CN112422650B - Building positioning method, building positioning device, building positioning equipment and computer readable storage medium - Google Patents
Building positioning method, building positioning device, building positioning equipment and computer readable storage medium Download PDFInfo
- Publication number
- CN112422650B CN112422650B CN202011227572.5A CN202011227572A CN112422650B CN 112422650 B CN112422650 B CN 112422650B CN 202011227572 A CN202011227572 A CN 202011227572A CN 112422650 B CN112422650 B CN 112422650B
- Authority
- CN
- China
- Prior art keywords
- building
- data
- user
- eci
- target
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/52—Network services specially adapted for the location of the user terminal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Navigation (AREA)
Abstract
The invention discloses a building positioning method, a device, equipment and a computer readable storage medium, wherein the building positioning method comprises the following steps: acquiring basic position data of a full amount of users, and acquiring rough position data of the full amount of users based on the basic position data; obtaining static state data when the user states of all users in the rough position data are static states, and associating the static state data with a preset building user sample to obtain a learning sample set; performing model training on all ECIs in the learning sample set to obtain building positioning decision models corresponding to the ECIs; and determining a target building positioning decision model corresponding to a target user in the total number of users based on each building positioning decision model, and inputting the basic position data of the target user into the target building positioning decision model for training to obtain a building positioning result of the target user. Thereby improving the accuracy of the user's location to the building.
Description
Technical Field
The present invention relates to the field of communications network technologies, and in particular, to a building positioning method, apparatus, device, and computer-readable storage medium.
Background
The current method for positioning a user to a building mainly includes a positioning method based on terminal position information Measurement and a fingerprint database positioning method based on MR (Measurement Report) data. The positioning method based on terminal position information measurement is to initiate a positioning request through a terminal side, and a position service provider performs network operation to position a user to a building. The fingerprint database positioning method based on the MR data is characterized in that wireless signal simulation is carried out by adopting a ray tracing model through base station cell configuration of a communication operator in combination with a three-dimensional map, a three-dimensional coverage fingerprint database of a whole network building is constructed, and then fingerprint matching is carried out on the MR of a user according to a fingerprint positioning model, so that the user is positioned to the building. Therefore, how to improve the accuracy of positioning the user to the building becomes a technical problem to be solved urgently at present.
Disclosure of Invention
The invention mainly aims to provide a building positioning method, a building positioning device, building positioning equipment and a computer readable storage medium, and aims to solve the technical problem of improving the accuracy of positioning a user to a building.
In order to achieve the above object, the present invention provides a building positioning method, comprising the steps of:
acquiring basic position data of a full amount of users, and acquiring rough position data of the full amount of users based on the basic position data;
obtaining static state data when the user states of the full amount of users in the rough position data are static states, and associating the static state data with a preset building user sample to obtain a learning sample set;
performing model training on all ECIs in the learning sample set to obtain building positioning decision models corresponding to the ECIs;
and determining a target building positioning decision model corresponding to a target user in the total number of users based on each building positioning decision model, and inputting the basic position data of the target user into the target building positioning decision model for training to obtain a building positioning result of the target user.
Optionally, the step of performing model training on all ECIs in the learning sample set to obtain a building positioning decision model corresponding to each ECI includes:
constructing a feature vector for all ECIs in the learning sample set to obtain a feature sequence sample corresponding to each ECI;
and traversing each ECI, and performing model training according to the characteristic sequence sample corresponding to the traversed ECI to obtain a building positioning decision model corresponding to the traversed ECI.
Optionally, the step of performing model training according to the feature sequence sample corresponding to the traversed ECI to obtain the building positioning decision model corresponding to the traversed ECI includes:
acquiring the number of samples and all features in the feature sequence sample corresponding to the traversed ECI, and if the number of samples is greater than a first preset threshold, calculating the Gini coefficient of each feature according to a preset Gini coefficient calculation formula and the number of samples;
obtaining a minimum kini coefficient with the minimum value in each kini coefficient, determining an optimal feature corresponding to the minimum kini coefficient in each feature, and dividing a feature sequence sample corresponding to the traversed ECI according to the optimal feature to obtain a left node data set and a right node data set;
and constructing a building positioning decision model corresponding to the traversed ECI based on the left node data set and the right node data set.
Optionally, the step of calculating the kini coefficient of each feature according to a preset kini coefficient calculation formula and the number of samples includes:
calculating a target kini coefficient of the characteristic sequence sample corresponding to the traversed ECI according to a preset kini coefficient calculation formula and the sample number;
and if the target damping coefficient is larger than a second preset threshold value, obtaining the characteristic value of each characteristic, and calculating the damping coefficient of each characteristic according to each characteristic value and the damping coefficient calculation formula.
Optionally, the step of constructing a feature vector for all ECIs in the learning sample set to obtain a feature sequence sample corresponding to each ECI includes:
sequentially traversing all ECIs in the learning sample set, determining all adjacent cells corresponding to the traversed ECIs, acquiring a preset number of target adjacent cells in each adjacent cell, and determining an adjacent cell identifier of each target adjacent cell;
and sequencing each adjacent cell identifier according to a preset sequencing rule, and determining a characteristic sequence sample corresponding to the traversed ECI based on the sequencing result of the sequencing.
Optionally, the step of associating the static state data with a preset building user sample to obtain a learning sample set includes:
performing information association on the static state data and a preset building user sample, and acquiring an associated data set based on an associated result;
and carrying out data cleaning on the associated data set based on a preset threshold value to obtain a learning sample set.
