probability is road transportation work style Dangerous identification model carries out identifying acquired risk probability to the traffic data to be assessed in a certain navigation section.

In the present embodiment, road risk identification model is specially that decision Tree algorithms is used to be trained model training data Acquired model, i.e. the road risk identification model are general with the corresponding risk of the different model training data of tree storage Rate.It include root node, leaf node and connection root node and leaf node in the taxonomic structure of road risk identification model Intermediate node, wherein without father node on root node, without child node under leaf node.Root node and each intermediate node pair A sorted logic is answered, the corresponding risk probability of each leaf node, the risk probability is according to all classification for meeting respective branch The quantity of the quantity of the model training data of logic and all model training data determines, can be the quotient of the two quantity.Tool Body, which is input to this after obtaining the corresponding traffic data to be assessed in navigation section by server Road risk identification model, the sorted logic according to decision Tree algorithms carry out wind to the corresponding traffic data to be assessed in navigation section When the assessment of danger, first the sorted logic of root node is called to classify traffic data to be assessed, determine the intermediate node belonging to it； Further classified according to the sorted logic of the intermediate node again, so that it is determined that the intermediate node ... of next stage class according to this Push away, until determine its corresponding leaf node, and by the corresponding risk probability of the leaf node be determined as navigate section it is corresponding Section risk probability.

S205: it based on the corresponding section risk probability at least one of each original navigation route navigation section, obtains The corresponding overall risk probability of original navigation route.

Since each original navigation route includes at least one navigation section, and each navigation section corresponds to a road section wind Dangerous probability, in the present embodiment, server is general by least one corresponding section risk in navigation section in each original navigation route Rate is added, and can obtain the corresponding overall risk probability of the original navigation route.It is to be appreciated that the overall risk probability is on the whole Reflect the driving risk of original navigation route.It is to be appreciated that the corresponding overall risk probability in the original navigation section is by least The section risk probability of one navigation routine, which calculates, to be obtained, so that the objective circumstances for comprehensively considering each navigation section are (such as each The road section information to be assessed in section, the temporal information to be assessed of navigating and it is based on location information to be assessed and temporal information to be assessed Determining weather information to be assessed) difference, the overall risk probability of acquisition has more objectivity.

S206: it is to recommend navigation routine by the smallest original navigation route determination of overall risk probability, navigation routine will be recommended It is sent to client.

It is corresponding to all original navigation routes total when navigation type in route guidance request is safe type of recommendation Risk probability is ranked up, and is to recommend navigation routine for the smallest original navigation route determination of overall risk probability, and by the recommendation Navigation routine is sent to client, so that user drives vehicle according to the recommendation navigation routine that client is shown.It is to be appreciated that Recommending navigation routine is the navigation routine slave origin to target location finally recommended to user.

It, can according to origin and target location in intelligent navigation method based on risk assessment provided by the present embodiment Quickly determine at least one original navigation route.When navigation type is safe type of recommendation, according in original navigation route The traffic data to be assessed in each navigation section, inquiry roads risk identification model obtain the corresponding section risk in navigation section Probability guarantees that the section risk probability got has objectivity.Section risk probability based at least one navigation section obtains Corresponding overall risk probability is taken, so that the overall risk probability reflects the driving risk of original navigation route on the whole, guarantees knot The objectivity of fruit.Finally, being to recommend navigation routine by the smallest original navigation route determination of overall risk probability, peace is based on to reach The purpose of total factor progress intelligent navigation.

In one embodiment, as shown in figure 3, the route guidance request that step S201, i.e. acquisition client are sent, route are led Boat request includes origin, target location and navigation type, is specifically comprised the following steps:

S301: the Voice Navigation request that client is sent is obtained, Voice Navigation request includes voice data to be identified.

Wherein, Voice Navigation request is the voice for being used to carry out route planning that user is sent by user end to server The request of form.Voice data to be identified is number of the user by reflection its navigation needs and navigation type of client typing According to.For example, user can " I wants from Guangzhou to Shenzhen, is safely by voice capture device (such as microphone) typing of client On ".

S302: speech recognition is carried out to voice data to be identified using speech recognition modeling, obtains text data to be identified.

Wherein, speech recognition modeling is the model of the content of text in preparatory trained voice data for identification.This Static voice decoding network can be used in speech recognition modeling in embodiment, since static decoding network is complete search space Portion's expansion, therefore it when carrying out text translation, decoding speed is very fast, so as to quick obtaining text data to be identified.It should Static voice decoding network is that acquired static decoding network, the spy are trained using the training voice data of specific area The training voice data for determining field can be understood as the corresponding voice data of pre-stored historical navigation data.Since voice is quiet State decoding network is that the training voice data based on specific area is trained acquired static decoding network, so that it is right The voice data to be identified of specific area is with strong points when being identified, so that decoding accuracy rate is higher.

S303: extracting text data to be identified using keyword extraction algorithm, obtains target keyword.

In, keyword extraction algorithm refers to the algorithm that keyword therein is extracted from text data.In the present embodiment, clothes Business device first uses participle tool to segment text data to be identified, then uses and deactivated word algorithm is gone to go word segmentation result Stop words processing, to obtain target keyword.The target keyword is the keyword that can reflect navigation needs and navigation type.

S304: matching predetermined keyword library based on target keyword, obtain the origin of successful match, target location and Navigation type.

Wherein, the database of the keyword with navigation needs and navigation type is stored in predetermined keyword library.Specifically, it takes Device be engaged in based on target keyword matching predetermined keyword library, obtains origin, target location and the navigation type of successful match. For example, the target keyword recognized in " I wants from Guangzhou to Shenzhen, selects safest route " is " I ", " thinking ", " from ", " Guangzhou ", " arriving ", " Shenzhen ", " selection ", " most ", " safety " and " route ".In predetermined keyword library configure " from C to D " this In one form of presentation, C is origin, and D is target location, so that it is determined that Guangzhou is origin, Shenzhen is target location.If When in the target keyword recognized including safe or security-related keyword, then assert that its navigation type pushes away for safety Recommend type；If assert its navigation type when in the target keyword recognized comprising expense or with costs related keyword Type is expense type of recommendation；If assert when in the target keyword recognized comprising distance and to apart from relevant keyword Its navigation type is apart from type of recommendation.

In intelligent navigation method based on risk assessment provided by the present embodiment, using speech form typing Voice Navigation Request, identifies voice data to be identified using speech recognition modeling, can quick obtaining text data to be identified；Then, Keyword extraction is carried out to text data to be identified using keyword extraction algorithm to close to obtain target keyword using target Keyword and predetermined keyword library, can the corresponding origin of quick obtaining, target location and navigation type, so that it is double to liberate user Hand can also carry out route guidance without being manually entered, and avoiding user in driving process needs to be manually entered route guidance request.

