CN109523390A - A method of air control model is established according to running car rule - Google Patents
A method of air control model is established according to running car rule Download PDFInfo
- Publication number
- CN109523390A CN109523390A CN201811140159.8A CN201811140159A CN109523390A CN 109523390 A CN109523390 A CN 109523390A CN 201811140159 A CN201811140159 A CN 201811140159A CN 109523390 A CN109523390 A CN 109523390A
- Authority
- CN
- China
- Prior art keywords
- data
- risk
- equipment
- vehicle
- speed
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of method for establishing air control model according to running car rule, the method for establishing air control model includes data acquisition, data cleansing, data analysis, factor identification, model construction, eight risk identification, Risk-warning and risk identification steps.The present invention is for equipment requirement and relies on not high: any equipment that can collect position data is ok, it not only include GPS, the pinpoint equipment of the energy such as Beidou, it further include base station, the equipment that WIFI etc. is non-precisely positioned, and it is wider to be applicable in scene, pattern drawing can be established as long as it can extract position data, and this model can be easy to be generalized to other scenes, it can not only take precautions against in advance, model can also be constructed while running data, the dimension of evaluation is according to data attribute and quality, equipment form and the difference of operating mode can be adaptive, finally provide the consistent evaluation method of standard.
Description
Technical field
The present invention relates to auto metal halide lamp technical field more particularly to a kind of air control model is established according to running car rule
Method.
Background technique
In auto metal halide lamp field, the main means of air control at present are by reference, the i.e. body of debtor by inquiry before borrowing
Part information, assesses the repaying ability of debtor according to previous transaction record and financial status, once completion of making loans, maturity
Between air control means be also it is few, most of is all that collection is just carried out after overdue.Insurance is also the same, is also all according to previous
Claims Resolution situation determines the premium of next year, and in recent years, although UBI also slowly rises at home, most of is also to need
Traveling a period of time could be assessed after being mounted with task equipment on vehicle, and belong to " subsequent ".
With the rise of car networking, various smart machines acquire data abundant from vehicle, us is made to start slowly to close
The portrait for vehicle is infused, the Novel presentation of " people " is transmitted to the variation of the behavior of " vehicle " before promise breaking, it is possible to identify ahead of time
It risk and is taken precautions against out.
Existing evaluation model is for equipment requirement and relies on height, and cost is relatively high, and the applicable scene ratio of evaluation model
It is relatively narrow, for this purpose, the present invention proposes a kind of method for establishing air control model according to running car rule.
Summary of the invention
The purpose of the present invention is to solve the problems of the prior art, and a kind of of proposition builds according to running car rule
The method of vertical air control model.
To achieve the goals above, present invention employs following technical solutions:
A method of air control model being established according to running car rule, the method for establishing air control model includes data
Acquisition, data cleansing, data analysis, factor identification, model construction, eight risk identification, Risk-warning and risk identification steps;
The data acquisition: data as much as possible are extracted by the equipment on vehicle, the attribute of data must include warp
Three degree, latitude and time most basic parameters, are also possible to that other data such as base station information of these three parameters can be calculated
Deng;
The data cleansing: removal interference and unworthy data, and these data are reasonably stored;
The data analysis: being processed data, analyzes the logical relation between data, and processing generates new dimension
The data of degree;
The factor identification: the possible evaluative dimension of vehicle is determined according to data with existing;
The model construction: evaluation model is constructed according to the factor of a model identified;
The risk identification: identification scene and risk are gone according to evaluation model;
The Risk-warning: early warning is carried out for the risk identified;
The risk identification: it is marked for the risk having been acknowledged.
Preferably, generally there are two types of approach, the equipment for not networking can pass through the data source of the data acquisition
Card reader directly extracts data from the storage equipment of equipment;And for the equipment of networking, can storage file beyond the clouds such as
It is directly extracted in database, the data of extraction include: location dependent data (time, longitude, latitude, speed, direction, base station ginseng
Number, WIFI parameter) and vehicle status data (starting/flame-out, driving behavior etc.).
Preferably, the data for needing to wash in the data cleansing specifically include that
The location point of drift: a) speed is more than 180km/h;B) distance between two points are greater than 25 kilometers;C) vehicle is in and puts out
Fiery state but position changes;
The illegal time: not at the extraction between section location point;
The base station that can not be positioned or WIFI parameter.
Preferably, it needs to handle in the data analysis and resolves following data: including longitude and latitude, speed, orientation
Angle, starting/flameout state calculate, family and business address analyze, are used to walk circuit analysis, the analysis of consumption place, equipment people
For anomaly analysis;
(1), it the longitude, latitude: for base station and WIFI, needs to calculate longitude and latitude number by relevant parametric solution
According to typically being realized by third party's interface;
(2), the speed: for the scene for not extracting speed, needing to calculate speed by longitude and latitude and time,
Spherical distance is first gone out according to calculation of longitude & latitude, then speed is calculated by distance;
Spherical distance (D) is calculated by Haversine formula:
Wherein
Haversin (θ)=sin2 (θ/2)=(1-cos (θ))/2
θ=d/R
R is earth radius, can be averaged 6371km;
The latitude of the expression two o'clock of φ 1, φ 2;
The difference of Δ λ expression two o'clock longitude.
So: V=D/ Δ T, Δ T are the time difference between two anchor points;
(3), the azimuth: i.e. the direction of vehicle advance, due north 0 rotate clockwise, and calculation method is as follows:
A, B, C: the angle folded by " camber line " is at this point on three points and spherical surface on spherical surface is indicated
A, b, c: indicating A, 3 points of B, C to angle folded by " arc " two-end-point and the earth's core line.