Optionally, the step of performing data cleaning on the associated data set based on a preset threshold value to obtain a learning sample set includes:
traversing all buildings in the associated data set, determining ECIs occupied by all users in the traversed buildings, and calculating absolute differences between position data of the ECIs and position data of the traversed buildings;
and if the target absolute difference value is larger than a preset threshold value in each absolute difference value, performing data cleaning on the ECI corresponding to the target absolute difference value, and taking the associated data set subjected to data cleaning as a learning sample set.
In addition, to achieve the above object, the present invention provides a building positioning device, including:
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring basic position data of a full amount of users and acquiring rough position data of the full amount of users based on the basic position data;
the acquisition module is used for acquiring static state data when the user states of the full amount of users in the rough position data are static states, and associating the static state data with a preset building user sample to acquire a learning sample set;
the training module is used for carrying out model training on all ECIs in the learning sample set so as to obtain building positioning decision models corresponding to the ECIs;
and the input module is used for determining a target building positioning decision model corresponding to a target user in the full users based on each building positioning decision model, and inputting the basic position data of the target user into the target building positioning decision model for training so as to obtain a building positioning result of the target user.
Further, to achieve the above object, the present invention also provides a building positioning apparatus including: a memory, a processor and a building positioning program stored on the memory and executable on the processor, the building positioning program when executed by the processor implementing the steps of the building positioning method as described above.
Furthermore, to achieve the above object, the present invention also provides a computer readable storage medium having stored thereon a building positioning program, which when executed by a processor implements the steps of the building positioning method as described above.
Acquiring basic position data of a full amount of users, and acquiring rough position data of the full amount of users based on the basic position data; obtaining static state data when the user states of the full amount of users in the rough position data are static states, and associating the static state data with a preset building user sample to obtain a learning sample set; performing model training on all ECIs in the learning sample set to obtain building positioning decision models corresponding to the ECIs; and determining a target building positioning decision model corresponding to a target user in the total number of users based on each building positioning decision model, and inputting the basic position data of the target user into the target building positioning decision model for training to obtain a building positioning result of the target user. The method comprises the steps of acquiring rough position data by collecting basic position data of all users, acquiring a learning sample set according to static state data in the rough position data and building user samples, acquiring building positioning decision models corresponding to all ECIs in the learning sample set, inputting basic position data of a target user into a target building positioning decision model for training to acquire a building positioning result, and therefore the phenomenon that positioning results of a positioning method adopting terminal position information measurement and a fingerprint library positioning method based on MR data in the prior art are low in accuracy is avoided.
Drawings
FIG. 1 is a schematic diagram of a building location device architecture for a hardware operating environment in accordance with an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a building location method according to a first embodiment of the present invention;
FIG. 3 is a schematic view of a device module of the building locating device of the present invention;
fig. 4 is a schematic flow chart of the building positioning method of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As shown in fig. 1, fig. 1 is a schematic structural diagram of a building location device in a hardware operating environment according to an embodiment of the present invention.
The building positioning device in the embodiment of the invention can be a terminal device such as a PC or a server (such as an X86 server) which is provided with a virtualization platform.
As shown in fig. 1, the building location device may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, a communication bus 1002. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a non-volatile memory (e.g., a magnetic disk memory). The memory 1005 may alternatively be a storage device separate from the processor 1001.
Those skilled in the art will appreciate that the terminal structure shown in fig. 1 is not intended to be limiting and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
As shown in fig. 1, the memory 1005, which is a type of computer storage medium, may include an operating system, a network communication module, a user interface module, and a building location program therein.
In the terminal shown in fig. 1, the network interface 1004 is mainly used for connecting to a backend server and performing data communication with the backend server; the user interface 1003 is mainly used for connecting a client (user side) and performing data communication with the client; and the processor 1001 may be configured to invoke the building location program stored in the memory 1005 and perform the operations of the privilege configuration method embodiments of the security component below.
Based on the hardware structure, the embodiment of the building positioning method is provided as follows.
Referring to fig. 2, fig. 2 is a schematic flow chart of a first embodiment of a data positioning method of the present invention, where the data positioning method includes:
step S10, collecting basic position data of a full amount of users, and acquiring rough position data of the full amount of users based on the basic position data;
in this embodiment, in order to avoid the positioning method based on terminal location information Measurement and the fingerprint library positioning method based on MR (Measurement Report) data in the prior art, a building accurate positioning method based on a decision tree, that is, a data positioning method, is proposed, which does not require to install special software for calculation on a user side or to be in a special environment, and can identify a home building of a user according to network side data as long as the user terminal is in a mobile communication network connection state. Meanwhile, the method does not depend on network working parameter data, does not depend on building materials and internal structures, and can effectively guarantee the accuracy of building positioning data.
In this embodiment, the configuration data, the network parameter data, the XDR (detailed record) data acquired by the network side, the MR data, and the building base map layer data are used to perform classification organization and correlation operation on various data, perform building positioning model modeling training with the base station cell as the dimension, form a decision tree-based building accurate positioning model, and perform building positioning on users in the whole network based on the model. For example, as shown in fig. 4, the data localization method includes 100, full-user MR and XDR acquisition correlation; 200, extracting user stationing state data; 300, acquiring a building user sample; 400, extracting a learning sample; 500, building decision model construction and training; 600, full user building location.