In one embodiment, it in order to further liberate user's both hands, is manually operated in driving vehicle processes without user Therefore to input Voice Navigation request, voice arousal function need to be arranged in client in intelligent guidance system.As shown in figure 4, Before step S201, i.e., in the Voice Navigation request for obtaining client transmission, Voice Navigation request includes voice number to be identified According to before, the intelligent navigation method based on risk assessment further includes following steps:

S401: the primary voice data that client acquires in real time is obtained.

Wherein, primary voice data is the voice data that client actual acquisition arrives.The primary voice data is understood that Collected voice data when for client in a dormant state, it is whether intentional for detecting user according to the primary voice data To progress Voice Navigation.In the present embodiment, whether user has intention to carry out Voice Navigation especially by detection primary voice data In whether realized comprising waking up keyword.

S402: end-point detection and feature extraction are carried out to primary voice data, obtain target acoustical feature.

Speech terminals detection (Voice Activity Detection, VAD), also known as voice activity detection or voice side Boundary's detection refers to the process of the presence or absence that voice is detected in noise circumstance.In the present embodiment, client is built-in with for real The voice activity detector (Voice Activity Detector) of existing speech terminals detection is obtained for excluding ambient noise Voice data described in user.Speech feature extraction, which refers to, comes out the constituents extraction in audio signal with identification, to go Except ambient noise or the process of other information unrelated with identification process.In the present embodiment, server passes through to original Voice data carries out end-point detection and feature extraction, to exclude the noise in primary voice data, obtains target acoustical feature.It should Target acoustical feature can be mel cepstrum coefficients or other phonetic features.

S403: identifying target acoustical feature using speech recognition modeling, obtains target text data.

Wherein, same speech recognition modeling is acquired in step S403 and step S302, known without in addition two voices of training Other model simplifies process flow.And the realization process of step S403 is similar to step S302, details are not described herein again.

S404: it if target text data include to wake up keyword, controls client and enters voice within a preset period of time Navigation interface receives Voice Navigation request.

Wherein, which is that server is pre-set for waking up the pass that client enters Voice Navigation interface Keyword.The preset time period is the pre-set period, for limiting the time of voice wake-up.In the present embodiment, work as service When device recognizes in target text data comprising waking up keyword, client can control to carry out Voice Navigation interface, to obtain language Sound navigation requests.Further, server control client is default from being recognized when waking up keyword in primary voice data Enter Voice Navigation interface in period, so that client only obtains the request of the Voice Navigation in preset time period.It is understood that Ground, client enter Voice Navigation interface within a preset period of time, if obtaining the language that client is sent in the preset time period Sound navigation requests then execute subsequent step S302；If the Voice Navigation of client transmission has not been obtained in the preset time period Request, then reenter dormant state, repeat step S401-S403, avoids waiting in Voice Navigation interface for a long time The problem for receiving Voice Navigation request and causing power consumption big.

Intelligent guidance system is also configured with voice arousal function, for working as service according to pre-set wake-up keyword Device detect client in real time collected target speech data include the wake-up keyword when, can start receive Voice Navigation Request.Voice, which wakes up, is otherwise referred to as keyword detection (Keyword spotting), that is, in continuously voice Keyword detection will be waken up to come out, the general number for waking up keyword is fewer, and (1-2 in the majority, and special circumstances can also extend To more several).Voice wake-up is to handle continuously voice flow, such as voice switch 24 hours continual detection wheats Target keyword in gram wind recording (i.e. target speech data)；Voice wake-up can be combined with speech recognition technology, for examining The position that voice starts is surveyed, replaces in key, such as Amazon Echo, uses " alexa " is as wake-up keyword, once it examines Wake-up keyword is measured, then controls client and carries out Voice Navigation interface, starts recording and carries out speech recognition.

In intelligent navigation method based on risk assessment provided by the present embodiment, by the raw tone acquired in real time Data carry out end-point detection and feature extraction, obtain target acoustical feature；Speech recognition modeling is acquired again to be identified, mesh is obtained Text data is marked, when target text data include to wake up keyword, control client is led into voice within a preset period of time Navigate interface, both guaranteed can to acquire Voice Navigation request, to avoid long-time in etc. Voice Navigation request to be received state, from And the amount of saving energy.

In one embodiment, due to being needed in step 204 using road risk identification model trained in advance to navigation road The corresponding traffic data to be assessed of section carries out risk assessment, therefore, carries out road in the route guidance request sent based on client Before line gauge is drawn, the road risk identification model that risk assessment can be achieved need to be trained in advance.As shown in figure 5, step 201 it Before, i.e., before obtaining the route guidance request that client is sent, the intelligent navigation method based on risk assessment further includes as follows Step:

S501: historical traffic casualty data, historical traffic thing are obtained from traffic police's platform database by third party's interface Therefore data include history driver information, history information of vehicles, history road section information, history location information and historical time letter Breath.

Wherein, third party's interface be arranged in intelligent guidance system for connect third-party platform (including but Be not limited to traffic police's platform database and weather platform database) interface.Traffic police's platform database is for storing traffic police's platform The database of the traffic accident data of typing and driving behavior data corresponding with each driver information.Historical traffic accident Data are stored in the number for log history traffic accident (traffic accident having occurred and that) in traffic police's platform database According to.In the present embodiment, server can obtain before the current time in system by third party's interface from traffic police's platform database It is entered into the historical traffic casualty data of traffic police's platform database.

Wherein, historical traffic casualty data includes history driver information, history information of vehicles, history road section information, goes through History location information and historical time information.History driver information is used to indicate that the letter of driver identity in historical traffic accident Breath, can specifically include name, ID card No. and the drivers license number of driver.History information of vehicles is used to indicate that history The information of vehicle in traffic accident, the including but not limited to model of vehicle and mileage travelled number.History road section information is to be used for Show the information in section locating for historical traffic accident.History location information is used to indicate that the letter of historical traffic accident Location Breath.Historical time information is used to indicate that the information of historical traffic accident corresponding time.

S502: traffic police's platform database is inquired based on history driver information, is obtained corresponding with history driver information Historical driving behavior data, be based on historical driving behavior data query behavioural information tables of data, obtain driver evaluation's index.

Wherein, historical driving behavior data are that traffic police's platform database is recorded whether the driver violates friendship rule, violates Hand over the behavioral datas such as the severity of rule.Since people is the most important factor that traffic accident occurs, historical driving behavior Data can reflect the driving technology and quality of the driver, for example, driver do not observe surface conditions drive directly into runway can Cause vehicle rear-end collision, habit to be stepped on the gas, unreasonable overtake other vehicles, is not concerned with the reasons such as traffic sign and graticule and is all easy to cause traffic thing Therefore occur.It is to be appreciated that the historical driving behavior data are for reflecting that the driver of the historical traffic casualty data is being It whether there is the behavior for violating friendship rule before system current time or violate the behavioral datas such as the severity of behavior for handing over rule.It should Historical driving behavior data can specifically pass through penalty note data, deduction of points data and other historical traffic casualty datas etc..

