CN105426553A - Target real-time tracking and early warning method and system based on intelligent equipment - Google Patents

Abstract

The invention discloses a target real-time tracking and early warning method and a target real-time tracking and early warning system based on intelligent equipment, which belong to a novel target behavior identifying technique. A principle is that peripheral information of a geographical position in which a target is positioned is determined by utilizing a positioning technique and location based service (LBS), then the target behavior probability is calculated according to database records, and when a certain behavior probability of the target is greater than a set threshold, early warning is given to the target or the receiving end of the target. According to the method and the system, on the basis that all possible behaviors of the target are analyzed, the target motion diversity problem (which is solved in a user position module) and the target behavior diversity problem (which is solved in a user behavior prediction module and an early warning judgment module) are respectively solved by utilizing the three modules.

Description

A kind of object real-time tracking method for early warning based on smart machine and system

Technical field

The present invention relates to a kind of object real-time tracking method for early warning based on smart machine and system, belong to civilian areas of information technology.

Background technology

In recent years, with economic development, population increasing, crime of kidnaping and selling people particularly is abducted children's crime and is presented increasing trend.One pieces is abducted children's event and is touched the nerve of people again and again, for society and the common people bring huge injury.But the phenomenon of abducting children remains incessant after repeated prohibition, and missing child is increasing, make the family of abducted child therefore incoherent, havoc society stable and harmonious, become a great factor leading to social instability.More allow people can not put up with, some offender even forces abducted next child to go to beg in the streets, be engaged in the illegal activity such as prostitution, traffic in drugs, causes these child's bodies and minds seriously to be wrecked.

Up to the present, there is no effective ways can prevent from abducting the generation of children's phenomenon.

Summary of the invention

Goal of the invention of the present invention is: for above-mentioned Problems existing, a kind of object real-time tracking method for early warning based on smart machine and system are provided, it is a kind of brand-new goal behavior discrimination technology, its principle utilizes location technology and geo-location service (LBS) to determine target geographic location peripheral information, then goal behavior probability is calculated by database data, when target behavior probability is greater than the threshold value of setting, to the receiving end early warning of target or target.

Object real-time tracking method for early warning based on smart machine provided by the invention relates to customer location judge module, user's behavior prediction module and early-warning judgment module, comprises the following steps:

The first step, in customer location judge module, obtains the position data of user within a period of time using and have the smart machine of positioning function and forms customer location vector, calculate the user velocity matrix of this customer location vector and carry out binary conversion treatment to it; If the user velocity matrix value after process is not 0, then judge that user just passes by, then do not process especially; If the user velocity matrix value after process is 0, then judge that user stops in this region and user's latest position is outputted to user's behavior prediction module;

Second step, in user's behavior prediction module, the facilities information in user neighboring area is obtained from database, matrix is set up for every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often in row, the probability that enter this concrete facility's objectives of first element for calculating according to data with existing in database, after other elements be the weighting weight of each influence factor (such as, user is from the distance of concrete facility's objectives); Calculate the probability-weighted of each concrete facility's objectives and get its maximal value enters such facility prediction probability as this user; After matrix computations to every class facility, the behavior prediction probability vector of user can be obtained and be processed into output vector, then outputting to early-warning judgment module;

3rd step, in early-warning judgment module, each element of output vector and corresponding threshold values are compared, judge whether early warning, and by its behavior record in database: should early warning if be determined as, then judge whether be provided with warning function in data further, if be provided with warning function, then send early warning, if do not arrange warning function, then do not send early warning.