(5), the starting/flameout state calculates: the case where for not collecting starting/flameout state, it is also necessary to count
Its state is calculated, calculated state does not need very accurate, it is desirable that there is the inconsistent of front and back 10 minutes or so with actual conditions,
Namely vehicle occurs in the movement that starts or stop working at algorithm can monitor within 10 minutes, due to some special feelings
Condition, data provided and less than calculating the state, so the calculating of the state be it is non-necessary,
It should be noted that since there may be missings for acquisition Data Styles, so algorithm needs to do time-out in judgement
Processing was then determined as time-out without continuous data more than 25 minutes, and pressure vehicle is flameout state;
Driving behavior: driving behavior refers mainly to three and suddenly one surpasses, i.e., anxious acceleration, anxious deceleration, sudden turn of events road and hypervelocity;
Anxious to accelerate: judgement of the acceleration greater than 13km/hs is anxious accelerate;
Anxious to slow down: judgement of the acceleration less than -15km/hs is anxious slow down;
Sudden turn of events road: a) speed is greater than 120km/h, and direction change is greater than 3 °/s, is determined as sudden turn of events road;
B) speed is greater than 5 °/s, is determined as sudden turn of events road in 80-120km/h, direction change;
C) speed is greater than 10 °/s, is determined as sudden turn of events road in 60-80km/h, direction change;
D) speed is greater than 25 °/s, is determined as sudden turn of events road in 45-60km/h, direction change;
Hypervelocity: according to place category of roads difference, different speed limit values is taken;
A) high speed: greater than 120km/h is judged to exceeding the speed limit;
B) Class I highway: greater than 80km/h is judged to exceeding the speed limit;
C) Class II highway: greater than 60km/h is judged to exceeding the speed limit;
D) Class III highway: greater than 40km/h is judged to exceeding the speed limit;
E) Class IV highway: greater than 30km/h is judged to exceeding the speed limit;
Category of roads for obtaining vehicle position is had any problem, and a kind of alternative is be all set to 80km/h
It is judged to exceeding the speed limit, by engineering verification, the influence for model is in controlled range;
(5), it the family and business address analysis: according to longitude and latitude and starting/flameout state and time, uses
DBSCAN algorithm calculates family and work unit's specific address, the specific steps are as follows:
Find out longitude and latitude and the time of all vehicle stalls;
Clustering is done using DBSCAN;
Two clusters for taking location point most are analyzed again;
It is write using third party's interface by the acquisition of work unit's cluster center longitude according to the central point longitude and latitude of cluster
Building, commercialization are determined as business address with the POI of room type point, obtain residential quarters by home address cluster center longitude
POI point be determined as home address;
(6), described to be used to walk circuit analysis: frequent vehicle line is analyzed, the specific method is as follows:
The longitude and latitude that tracing point (speed is greater than 5km/h) when finding out all vehicle drivings is put;
These longitudes and latitudes do clustering using DBSCAN algorithm;
Access is according to most preceding 5 clusters;
It calls third-party map interface to obtain road information where these clusters, forms route;
Another method is first to obtain road, then do clustering to road, can also obtain effect same.
(7), the consumption place analysis: different from family and work unit's analysis, family and business address are first
It does clustering and goes out central point, further according to the type for obtaining POI after time judgement.The analysis of consumption place is also first to do cluster point
Central point is precipitated, but is the type for removing to match position POI again according to the position of central point, i.e., given (X, Y) needs to pass through
It is school, KTV, hotel, shopping center etc. that third-party map interface analysis, which is the position,;
(8), the artificial anomaly analysis of the equipment: before premeditated promise breaking, car owner generally in advance can artificially be broken equipment
It is bad, to escape or cover its subsequent behavior or whereabouts, can generally it analyze following situations:
Equipment Removal: equipment has dismounting alarm to report or equipment does not have any data to report within 15 days;
Equipment separation: will be installed more positioning devices on Some vehicles, when equipment has in separation alarm or same
Distance in all devices 5 minutes is greater than 10KM;
Device mask: vehicle in the process of moving (speed in 20km/h or more), interrupt 5 minutes or more suddenly by data, it
Afterwards again self-recovery the case where.It should be noted that device mask influenced by the external world it is more, such as cross tunnel, actually using
Process needs to carry out external environment locating for auxiliary judgment vehicle according to the map, excludes externalities.
Preferably, the factor identification is according to the above analysis data, those needs of the data of the above dimension of comprehensive judgement
Participating in model construction and weight, weight, that is, data confidence level can be assessed, each according to the quality of data and concentration degree
The calculation method of dimension is different.The specific method is as follows:
A), park risk for a long time: according to the state for starting/stopping working, the vehicle for judging that 15 days or more do not start is risky,
Weight=park number of days -15;
B), home address risk: according to home address, vehicle is parked secondary within the scope of home address 1.5KM in 7 days
It is judged as home address risk, weight=risk frequency less than 2 times;
C), business address risk: according to unit address, vehicle stops within the scope of the 1.5KM of business address in 7 days
Secondary be less than 3 times put is judged as business address risk, weight=risk frequency;
D), route deviates risk: the number occurred on used cabling road in 10 days is judged as that route deviates wind less than 3 times
Danger, weight=route deviate lasting number of days;
Damage of facilities risk: artificially determined extremely according to equipment, the frequency within weight=1 month;
E), consumer behavior abnormal risk: consumption place occur big variation (arrive at number not as good as original 30%), weight=
All changes continuous days and summation.