Behavioural information tables of data is that each historical driving behavior data that are used to record being stored in advance in the database correspond to Driver evaluation's index tables of data.It is current that according to rule are handed over, to delay penalty, there are following several situations: directly revoke driving license, It writes 12 points (including primary and n times) all over, 6 points of note, 3 points of note, 2 points of note, remember 1 point and do not score and only impose a fine.Behavior information data Table is several evaluation grades being configured according to the punishment situation of above-mentioned delay penalty, such as can be divided into P1, P2, P3 and P4 These types of type, wherein P1 is to be directed to directly to revoke driving license and write the types such as 12 points all over, and P2 is to remember 6 points and 1 in the presence of primary 12 type is not write year all over, P3 is at most 3 points of note but punishment number reaches the type of n times, and P4 is that primary punishment is less than 3 Divide but punish the type that number reaches n times.

S503: being based on history information of vehicles enquiring vehicle information data table, obtains vehicle corresponding with history information of vehicles Evaluation index.

Vehicle information data table is that commenting for recording the corresponding vehicle of each information of vehicles in the database is stored in advance Estimate the tables of data of index.In vehicle information data table for record influence vehicle evaluation index various vehicle assessment factors with And any two kinds or the corresponding several evaluation grades of two kinds of vehicle assessment factor combination of the above, respectively with C1, C2, C3 and C4 come table Show.Wherein, vehicle assessment factor includes vehicle power parameter, vehicle service life (newness degree) and vehicle maintenance number.Example Such as, can be divided according to vehicle service life four grades be respectively 8 years or more, 5-8,2-5 perhaps within 2 years or Vehicle maintenance number is divided into 8 times, 5-8 times, 2-5 times or 2 times.

S504: road section information tables of data is inquired based on history road section information, obtains road corresponding with history road section information Section evaluation index.

Road section information tables of data is that the corresponding section assessment of every a road section that is used to record being stored in advance in the database refers to Target tables of data.For recording the various section assessment factors of influence section evaluation index and appointing in road section information tables of data At least one section assessment factor of anticipating combines corresponding several evaluation grades, and respectively L1, L2, L3 and L4 is indicated.For example, should Section assessment factor includes section attribute, which includes mixing road, vehicle one-way road, vehicle multirow road and high speed road Road.Super expressway includes high speed fast traffic lane (such as 100km/h or more) and high speed slow lane.Mixing road refers to no road-center The road of line, i.e. road center do not have yellow line to distinguish the road in opposite lane.There is no any graticule on Ordinary Rd, is motor-driven Vehicle, non-motor vehicle, people's vehicle mix trade road.

S505: it through third party's interface from meteorological platform database, obtains and believes with history location information and historical time The corresponding history weather information of manner of breathing is inquired weather information data table based on history weather information, is obtained and history meteorological data Corresponding meteorology evaluation index.

Wherein, history weather information refers to weather information when a certain historical traffic accident occurs.Weather information data table It is the tables of data for storing the corresponding meteorological evaluation index of each weather information.Recording in weather information data table influences The various meteorological assessment factors and at least one meteorological assessment factor of meteorological evaluation index combine corresponding several evaluation grades, Respectively Q1, Q2, Q3 and Q4.It, can also be according to the evaluation grade that intelligent guidance system is independently arranged in current weather evaluation process Classification chart determines its evaluation grade, can also be according to typhoon, heavy rain, high temperature, cold wave, dense fog, thunderstorm gale, strong wind, sandstorm, ice The warning grades such as hail, snow disaster and icy road determine its evaluation grade.For example, Q1, Q2, Q3 and Q4 correspond respectively to typhoon early warning Red, orange, yellow and blue early warning.

S506: inquiring year ephemeris based on historical time information, obtains time assessment corresponding with historical time information and refers to Mark.

The tables of data of annual festivals or holidays situation is recorded in year ephemeris.In the present embodiment, it is based on temporal information query history table, To determine whether the historical traffic accident occurred in festivals or holidays, if in information such as peak periods on and off duty.For example, can will at that time Between evaluation index be divided into festivals or holidays, weekday rush phase, working day non-peak period and weekend, respectively correspond T1, T2, T3 and T4。

S507: referred to based on the corresponding driver evaluation's index of historical traffic casualty data, vehicle evaluation index, section assessment Mark, meteorological evaluation index and time evaluation index, obtain model training data corresponding with historical traffic casualty data.

For each historical traffic casualty data in traffic police's platform database, carried out according to above-mentioned steps S502-S506 Processing obtains its corresponding driver evaluation's index, vehicle evaluation index, section evaluation index, meteorological evaluation index and time Evaluation index can be respectively adopted the expression of the parameter identifications such as P, C, L, Q and T, that is, it is corresponding to obtain each historical traffic casualty data Model training data are represented by (P, C, L, Q, T), the corresponding concrete condition of each parameter identification of one list of following table.It is understood that The corresponding each factor of each historical traffic casualty data is converted into respectively in corresponding evaluation index, in order to carry out by ground Data analysis.

Parameter declaration in one model training data of table

S508: model training is carried out to model training data using decision Tree algorithms, obtains the road transportation work style based on decision tree Dangerous identification model.

Wherein, decision Tree algorithms are a kind of common sorting algorithms, by more comprising one group of attribute and classification A sample is learnt, and a classifier is got, so that the classifier is by normally classifying to emerging object. I.e. decision Tree algorithms are a kind of methods for approaching discrete function value, and substantially decision tree is to be carried out by series of rules to data The process of classification.C4.5 decision Tree algorithms can be used in the present embodiment.

Specifically, when carrying out model training to model training data using decision Tree algorithms, tree-shaped taxonomic structure Important journey including root node, leaf node and the intermediate node for connecting root node and leaf node, according to each evaluation index Degree, determines the sorted logic of its root node and intermediate node, to form raw risk identification model；By all model training data Classify according to the raw risk identification model, all model training data classifications are stored to the leaf node belonging to it It is in corresponding data set, the quantity of the model training data in the data set and the quotient of the quantity of all model training data is true It is set to the risk probability of the leaf node, to form road risk identification model.

In intelligent navigation method based on risk assessment provided by the present embodiment, by all historical traffic accident numbers History driver information, history information of vehicles, history road section information, history location information and historical time information in carry out Conversion process obtains its corresponding driver evaluation's index respectively, vehicle evaluation index, section evaluation index, meteorological assessment refer to Mark and time evaluation index form model training data to determine a kind of categorical attribute based on each evaluation index.It uses again Decision Tree algorithms are trained to model training data are formed by, to obtain corresponding road risk identification model, so that should Road risk identification model carries out risk identification.