As preferably, the first step is in customer location judge module, and the customer location vector of formation is $A=\left[\begin{array}{c}{x}_1,x_2...x_n\\ y_1,y_2...y_n\end{array}\right],$ According to the user velocity matrix that this vector calculation goes out be $A^{\&prime;}=\left[\begin{array}{c}{x^{\&prime;}}_1,{x^{\&prime;}}_2\mathrm{...}{x^{\&prime;}}_{n-1}\\ {y^{\&prime;}}_1,{y^{\&prime;}}_2\mathrm{...}{y^{\&prime;}}_{n-1}\end{array}\right],$ Wherein to each group column vector $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]$ Computing formula be $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]=\left[\begin{array}{c}{x^2}_{i+1}\\ {y^2}_{i+1}\end{array}\right]-\left[\begin{array}{c}{x^2}_i\\ {y^2}_i\end{array}\right];$ Pre-set threshold value T is utilized to the process that user velocity matrix carries out binary conversion treatment to be: if x' _i< T, then make x' _i=0, if y' _iduring < T, then make y' _i=0, if the user velocity matrix after process meets A' ≠ 0, be then judged as that user just passes by this place, then do not process especially, if meet A'=0, be then judged as that user is in the stop of this region and by the column vector of user's latest position $\left[\begin{array}{c}{x}_n\\ y_n\end{array}\right]$ Output to user's behavior prediction module, user behavior is predicted.

As preferably, second step is in user's behavior prediction module, from database, obtain the facilities information in border circular areas that its periphery radius is R according to the latest position of user, the statistics species number m of peripheral facility and the number k of often kind of facility, wherein the number of m kind is denoted as k _m; Set up matrix to every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often row first element is the probability that user enters this concrete facility's objectives, passes through formula determine, if often go other element positive correlation elements below, then its computing formula is: if negative correlation element, then its computing formula is: wherein, I _krepresent that a kth target of data-base recording is by the total degree entered, N _krepresent that in database, a kth target is not by the total degree entered, D _irepresent the weighted data preset; Then formula is passed through calculate the probability-weighted of each concrete facility's objectives in every class facility, and get its maximum probability-weighted wp _j, and record the position of j, and make vectorial op _i=wp _j, wherein, op _irepresent that user enters the probability of the i-th class facility generation under global context; After matrix computations to all kinds of facility, the behavior prediction probability vector op of this user can be obtained.

As preferably, second step is in user's behavior prediction module, and the behavior prediction probability vector op of user is by formula p _i=(1-op ₁) (1-op ₂) ... (1-op _i-1) op _i(1-op _i+1) (1-op _m) process, obtain output vector P, before then outputting to early-warning judgment module.

As preferably, the 3rd step, in early-warning judgment module, before differentiation, reads presetting information vector S, if s _i=0, illustrate that the i-th class facility does not set warning function, if s _i=1, illustrate that the i-th class facility sets warning function; If s _i=0 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=0 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, should early warning be sent but not send early warning; If s _i=1 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=1 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, send early warning.Carry out further preferably, the 3rd step is in early-warning judgment module, if differentiate, result is not for should send early warning, then in a database all facilities of the i-th class facility always not being entered number of times increases by 1 time; If differentiate, result is for should send early warning, then increase by 1 time by the number of times that always enters of facility maximum for probability-weighted under the i-th class facility in a database, always not entering number of times and increase by 1 time all the other all facilities in region.

Object real-time tracking early warning system based on smart machine provided by the invention comprises customer location judge module, user's behavior prediction module and early-warning judgment module, and the function of each module is as follows:

The function of customer location judge module: obtain the position data of user within a period of time using and there is the smart machine of positioning function and form customer location vector, calculate the user velocity matrix of this customer location vector and binary conversion treatment is carried out to it; If the user velocity matrix value after process is not 0, then judge that user just passes by, then do not process especially; If the user velocity matrix value after process is 0, then judge that user stops in this region and user's latest position is outputted to user's behavior prediction module.