6. a kind of method for establishing air control model according to running car rule according to claim 5, feature exist
In the model construction is the scorecard model that entire model uses logic-based to return, and based on above scoring factor, is used
Following methods are modeled:
(1), risk quantification: the determination data (E) of each risks and assumptions is not determined as and if so, being judged as 1
0, weight R, so the Risk Calculation value xi=E*R of each factor;
(2), normalize: since the numerical difference of x is larger away from possibility, partial factors may significantly affect other factors, institute
To need that x value is normalized.Here using 0 mean value standardization (Z-score standardization), letter is converted
Number are as follows:
Wherein μ is wherein mean value, and δ is standard deviation;
(3), logistic regression: being expressed as p for the probability of customer default, then normal probability is 1-p, therefore, available:
The Probability p of customer default may be expressed as:
The score value scale of scorecard setting can be defined by the way that score value to be expressed as to the linear representation of ratio logarithm, i.e.,
It is represented by following formula:
Score=A-Blog (Odds)
Logic Regression Models calculating ratio is as follows:
Log (Odds)=β0+β1x1+…+βnxn
Wherein, with the available model parameter β 0 of modeling parameters model of fit, β 1 ..., β n;
The value of constant A, B in formula can be by will be known to two or the score value assumed is brought into and is calculated.
Preferably, the risk identification and Risk-warning are the model to above building, are carried out by scoring to risk
Classification.
Preferably, the risk identification is the risk for identifying, needs to carry out the forms such as carry out background check, visit to the parents of schoolchildren or young workers
Confirmed, it is thus identified that after, it needs to be marked in systems.For the risk having been acknowledged, the weight of correlation factor is needed
It adjusts, according to engineering practice, generally increases by 30%, i.e., multiplied by 1.3 on the basis of original weight, entirely commented to influence
Valence model.
Compared with prior art, the present invention provides a kind of method for establishing air control model according to running car rule, tools
It is standby following the utility model has the advantages that
For equipment requirement and rely on it is not high: any equipment that can collect position data is ok, not only include GPS,
The pinpoint equipment of the energy such as Beidou, further includes the equipment that base station, WIFI etc. are non-precisely positioned, and applicable scene is wider,
Pattern drawing can be established as long as it can extract position data, and this model can be easy to be generalized to other scenes, not only
It can take precautions against in advance, model can also be constructed while running data, the dimension of evaluation is according to data attribute and quality, equipment shape
State and the difference of operating mode can be adaptive, finally provide the consistent evaluation method of standard.
It is not directed to part in the device to be the same as those in the prior art or can be realized by using the prior art, structure of the invention
Simply, easy to operate.
Detailed description of the invention
Fig. 1 is a kind of structural schematic diagram for the method that air control model is established according to running car rule proposed by the present invention.
Fig. 2 is that a kind of azimuthal distribution for the method that air control model is established according to running car rule proposed by the present invention is shown
It is intended to.
Fig. 3 is a kind of vehicle stall cluster of method that air control model is established according to running car rule proposed by the present invention
Analysis chart.
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 only a part of the embodiments of the present invention, instead of all the embodiments.
In the description of the present invention, it is to be understood that, term " on ", "lower", "front", "rear", "left", "right", "top",
The orientation or positional relationship of the instructions such as "bottom", "inner", "outside" is to be based on the orientation or positional relationship shown in the drawings, merely to just
In description the present invention and simplify description, rather than the device or element of indication or suggestion meaning must have a particular orientation, with
Specific orientation construction and operation, therefore be not considered as limiting the invention.
A method of air control model being established according to running car rule, the method for establishing air control model includes data
Acquisition, data cleansing, data analysis, factor identification, model construction, eight risk identification, Risk-warning and risk identification steps;
The data acquisition: data as much as possible are extracted by the equipment on vehicle, the attribute of data must include warp
Three degree, latitude and time most basic parameters, are also possible to that other data such as base station information of these three parameters can be calculated
Deng;
The data cleansing: removal interference and unworthy data, and these data are reasonably stored;
The data analysis: being processed data, analyzes the logical relation between data, and processing generates new dimension
The data of degree;
The factor identification: the possible evaluative dimension of vehicle is determined according to data with existing;
The model construction: evaluation model is constructed according to the factor of a model identified;
The risk identification: identification scene and risk are gone according to evaluation model;
The Risk-warning: early warning is carried out for the risk identified;
The risk identification: it is marked for the risk having been acknowledged.
For the data source of the data acquisition generally there are two types of approach, the equipment for not networking can pass through card reader
Directly data are extracted from the storage equipment of equipment;It, can storage file such as database beyond the clouds and for the equipment of networking
In directly extract, the data of extraction include: location dependent data (time, longitude, latitude, speed, direction, base station parameter, WIFI
Parameter) and vehicle status data (starting/flame-out, driving behavior etc.).
The data for needing to wash in the data cleansing specifically include that
The location point of drift: a) speed is more than 180km/h;B) distance between two points are greater than 25 kilometers;C) vehicle is in and puts out
Fiery state but position changes;
The illegal time: not at the extraction between section location point;
The base station that can not be positioned or WIFI parameter.
Need to handle in data analysis and resolve following data: including longitude and latitude, speed, azimuth, starting/
Flameout state calculates, family and business address analysis, be used to walk circuit analysis, the analysis of consumption place, equipment are artificially divided extremely
Analysis;
(1), it the longitude, latitude: for base station and WIFI, needs to calculate longitude and latitude number by relevant parametric solution
According to typically being realized by third party's interface;
(2), the speed: for the scene for not extracting speed, needing to calculate speed by longitude and latitude and time,
Spherical distance is first gone out according to calculation of longitude & latitude, then speed is calculated by distance;
Spherical distance (D) is calculated by Haversine formula:
Wherein
Haversin (θ)=sin2(θ/2)=(1-cos (θ))/2
θ=d/R
R is earth radius, can be averaged 6371km;
The latitude of the expression two o'clock of φ 1, φ 2;
The difference of Δ λ expression two o'clock longitude.