In one embodiment, as shown in fig. 6, step S508, i.e., carry out mould to model training data using decision Tree algorithms Type training obtains the road risk identification model based on decision tree, specifically comprises the following steps:

S601: determine that its corresponding class label feature and at least two training attributes are special based on each model training data Model training data are stored in training data and concentrated by sign.

The training attributive character can be understood as each evaluation index of model training data.Category label characteristics can be with It is interpreted as different risk class.If all model training data can be specifically divided into according to the intersection of its all evaluation index Dry set, determines its risk probability, foundation based on the quotient of model training data in each set and all model training data The risk probability carries out grade classification, such as can be divided into high risk, risk or low-risk risk class.

S602: class label feature and training attributive character to model training data carry out information gain-ratio calculating, obtain Take the corresponding information gain-ratio of each trained attributive character.

Wherein, information gain-ratio calculating is the process for calculating the information gain-ratio of trained attributive character.Information gain Rate is a kind of ratio of profit increase, is a mathematics noun, specially the ratio of information gain and division Information Meter.Wherein, information gain is The degree of information uncertainty reduction, information gain are partial to the more feature of value.Division Information Meter, which is characterized, to be divided When branch quantity.Information gain-ratio is the ratio of information gain and division Information Meter, to be increased information using information fiber thinness Benefit standardization, so that accuracy rate is higher when being classified based on information gain-ratio.In the present embodiment, server is to model training number According to class label feature and training attributive character carry out information gain-ratio calculating, it is corresponding to obtain each trained attributive character Information gain-ratio, so as to using classification foundation of the information gain-ratio as model training, can make its assorting process accuracy rate compared with Height helps to improve the recognition rate of trained model.

S603: target classification feature of the maximum trained attributive character of information gain-ratio as training dataset is chosen, is adopted Training dataset is divided at least two training subsets with target classification feature.

Wherein, target classification is characterized in the feature divided to training dataset.In the present embodiment, information gain is chosen Target classification feature of the maximum trained attributive character of rate as the training dataset.Training subset is based on target classification feature The set that training dataset is divided.For example, calculating driver evaluation's index in a model training data, vehicle is commented Estimate index, section evaluation index, meteorological evaluation index and the corresponding trained attributive character of time evaluation index information gain-ratio, The maximum trained attributive character of information gain-ratio is chosen as a target classification feature, to carry out based on the target classification feature Classification.If the corresponding information gain-ratio of driver evaluation's index is maximum, first as target classification feature to all models Training data is classified, to be divided at least two training subsets.

S604: the training attributive character for not being determined as target classification feature is judged whether there is.

In the present embodiment, it whether there is in the corresponding trained attributive character of judgment models training data and be not determined as target point The training attributive character of category feature is not determined as mesh to determine that can the training subset carry out further division if it exists The training attributive character for marking characteristic of division, then illustrate that the training subset can not carry out further division；If training subset is not present It is not determined as the training attributive character of target classification feature, then illustrates that the training subset can carry out further division.

S605: it is not determined as the training attributive character of target classification feature if it exists, is then updated to train by training subset Data set repeats and carries out information gain-ratio calculating to the class label feature and training attributive character of model training data, The step of obtaining each trained attributive character corresponding information gain-ratio.

In the present embodiment, after training dataset is divided at least two training subsets using target classification feature, if In the presence of the training attributive character for not being determined as target classification feature, illustrate that the training subset can carry out further division, at this point, will The step of training subset is updated to training dataset, repeats step S602-S603, until there is no be not determined as target point The training attributive character of category feature, to complete the training process of the road risk identification model based on decision Tree algorithms.

S606: not being determined as the training attributive character of target classification feature if it does not exist, then is formed and be based on decision Tree algorithms Raw risk identification model, obtain the corresponding data volume of each leaf node in raw risk identification model, be based on data volume Raw risk identification model is cut out, the road risk identification model based on decision tree is obtained.

In the present embodiment, after training dataset is divided at least two training subsets using target classification feature, if There is no the training attributive character for not being determined as target classification feature, then illustrate that at least two training subsets can not further be drawn Point, the raw risk identification model based on decision Tree algorithms can be obtained, at this point, training subset, which corresponds to the raw risk, identifies mould The leaf node of the tree shaped model framework of type, and target classification feature is the upper level intermediate node or root node of the leaf node Characteristic of division.

Specifically, after forming the raw risk identification model based on decision tree, raw risk identification mould can be used Type concentrates all model training data to classify training data, with the corresponding model training data of each leaf node of determination Data volume.Then, the data volume of the corresponding model training data of each leaf node is compared with preset threshold；If should Data volume is greater than preset threshold, then retains the leaf node；If the data volume is not more than preset threshold, it is found upwards in turn The corresponding model training data of corresponding superior node (i.e. according to father's node, the grandparent node ... for obtaining the leaf node) Data volume, until superior node data volume be greater than preset threshold, then by the superior node be determined as be truncated node, based on should Truncation node cuts raw risk identification model, to obtain the road risk identification model based on decision tree.Wherein, in advance If threshold value is pre-set for carrying out cutting processing minimum threshold.For example, the preset threshold is set as 5, if any leaf The data volume of the corresponding model training data of node is greater than 5, then retains the leaf node；If the corresponding model of any leaf node The data volume of training data is not more than 5, then the data volume of the corresponding model training data of its superior node is found upwards, if higher level The data volume of node is greater than 5, it is determined that for truncation node, raw risk identification model cut according to the truncation node, To obtain the road risk identification model simplified, road risk identification model excessively redundancy is avoided.Further, in road risk In identification model, the corresponding risk probability of also settable each truncation node is low probability risk；Its leaf remained Node is determined according to the ratio of class label feature or the data volume of the data volume and training dataset of leaf node.

In intelligent navigation method based on risk assessment provided by the present embodiment, chooses the training that training data is concentrated and belong to Property feature as classification foundation, pass through the information gain-ratio for calculating each trained attributive character；It is maximum to choose information gain-ratio Target classification feature of the training attributive character as the training dataset, to be drawn training dataset using the target classification feature It is divided at least two training subsets；When there is no the training attributive character for not being determined as target classification feature, formed based on certainly The raw risk identification model of plan tree algorithm obtains the corresponding data volume of each leaf node, base in raw risk identification model Raw risk identification model is cut out in data volume, obtain the road risk identification model based on decision tree, with obtain compared with The road risk identification model simplified, avoids model redundancy.The classification of road risk identification model acquired in the present embodiment is quasi- True rate is higher, and classifying rules is clear, when so that the later use road risk identification model being identified, recognition efficiency compared with Fastly.

In one embodiment, route guidance request further includes user account number, which triggered for unique identification The identity of the user of route guidance request.As shown in fig. 7, after step S203, if being safety in navigation type Type of recommendation, then after obtaining the corresponding traffic data to be assessed in each navigation section, the intelligent navigation side based on risk assessment Method further includes following steps:

S701: being based on user account number enquiry navigation system database, obtains historical navigation number corresponding with user account number According to the corresponding history type of recommendation of each historical navigation data.