As preferably, in customer location judge module, the customer location vector of formation is $A=\left[\begin{array}{c}{x}_1,x_2...x_n\\ y_1,y_2...y_n\end{array}\right],$ According to the user velocity matrix that this vector calculation goes out be $A^{\&prime;}=\left[\begin{array}{c}{x^{\&prime;}}_1,{x^{\&prime;}}_2\mathrm{...}{x^{\&prime;}}_{n-1}\\ {y^{\&prime;}}_1,{y^{\&prime;}}_2\mathrm{...}{y^{\&prime;}}_{n-1}\end{array}\right],$ Wherein to each group column vector $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]$ Computing formula be $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]=\left[\begin{array}{c}{x^2}_{i+1}\\ {y^2}_{i+1}\end{array}\right]-\left[\begin{array}{c}{x^2}_i\\ {y^2}_i\end{array}\right];$ Pre-set threshold value T is utilized to the process that user velocity matrix carries out binary conversion treatment to be: if x' _i< T, then make x' _i=0, if y' _iduring < T, then make y' _i=0, if the user velocity matrix after process meets A' ≠ 0, be then judged as that user just passes by this place, then do not process especially, if meet A'=0, be then judged as that user is in the stop of this region and by the column vector of user's latest position $\left[\begin{array}{c}{x}_n\\ y_n\end{array}\right]$ Output to user's behavior prediction module, user behavior is predicted.

The function of user's behavior prediction module: obtain the facilities information in user neighboring area from database, matrix is set up for every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often in row, the probability that enter this concrete facility's objectives of first element for calculating according to data with existing in database, after other elements be the weighting weight of each influence factor; Calculate the probability-weighted of each concrete facility's objectives and get its maximal value enters such facility prediction probability as this user; After matrix computations to every class facility, the behavior prediction probability vector of user can be obtained and be processed into output vector, then outputting to early-warning judgment module.

As preferably, in user's behavior prediction module, obtain the facilities information in border circular areas that its periphery radius is R according to the latest position of user from database, the statistics species number m of peripheral facility and the number k of often kind of facility, wherein the number of m kind is denoted as k _m; Set up matrix to every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often row first element is the probability that user enters this concrete facility's objectives, passes through formula determine, if often go other element positive correlation elements below, then its computing formula is: if negative correlation element, then its computing formula is: wherein, I _krepresent that a kth target of data-base recording is by the total degree entered, N _krepresent that in database, a kth target is not by the total degree entered, D _irepresent the weighted data preset; Then formula is passed through calculate the probability-weighted of each concrete facility's objectives in every class facility, and get its maximum probability-weighted wp _j, and record the position of j, and make vectorial op _i=wp _j, wherein, op _irepresent that user enters the probability of the i-th class facility generation under global context; After matrix computations to all kinds of facility, can obtain the behavior prediction probability vector op of this user, the behavior prediction probability vector op of user is by formula p _i=(1-op ₁) (1-op ₂) ... (1-op _i-1) op _i(1-op _i+1) (1-op _m) process, obtain output vector P, before then outputting to early-warning judgment module.

The function of early-warning judgment module: each element of output vector and corresponding threshold values are compared, judge whether early warning, and by its behavior record in database: should early warning if be determined as, then judge whether be provided with warning function in data further, if be provided with warning function, then send early warning, if do not arrange warning function, then do not send early warning.

As preferably, in early-warning judgment module, before differentiation, read presetting information vector S, if s _i=0, illustrate that the i-th class facility does not set warning function, if s _i=1, illustrate that the i-th class facility sets warning function; If s _i=0 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=0 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, should early warning be sent but not send early warning; If s _i=1 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=1 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, send early warning; If differentiate, result is not for should send early warning, then in a database all facilities of the i-th class facility always not being entered number of times increases by 1 time; If differentiate, result is for should send early warning, then increase by 1 time by the number of times that always enters of facility maximum for probability-weighted under the i-th class facility in a database, always not entering number of times and increase by 1 time all the other all facilities in region.

In sum, the present invention, on the basis analyzing all probable behaviors of target, utilizes three large modules to solve target respectively and moves diverse problems (this problem is solved in customer location module) and goal behavior diverse problems (this problem is solved in user's behavior prediction module and early-warning judgment module).Tool of the present invention has the following advantages:

1, by target location coordinate, certain judging that target passes by that this region or preparation enter this region is arranged, and effectively can reduce system load;

2, utilize the contingent behavior of probabilistic method target of prediction, can not make because of increasing of user data the variation that predicts the outcome, make on the contrary to predict the outcome more accurate.

3, calculated amount needed for system is little, has stronger engineering realizability.