So: V=D/ Δ T, Δ T are the time difference between two anchor points;
(3), the azimuth: i.e. the direction of vehicle advance, due north 0 rotate clockwise, and calculation method is as follows:
It is as shown in Figure 2:
A, B, C: the angle folded by " camber line " is at this point on three points and spherical surface on spherical surface is indicated
A, b, c: indicating A, and 3 points of B, C (are construed to " arc " two-end-point and angle folded by the line of the earth's core here in fact
3 points of the ABC radians to arc are more convenient)
O: the centre of sphere
Aj:A point longitude
Aw:A point latitude
Bj:B point longitude
Bw:B point latitude
Discussed on four quadrants, two axis according to B relative to the position of A, according to different situations to calculated result into
Row different disposal, final calculation result:
B point is in first quartile, A=A ';
B is in the second quadrant, A=360+A ';
B is in third four-quadrant, A=180-A '.
(6), the starting/flameout state calculates: the case where for not collecting starting/flameout state, it is also necessary to count
Its state is calculated, calculated state does not need very accurate, it is desirable that there is the inconsistent of front and back 10 minutes or so with actual conditions,
Namely vehicle occurs in the movement that starts or stop working at algorithm can monitor within 10 minutes, due to some special feelings
Condition, data provided and less than calculating the state, so the calculating of the state be it is non-necessary,
It should be noted that since there may be missings for acquisition Data Styles, so algorithm needs to do time-out in judgement
Processing was then determined as time-out without continuous data more than 25 minutes, and pressure vehicle is flameout state;
Driving behavior: driving behavior refers mainly to three and suddenly one surpasses, i.e., anxious acceleration, anxious deceleration, sudden turn of events road and hypervelocity;
Anxious to accelerate: judgement of the acceleration greater than 13km/hs is anxious accelerate;
Anxious to slow down: judgement of the acceleration less than -15km/hs is anxious slow down;
Sudden turn of events road: a) speed is greater than 120km/h, and direction change is greater than 3 °/s, is determined as sudden turn of events road;
B) speed is greater than 5 °/s, is determined as sudden turn of events road in 80-120km/h, direction change;
C) speed is greater than 10 °/s, is determined as sudden turn of events road in 60-80km/h, direction change;
D) speed is greater than 25 °/s, is determined as sudden turn of events road in 45-60km/h, direction change;
Hypervelocity: according to place category of roads difference, different speed limit values is taken;
A) high speed: greater than 120km/h is judged to exceeding the speed limit;
B) Class I highway: greater than 80km/h is judged to exceeding the speed limit;
C) Class II highway: greater than 60km/h is judged to exceeding the speed limit;
D) Class III highway: greater than 40km/h is judged to exceeding the speed limit;
E) Class IV highway: greater than 30km/h is judged to exceeding the speed limit;
Category of roads for obtaining vehicle position is had any problem, and a kind of alternative is be all set to 80km/h
It is judged to exceeding the speed limit, by engineering verification, the influence for model is in controlled range;
(5), it the family and business address analysis: according to longitude and latitude and starting/flameout state and time, uses
DBSCAN algorithm calculates family and work unit's specific address, the specific steps are as follows:
Find out longitude and latitude and the time of all vehicle stalls;
Clustering is done using DBSCAN;
Two clusters for taking location point most are analyzed again, as shown in Figure 3:
According to the time of data in cluster, the point occurred between 7-10 point is the cluster of work unit, is occurred between 18-21 point
Point is the cluster of home location;
It is write using third party's interface by the acquisition of work unit's cluster center longitude according to the central point longitude and latitude of cluster
Building, commercialization are determined as business address with the POI of room type point, obtain residential quarters by home address cluster center longitude
POI point be determined as home address;
(6), described to be used to walk circuit analysis: frequent vehicle line is analyzed, the specific method is as follows:
The longitude and latitude that tracing point (speed is greater than 5km/h) when finding out all vehicle drivings is put;
These longitudes and latitudes do clustering using DBSCAN algorithm;
Access is according to most preceding 5 clusters;
It calls third-party map interface to obtain road information where these clusters, forms route;
Another method is first to obtain road, then do clustering to road, can also obtain effect same.
(7), the consumption place analysis: different from family and work unit's analysis, family and business address are first
It does clustering and goes out central point, further according to the type for obtaining POI after time judgement.The analysis of consumption place is also first to do cluster point
Central point is precipitated, but is the type for removing to match position POI again according to the position of central point, i.e., given (X, Y) needs to pass through
It is school, KTV, hotel, shopping center etc. that third-party map interface analysis, which is the position,;
(8), the artificial anomaly analysis of the equipment: before premeditated promise breaking, car owner generally in advance can artificially be broken equipment
It is bad, to escape or cover its subsequent behavior or whereabouts, can generally it analyze following situations:
Equipment Removal: equipment has dismounting alarm to report or equipment does not have any data to report within 15 days;
Equipment separation: will be installed more positioning devices on Some vehicles, when equipment has in separation alarm or same
Distance in all devices 5 minutes is greater than 10KM;
Device mask: vehicle in the process of moving (speed in 20km/h or more), interrupt 5 minutes or more suddenly by data, it
Afterwards again self-recovery the case where.It should be noted that device mask influenced by the external world it is more, such as cross tunnel, actually using
Process needs to carry out external environment locating for auxiliary judgment vehicle according to the map, excludes externalities.