Wherein, historical navigation data referred to before the current time in system, and the route guidance request of user's triggering is recorded Navigation data.Due to historical navigation data be based on the current time in system before route guidance request be formed by navigation number According to therefore, which also can record its route guidance and request corresponding navigation type, which is determined as The corresponding history type of recommendation of historical navigation data.It is to be appreciated that the history type of recommendation includes but is not limited to that expense is recommended Type, apart from type of recommendation and safe type of recommendation.

S702: in the history type of recommendation of statistical history navigation data, the corresponding history recommended frequency of safe type of recommendation.

Since each historical navigation data only correspond to a kind of history type of recommendation, and history type of recommendation includes but is not limited to Expense type of recommendation, apart from the types such as type of recommendation and safe type of recommendation, can be by counting each history type of recommendation pair The history recommended frequency answered, understands the driving habit of user.In the present embodiment, the history type of recommendation of statistical history navigation data In, it the corresponding history recommended frequency of safe type of recommendation, in particular to calculates in the history type of recommendation of historical navigation data, peace The quotient of the quantity of the quantity and all historical navigation data of the corresponding historical navigation data of full type of recommendation.

S703: it if the corresponding history recommended frequency of safe type of recommendation is less than predeterminated frequency threshold value, obtains stroke and nearly believes Breath, and stroke danger information is sent to client.

Wherein, predeterminated frequency threshold value is pre-set for assessing whether to need to carry out the probability threshold that stroke is nearly recommended Value.Stroke danger information refers to the information of insurance relevant to stroke.Specifically, when the corresponding history of safe type of recommendation recommends frequency When rate is less than predeterminated frequency threshold value, illustrate fewer safe this navigation type of type of recommendation of selection of the driving habit of the user, And safe type of recommendation is selected in the request of this route guidance, and illustrate that this trip is not identical with previous driving habit, it may Colleague has the special circumstances such as child or old man.In order to further ensure user security, guarantor relevant to stroke can be carried out at this moment Danger is recommended, i.e., to client push stroke danger information.It is to be appreciated that the trip danger information is closed with intelligent guidance system Insurance information relevant to stroke provided by the insurance company of work.Specifically, the trip danger information includes and intelligent navigation system The tourism personal accident provided there are the insurance company of cooperative relationship of uniting is nearly and vehicles personal accidental death and injury insurance etc. is related to stroke Insurance information, for example, the trip danger information may include if the trip is nearly the self-driving tourist insurance that provides of safety peace way The information such as this corresponding premium of self-driving tourist insurance and Claims Resolution amount.The trip danger information includes the purchase of corresponding stroke danger Linking inlet ports are nearly bought with facilitating user to carry out stroke.

In intelligent navigation method based on risk assessment provided by the present embodiment, gone through by the way that counting user account number is corresponding In history navigation data, the history recommended frequency of this history type of recommendation of safe type of recommendation, to understand the driving habit of user In the frequency navigated based on safe type of recommendation；And when history recommended frequency is less than predeterminated frequency threshold value, illustrate current Driving habit different from the past is driven, stroke can be carried out and nearly recommended, so that nearly recommendation is more targeted for stroke.

In one embodiment, each recommendation navigation routine includes at least one Frequent Accidents place, and Frequent Accidents place is Refer to the place that traffic accident frequently occurs.As shown in figure 8, navigation routine will be recommended to be sent to client after step S206 After end, the intelligent navigation method based on risk assessment further includes following steps:

S801: in vehicle according in recommendation navigation routine driving process, the Current traffic number of client transmission is obtained in real time According to current steering position.

Wherein, Current traffic data refer to according to recommendation navigation routine driving process, collected in real time for commenting Estimate its data needed for risk of driving a vehicle.The current traffic data include but is not limited to current driver's information, present vehicle information, Current road segment information, current location information and current time information.Wherein, current driver's information is used to indicate that client is closed The information of the driver identity of connection can specifically include name, ID card No. and the drivers license number of driver.Current vehicle Information is used to indicate that the information of vehicle associated by client, the including but not limited to model of vehicle and mileage travelled number. Current road segment information is used to indicate that the information in the corresponding section in navigation section being presently in.Current location information is for table The information in place locating for the bright navigation section being presently in.Current time information is used to indicate that the corresponding letter of current time Breath.Current steering position is the position that the vehicle of the GPS module acquisition carried by client is currently located.Specifically, in vehicle According to recommend navigation routine driving process in, server obtain in real time client transmission Current traffic data, to carry out Road risk assessment；And current steering position is actually obtained, to carry out route tracking and risk assessment.

S802: carrying out risk identification to Current traffic data using road risk identification model, obtain current risk probability, If current risk probability is greater than default risk threshold value, the first risk data is obtained.

Default risk threshold value is pre-set for assessing whether the threshold value for needing to carry out risk prompting.Specifically, it takes When business device carries out risk identification to Current traffic data using road risk identification model, which can be input to Road risk identification model is identified, determines current traffic data leaf node locating in tree-shaped taxonomic structure, The corresponding risk probability of the leaf node is determined as current risk probability.In the present embodiment, if Current traffic data are corresponding When current risk probability is greater than default risk threshold value, illustrate that user drives vehicle foundation and recommends in navigation routine driving process, wind Danger is larger, at this point, the first risk data need to be generated, to carry out risk prompting based on first risk data.It is to be appreciated that First risk data can be the risk class according to its current risk determine the probability, such as high risk, risk and low-risk.

S803: according to the direction of traffic for recommending navigation routine, if existing within the scope of the pre-determined distance of current steering position Frequent Accidents place or burst accident information, then obtain the second risk data.

Pre-determined distance range refers to pre-set distance range.Burst accident information refers to that user drives vehicle foundation and pushes away It recommends in navigation routine driving process, the accident information that intelligent guidance system is known.Specifically, vehicle foundation is driven in user to push away Recommend navigation routine when driving, according to recommend navigation routine direction of traffic, vehicle current steering position preset range (such as In 1km), there are Frequent Accidents place and burst accident information, (there are accidents such as in the front 1km of the direction of traffic of vehicle frequently Send out place and burst accident information), the second risk data need to be generated, to carry out risk prompting based on second risk data.

S804: being filled into default risk for the first risk data and the second risk data and remind in template, obtains risk and mentions Awake text data.

Wherein, presetting risk and reminding template is pre-set for carrying out the template of risk prompting.For example, the default wind Template is reminded to be " at road ahead ... rice, there are Frequent Accidents place, careful drivings ", " at road ahead ... rice, send out in danger Raw prominent accident ... ... lane blocks, careful driving " or " it is high risk section that vehicle, which currently drives locating section, carefully Drive " etc..First risk data and the second risk data are filled into the default risk and reminded in template by server, to form wind Remind text data in danger.