Therefore, when building location is performed, basic location data of a full amount of users needs to be collected first, where the basic location data includes MR data and XDR data of the full amount of users, that is, MR data of a communication carrier is collected, XDR data of S1-MMR is collected (a signaling monitoring system is established by a mobile communication carrier, and control plane and user plane signaling when users use are collected from an S1 interface of an LTE (Long-Term Evolution, wireless network) network core network to form XDR data), and the MR data and the XDR data are correlated to form user rough location information including user identification and user MR characteristics, that is, rough location data. And the coarse location data includes Time, MSISDN (telephone number), ECI (network cell identity), home service cell RSRP (reference signal received power), home service cell RSRQ (reference signal received quality), AOA (Angle of Arrival), TA (Time Advance), neighbor 1 identity, neighbor 1RSRP, neighbor 2 identity, neighbor 2RSRP, neighbor 3 identity, neighbor 3RSRP, neighbor 4 identity, neighbor 4RSRP, neighbor 5 identity, neighbor 5RSRP, neighbor 6 identity, neighbor 6RSRP, and other auxiliary information.
Step S20, obtaining static state data when the user states of the full amount of users in the rough position data are static states, and associating the static state data with a preset building user sample to obtain a learning sample set;
after the rough position data is acquired, user states of all users can be identified, wherein the user states comprise static states, namely stagnation point states and road motion states, data of the user states in the static states, namely stagnation point state data, are extracted from the rough position, and the extracted stagnation point state data are used as static state data.
The preset building user sample is obtained, and because the home wide configuration data comprises the MSISDN (telephone number) of the home wide user, the residential community where the user is located and the building information, the configuration information of the home wide user can be associated with the building map layer in advance to form the building user sample. The building user sample comprises a building ID (Identity), a building longitude and latitude, an MSISDN (Mobile station identifier) and other auxiliary information.
And then, associating the static state data with a building user sample to obtain the MR characteristics of the sample user in the building to form a learning sample set for building positioning, wherein the learning sample set can comprise time, building ID, MSISDN, ECI, main service cell RSRP, main service cell RSRQ, AOA, TA, neighbor 1 identifier, neighbor 1RSRP, neighbor 2 identifier, neighbor 2RSRP, neighbor 3 identifier, neighbor 3RSRP, neighbor 4 identifier, neighbor 4RSRP, neighbor 5 identifier, neighbor 5RSRP, neighbor 6 identifier, neighbor 6RSRP and other auxiliary information.
Step S30, performing model training on all ECIs in the learning sample set to obtain building positioning decision models corresponding to the ECIs;
after the learning sample set is obtained, a building positioning decision model can be constructed by taking the ECI as an object, namely, model training is carried out on all ECIs in the learning sample set to obtain building positioning decision models of building positioning corresponding to all ECIs. The building positioning decision model corresponding to the ECI may be obtained by first constructing a feature vector for each ECI to obtain a feature sequence sample corresponding to each ECI. And because the adjacent regions corresponding to each ECI are different, the characteristics of the decision tree need to be independently constructed.
After the feature sequence samples corresponding to each ECI are constructed, decision tree training can be performed on each ECI to obtain each building positioning decision model. That is, when a decision tree is constructed for a certain ECI, the feature sequence sample corresponding to the ECI may be used as the sample set D. In the embodiment, the decision tree is a mature CART (Classification and Regression Trees) Classification tree model, each ECI corresponds to a CART Classification tree, and for the construction of the CART Classification tree, the algorithm starts from the root node and recursively establishes the CART tree according to the sample set D.
Therefore, after the sample set D is obtained, the root node in the sample set D may be selected as the current node, and it is detected whether the number of samples in the sample set D is smaller than the first preset threshold, if the number of samples is smaller than the first preset threshold, or there is no feature, the decision tree is returned, and the current node stops recursion. And if the number of the samples is greater than a first preset threshold value, calculating a target damping coefficient of the sample set according to a preset damping coefficient calculation formula, detecting whether the target damping coefficient is less than a second preset threshold value, returning to the decision tree, and stopping recursion of the current node. Wherein the equation for calculating the kini coefficient of the sample set D may be
k represents the number of sample classes in the sample set D, and the number of k-th classes is ck。
If the target kini coefficient is larger than or equal to a second preset threshold value, calculating the current characteristic value of each characteristic of the current node to the kini coefficient of the data set, namely the current characteristic value of each sample type. And selecting the feature a (such as serving cell RSRP, neighboring cell RSRP, TA, AOA, etc.) with the smallest kini coefficient and the corresponding feature value a (specific feature value) from the calculated kini coefficients of each feature value pair data set D of each feature. According to the optimal characteristic and the optimal characteristic value, a data set is divided into two parts, namely D1 and D2, the left node and the right node of the current node are simultaneously established, the data set D of the left node is D1, the data set D of the right node is D2, and then recursive operation is carried out on the left node and the right node, namely, decision tree training is continuously carried out, so that a building positioning decision model corresponding to the ECI is obtained.
Wherein, the calculation formula of the kini coefficient of the sample set D under the condition of the characteristic A is as follows:
step S40, determining a target building positioning decision model corresponding to a target user in the total number of users based on each building positioning decision model, and inputting the basic position data of the target user into the target building positioning decision model for training to obtain a building positioning result of the target user.
In this embodiment, after obtaining each building location decision model, building location may be performed on each full-scale user according to each building location decision model. That is, a target user of the total number of users may be determined, basic position data corresponding to the target user, such as the target rough position data, may be determined, a target building positioning decision model may be determined in each building positioning decision model according to the target rough position data, and then the basic position data of the target user may be input to the target building positioning decision model for training to obtain the building positioning result of the target user.