Factor identification be according to the above analysis data, the data of the above dimension of comprehensive judgement those need to participate in model
Building and weight, weight, that is, data confidence level can be assessed, the meter of each dimension according to the quality of data and concentration degree
Calculation method is different.The specific method is as follows:
A), park risk for a long time: according to the state for starting/stopping working, the vehicle for judging that 15 days or more do not start is risky,
Weight=park number of days -15;
B), home address risk: according to home address, vehicle is parked secondary within the scope of home address 1.5KM in 7 days
It is judged as home address risk, weight=risk frequency less than 2 times;
C), business address risk: according to unit address, vehicle stops within the scope of the 1.5KM of business address in 7 days
Secondary be less than 3 times put is judged as business address risk, weight=risk frequency;
D), route deviates risk: the number occurred on used cabling road in 10 days is judged as that route deviates wind less than 3 times
Danger, weight=route deviate lasting number of days;
Damage of facilities risk: artificially determined extremely according to equipment, the frequency within weight=1 month;
E), consumer behavior abnormal risk: consumption place occur big variation (arrive at number not as good as original 30%), weight=
All changes continuous days and summation.
6. a kind of method for establishing air control model according to running car rule according to claim 5, feature exist
In the model construction is the scorecard model that entire model uses logic-based to return, and based on above scoring factor, is used
Following methods are modeled:
(1), risk quantification: the determination data (E) of each risks and assumptions is not determined as and if so, being judged as 1
0, weight R, so the Risk Calculation value xi=E*R of each factor;
(2), normalize: since the numerical difference of x is larger away from possibility, partial factors may significantly affect other factors, institute
To need that x value is normalized.Here using 0 mean value standardization (Z-score standardization), letter is converted
Number are as follows:
Wherein μ is wherein mean value, and δ is standard deviation;
(3), logistic regression: being expressed as p for the probability of customer default, then normal probability is 1-p, therefore, available:
The Probability p of customer default may be expressed as:
The score value scale of scorecard setting can be defined by the way that score value to be expressed as to the linear representation of ratio logarithm, i.e.,
It is represented by following formula:
Score=A-Blog (Odds)
Logic Regression Models calculating ratio is as follows:
Log (Odds)=β0+β1x1+…+βnxn
Wherein, with the available model parameter β 0 of modeling parameters model of fit, β 1 ..., β n;
The value of constant A, B in formula can be by will be known to two or the score value assumed is brought into and is calculated;
By the engineering practice of available data, following empirical value is obtained:
The risk identification and Risk-warning are the model to above building, are classified by scoring to risk, point
The foundation of grade is as follows:
Scoring | Risk class |
>90 | Devoid of risk |
75~90 | Low-risk |
65~75 | Risk |
50-65 | Medium or high risk |
<50 | High risk |
It in the case of more than risk, needs to carry out early warning in systems, air control personnel is reminded to confirm and handle.
The risk identification is the risk for identifying, needs to carry out the forms such as carry out background check, visit to the parents of schoolchildren or young workers and carries out really
Recognize, it is thus identified that after, it needs to be marked in systems.For the risk having been acknowledged, the weight of correlation factor needs to adjust,
According to engineering practice, generally increase by 30%, i.e., multiplied by 1.3 on the basis of original weight, to influence entire evaluation mould
Type.
Method of the invention is modeled by extracting the various simple position datas of equipment on vehicle in advance, this is passed through
A evaluation model carries out the risk identification in thing.The present invention can establish pattern drawing, and this as long as it can extract position data
Kind model can be easy to be generalized to other scenes, can not only take precautions against in advance, can also construct model while running data,
The dimension of evaluation can be adaptive according to the difference of data attribute and quality, equipment form and operating mode, finally provides standard
Consistent evaluation method.
Embodiment 1
As shown in Figure 1, a kind of method for establishing air control model according to running car rule, the side for establishing air control model
Method includes that data acquisition, data cleansing, data analysis, factor identification, model construction, risk identification, Risk-warning and risk are true
Recognize eight steps;
The data acquisition: data as much as possible are extracted by the equipment on vehicle, the attribute of data must include warp
Three degree, latitude and time most basic parameters, are also possible to that other data such as base station information of these three parameters can be calculated
Deng;
The data cleansing: removal interference and unworthy data, and these data are reasonably stored;
The data analysis: being processed data, analyzes the logical relation between data, and processing generates new dimension
The data of degree;
The factor identification: the possible evaluative dimension of vehicle is determined according to data with existing;
The model construction: evaluation model is constructed according to the factor of a model identified;
The risk identification: identification scene and risk are gone according to evaluation model;
The Risk-warning: early warning is carried out for the risk identified;
The risk identification: it is marked for the risk having been acknowledged.
Equipment of the data source of the data acquisition for not networking, can be by card reader directly from the storage of equipment
Data are extracted in equipment;The data of extraction include: location dependent data (time, longitude, latitude, speed, direction, base station parameter,
WIFI parameter) and vehicle status data (starting/flame-out, driving behavior etc.).
Embodiment 2
As shown in Figure 1, a kind of method for establishing air control model according to running car rule, the side for establishing air control model
Method includes that data acquisition, data cleansing, data analysis, factor identification, model construction, risk identification, Risk-warning and risk are true
Recognize eight steps;
The data acquisition: data as much as possible are extracted by the equipment on vehicle, the attribute of data must include warp
Three degree, latitude and time most basic parameters, are also possible to that other data such as base station information of these three parameters can be calculated
Deng;
The data cleansing: removal interference and unworthy data, and these data are reasonably stored;
The data analysis: being processed data, analyzes the logical relation between data, and processing generates new dimension
The data of degree;
The factor identification: the possible evaluative dimension of vehicle is determined according to data with existing;
The model construction: evaluation model is constructed according to the factor of a model identified;
The risk identification: identification scene and risk are gone according to evaluation model;
The Risk-warning: early warning is carried out for the risk identified;
The risk identification: it is marked for the risk having been acknowledged.
The data source of data acquisition, can be straight in storage file beyond the clouds such as database for the equipment of networking
Extraction is connect, the data of extraction include: location dependent data (time, longitude, latitude, speed, direction, base station parameter, WIFI ginseng
Number) and vehicle status data (starting/flame-out, driving behavior etc.);
Embodiment 1- embodiment 2 is withdrawn deposit of the invention not high for equipment requirement and dependence: any to collect position data
Equipment be ok, not only include GPS, Beidou etc. can pinpoint equipment, further include that base station, WIFI etc. are non-precisely positioned
Equipment.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Anyone skilled in the art in the technical scope disclosed by the present invention, according to the technique and scheme of the present invention and its
Inventive concept is subject to equivalent substitution or change, should be covered by the protection scope of the present invention.
Claims (8)
1. a kind of method for establishing air control model according to running car rule, which is characterized in that the side for establishing air control model
Method includes that data acquisition, data cleansing, data analysis, factor identification, model construction, risk identification, Risk-warning and risk are true
Recognize eight steps;
The data acquisition: data as much as possible are extracted by the equipment on vehicle, the attribute of data must include longitude, latitude
Degree and three most basic parameters of time, are also possible to that other data such as base station information of these three parameters etc. can be calculated;
The data cleansing: removal interference and unworthy data, and these data are reasonably stored;
The data analysis: being processed data, analyzes the logical relation between data, and processing generates new dimension
Data;
The factor identification: the possible evaluative dimension of vehicle is determined according to data with existing;
The model construction: evaluation model is constructed according to the factor of a model identified;
The risk identification: identification scene and risk are gone according to evaluation model;
The Risk-warning: early warning is carried out for the risk identified;
The risk identification: it is marked for the risk having been acknowledged.
2. a kind of method for establishing air control model according to running car rule according to claim 1, which is characterized in that institute
Stating the data source that data acquire, generally there are two types of approach, and the equipment for not networking can be by card reader directly from equipment
Storage equipment on extract data;And for the equipment of networking, can directly be extracted in storage file beyond the clouds such as database,
The data of extraction include: location dependent data (time, longitude, latitude, speed, direction, base station parameter, WIFI parameter) and vehicle
Status data (starting/flame-out, driving behavior etc.).
3. a kind of method for establishing air control model according to running car rule according to claim 1, which is characterized in that institute
The data for needing to wash in data cleansing are stated to specifically include that
The location point of drift: a) speed is more than 180km/h;B) distance between two points are greater than 25 kilometers;C) vehicle is in flame-out shape
State but position changes;
The illegal time: not at the extraction between section location point;
The base station that can not be positioned or WIFI parameter.
4. a kind of method for establishing air control model according to running car rule according to claim 1, which is characterized in that institute
It states and needs to handle in data analysis and resolve following data: including longitude and latitude, speed, azimuth, starting/flameout state meter
It calculates, family and business address analyze, are used to walk circuit analysis, the analysis of consumption place, the artificial anomaly analysis of equipment;
(1), the longitude, latitude: for base station and WIFI, needing to calculate longitude and latitude data by relevant parametric solution,
Typically realized by third party's interface;
(2), it the speed: for the scene for not extracting speed, needs to calculate speed by longitude and latitude and time, i.e., first
Go out spherical distance according to calculation of longitude & latitude, then speed is calculated by distance;
Spherical distance (D) is calculated by Haversine formula:
Wherein
Haversin (θ)=sin2(θ/2)=(1-cos (θ))/2
θ=d/R
R is earth radius, can be averaged 6371km;
The latitude of the expression two o'clock of φ 1, φ 2;
The difference of Δ λ expression two o'clock longitude.
So: V=D/ Δ T, Δ T are the time difference between two anchor points;
(3), the azimuth: i.e. the direction of vehicle advance, due north 0 rotate clockwise, and calculation method is as follows:
A, B, C: the angle folded by " camber line " is at this point on three points and spherical surface on spherical surface is indicated
A, b, c: indicating A, 3 points of B, C to angle folded by " arc " two-end-point and the earth's core line.