S805: risk prompting text data is converted by risk using TTS technology and reminds voice data, and passes through client It broadcasts risk and reminds voice data.

Wherein, TTS (abbreviation of Text To Speech, i.e., " from Text To Speech "), is interactive a part, allows Machine can speak.Specifically, server is proposed the risk using TTS technology after getting risk and reminding text data Awake text data is converted into risk and reminds voice data, and reminds voice data by the client terminal playing risk, so that user It can hear that risk reminds voice data, to reach risk prompting, remind the purpose of user's careful driving.

In intelligent navigation method based on risk assessment provided by the present embodiment, using road risk identification model to working as Preceding traffic data identified, can the first risk data of quick obtaining, the acquisition of first risk data has objectivity.Again according to According to the running mode for recommending navigation routine, there are Frequent Accidents place or prominent things in the preset range of current steering position Therefore when information, the second risk data is generated, to carry out corresponding risk prompting.By the first risk data and the second risk data It is filled into default risk to remind template and converted using TTS technology, risk is formed by with casting and reminds voice data, with User is set to hear that risk reminds voice data in real time in driving procedure, to achieve the purpose that remind user's careful driving.

It should be understood that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment, each process Execution sequence should be determined by its function and internal logic, the implementation process without coping with the embodiment of the present invention constitutes any limit It is fixed.

In one embodiment, a kind of intelligent navigation device based on risk assessment is provided, it should the intelligence based on risk assessment Intelligent navigation method in navigation device and above-described embodiment based on risk assessment corresponds.As shown in figure 9, risk should be based on The intelligent navigation device of assessment includes route guidance request module 901, original navigation route acquiring module 902, to be assessed Traffic data obtains module 903, section risk probability obtains module 904, overall risk probability obtains module 905 and recommends navigation road Wire module 906.Detailed description are as follows for each functional module:

Route guidance request module 901, for obtaining the route guidance request of client transmission, route guidance request Including origin, target location and navigation type.

Original navigation route acquiring module 902 obtains at least one and original leads for being based on origin and target location Air route line, each original navigation route include at least one navigation section.

Traffic data to be assessed obtains module 903, if being safe type of recommendation for navigation type, obtains each navigation The corresponding traffic data to be assessed in section.

Section risk probability obtains module 904, adopts in advance for the corresponding traffic data input to be assessed in the section that will navigate In road risk identification model with decision Tree algorithms training, the sorted logic according to decision Tree algorithms is corresponding to navigation section Traffic data to be assessed carries out risk assessment, obtains the corresponding section risk probability in navigation section.

Overall risk probability obtains module 905, for based at least one of each original navigation route navigation section pair The section risk probability answered obtains the corresponding overall risk probability of original navigation route.

Recommend navigation routine module 906, for being to recommend navigation by the smallest original navigation route determination of overall risk probability Route will recommend navigation routine to be sent to client.

Preferably, route guidance request module 901 includes Voice Navigation request unit, voice recognition processing list Member, keyword extracting unit and Keywords matching unit.

Voice Navigation request unit, for obtaining the Voice Navigation request of client transmission, Voice Navigation request packet Include voice data to be identified.

Voice recognition processing unit is obtained for carrying out speech recognition to voice data to be identified using speech recognition modeling Take text data to be identified.

Keyword extracting unit obtains mesh for extracting using keyword extraction algorithm to text data to be identified Mark keyword.

Keywords matching unit obtains the starting of successful match for matching predetermined keyword library based on target keyword Place, target location and navigation type.

Preferably, before Voice Navigation request unit, the intelligent navigation device based on risk assessment further includes original Beginning voice data acquiring unit, target acoustical feature acquiring unit, target text data capture unit and text data acquisition obtain Take unit.

Primary voice data acquiring unit, the primary voice data acquired in real time for obtaining client.

Target acoustical feature acquiring unit obtains mesh for carrying out end-point detection and feature extraction to primary voice data Mark acoustic feature.

Target text data capture unit is obtained for being identified using speech recognition modeling to target acoustical feature Target text data.

Text data acquires acquiring unit, if including to wake up keyword for target text data, controls client and exists Enter Voice Navigation interface in preset time period, receives Voice Navigation request.

Preferably, before route guidance request module 901, the intelligent navigation device based on risk assessment further includes Historical traffic casualty data acquiring unit, historical driving behavior data capture unit, vehicle evaluation index acquiring unit, section are commented Estimate index selection unit, meteorological evaluation index acquiring unit, time evaluation index acquiring unit, model training data capture unit With road risk identification model acquiring unit.

Historical traffic casualty data acquiring unit, for obtaining history from traffic police's platform database by third party's interface Traffic accident data, historical traffic casualty data include history driver information, history information of vehicles, history road section information, go through History location information and historical time information.

Historical driving behavior data capture unit is obtained for inquiring traffic police's platform database based on history driver information Historical driving behavior data corresponding with history driver information are taken, historical driving behavior data query behavioural information number is based on According to table, driver evaluation's index is obtained.

Vehicle evaluation index acquiring unit is obtained and is gone through for being based on history information of vehicles enquiring vehicle information data table The corresponding vehicle evaluation index of history information of vehicles.

Section evaluation index acquiring unit is obtained and is gone through for inquiring road section information tables of data based on history road section information The corresponding section evaluation index of history road section information.

Meteorological evaluation index acquiring unit, for passing through third party's interface from meteorological platform database, acquisition and history Location information and the corresponding history weather information of historical time information inquire weather information data based on history weather information Table obtains meteorological evaluation index corresponding with history meteorological data.

Time evaluation index acquiring unit obtains and believes with historical time for inquiring year ephemeris based on historical time information The corresponding time evaluation index of manner of breathing.

Model training data acquisition acquiring unit is based on the corresponding driver evaluation's index of historical traffic casualty data, vehicle Evaluation index, section evaluation index, meteorological evaluation index and time evaluation index, obtain opposite with historical traffic casualty data The model training data answered.

Road risk identification model acquiring unit carries out model training to model training data using decision Tree algorithms, obtains By way of transportation work style danger identification model.

Preferably, road risk identification model acquiring unit includes determining that subelement, ratio of profit increase obtain subelement, instruction Practice subset division subelement, attributive character judgment sub-unit, compute repeatedly subelement and identification model acquisition subelement.

Subelement is determined, for determining its corresponding class label feature and at least based on each model training data Model training data are stored in training data and concentrated by two trained attributive character.

Ratio of profit increase obtains subelement, for the class label feature to model training data and attributive character is trained to carry out letter It ceases ratio of profit increase to calculate, obtains the corresponding information gain-ratio of each trained attributive character.

Training subset divides subelement, for choosing the maximum trained attributive character of information gain-ratio as training dataset Target classification feature, training dataset is divided by least two training subsets using target classification feature.

Attributive character judgment sub-unit, for judging whether there is the training attribute spy for not being determined as target classification feature Sign.

Subelement is computed repeatedly, for not being determined as the training attributive character of target classification feature if it exists, then will be trained Subset is updated to training dataset, repeats and carries out letter to the class label feature and training attributive character of model training data The step of ceasing ratio of profit increase to calculate, obtaining each trained attributive character corresponding information gain-ratio.

Identification model obtains subelement, for not being determined as the training attributive character of target classification feature if it does not exist, then The raw risk identification model based on decision Tree algorithms is formed, it is corresponding to obtain each leaf node in raw risk identification model Data volume is cut out raw risk identification model based on data volume, obtains the road risk identification model based on decision tree.

Preferably, route guidance request further includes user account number；After traffic data to be assessed obtains module 903, base It further include that historical navigation data cell, history recommended frequency unit and stroke danger information obtain in the intelligent navigation device of risk assessment Take unit.

Historical navigation data cell obtains and user account number phase for being based on user account number enquiry navigation system database Corresponding historical navigation data, the corresponding history type of recommendation of each historical navigation data.

History recommended frequency unit, in the history type of recommendation of statistical history navigation data, safe type of recommendation pair The history recommended frequency answered.

Stroke danger information acquisition unit, if being less than predeterminated frequency threshold for the corresponding history recommended frequency of safe type of recommendation Value, then obtain stroke danger information, and stroke danger information is sent to client.

Preferably, recommending navigation routine module 906 includes at least one Frequent Accidents place.Recommending navigation routine module After 906, the intelligent navigation device based on risk assessment further include: current data acquiring unit, the first risk data obtain single Member, the second risk data acquiring unit remind text data acquiring unit and remind voice datacast unit.

Current data acquiring unit, for, according in recommendation navigation routine driving process, obtaining client in real time in vehicle The Current traffic data of transmission and current steering position.

First risk data acquiring unit, for carrying out risk knowledge to Current traffic data using road risk identification model Not, current risk probability is obtained, if current risk probability is greater than default risk threshold value, obtains the first risk data.

Second risk data acquiring unit, for the direction of traffic according to recommendation navigation routine, if in current steering position Pre-determined distance within the scope of there are Frequent Accidents place or burst accident information, then obtain the second risk data.

Text data acquiring unit is reminded, is mentioned for the first risk data and the second risk data to be filled into default risk In template of waking up, obtains risk and remind text data.

Voice datacast unit is reminded, is reminded for risk prompting text data to be converted into risk using TTS technology Voice data, and risk is broadcasted by client and reminds voice data.

Specific restriction about the intelligent navigation device based on risk assessment may refer to comment above for based on risk The restriction for the intelligent navigation method estimated, details are not described herein.Each mould in the above-mentioned intelligent navigation device based on risk assessment Block can be realized fully or partially through software, hardware and combinations thereof.Above-mentioned each module can be embedded in the form of hardware or independence In processor in computer equipment, it can also be stored in a software form in the memory in computer equipment, in order to locate It manages device and calls the corresponding operation of the above modules of execution.

In one embodiment, a kind of computer equipment is provided, which can be server, internal junction Composition can be as shown in Figure 10.The computer equipment include by system bus connect processor, memory, network interface and Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating The database of machine equipment, which is used to store, executes what the above-mentioned intelligent navigation method based on risk assessment was used or generated in the process Data, such as traffic data to be assessed.The network interface of the computer equipment is used to communicate with external terminal by network connection. To realize a kind of intelligent navigation method based on risk assessment when the computer program is executed by processor.

In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory And the computer program that can be run on a processor, processor are realized in above-described embodiment when executing computer program based on risk The step of intelligent navigation method of assessment, such as step S201-S206 or Fig. 3 shown in Fig. 2 is to step shown in fig. 8, To avoid repeating, which is not described herein again.Alternatively, processor realizes the intelligent navigation based on risk assessment when executing computer program The function of each module/unit in this embodiment of device, for example, it is route guidance request module 901 shown in Fig. 9, original Navigation routine obtains module 902, traffic data to be assessed obtains module 903, section risk probability obtains module 904, overall risk Probability obtains module 905 and recommends the function of navigation routine module 906, and to avoid repeating, which is not described herein again.

In one embodiment, a computer readable storage medium is provided, meter is stored on the computer readable storage medium Calculation machine program, the computer program realize the intelligent navigation method in above-described embodiment based on risk assessment when being executed by processor The step of, such as step S201-S206 or Fig. 3 shown in Fig. 2 is to step shown in fig. 8, to avoid repeating, here not It repeats again.Alternatively, the computer program realized when being executed by processor the above-mentioned intelligent navigation device based on risk assessment this The function of each module/unit in embodiment, such as route guidance request module 901 shown in Fig. 9, original navigation route Obtain module 902, traffic data to be assessed obtains module 903, section risk probability obtains module 904, overall risk probability obtains Module 905 and the function of recommending navigation routine module 906, to avoid repeating, which is not described herein again.

Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with Instruct relevant hardware to complete by computer program, computer program to can be stored in a non-volatile computer readable It takes in storage medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, this Shen Please provided by any reference used in each embodiment to memory, storage, database or other media, may each comprise Non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM (PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms, Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..

It is apparent to those skilled in the art that for convenience of description and succinctly, only with above-mentioned each function Can unit, module division progress for example, in practical application, can according to need and by above-mentioned function distribution by different Functional unit, module are completed, i.e., the internal structure of device are divided into different functional unit or module, to complete above description All or part of function.

The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although with reference to the foregoing embodiments Invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each implementation Technical solution documented by example is modified or equivalent replacement of some of the technical features；And these modification or Replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution should all include Within protection scope of the present invention.

Claims (10)

1. a kind of intelligent navigation method based on risk assessment characterized by comprising

The route guidance request that client is sent is obtained, the route guidance request includes origin, target location and navigation Type；

Based on the origin and the target location, at least one original navigation route, each original navigation are obtained Route includes at least one navigation section；

If the navigation type is safe type of recommendation, the corresponding traffic data to be assessed in each navigation section is obtained；

The corresponding traffic data input to be assessed in the navigation section is known using the road risk of decision Tree algorithms training in advance In other model, the sorted logic traffic data to be assessed corresponding to the navigation section according to decision Tree algorithms carries out risk and comments Estimate, obtains the corresponding section risk probability in the navigation section；

Based on the corresponding section risk probability in navigation section described at least one of each described original navigation route, institute is obtained State the corresponding overall risk probability of original navigation route；

It is to recommend navigation routine by the smallest original navigation route determination of overall risk probability, the recommendation navigation routine is sent to The client.