In the embodiment, a building location decision model is constructed by taking ECIs (cell ECIs) of a base station as objects, namely, each ECI corresponds to one location decision model, characteristic conversion is carried out on learning samples, the models are trained and modeled, and building attribution location based on a decision tree is carried out on an MR (magnetic resonance) of a user, so that the problem of accurately locating the user to a building is effectively solved. And when the learning sample is obtained, the accurate learning sample for the MR characteristics of the building is extracted according to the home wide configuration data of the communication operator and the characteristic of the location of the stagnation point of the user, so that the problem of collecting the learning sample for the building decision tree positioning model is solved with low cost and high quality.
In this embodiment, by collecting basic location data of a full amount of users, rough location data of the full amount of users is obtained based on the basic location data; obtaining static state data when the user states of the full amount of users in the rough position data are static states, and associating the static state data with a preset building user sample to obtain a learning sample set; performing model training on all ECIs in the learning sample set to obtain building positioning decision models corresponding to the ECIs; and determining a target building positioning decision model corresponding to a target user in the total number of users based on each building positioning decision model, and inputting the basic position data of the target user into the target building positioning decision model for training to obtain a building positioning result of the target user. The method comprises the steps of acquiring rough position data by collecting basic position data of all users, acquiring a learning sample set according to static state data in the rough position data and building user samples, acquiring building positioning decision models corresponding to all ECIs in the learning sample set, inputting basic position data of a target user into a target building positioning decision model for training to acquire a building positioning result, and therefore the phenomenon that positioning results of a positioning method adopting terminal position information measurement and a fingerprint library positioning method based on MR data in the prior art are low in accuracy is avoided.
Further, based on the first embodiment of the present invention, a second embodiment of the building location method of the present invention is provided, in this embodiment, in step S30 in the above embodiment, the step of performing model training on all ECIs in the learning sample set to obtain a building location decision model corresponding to each ECI includes:
step a, constructing a feature vector for all ECIs in the learning sample set to obtain a feature sequence sample corresponding to each ECI;
in this embodiment, after a learning sample set is obtained, feature vector construction may be performed on all ECIs in the learning sample set, and the same feature vector construction manner is adopted for each ECI, that is, the number of occupied neighbor cells is counted according to the ECI to form a data set, where the data set includes ECI-neighbor cell identifier-sample number, and then sorting is performed according to the sample number, to obtain neighbor cell identifiers of a preset number (any number, e.g., 50, set in advance by a user), and sort the neighbor cell identifiers, where the sorting manner may be sorting according to names to form a fixed field sequence, and perform feature conversion according to the fixed field sequence. And for each specific ECI, after a preset number of adjacent cells are obtained through statistics, the converted characteristic field is fixed, and the learning sample is converted into a characteristic sequence according to the adjacent cell identifier to obtain a characteristic sequence sample corresponding to each ECI. For example, the neighbor a of the sample S1 corresponds to neighbor 2, the neighbor B corresponds to the feature sequence neighbor 8, and if the neighbor RSRP of the feature sequence is not present in the sample, NULL is assigned.
And b, traversing each ECI, and performing model training according to the characteristic sequence sample corresponding to the traversed ECI to obtain a building positioning decision model corresponding to the traversed ECI.
After the feature sequence samples corresponding to the ECIs are obtained, the ECIs can be traversed, model training, namely decision tree training, is carried out according to the feature sequence samples corresponding to the traversed ECIs, and the building positioning decision model corresponding to the traversed ECIs is obtained. The method comprises the steps that a mature CART classification tree model is selected for training a decision tree, each ECI corresponds to one CART classification tree, and for the construction of the CART classification trees, an algorithm starts from a root node and establishes the CART trees according to a sample set D recursion.
In this embodiment, the feature sequence samples of all the ECIs in the learning sample set are constructed, then all the ECIs are traversed, and model training is performed according to the feature sequence samples corresponding to the traversed ECIs to obtain the building positioning decision model corresponding to the traversed ECIs, so that the effectiveness of the obtained building positioning decision model is ensured.
Specifically, the step of performing model training according to the feature sequence sample corresponding to the traversed ECI to obtain the building positioning decision model corresponding to the traversed ECI includes:
step c, acquiring the number of samples and all the characteristics in the characteristic sequence sample corresponding to the traversed ECI, and if the number of samples is greater than a first preset threshold value, calculating the Gini coefficient of each characteristic according to a preset Gini coefficient calculation formula and the number of samples;
in this embodiment, when performing model training on the traversed ECI, the number of samples and all features in the feature sequence sample corresponding to the traversed ECI may be obtained first, and it is detected whether the number of samples is greater than a first preset threshold (any threshold set in advance by a user), if the number of samples is less than the first preset threshold, or there is no feature, the decision tree is returned, and the current node stops recursion. If the number of the samples is larger than a first preset threshold, calculating a target kini coefficient of the sample set according to a preset kini coefficient calculation formula, detecting whether the target kini coefficient is smaller than a second preset threshold (any threshold set in advance by a user can be the same as or different from the first preset threshold), and if so, returning to the decision tree and stopping recursion of the current node. Wherein the equation for calculating the kini coefficient of the sample set D may be
k represents the number of sample classes in the sample set D, and the number of k-th classes is Ck。
And if the target kini coefficient is larger than or equal to a second preset threshold value, calculating the current characteristic value of each characteristic of the current node and the kini coefficient of the data set.
Step d, obtaining the minimum kini coefficient with the minimum value in the kini coefficients, determining the optimal feature corresponding to the minimum kini coefficient in the features, and dividing the feature sequence samples corresponding to the traversed ECI according to the optimal feature to obtain a left node data set and a right node data set;
and e, building a building positioning decision model corresponding to the traversed ECI based on the left node data set and the right node data set.