(4), the starting/flameout state calculates: the case where for not collecting starting/flameout state, it is also necessary to calculate
Its state, calculated state do not need very accurate, it is desirable that have the inconsistent of front and back 10 minutes or so with actual conditions, yet
It is that vehicle occurs in the movement that starts or stop working at algorithm can monitor within 10 minutes, due to some special circumstances, mention
Data of confession and less than calculating the state, thus the calculating of the state be it is non-necessary,
It should be noted that since there may be missings for acquisition Data Styles, so algorithm needs to do timeout treatment in judgement,
Then it was determined as time-out without continuous data more than 25 minutes, pressure vehicle is flameout state;
Driving behavior: driving behavior refers mainly to three and suddenly one surpasses, i.e., anxious acceleration, anxious deceleration, sudden turn of events road and hypervelocity;
Anxious to accelerate: judgement of the acceleration greater than 13km/hs is anxious accelerate;
Anxious to slow down: judgement of the acceleration less than -15km/hs is anxious slow down;
Sudden turn of events road: a) speed is greater than 120km/h, and direction change is greater than 3 °/s, is determined as sudden turn of events road;
B) speed is greater than 5 °/s, is determined as sudden turn of events road in 80-120km/h, direction change;
C) speed is greater than 10 °/s, is determined as sudden turn of events road in 60-80km/h, direction change;
D) speed is greater than 25 °/s, is determined as sudden turn of events road in 45-60km/h, direction change;
Hypervelocity: according to place category of roads difference, different speed limit values is taken;
A) high speed: greater than 120km/h is judged to exceeding the speed limit;
B) Class I highway: greater than 80km/h is judged to exceeding the speed limit;
C) Class II highway: greater than 60km/h is judged to exceeding the speed limit;
D) Class III highway: greater than 40km/h is judged to exceeding the speed limit;
E) Class IV highway: greater than 30km/h is judged to exceeding the speed limit;
Category of roads for obtaining vehicle position is had any problem, and a kind of alternative is to be all set to 80km/h judgement
For hypervelocity, by engineering verification, the influence for model is in controlled range;
(5), it the family and business address analysis: according to longitude and latitude and starting/flameout state and time, uses
DBSCAN algorithm calculates family and work unit's specific address, the specific steps are as follows:
Find out longitude and latitude and the time of all vehicle stalls;
Clustering is done using DBSCAN;
Two clusters for taking location point most are analyzed again;
According to the central point longitude and latitude of cluster, using third party's interface, office building, quotient are obtained by work unit's cluster center longitude
It is determined as business address with the POI of room type point, the POI of residential quarters is obtained by home address cluster center longitude
Point is determined as home address;
(6), described to be used to walk circuit analysis: frequent vehicle line is analyzed, the specific method is as follows:
The longitude and latitude that tracing point (speed is greater than 5km/h) when finding out all vehicle drivings is put;
These longitudes and latitudes do clustering using DBSCAN algorithm;
Access is according to most preceding 5 clusters;
It calls third-party map interface to obtain road information where these clusters, forms route;
Another method is first to obtain road, then do clustering to road, can also obtain effect same.
(7), consumption place analysis: it is different from family and work unit's analysis, family and business address be first do it is poly-
Alanysis goes out central point, further according to the type for obtaining POI after time judgement.Consumption place analysis be also first do clustering go out in
Heart point, but be the type for removing to match position POI again according to the position of central point, i.e., given (X, Y) needs to pass through third party
Map interface analysis be the position be school, KTV, hotel, shopping center etc.;
(8), the artificial anomaly analysis of the equipment: before premeditated promise breaking, car owner generally can carry out artificial destruction to equipment in advance, with
Its subsequent behavior or whereabouts are escaped or are covered, can generally analyze following situations:
Equipment Removal: equipment has dismounting alarm to report or equipment does not have any data to report within 15 days;
Equipment separation: will be installed more positioning devices on Some vehicles, when equipment has owning in separation alarm or same
Distance in equipment 5 minutes is greater than 10KM;
Device mask: (speed interrupts 5 minutes or more in 20km/h or more), data to vehicle suddenly, Zhi Houyou in the process of moving
The case where self-recovery.It should be noted that device mask influenced by the external world it is more, such as cross tunnel, actual use process,
It needs to carry out external environment locating for auxiliary judgment vehicle according to the map, excludes externalities.
5. a kind of method for establishing air control model according to running car rule according to claim 4, which is characterized in that institute
State factor identification be according to the above analysis data, the data of the above dimension of comprehensive judgement those need to participate in model construction and power
Weight, weight, that is, data confidence level can be assessed according to the quality of data and concentration degree, and the calculation method of each dimension is not
Together.The specific method is as follows:
A), park risk for a long time: according to the state for starting/stopping working, the vehicle for judging that 15 days or more do not start is risky, weight
=park number of days -15;
B), home address risk: according to home address, vehicle is parked secondary less than 2 within the scope of home address 1.5KM in 7 days
It is secondary to be judged as home address risk, weight=risk frequency;
C), business address risk: according to unit address, vehicle is parked within the scope of the 1.5KM of business address in 7 days
It is secondary to be less than 3 times and be judged as business address risk, weight=risk frequency;
D), route deviates risk: the number occurred on used cabling road in 10 days is judged as that route deviates risk less than 3 times,
Weight=route deviates lasting number of days;
Damage of facilities risk: artificially determined extremely according to equipment, the frequency within weight=1 month;
E), consumer behavior abnormal risk: consumption place occurs big variation and (arrives at number not as good as original 30%), weight=all
Change continuous days and summation.
6. a kind of method for establishing air control model according to running car rule according to claim 5, which is characterized in that institute
Stating model construction is the scorecard model that entire model uses logic-based to return, and based on above scoring factor, use is following
Method is modeled:
(1), risk quantification: for the determination data (E) of each risks and assumptions, and if so, being judged as 1, it is not determined as 0, power
Weight is R, so the Risk Calculation value xi=E*R of each factor;
(2), normalize: since the numerical difference of x is larger away from possibility, partial factors may significantly affect other factors, so needing
X value is normalized.Here using 0 mean value standardization (Z-score standardization), function is converted
Are as follows:
Wherein μ is wherein mean value, and δ is standard deviation;
(3), logistic regression: being expressed as p for the probability of customer default, then normal probability is 1-p, therefore, available:
The Probability p of customer default may be expressed as:
The score value scale of scorecard setting can be defined by the way that score value to be expressed as to the linear representation of ratio logarithm, Ji Kebiao
It is shown as following formula:
Score=A-Blog (Odds)
Logic Regression Models calculating ratio is as follows:
Log (Odds)=β0+β1x1+…+βnxn
Wherein, with the available model parameter β 0 of modeling parameters model of fit, β 1 ..., β n;
The value of constant A, B in formula can be by will be known to two or the score value assumed is brought into and is calculated.