4, system considers the multiple factors such as distance, and exploitation right is vectorial strengthens to original probability weighting the accuracy predicted the outcome.

5, the multiple behavior of early warning can be judged whether according to user's setting.

Accompanying drawing explanation

Fig. 1 is the construction module figure of present system.

Fig. 2 is the decision flow chart of the customer location judge module of present system.

Fig. 3 is the prediction process flow diagram of the user's behavior prediction module of present system.

Fig. 4 is the early warning decision flow chart of the early-warning judgment module of present system.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in detail.

Embodiment 1:

Based on an object real-time tracking method for early warning for smart machine, relate to customer location judge module, user's behavior prediction module and early-warning judgment module, its cardinal principle flow process as shown in Figure 1.This method comprises the following steps:

The first step as shown in Figure 2, in customer location judge module, obtain the position data of user within a period of time using and there is the smart machine of positioning function and form customer location vector, calculate the user velocity matrix of this customer location vector and binary conversion treatment is carried out to it; If the user velocity matrix value after process is not 0, then judge that user just passes by, then do not process especially; If the user velocity matrix value after process is 0, then judge that user stops in this region and user's latest position is outputted to user's behavior prediction module.

The first step is in customer location judge module, and the customer location vector of formation is $A=\left[\begin{array}{c}{x}_1,x_2...x_n\\ y_1,y_2...y_n\end{array}\right],$ According to the user velocity matrix that this vector calculation goes out be $A^{\&prime;}=\left[\begin{array}{c}{x^{\&prime;}}_1,{x^{\&prime;}}_2\mathrm{...}{x^{\&prime;}}_{n-1}\\ {y^{\&prime;}}_1,{y^{\&prime;}}_2\mathrm{...}{y^{\&prime;}}_{n-1}\end{array}\right],$ Wherein to each group column vector $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]$ Computing formula be $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]=\left[\begin{array}{c}{x^2}_{i+1}\\ {y^2}_{i+1}\end{array}\right]-\left[\begin{array}{c}{x^2}_i\\ {y^2}_i\end{array}\right];$ Pre-set threshold value T is utilized to the process that user velocity matrix carries out binary conversion treatment to be: if x' _i< T, then make x' _i=0, if y' _iduring < T, then make y' _i=0, if the user velocity matrix after process meets A' ≠ 0, be then judged as that user just passes by this place, then do not process especially, if meet A'=0, be then judged as that user is in the stop of this region and by the column vector of user's latest position $\left[\begin{array}{c}{x}_n\\ y_n\end{array}\right]$ Output to user's behavior prediction module, user behavior is predicted.

Second step as shown in Figure 3, in user's behavior prediction module, the facilities information in user neighboring area is obtained from database, matrix is set up for every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often in row, the probability that enter this concrete facility's objectives of first element for calculating according to data with existing in database, after other elements be the weighting weight of each influence factor (such as, user is from the distance of concrete facility's objectives); Calculate the probability-weighted of each concrete facility's objectives and get its maximal value enters such facility prediction probability as this user; After matrix computations to every class facility, the behavior prediction probability vector of user can be obtained and be processed into output vector, then outputting to early-warning judgment module.

Second step, in user's behavior prediction module, obtains the facilities information in border circular areas that its periphery radius is R according to the latest position of user from database, and the statistics species number m of peripheral facility and the number k of often kind of facility, wherein the number of m kind is denoted as k _m; Set up matrix to every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often row first element is the probability that user enters this concrete facility's objectives, passes through formula determine, if often go other element positive correlation elements below, then its computing formula is: if negative correlation element, then its computing formula is: wherein, I _krepresent that a kth target of data-base recording is by the total degree entered, N _krepresent that in database, a kth target is not by the total degree entered, D _irepresent the weighted data preset; Then formula is passed through calculate the probability-weighted of each concrete facility's objectives in every class facility, and get its maximum probability-weighted wp _j, and record the position of j, and make vectorial op _i=wp _j, wherein, op _irepresent that user enters the probability of the i-th class facility generation under global context; After matrix computations to all kinds of facility, the behavior prediction probability vector op of this user can be obtained.Second step is in user's behavior prediction module, and the behavior prediction probability vector op of user is by formula p _i=(1-op ₁) (1-op ₂) ... (1-op _i-1) op _i(1-op _i+1) (1-op _m) process, obtain output vector P, before then outputting to early-warning judgment module.