Among the calculated kini coefficients of each feature value pair data set D of each feature, a feature a (such as serving cell RSRP, neighbor cell RSRP, TA, AOA, etc.) and a corresponding feature value a (specific feature value) with the smallest kini coefficient are selected. According to the optimal characteristic and the optimal characteristic value, a data set is divided into two parts, namely D1 and D2, a left node and a right node of the current node are simultaneously established, the data set D of the left node is D1 (namely a left node data set), the data set D of the right node is D2 (namely a right node data set), and then recursive operation is carried out on the left node and the right node, namely decision tree training is continuously carried out, so that a building positioning decision model corresponding to the ECI is obtained.
Wherein, the calculation formula of the kini coefficient of the sample set D under the condition of the characteristic A is as follows:
in this embodiment, when the number of samples in the feature sequence samples corresponding to the traversed ECI is greater than a first preset threshold, the kini coefficients of the features are calculated, the optimal feature corresponding to the minimum kini coefficient is determined, and the feature sequence samples corresponding to the traversed ECI are divided according to the optimal feature to obtain a left node data set and a right node data set, so that a building positioning decision model corresponding to the traversed ECI is constructed, and the effectiveness of the constructed building positioning decision model is ensured.
Specifically, the step of calculating the kini coefficient of each feature according to a preset kini coefficient calculation formula and the number of samples includes:
step f, calculating the target kini coefficient of the characteristic sequence sample corresponding to the traversed ECI according to a preset kini coefficient calculation formula and the sample number;
in this embodiment, a kini coefficient calculation formula set in advance may be obtained first, and then the target kini coefficient of the feature sequence sample corresponding to the traversed ECI may be calculated according to the sample number k. For example, assume that there are K classes, the number of kth classes being CkThen the expression for the kini coefficient of sample D is:
and g, if the target damping coefficient is larger than a second preset threshold value, obtaining a characteristic value of each characteristic, and calculating the damping coefficient of each characteristic according to each characteristic value and the damping coefficient calculation formula.
And after the target damping coefficient is obtained through calculation, detecting whether the target damping coefficient is larger than a second preset threshold, if the target damping coefficient is smaller than the second preset threshold, returning to the decision tree, and stopping recursion of the current node. And if the target kini coefficient is larger than a second preset threshold value, calculating the current characteristic value of each characteristic of the current node and the kini coefficient of the data set.
For example, for sample D, if D is divided into two parts, D1 and D2, according to a certain value a of feature a, the expression of the kini coefficient of D under the condition of feature a is:
in this embodiment, the target damping coefficient is calculated according to the damping coefficient calculation formula and the number of samples, and when the target damping coefficient is greater than the second preset threshold, the damping coefficient of each feature is calculated, so that a basis is provided for subsequent building positioning.
Further, the step of constructing a feature vector for all the ECIs in the learning sample set to obtain a feature sequence sample corresponding to each ECI includes:
step h, sequentially traversing all ECIs in the learning sample set, determining all adjacent cells corresponding to the traversed ECIs, acquiring a preset number of target adjacent cells in each adjacent cell, and determining an adjacent cell identifier of each target adjacent cell;
in this embodiment, when feature vectors are constructed for all the ECIs in the learning sample set, all the ECIs in the learning sample set may be sequentially traversed, and all the neighboring cells corresponding to the traversed ECI are determined, that is, all the neighboring cells are determined according to the number of times that the neighboring cells are occupied by the ECI statistics, so as to form a corresponding data set, where the data set includes the ECI, the neighboring cell identifier, and the sample number. Then, a preset number (any number preset by the user, for example, 50) of target neighboring cells are obtained from the data set, and the neighboring cell identifier of the target neighboring cell in the data set is determined.
And k, sequencing the identifiers of the adjacent cells according to a preset sequencing rule, and determining a characteristic sequence sample corresponding to the traversed ECI based on the sequencing result of the sequencing.
And then, sequencing the identifiers of the adjacent cells according to a preset sequencing rule, for example, sequencing according to the names of the adjacent cells to form a fixed field sequence, wherein the field sequence can be used as a characteristic sequence sample corresponding to the traversed ECI. For example, if the sample of the traversed ECI is as shown in Table 1
TABLE 1
When the neighbor cell identifiers in table 1 are sorted according to a preset sorting rule and subjected to feature conversion, a feature sequence sample can be obtained, as shown in table 2.
TABLE 1
In this embodiment, all the ECIs in the learning sample set are traversed, a preset number of target neighboring cells are obtained from all the neighboring cells corresponding to the traversed ECI, the neighboring cell identifiers of the target neighboring cells are sorted according to a preset sorting rule, and the feature sequence sample corresponding to the traversed ECI is determined according to the basic sorting result, so that the accuracy of the obtained feature sequence sample corresponding to the traversed ECI is ensured.
Further, the step of associating the static state data with a preset building user sample to obtain a learning sample set includes:
m, performing information association on the static state data and a preset building user sample, and acquiring an associated data set based on an associated result;
in this embodiment, after the static state data is acquired and the preset building user sample is determined, information association may be performed to acquire an associated data set. The associated data set comprises time, building ID, MSISDN, ECI, service cell RSRP, service cell RSRQ, AOA, TA, neighbor 1 identifier, neighbor 1RSRP, neighbor 2 identifier, neighbor 2RSRP, neighbor 3 identifier, neighbor 3RSRP, neighbor 4 identifier, neighbor 4RSRP, neighbor 5 identifier, neighbor 5RSRP, neighbor 6 identifier, neighbor 6RSRP and other auxiliary information.