2. as described in claim 1 based on the intelligent navigation method of risk assessment, which is characterized in that the acquisition client hair The route guidance request sent, the route guidance request includes origin, target location and navigation type, comprising:

The Voice Navigation request that client is sent is obtained, the Voice Navigation request includes voice data to be identified；

Speech recognition is carried out to the voice data to be identified using speech recognition modeling, obtains text data to be identified；

The text data to be identified is extracted using keyword extraction algorithm, obtains target keyword；

Predetermined keyword library is matched based on the target keyword, obtains origin, target location and the navigation of successful match Type.

3. as described in claim 1 based on the intelligent navigation method of risk assessment, which is characterized in that in the acquisition client The Voice Navigation of transmission is requested, described based on risk assessment before the Voice Navigation request includes voice data to be identified Intelligent navigation method further include:

Obtain the primary voice data that client acquires in real time；

End-point detection and feature extraction are carried out to the primary voice data, obtain target acoustical feature；

The target acoustical feature is identified using the speech recognition modeling, obtains target text data；

If the target text data include to wake up keyword, controls the client and led within a preset period of time into voice Boat interface receives the Voice Navigation request.

4. as described in claim 1 based on the intelligent navigation method of risk assessment, which is characterized in that in the acquisition client Before the route guidance request of transmission, the intelligent navigation method based on risk assessment further include:

Historical traffic casualty data, the historical traffic casualty data are obtained from traffic police's platform database by third party's interface Including history driver information, history information of vehicles, history road section information, history location information and historical time information；

Traffic police's platform database is inquired based on the history driver information, is obtained opposite with the history driver information The historical driving behavior data answered are based on the historical driving behavior data query behavioural information tables of data, obtain driver and comment Estimate index；

Based on the history information of vehicles enquiring vehicle information data table, vehicle corresponding with the history information of vehicles is obtained Evaluation index；

Road section information tables of data is inquired based on the history road section information, obtains section corresponding with the history road section information Evaluation index；

Through third party's interface from meteorological platform database, obtain and the history location information and the historical time information Corresponding history weather information is inquired weather information data table based on history weather information, is obtained and history meteorological data phase Corresponding meteorology evaluation index；

Year ephemeris is inquired based on the historical time information, time assessment corresponding with the historical time information is obtained and refers to Mark；

Based on the corresponding driver evaluation's index of the historical traffic casualty data, the vehicle evaluation index, the road Section evaluation index, the meteorological evaluation index and the time evaluation index, obtain opposite with the historical traffic casualty data The model training data answered；

Model training is carried out to the model training data using decision Tree algorithms, obtains the road risk identification based on decision tree Model.

5. as claimed in claim 4 based on the intelligent navigation method of risk assessment, which is characterized in that use decision Tree algorithms pair The model training data carry out model training, obtain the road risk identification model based on decision tree, comprising:

Its corresponding class label feature and at least two training attributive character are determined based on each model training data, it will The model training data are stored in training data concentration；

Class label feature and training attributive character to the model training data carry out information gain-ratio calculating, obtain each The corresponding information gain-ratio of the trained attributive character；

Target classification feature of the maximum trained attributive character of the information gain-ratio as the training dataset is chosen, is used The training dataset is divided at least two training subsets by the target classification feature；

Judge whether there is the training attributive character for not being determined as target classification feature；

It is not determined as the training attributive character of target classification feature if it exists, then the training subset is updated to the trained number According to collection, repeats and information gain-ratio meter is carried out to the class label feature and training attributive character of the model training data The step of calculating, obtaining each trained attributive character corresponding information gain-ratio；

It is not determined as the training attributive character of target classification feature if it does not exist, then forms the raw risk based on decision Tree algorithms Identification model obtains the corresponding data volume of each leaf node in the raw risk identification model, based on the data amount pair The raw risk identification model is cut out, and obtains the road risk identification model based on decision tree.

6. as described in claim 1 based on the intelligent navigation method of risk assessment, which is characterized in that the route guidance request It further include user account number；

If being safe type of recommendation in the navigation type, the corresponding traffic to be assessed in each navigation section is obtained After data, the intelligent navigation method based on risk assessment further include:

Based on the user account number enquiry navigation system database, historical navigation number corresponding with the user account number is obtained According to each corresponding history type of recommendation of the historical navigation data；

In the history type of recommendation for counting the historical navigation data, the corresponding history recommended frequency of safe type of recommendation；

If the corresponding history recommended frequency of the safe type of recommendation is less than predeterminated frequency threshold value, stroke danger information is obtained, and Stroke danger information is sent to the client.

7. as described in claim 1 based on the intelligent navigation method of risk assessment, which is characterized in that each recommendation navigation Route includes at least one Frequent Accidents place；

It is described the recommendation navigation routine is sent to the client after, the intelligent navigation method based on risk assessment Further include:

In vehicle according in the recommendation navigation routine driving process, the Current traffic data and work as that client is sent are obtained in real time Preceding steering position；

Risk identification is carried out to the Current traffic data using the road risk identification model, obtains current risk probability, If the current risk probability is greater than default risk threshold value, the first risk data is obtained；

According to the direction of traffic for recommending navigation routine, if there are institutes within the scope of the pre-determined distance of the current steering position Frequent Accidents place or burst accident information are stated, then obtains the second risk data；

First risk data and second risk data are filled into default risk to remind in template, risk is obtained and reminds Text data；

Risk prompting text data is converted by risk using TTS technology and reminds voice data, and passes through the client It broadcasts the risk and reminds voice data.

8. a kind of intelligent navigation device based on risk assessment characterized by comprising

Route guidance request module, for obtaining the route guidance request of client transmission, route guidance request includes Beginning place, target location and navigation type；

Original navigation route acquiring module, for obtaining at least one original navigation route based on origin and target location, Each original navigation route includes at least one navigation section；

Traffic data to be assessed obtains module, if being safe type of recommendation for navigation type, obtains each navigation section pair The traffic data to be assessed answered；

Section risk probability obtains module, determines for using the corresponding traffic data input to be assessed in the navigation section in advance In the road risk identification model of plan tree algorithm training, the sorted logic according to decision Tree algorithms is corresponding to the navigation section Traffic data to be assessed carries out risk assessment, obtains the corresponding section risk probability in the navigation section；

Overall risk probability obtains module, for based on the corresponding section at least one of each original navigation route navigation section Risk probability obtains the corresponding overall risk probability of original navigation route；

Recommend navigation routine module, it, will for being to recommend navigation routine by the smallest original navigation route determination of overall risk probability Navigation routine is recommended to be sent to client.

9. a kind of computer equipment, including memory, processor and storage are in the memory and can be in the processor The computer program of upper operation, which is characterized in that the processor realized when executing the computer program as claim 1 to The step of intelligent navigation method described in 7 any one based on risk assessment.

10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists In realization is as described in any one of claim 1 to 7 based on the intelligence of risk assessment when the computer program is executed by processor The step of air navigation aid.