7. a kind of method for establishing air control model according to running car rule according to claim 6, which is characterized in that institute
The model of risk identification and Risk-warning i.e. to above building is stated, risk is classified by scoring.
8. a kind of method for establishing air control model according to running car rule according to claim 7, which is characterized in that institute
Stating risk identification is the risk for identifying, needs to carry out the forms such as carry out background check, visit to the parents of schoolchildren or young workers and is confirmed, it is thus identified that with
Afterwards, it needs to be marked in systems.For the risk having been acknowledged, the weight of correlation factor needs to adjust, according to engineering reality
It tramples, generally increases by 30%, i.e., multiplied by 1.3 on the basis of original weight, to influence entire evaluation model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811140159.8A CN109523390A (en) | 2018-09-28 | 2018-09-28 | A method of air control model is established according to running car rule |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811140159.8A CN109523390A (en) | 2018-09-28 | 2018-09-28 | A method of air control model is established according to running car rule |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109523390A true CN109523390A (en) | 2019-03-26 |
Family
ID=65769847
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811140159.8A Pending CN109523390A (en) | 2018-09-28 | 2018-09-28 | A method of air control model is established according to running car rule |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109523390A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110254435A (en) * | 2019-06-28 | 2019-09-20 | 同济大学 | A kind of driving style recognition methods |
CN110633729A (en) * | 2019-08-06 | 2019-12-31 | 清华大学苏州汽车研究院(相城) | Driving risk hierarchical clustering method for intelligent networking vehicle group test |
CN112802231A (en) * | 2021-03-19 | 2021-05-14 | 四川万网鑫成信息科技有限公司 | Vehicle risk assessment method, device, equipment and medium based on GPS data |
CN113064883A (en) * | 2020-09-28 | 2021-07-02 | 开鑫金服(南京)信息服务有限公司 | Method for constructing logistics wind control model, computer equipment and storage medium |
CN113888302A (en) * | 2021-08-09 | 2022-01-04 | 深圳市麦谷科技有限公司 | Risk early warning method and system for automobile financial vehicle loan |
-
2018
- 2018-09-28 CN CN201811140159.8A patent/CN109523390A/en active Pending
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110254435A (en) * | 2019-06-28 | 2019-09-20 | 同济大学 | A kind of driving style recognition methods |
CN110633729A (en) * | 2019-08-06 | 2019-12-31 | 清华大学苏州汽车研究院(相城) | Driving risk hierarchical clustering method for intelligent networking vehicle group test |
CN113064883A (en) * | 2020-09-28 | 2021-07-02 | 开鑫金服(南京)信息服务有限公司 | Method for constructing logistics wind control model, computer equipment and storage medium |
CN112802231A (en) * | 2021-03-19 | 2021-05-14 | 四川万网鑫成信息科技有限公司 | Vehicle risk assessment method, device, equipment and medium based on GPS data |
CN112802231B (en) * | 2021-03-19 | 2021-07-06 | 四川万网鑫成信息科技有限公司 | Vehicle risk assessment method, device, equipment and medium based on GPS data |
CN113888302A (en) * | 2021-08-09 | 2022-01-04 | 深圳市麦谷科技有限公司 | Risk early warning method and system for automobile financial vehicle loan |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109523390A (en) | A method of air control model is established according to running car rule | |
Jiang et al. | Measuring urban sprawl in Beijing with geo-spatial indices | |
CN107067781B (en) | GIS road black spot map generation method for advanced driving assistance system application | |
CN111539450B (en) | Landslide susceptibility evaluation method in earthquake region | |
CN106056447A (en) | Vehicle secondary mortgage risk early warning method and system | |
Lawson et al. | Applications of extraction mapping in environmental epidemiology | |
CN104900057A (en) | City expressway main and auxiliary road floating vehicle map matching method | |
Ding et al. | RTVEMVS: Real-time modeling and visualization system for vehicle emissions on an urban road network | |
CN109447478A (en) | A kind of industrial thermal pollution area extracting method based on eight period in arithmetric | |
CN105355047A (en) | Data fusion processing method for dynamic time granularity of multiple traffic detection sources | |
Liang et al. | Planning control over rural land transformation in Hong Kong: A remote sensing analysis of spatio-temporal land use change patterns | |
WO2023034141A1 (en) | System and method for monitoring a vehicle | |
Xie et al. | A least cumulative ventilation cost method for urban ventilation environment analysis | |
Fu et al. | Spatial heterogeneity and migration characteristics of traffic congestion—A quantitative identification method based on taxi trajectory data | |
CN104794164B (en) | Method based on the social parking demand of data identification settlement parking stall matching of increasing income | |
Fu et al. | Spatial analysis of taxi speeding event using GPS trajectory data | |
Tian et al. | Suburban sprawl measurement and landscape analysis of cropland and ecological land: A case study of Jiangsu Province, China | |
Chung et al. | Investigating the effects of POI-based land use on traffic accidents in Suzhou Industrial Park, China | |
US20230394964A1 (en) | Method, electronic device, and system for detecting overspeeding | |
US11823572B2 (en) | Method, electronic device, and system for predicting future overspeeding | |
CN115358507A (en) | Production and construction project disturbance pattern spot water and soil loss risk identification and evaluation method | |
Chen et al. | Research on the influencing factors of elderly pedestrian traffic accidents considering the built environment | |
Levine | Hot spot analysis of zones | |
Yu et al. | Research on site classification method based on BP neural network | |
CN107590998B (en) | Road node running state recognition system based on floating car data |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190326 |
|
RJ01 | Rejection of invention patent application after publication |