And n, performing data cleaning on the associated data set based on a preset threshold value to obtain a learning sample set.
After the associated data set is obtained, the ECI occupied by each user MSISDN can be counted according to the building ID and the MSISDN, the ECI which occupies the most specific building ID and MSISDN is identified, and the corresponding record is extracted to be used as training sample data. And then, extracting longitude and latitude POS _ ECI corresponding to the cell ECI and longitude and latitude POS _ Building corresponding to the Building ID in the training sample data, calculating the distance between the POS _ ECI and the POS _ Building, filtering records of which the distance exceeds a preset threshold value (any value set in advance by a user), and finally obtaining learning sample data, wherein the field content is the same as S _ 403. The distance threshold is distinguished according to the types of base station cells, wherein the distance threshold is 500 meters in urban areas, 800 meters in suburban areas and 3000 meters in rural areas. And finally, collecting and accumulating learning sample data to obtain a learning sample set.
In this embodiment, the static state data and the building user sample are associated with each other to obtain an associated data set, and the associated data set is subjected to data cleaning to obtain a learning sample set, so that the effectiveness of the obtained learning sample set is guaranteed.
Specifically, the step of performing data cleaning on the associated data set based on a preset threshold value to obtain a learning sample set includes:
step x, traversing all buildings in the associated data set, determining ECIs occupied by all users in the traversed buildings, and calculating absolute differences between the position data of the ECIs and the position data of the traversed buildings;
in this embodiment, when performing data cleansing on the associated data set, all building IDs in the associated data set may be traversed first, and the ECI occupied by all users in the traversed building may be determined. It should be noted that the same user may occupy multiple ECIs, and then obtain the location data of each ECI and the location data of the traversed Building, that is, the longitude and latitude POS _ ECI corresponding to each ECI and the longitude and latitude POS _ Building corresponding to the Building ID. An absolute difference between the location data for each ECI and the location data for the traversed building is calculated.
And step y, if the target absolute difference value is larger than a preset threshold value in each absolute difference value, performing data cleaning on the ECI corresponding to the target absolute difference value, and taking the associated data set subjected to data cleaning as a learning sample set.
After all the absolute difference values are obtained, each absolute difference value needs to be compared with a preset threshold value, if a target absolute difference value is larger than the preset threshold value in each absolute difference value, data cleaning can be performed on records related to the ECI corresponding to the target absolute difference value, all the remaining records related to the ECI are collected and accumulated, namely, a data-cleaned associated data set is obtained, and the data-cleaned associated data set learns a sample set at last night.
In this embodiment, all buildings in the association data set are traversed, an absolute difference between the position data of each ECI in the traversed buildings and the position data of the traversed buildings is calculated, when a target absolute difference larger than a preset threshold exists, data cleaning is performed on the ECI corresponding to the target absolute difference, and the association data set subjected to data cleaning is used as a learning sample set, so that the effectiveness of the acquired learning sample set is ensured.
Referring to fig. 3, the present invention further provides a building positioning device, in this embodiment, the building positioning device includes:
the acquisition module A10 is used for acquiring basic position data of a full amount of users and acquiring rough position data of the full amount of users based on the basic position data;
an obtaining module a20, configured to obtain static state data when the user states of the full number of users in the rough location data are static states, and associate the static state data with a preset building user sample to obtain a learning sample set;
the training module A30 is used for performing model training on all ECIs in the learning sample set to obtain building positioning decision models corresponding to the ECIs;
an input module a40, configured to determine, based on each building positioning decision model, a target building positioning decision model corresponding to a target user among the full users, and input basic position data of the target user into the target building positioning decision model for training to obtain a building positioning result of the target user.
Optionally, the training module a30 is configured to:
constructing a feature vector for all ECIs in the learning sample set to obtain a feature sequence sample corresponding to each ECI;
and traversing each ECI, and performing model training according to the characteristic sequence sample corresponding to the traversed ECI to obtain a building positioning decision model corresponding to the traversed ECI.
Optionally, the training module a30 is configured to:
acquiring the number of samples and all features in the feature sequence sample corresponding to the traversed ECI, and if the number of samples is greater than a first preset threshold, calculating the Gini coefficient of each feature according to a preset Gini coefficient calculation formula and the number of samples;
obtaining a minimum kini coefficient with the minimum value in each kini coefficient, determining an optimal feature corresponding to the minimum kini coefficient in each feature, and dividing a feature sequence sample corresponding to the traversed ECI according to the optimal feature to obtain a left node data set and a right node data set;
and constructing a building positioning decision model corresponding to the traversed ECI based on the left node data set and the right node data set.
Optionally, the training module a30 is configured to:
calculating a target kini coefficient of the characteristic sequence sample corresponding to the traversed ECI according to a preset kini coefficient calculation formula and the sample number;
and if the target damping coefficient is larger than a second preset threshold value, obtaining the characteristic value of each characteristic, and calculating the damping coefficient of each characteristic according to each characteristic value and the damping coefficient calculation formula.
Optionally, the training module a30 is configured to:
sequentially traversing all ECIs in the learning sample set, determining all adjacent cells corresponding to the traversed ECIs, acquiring a preset number of target adjacent cells in each adjacent cell, and determining an adjacent cell identifier of each target adjacent cell;
and sequencing each adjacent cell identifier according to a preset sequencing rule, and determining a characteristic sequence sample corresponding to the traversed ECI based on the sequencing result of the sequencing.