3rd step as shown in Figure 4, in early-warning judgment module, each element of output vector and corresponding threshold values are compared, judge whether early warning, and by its behavior record in database: should early warning if be determined as, then judge whether be provided with warning function in data further, if be provided with warning function, then send early warning, if do not arrange warning function, then do not send early warning.

3rd step, in early-warning judgment module, before differentiation, reads presetting information vector S, if s _i=0, illustrate that the i-th class facility does not set warning function, if s _i=1, illustrate that the i-th class facility sets warning function; If s _i=0 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=0 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, should early warning be sent but not send early warning; If s _i=1 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=1 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, send early warning.Carry out further preferably, the 3rd step is in early-warning judgment module, if differentiate, result is not for should send early warning, then in a database all facilities of the i-th class facility always not being entered number of times increases by 1 time; If differentiate, result is for should send early warning, then increase by 1 time by the number of times that always enters of facility maximum for probability-weighted under the i-th class facility in a database, always not entering number of times and increase by 1 time all the other all facilities in region.

Embodiment 2:

Based on an object real-time tracking early warning system for smart machine, comprise customer location judge module, user's behavior prediction module and early-warning judgment module, the function of each module is as follows:

The function of customer location judge module: obtain the position data of user within a period of time using and there is the smart machine of positioning function and form customer location vector, calculate the user velocity matrix of this customer location vector and binary conversion treatment is carried out to it; If the user velocity matrix value after process is not 0, then judge that user just passes by, then do not process especially; If the user velocity matrix value after process is 0, then judge that user stops in this region and user's latest position is outputted to user's behavior prediction module.

In customer location judge module, the customer location vector of formation is $A=\left[\begin{array}{c}{x}_1,x_2...x_n\\ y_1,y_2...y_n\end{array}\right],$ According to the user velocity matrix that this vector calculation goes out be $A^{\&prime;}=\left[\begin{array}{c}{x^{\&prime;}}_1,{x^{\&prime;}}_2\mathrm{...}{x^{\&prime;}}_{n-1}\\ {y^{\&prime;}}_1,{y^{\&prime;}}_2\mathrm{...}{y^{\&prime;}}_{n-1}\end{array}\right],$ Wherein to each group column vector $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]$ Computing formula be $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]=\left[\begin{array}{c}{x^2}_{i+1}\\ {y^2}_{i+1}\end{array}\right]-\left[\begin{array}{c}{x^2}_i\\ {y^2}_i\end{array}\right];$ Pre-set threshold value T is utilized to the process that user velocity matrix carries out binary conversion treatment to be: if x' _i< T, then make x' _i=0, if y' _iduring < T, then make y' _i=0, if the user velocity matrix after process meets A' ≠ 0, be then judged as that user just passes by this place, then do not process especially, if meet A'=0, be then judged as that user is in the stop of this region and by the column vector of user's latest position $\left[\begin{array}{c}{x}_n\\ y_n\end{array}\right]$ Output to user's behavior prediction module, user behavior is predicted.

The function of user's behavior prediction module: obtain the facilities information in user neighboring area from database, matrix is set up for every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often in row, the probability that enter this concrete facility's objectives of first element for calculating according to data with existing in database, after other elements be the weighting weight of each influence factor; Calculate the probability-weighted of each concrete facility's objectives and get its maximal value enters such facility prediction probability as this user; After matrix computations to every class facility, the behavior prediction probability vector of user can be obtained and be processed into output vector, then outputting to early-warning judgment module.

In user's behavior prediction module, obtain the facilities information in border circular areas that its periphery radius is R according to the latest position of user from database, the statistics species number m of peripheral facility and the number k of often kind of facility, wherein the number of m kind is denoted as k _m; Set up matrix to every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often row first element is the probability that user enters this concrete facility's objectives, passes through formula determine, if often go other element positive correlation elements below, then its computing formula is: if negative correlation element, then its computing formula is: wherein, I _krepresent that a kth target of data-base recording is by the total degree entered, N _krepresent that in database, a kth target is not by the total degree entered, D _irepresent the weighted data preset; Then formula is passed through calculate the probability-weighted of each concrete facility's objectives in every class facility, and get its maximum probability-weighted wp _j, and record the position of j, and make vectorial op _i=wp _j, wherein, op _irepresent that user enters the probability of the i-th class facility generation under global context; After matrix computations to all kinds of facility, can obtain the behavior prediction probability vector op of this user, the behavior prediction probability vector op of user is by formula p _i=(1-op ₁) (1-op ₂) ... (1-op _i-1) op _i(1-op _i+1) (1-op _m) process, obtain output vector P, before then outputting to early-warning judgment module.

The function of early-warning judgment module: each element of output vector and corresponding threshold values are compared, judge whether early warning, and by its behavior record in database: should early warning if be determined as, then judge whether be provided with warning function in data further, if be provided with warning function, then send early warning, if do not arrange warning function, then do not send early warning.

In early-warning judgment module, before differentiation, read presetting information vector S, if s _i=0, illustrate that the i-th class facility does not set warning function, if s _i=1, illustrate that the i-th class facility sets warning function; If s _i=0 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=0 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, should early warning be sent but not send early warning; If s _i=1 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=1 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, send early warning; If differentiate, result is not for should send early warning, then in a database all facilities of the i-th class facility always not being entered number of times increases by 1 time; If differentiate, result is for should send early warning, then increase by 1 time by the number of times that always enters of facility maximum for probability-weighted under the i-th class facility in a database, always not entering number of times and increase by 1 time all the other all facilities in region.

Claims (10)

1., based on an object real-time tracking method for early warning for smart machine, relate to customer location judge module, user's behavior prediction module and early-warning judgment module, it is characterized in that comprising the following steps:

The first step, in customer location judge module, obtains the position data of user within a period of time using and have the smart machine of positioning function and forms customer location vector, calculate the user velocity matrix of this customer location vector and carry out binary conversion treatment to it; If the user velocity matrix value after process is not 0, then judge that user just passes by, then do not process especially; If the user velocity matrix value after process is 0, then judge that user stops in this region and user's latest position is outputted to user's behavior prediction module;

Second step, in user's behavior prediction module, the facilities information in user neighboring area is obtained from database, matrix is set up for every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often in row, the probability that enter this concrete facility's objectives of first element for calculating according to data with existing in database, after other elements be the weighting weight of each influence factor; Calculate the probability-weighted of each concrete facility's objectives and get its maximal value enters such facility prediction probability as this user; After matrix computations to every class facility, the behavior prediction probability vector of user can be obtained and be processed into output vector, then outputting to early-warning judgment module;

3rd step, in early-warning judgment module, each element of output vector and corresponding threshold values are compared, judge whether early warning, and by its behavior record in database: should early warning if be determined as, then judge whether be provided with warning function in data further, if be provided with warning function, then send early warning, if do not arrange warning function, then do not send early warning.

2. a kind of object real-time tracking method for early warning based on smart machine according to claim 1, is characterized in that: the first step is in customer location judge module, and the customer location vector of formation is $A=\left[\begin{array}{c}{x}_1,x_2...x_n\\ y_1,y_2...y_n\end{array}\right],$ According to the user velocity matrix that this vector calculation goes out be $A^{\&prime;}=\left[\begin{array}{c}{x^{\&prime;}}_1,{x^{\&prime;}}_2...{x^{\&prime;}}_{n-1}\\ {y^{\&prime;}}_1,{y^{\&prime;}}_2...{y^{\&prime;}}_{n-1}\end{array}\right],$ Wherein to each group column vector $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]$ Computing formula be $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]=\left[\begin{array}{c}{x^2}_{i+1}\\ {y^2}_{i+1}\end{array}\right]-\left[\begin{array}{c}{x^2}_i\\ {y^2}_i\end{array}\right];$ Pre-set threshold value T is utilized to the process that user velocity matrix carries out binary conversion treatment to be: if x' _i< T, then make x' _i=0, if y' _iduring < T, then make y' _i=0, if the user velocity matrix after process meets A' ≠ 0, be then judged as that user just passes by this place, then do not process especially, if meet A'=0, be then judged as that user is in the stop of this region and by the column vector of user's latest position $\left[\begin{array}{c}{x}_n\\ y_n\end{array}\right]$ Output to user's behavior prediction module, user behavior is predicted.