Optionally, the obtaining module a20 is configured to:
performing information association on the static state data and a preset building user sample, and acquiring an associated data set based on an associated result;
and carrying out data cleaning on the associated data set based on a preset threshold value to obtain a learning sample set.
Optionally, the obtaining module a20 is configured to:
traversing all buildings in the associated data set, determining ECIs occupied by all users in the traversed buildings, and calculating absolute differences between position data of the ECIs and position data of the traversed buildings;
and if the target absolute difference value is larger than a preset threshold value in each absolute difference value, performing data cleaning on the ECI corresponding to the target absolute difference value, and taking the associated data set subjected to data cleaning as a learning sample set.
The method for implementing the functional modules can refer to the embodiment of the building positioning method, and is not described herein again.
The present invention also provides a building positioning apparatus, including: a memory, a processor, a communication bus, and a building location program stored on the memory:
the communication bus is used for realizing connection communication between the processor and the memory;
the processor is used for executing the building positioning program to realize the steps of the embodiments of the building positioning method.
The invention also provides a computer readable storage medium.
The inventive computer readable storage medium has stored thereon a building positioning program which, when executed by a processor, carries out the steps of the building positioning method as described above.
The method implemented when the building positioning program executed on the processor is executed may refer to each embodiment of the building positioning method of the present invention, and details are not described herein.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) as described above and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. A building positioning method is characterized by comprising the following steps:
acquiring basic position data of a full amount of users, acquiring rough position data of the full amount of users based on the basic position data, wherein the basic position data comprises full amount of user MR data and XDR data, and associating the MR data and the XDR data to form rough position data containing user identification and user MR characteristics;
obtaining static state data when the user states of all users in the rough position data are static states, and associating the static state data with a preset building user sample to obtain a learning sample set, wherein configuration information of the home-wide users is associated with a building map layer to form a building user sample, and the building user sample comprises a building ID, building longitude and latitude and a telephone number;
performing model training on all ECIs in the learning sample set to obtain building positioning decision models corresponding to the ECIs;
and determining a target building positioning decision model corresponding to a target user in the total number of users based on each building positioning decision model, and inputting the basic position data of the target user into the target building positioning decision model for training to obtain a building positioning result of the target user.
2. The building positioning method of claim 1, wherein the step of performing model training on all ECIs in the learning sample set to obtain a building positioning decision model corresponding to each ECI comprises:
constructing a feature vector for all ECIs in the learning sample set to obtain a feature sequence sample corresponding to each ECI;
and traversing each ECI, and performing model training according to the characteristic sequence sample corresponding to the traversed ECI to obtain a building positioning decision model corresponding to the traversed ECI.
3. The building positioning method as claimed in claim 2, wherein said step of performing model training based on the feature sequence samples corresponding to the traversed ECI to obtain the building positioning decision model corresponding to the traversed ECI comprises:
acquiring the number of samples and all features in the feature sequence sample corresponding to the traversed ECI, and if the number of samples is greater than a first preset threshold, calculating the Gini coefficient of each feature according to a preset Gini coefficient calculation formula and the number of samples;
obtaining a minimum kini coefficient with the minimum value in each kini coefficient, determining an optimal feature corresponding to the minimum kini coefficient in each feature, and dividing a feature sequence sample corresponding to the traversed ECI according to the optimal feature to obtain a left node data set and a right node data set;
and constructing a building positioning decision model corresponding to the traversed ECI based on the left node data set and the right node data set.
4. The building positioning method according to claim 3, wherein the step of calculating the kini coefficient of each of the features based on a preset kini coefficient calculation formula and the number of samples includes:
calculating a target kini coefficient of the characteristic sequence sample corresponding to the traversed ECI according to a preset kini coefficient calculation formula and the sample number;
and if the target damping coefficient is larger than a second preset threshold value, obtaining the characteristic value of each characteristic, and calculating the damping coefficient of each characteristic according to each characteristic value and the damping coefficient calculation formula.
5. The building positioning method as claimed in claim 2, wherein said step of performing feature vector construction on all ECIs in said learning sample set to obtain a feature sequence sample corresponding to each ECI comprises:
sequentially traversing all ECIs in the learning sample set, determining all adjacent cells corresponding to the traversed ECIs, acquiring a preset number of target adjacent cells in each adjacent cell, and determining an adjacent cell identifier of each target adjacent cell;
and sequencing each adjacent cell identifier according to a preset sequencing rule, and determining a characteristic sequence sample corresponding to the traversed ECI based on the sequencing result of the sequencing.
6. The building location method of any one of claims 1-5, wherein the step of correlating the stationary state data with a preset building user sample to obtain a learning sample set comprises:
performing information association on the static state data and a preset building user sample, and acquiring an associated data set based on an associated result;
and carrying out data cleaning on the associated data set based on a preset threshold value to obtain a learning sample set.
7. The building positioning method of claim 6, wherein the step of performing data washing on the associated data set based on a preset threshold value to obtain a learning sample set comprises:
traversing all buildings in the associated data set, determining ECIs occupied by all users in the traversed buildings, and calculating absolute differences between position data of the ECIs and position data of the traversed buildings;
and if the target absolute difference value is larger than a preset threshold value in each absolute difference value, performing data cleaning on the ECI corresponding to the target absolute difference value, and taking the associated data set subjected to data cleaning as a learning sample set.