3. a kind of object real-time tracking method for early warning based on smart machine according to claim 1, it is characterized in that: second step is in user's behavior prediction module, from database, the facilities information in its periphery radius is R border circular areas is obtained according to the latest position of user, the statistics species number m of peripheral facility and the number k of often kind of facility, wherein the number of m kind is denoted as k _m; Set up matrix to every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often row first element is the probability that user enters this concrete facility's objectives, passes through formula determine, if often go other element positive correlation elements below, then its computing formula is: if negative correlation element, then its computing formula is: wherein, I _krepresent that a kth target of data-base recording is by the total degree entered, N _krepresent that in database, a kth target is not by the total degree entered, D _irepresent the weighted data preset; Then formula is passed through calculate the probability-weighted of each concrete facility's objectives in every class facility, and get its maximum probability-weighted wp _j, and record the position of j, and make vectorial op _i=wp _j, wherein, op _irepresent that user enters the probability of the i-th class facility generation under global context; After matrix computations to all kinds of facility, the behavior prediction probability vector op of this user can be obtained.

4. a kind of object real-time tracking method for early warning based on smart machine according to claim 3, is characterized in that: second step is in user's behavior prediction module, and the behavior prediction probability vector op of user is by formula p _i=(1-op ₁) (1-op ₂) ... (1-op _i-1) op _i(1-op _i+1) (1-op _m) process, obtain output vector P, before then outputting to early-warning judgment module.

5. a kind of object real-time tracking method for early warning based on smart machine according to claim 1,3 or 4, is characterized in that: the 3rd step, in early-warning judgment module, before differentiation, reads presetting information vector S, if s _i=0, illustrate that the i-th class facility does not set warning function, if s _i=1, illustrate that the i-th class facility sets warning function; If s _i=0 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=0 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, should early warning be sent but not send early warning; If s _i=1 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=1 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, send early warning.

6. a kind of object real-time tracking method for early warning based on smart machine according to claim 5, it is characterized in that: the 3rd step is in early-warning judgment module, if differentiate, result is not for should send early warning, then in a database all facilities of the i-th class facility always not being entered number of times increases by 1 time; If differentiate, result is for should send early warning, then increase by 1 time by the number of times that always enters of facility maximum for probability-weighted under the i-th class facility in a database, always not entering number of times and increase by 1 time all the other all facilities in region.

7., based on an object real-time tracking early warning system for smart machine, it is characterized in that comprising customer location judge module, user's behavior prediction module and early-warning judgment module, the function of each module is as follows:

The function of customer location judge module: obtain the position data of user within a period of time using and there is the smart machine of positioning function and form customer location vector, calculate the user velocity matrix of this customer location vector and binary conversion treatment is carried out to it; If the user velocity matrix value after process is not 0, then judge that user just passes by, then do not process especially; If the user velocity matrix value after process is 0, then judge that user stops in this region and user's latest position is outputted to user's behavior prediction module;

The function of user's behavior prediction module: obtain the facilities information in user neighboring area from database, matrix is set up for every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often in row, the probability that enter this concrete facility's objectives of first element for calculating according to data with existing in database, after other elements be the weighting weight of each influence factor; Calculate the probability-weighted of each concrete facility's objectives and get its maximal value enters such facility prediction probability as this user; After matrix computations to every class facility, the behavior prediction probability vector of user can be obtained and be processed into output vector, then outputting to early-warning judgment module;

The function of early-warning judgment module: each element of output vector and corresponding threshold values are compared, judge whether early warning, and by its behavior record in database: should early warning if be determined as, then judge whether be provided with warning function in data further, if be provided with warning function, then send early warning, if do not arrange warning function, then do not send early warning.

8. a kind of object real-time tracking early warning system based on smart machine according to claim 7, it is characterized in that: in customer location judge module, the customer location vector of formation is $A=\left[\begin{array}{c}{x}_1,x_2...x_n\\ y_1,y_2...y_n\end{array}\right],$ According to the user velocity matrix that this vector calculation goes out be $A^{\&prime;}=\left[\begin{array}{c}{x^{\&prime;}}_1,{x^{\&prime;}}_2...{x^{\&prime;}}_{n-1}\\ {y^{\&prime;}}_1,{y^{\&prime;}}_2...{y^{\&prime;}}_{n-1}\end{array}\right],$ Wherein to each group column vector $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]$ Computing formula be $\left[\begin{array}{c}{x^{\&prime;}}_i\\ {y^{\&prime;}}_i\end{array}\right]=\left[\begin{array}{c}{x^2}_{i+1}\\ {y^2}_{i+1}\end{array}\right]-\left[\begin{array}{c}{x^2}_i\\ {y^2}_i\end{array}\right];$ Pre-set threshold value T is utilized to the process that user velocity matrix carries out binary conversion treatment to be: if x' _i< T, then make x' _i=0, if y' _iduring < T, then make y' _i=0, if the user velocity matrix after process meets A' ≠ 0, be then judged as that user just passes by this place, then do not process especially, if meet A'=0, be then judged as that user is in the stop of this region and by the column vector of user's latest position $\left[\begin{array}{c}{x}_n\\ y_n\end{array}\right]$ Output to user's behavior prediction module, user behavior is predicted.

9. a kind of object real-time tracking early warning system based on smart machine according to claim 7, it is characterized in that: in user's behavior prediction module, from database, the facilities information in its periphery radius is R border circular areas is obtained according to the latest position of user, the statistics species number m of peripheral facility and the number k of often kind of facility, wherein the number of m kind is denoted as k _m; Set up matrix to every class facility, in this matrix, the often information of a capable expression concrete facility's objectives, often row first element is the probability that user enters this concrete facility's objectives, passes through formula determine, if often go other element positive correlation elements below, then its computing formula is: if negative correlation element, then its computing formula is: wherein, I _krepresent that a kth target of data-base recording is by the total degree entered, N _krepresent that in database, a kth target is not by the total degree entered, D _irepresent the weighted data preset; Then formula is passed through calculate the probability-weighted of each concrete facility's objectives in every class facility, and get its maximum probability-weighted wp _j, and record the position of j, and make vectorial op _i=wp _j, wherein, op _irepresent that user enters the probability of the i-th class facility generation under global context; After matrix computations to all kinds of facility, can obtain the behavior prediction probability vector op of this user, the behavior prediction probability vector op of user is by formula p _i=(1-op ₁) (1-op ₂) ... (1-op _i-1) op _i(1-op _i+1) (1-op _m) process, obtain output vector P, before then outputting to early-warning judgment module.

10. a kind of object real-time tracking early warning system based on smart machine according to claim 7, is characterized in that: in early-warning judgment module, before differentiation, reads presetting information vector S, if s _i=0, illustrate that the i-th class facility does not set warning function, if s _i=1, illustrate that the i-th class facility sets warning function; If s _i=0 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=0 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, should early warning be sent but not send early warning; If s _i=1 and P _i< T _i, illustrate that target enters the probability of the i-th class facility less, then user behavior be recorded in database, should not send early warning; If s _i=1 and P _i>=T _i, illustrate that target enters the probability of the i-th class facility comparatively greatly, then user behavior be recorded in database, send early warning; If differentiate, result is not for should send early warning, then in a database all facilities of the i-th class facility always not being entered number of times increases by 1 time; If differentiate, result is for should send early warning, then increase by 1 time by the number of times that always enters of facility maximum for probability-weighted under the i-th class facility in a database, always not entering number of times and increase by 1 time all the other all facilities in region.