8. A building positioning device, the building positioning device comprising:
the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring basic position data of a full amount of users, acquiring rough position data of the full amount of users based on the basic position data, the basic position data comprises full amount of user MR data and XDR data, and associating the MR data and the XDR data to form rough position data containing user identification and user MR characteristics;
the acquisition module is used for acquiring static state data when the user states of all the users in the rough position data are static states, and associating the static state data with a preset building user sample to acquire a learning sample set, wherein configuration information of the home-wide user is associated with a building map layer to form the building user sample, and the building user sample comprises a building ID, building longitude and latitude and a telephone number;
the training module is used for carrying out model training on all ECIs in the learning sample set so as to obtain building positioning decision models corresponding to the ECIs;
and the input module is used for determining a target building positioning decision model corresponding to a target user in the full users based on each building positioning decision model, and inputting the basic position data of the target user into the target building positioning decision model for training so as to obtain a building positioning result of the target user.
9. A building location device, the building location device comprising: a memory, a processor and a building location program stored on the memory and executable on the processor, the building location program when executed by the processor implementing the steps of the building location method as claimed in any one of claims 1 to 7.
10. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a building positioning program, which when executed by a processor implements the steps of the building positioning method as claimed in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011227572.5A CN112422650B (en) | 2020-11-05 | 2020-11-05 | Building positioning method, building positioning device, building positioning equipment and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011227572.5A CN112422650B (en) | 2020-11-05 | 2020-11-05 | Building positioning method, building positioning device, building positioning equipment and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112422650A CN112422650A (en) | 2021-02-26 |
| CN112422650B true CN112422650B (en) | 2021-10-22 |
Family
ID=74827078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011227572.5A Active CN112422650B (en) | 2020-11-05 | 2020-11-05 | Building positioning method, building positioning device, building positioning equipment and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112422650B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115242677B (en) * | 2021-04-23 | 2023-09-01 | 中国移动通信集团四川有限公司 | Home-wide user state monitoring system, method and device |
| CN116709509B (en) * | 2023-08-09 | 2023-10-17 | 北京邮电大学 | User equipment sensing and positioning method, system, electronic equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106793067A (en) * | 2016-11-29 | 2017-05-31 | 上海斐讯数据通信技术有限公司 | A kind of many floor indoor orientation methods and server based on joint network |
| CN110012428A (en) * | 2019-05-22 | 2019-07-12 | 合肥工业大学 | A kind of indoor orientation method based on WiFi |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11276015B2 (en) * | 2017-04-20 | 2022-03-15 | Capital One Services, Llc | Machine learning artificial intelligence system for predicting hours of operation |
| CN108303093A (en) * | 2018-01-31 | 2018-07-20 | 中国联合网络通信集团有限公司 | A kind of indoor orientation method and device |
| CN109151750B (en) * | 2018-09-06 | 2020-11-27 | 哈尔滨工业大学 | LTE indoor positioning floor distinguishing method based on recurrent neural network model |
| CN110493720B (en) * | 2019-09-11 | 2021-02-19 | 深圳市名通科技股份有限公司 | Terminal positioning method and device and storage medium |
| CN111615048A (en) * | 2020-04-09 | 2020-09-01 | 汉海信息技术(上海)有限公司 | Positioning method, positioning device, electronic equipment and storage medium |
-
2020
- 2020-11-05 CN CN202011227572.5A patent/CN112422650B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106793067A (en) * | 2016-11-29 | 2017-05-31 | 上海斐讯数据通信技术有限公司 | A kind of many floor indoor orientation methods and server based on joint network |
| CN110012428A (en) * | 2019-05-22 | 2019-07-12 | 合肥工业大学 | A kind of indoor orientation method based on WiFi |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112422650A (en) | 2021-02-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105701123B (en) | The recognition methods of man-vehicle interface and device | |
| CN106790468B (en) | Distributed implementation method for analyzing WiFi (Wireless Fidelity) activity track rule of user | |
| CN112422650B (en) | Building positioning method, building positioning device, building positioning equipment and computer readable storage medium | |
| CN110337059B (en) | Analysis algorithm, server and network system for family relationship of user | |
| CN106646338A (en) | Rapidly accurate indoor location method | |
| CN111078818B (en) | Address analysis method and device, electronic equipment and storage medium | |
| CN109151890A (en) | A kind of mobile terminal locating method and device | |
| CN108834077B (en) | Tracking area division method and device based on user movement characteristics and electronic equipment | |
| CN107423742A (en) | The determination method and device of crowd's flow | |
| CN111294730B (en) | Method and device for processing network problem complaint information | |
| CN110298687B (en) | Regional attraction assessment method and device | |
| CN106817712B (en) | Positioning method and device and server | |
| CN108566620A (en) | A kind of indoor orientation method based on WIFI | |
| CN110727752B (en) | Position fingerprint database processing method, device and computer readable storage medium | |
| CN110348717B (en) | Base station value scoring method and device based on grid granularity | |
| CN113923706A (en) | Mobile network coverage quality evaluation method and device, electronic equipment and storage medium | |
| CN111368862A (en) | Method for distinguishing indoor and outdoor marks, training method and device of classifier and medium | |
| CN111475746B (en) | Point-of-interest mining method, device, computer equipment and storage medium | |
| CN113194474A (en) | Pseudo base station positioning method and device, electronic equipment and readable storage medium | |
| CN110992230A (en) | Full-scale demographic method, device and server based on terminal signaling data | |
| CN111343664A (en) | User positioning method, device, equipment and medium | |
| CN115623520A (en) | False positioning detection method and device and electronic equipment | |
| Shad et al. | Precise location acquisition of mobility data using cell-id | |
| CN111835541B (en) | Method, device, equipment and system for detecting aging of flow identification model | |
| CN109831737B (en) | Bluetooth positioning method, device, equipment and system based on confidence degree |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |