Abnormal base station detection and correction method and system based on multi-source data

Abstract

The invention discloses an abnormal base station detection and correction method and system based on multi-source data, and the method comprises the steps: obtaining signaling track data and heterogenous track data, carrying out the preprocessing of the data, constructing signaling terminal short-time switching sequence data, carrying out the statistics of the data, obtaining the short-time switching information of a base station, and constructing terminal-heterogenous matching pair data; performing multi-stage noise processing on the matching pair data to screen high-confidence matching pair data to obtain high-confidence matching pair co-location information and ectopic information data; calculating abnormal switching characteristic data of the base station based on the short-time switching information of the base station, and calculating abnormal ectopic characteristic data of the base station based on the ectopic information of the high-confidence matching pair; performing abnormal base station and high-confidence base station detection output based on combined features of the base station abnormal switching feature data and the base station abnormal ectopic feature data; and the high-confidence neighbor cell base station and the high-confidence neighbor cell different-source equipment are used for sequentially carrying out correction unit cleaning, correction unit weighting and joint weighted position correction, and position correction of the abnormal base station is completed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of abnormal base station detection and rectification, and more particularly to an abnormal base station detection and rectification method and system based on multi-source data. BACKGROUND

[0002] The signaling data generated by the user terminal attached to the base station is the cornerstone of communication big data application, and has important significance in the fields of municipal planning, economic construction, disaster analysis, population prediction, stability maintenance and technical reconnaissance.

[0003] In the application of signaling data, the abnormal position of the base station caused by the error of the input work parameter information has a great impact, and the error data generated by the abnormal base station presents the characteristics of high level, high influence and concealment.

[0004] At present, there are many algorithms for detecting abnormal base stations and calibration based on signaling data, including signaling switching relationship, switching angle, switching times, and adjacent base station polygon relationship information, using experience rules, traditional machine learning methods, and even deep learning methods to solve the problem.

[0005] However, the existing method only uses signaling data as the basis for detection and calibration, and the underlying technology lacks multi-source data verification, which has great detection difficulty in areas with frequent abnormal base stations and complex base station deployment, and high-precision detection and calibration becomes a difficulty.

[0006] Therefore, how to provide an abnormal base station automatic detection and rectification technology based on multi-source data and with higher robustness is a problem that needs to be solved by those skilled in the art. SUMMARY

[0007] Therefore, the present application provides an abnormal base station detection and rectification method and system based on multi-source data to solve some of the technical problems mentioned in the background art.

[0008] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0009] An abnormal base station detection and rectification method based on multi-source data, comprising the following steps:

[0010] S1. Obtain signaling trajectory data and heterogenous trajectory data and preprocess, construct signaling terminal short switching sequence data and obtain base station short switching information by statistics, and construct terminal-heterogeneous matching pair data;

[0011] S2. Multi-level noise processing is performed on the terminal-heterogeneous matching pair data to screen out high-confidence matching pair data, and high-confidence matching pair same-site information and different-site information data are obtained;

[0012] S3. Calculate base station abnormal switching feature data based on base station short-time switching information, and calculate base station abnormal abnormality feature data based on high-confidence matching pair abnormality information;

[0013] S4. Perform abnormal base station and high-confidence base station detection output based on the combined features of base station abnormal switching feature data and base station abnormal abnormality feature data;

[0014] S5. Use high-confidence neighbor base stations and high-confidence neighbor heterogeneous devices to perform rectification unit cleaning, rectification unit weighting, and joint weighted position rectification in turn to complete the position rectification of abnormal base stations.

[0015] Preferably, in step S1, the signaling trajectory data is:

[0016] Traj a = {Pos a1 , Pos a2 , …, Pos an}

[0017] Pos ai = <user a , uli, time, longtitude, latitude>

[0018] Wherein, Pos ai is the i-th point in the signaling trajectory Traj a of the signaling terminal user a, n is the point number of the signaling trajectory, user a is the username of the signaling terminal user a, uli is the base station serial number, time is the timestamp, longtitude is the longitude information, and latitude is the latitude information;

[0019] The heterogeneous trajectory data is:

[0020] Trck a = {Loc a1 , Loc a2 , …, Loc am}

[0021] Loc i = <id a , device, time, longtitude, latitude>

[0022] Wherein, Loc i is the i-th point in the heterogeneous trajectory Trck a of the signaling terminal user a, m is the number of heterogeneous trajectory points, id a is the heterogeneous user id, and device is the heterogeneous device id.

[0023] The statistical base station short-time switching information is:

[0024] {<uli1,c i1 ,d i1 >,…,<uli j ,c ij ,d ij >,…},j∈J

[0025] d ij =Haversine(Lon i ,Lat i ,Lon j ,Lat j ),j∈J

[0026] wherein c ij is the switching times of the base station i and the base station j, d ij is the switching distance, and J is the switching base station set;

[0027] The terminal-heterogeneous matching pair data is a matching relationship constructed by the signaling terminal user id and the plurality of heterogeneous user ids:

[0028] <user a ,id a1 ,id a2 ,…>.

[0029] Preferably, the specific content of the multi-level noise processing of the terminal-heterogeneous matching pair data in step S2 includes:

[0030] The longitude mean and the latitude mean of all points of the heterogeneous trajectory of the heterogeneous user are calculated, the longitude mean and the latitude mean of all points of the signaling trajectory of the signaling terminal user are calculated, and the matching pair daily average distance based on the full data is calculated;

[0031] The outlier analysis is performed on the signaling trajectory and the heterogeneous trajectory, the outliers are marked by using the longitude and latitude standard scores, the noise factor is calculated according to the matching pair daily average distance based on the full data and the filtering outlier matching pair daily average distance;

[0032] The accompanying features of the heterogeneous trajectory and the signaling trajectory are calculated, including the accompanying space proportion and the accompanying time proportion, the accompanying space proportion = the number of accompanying spaces / the number of heterogeneous trajectory spaces, and the accompanying time proportion = the number of accompanying times / the number of heterogeneous trajectory times.

[0033] Preferably, the high-confidence matching pair data is specifically:

[0034] For the high-confidence matching pair with the filtering outlier matching pair daily average distance greater than the preset threshold, all the heterogeneous trajectories in the calculation period are screened out;

[0035] For the matching pairs of full data, the high-confidence matching pairs with daily average distance greater than the preset threshold, accompanied by spatial proportion <0.1 and accompanied by time proportion <0.1, are excluded from all heterogeneous trajectories in the calculation period;

[0036] For the case that the noise factor is greater than the preset threshold, and the heterogeneous point is an outlier, the corresponding heterogeneous point is excluded.

[0037] Preferably, in step S3, the base station abnormal switching feature data includes weighted base station switching distance mean, high-frequency switching base station near-distance switching proportion, and high-frequency switching base station far-distance switching proportion; the base station abnormal alienation feature data includes same-site matching pair proportion, same-site heterogeneous device proportion, alien-site matching pair proportion, alien-site heterogeneous device proportion, and matching pair point distance mean.

[0038] Preferably, the weighted base station switching distance mean is:

[0039]

[0040] Wherein, D i is the weighted base station switching distance mean of base station i, d ij is the distance between base station i and base station j, c ij is the number of switching between base station i and base station j, w ij is the switching weight between base station i and base station j, C i is the total number of switching between base station i and all its switching base stations.

[0041] The near-distance switching proportion is:

[0042]

[0043] The far-distance switching proportion is:

[0044]

[0045] Wherein, N i is the high-frequency switching base station set, taking the k base stations with the highest switching weight and the number of switching greater than the preset threshold μ1 in the base station i switching base station set as the high-frequency switching base station set; Near i is the near-distance high-frequency switching base station set, the near-distance is the base station switching distance less than the preset threshold μ2; Far i is the far-distance high-frequency switching base station set, the far-distance is the base station switching distance greater than the preset threshold μ3.

[0046] Preferably, the calculation method of base station abnormal alienation feature data is:

[0047] For the matching pair set P i of the attached base station i, P ikid ik}, the number of matching pairs and the number of alien devices can be calculated according to the homologous information and the high-confidence mismatching pairs respectively; then:

[0048] The proportion of homologous matching pairs = the number of homologous matching pairs / the total number of matching pairs;

[0049] The proportion of homologous alien devices = the number of homologous alien devices / the total number of alien devices;

[0050] The proportion of mismatching pairs = the number of mismatching pairs / the total number of matching pairs;

[0051] The proportion of mismatching alien devices = the number of mismatching alien devices / the total number of alien devices;

[0052] The average distance of matching pair points = the average distance of all matching pair points within 60s.

[0053] Preferably, the combined features output by the abnormal base station detection in step S4 are specifically:

[0054] The total number of high-frequency switching base stations is greater than a set threshold value, the total number of matching pairs is greater than a set threshold value, and the total number of mismatching alien devices is greater than a set threshold value;

[0055] The weighted average distance of base station switching is greater than a set threshold value, the proportion of high-frequency switching base stations in close-range switching is less than a set threshold value, and the proportion of high-frequency switching base stations in long-range switching is greater than a set threshold value;

[0056] Or the average distance of matching pair points is greater than a set threshold value, the proportion of mismatching pairs is greater than a set threshold value, the proportion of mismatching alien devices is greater than a set threshold value, the proportion of homologous matching pairs is less than a set threshold value, and the proportion of homologous alien devices is less than a set threshold value;

[0057] The combined features output by the high-confidence base station detection are:

[0058] The total number of high-frequency switching base stations is greater than a set threshold value, the total number of matching pairs is greater than a set threshold value, and the total number of mismatching alien devices is greater than a set threshold value;

[0059] The weighted average distance of base station switching is less than a set threshold value, the proportion of high-frequency switching base stations in close-range switching is greater than a set threshold value, the proportion of high-frequency switching base stations in long-range switching is less than a set threshold value, the average distance of matching pair points is less than a set threshold value, the proportion of mismatching pairs is less than a set threshold value, the proportion of mismatching alien devices is less than a set threshold value, the proportion of homologous matching pairs is greater than a set threshold value, and the proportion of homologous alien devices is greater than a set threshold value.

[0060] Preferably, the final offset correction position in step S5 is specifically:

[0061] Final_lon = α * Cali_lonf + (1 - a) Cali_lon p

[0062] Final_Lat = a * Cali_Lat f + (1 - a) Cali_Lat p

[0063] wherein (Final_lon, Final_Lat) is the final rectified position; (Cali_lon f , Cali_Lat f ) is the rectified position of the alien device, obtained by summing the product of the position of the alien device and the weight of the alien device; (Cali_lon p , Cali_Lat p ) is the rectified position of the base station, obtained by summing the product of the position of the base station and the weight of the base station; a is the rectified coefficient of the alien device, 0 < a < 1.

[0064] An abnormal base station detection and rectification system based on multi-source data, based on the abnormal base station detection and rectification method based on multi-source data, comprising: a data acquisition and preprocessing module, a base station abnormal feature calculation module, an abnormal base station detection module and an abnormal base station rectification module;

[0065] The data acquisition and preprocessing module is used for acquiring signaling trajectory data and alien trajectory data and preprocessing, constructing signaling terminal short-time switching sequence data and obtaining base station short-time switching information, and constructing terminal-alien matching pair data. The terminal-alien matching pair data is processed by multiple levels of noise to screen out high-confidence matching pair data, and high-confidence matching pair parietal information and ectopic information data are obtained.

[0066] The base station abnormal feature calculation module is used for calculating base station abnormal switching feature data based on the base station short-time switching information, and calculating base station abnormal ectopic feature data based on the high-confidence matching pair ectopic information.

[0067] The abnormal base station detection module is used for detecting abnormal base stations and high-confidence base stations based on the combined features of the base station abnormal switching feature data and the base station abnormal ectopic feature data.

[0068] The abnormal base station rectification module is used for rectifying the units in sequence, weighting the units and jointly weighting the position rectification of the high-confidence adjacent area base station and the high-confidence adjacent area alien device, to complete the position rectification of the abnormal base station.

[0069] Compared with the prior art, the abnormal base station detection and deviation rectification method and system based on multi-source data provided by the technical solution can comprehensively depict the abnormal characteristics of the abnormal base station by combining multi-source data, rectify deviation by using multi-source data, break through the limitation of single data source of the abnormal base station detection and deviation rectification method, and effectively improve the abnormal base station detection accuracy and deviation rectification effect. BRIEF DESCRIPTION OF DRAWINGS

[0070] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of the provided drawings.

[0071] Figure 1 A schematic diagram of the abnormal base station detection and deviation rectification method based on multi-source data provided by the present application;

[0072] Figure 2 A schematic diagram of the abnormal base station detection and deviation rectification system based on multi-source data provided by the present application. DETAILED DESCRIPTION

[0073] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0074] The embodiments of the present application disclose an abnormal base station detection and deviation rectification method based on multi-source data, as shown in Figure 1 , comprising the following steps:

[0075] S1. Acquire signaling trajectory data and heterogeneous trajectory data and preprocess, construct signaling terminal short-time switching sequence data and obtain base station short-time switching information by statistics, and construct terminal-heterogeneous matching pair data;

[0076] S2. Perform multi-level noise processing on the terminal-heterogeneous matching pair data to screen out high-confidence matching pair data, and obtain high-confidence matching pair homoplasmy information and allotopy information data;

[0077] S3. Calculate base station abnormal switching feature data based on the base station short-time switching information, and calculate base station abnormal allotopy feature data based on the high-confidence matching pair allotopy information;

[0078] S4. Abnormal base station and high-confidence base station detection output based on the combined features of base station abnormal switching feature data and base station abnormal relocation feature data;

[0079] S5. Using high-confidence neighbor base stations and high-confidence neighbor heterogeneous devices to perform rectification unit cleaning, rectification unit weighting, and joint weighted position rectification in turn to complete the position rectification of abnormal base stations.

[0080] In this embodiment, the signaling trajectory data comes from the mounting behavior of the terminal and the base station, the heterogeneous trajectory refers to the trajectory data recorded by other devices, carrying the user id, geographic location, time, and other information of the recording time, common heterogeneous trajectories include GPS, traffic card, and other data, the heterogeneous trajectory has a homing relationship with a certain signaling terminal data, and can describe the moving path of the same target together with the signaling trajectory; The matching relationship constructed by the signaling terminal id and the heterogeneous user id is terminal-heterogeneous matching pair data.

[0081] In order to further implement the above technical solutions, in step S1, the signaling trajectory data refers to the time sequence of the terminal attaching the base station, the user a signaling trajectory Traj a is represented by a plurality of points Pos, each point contains a user name, a base station serial number, a timestamp, longitude, and latitude information, specifically:

[0082] Traj a = {Pos a1 , Pos a2 , …, Pos an}

[0083] Pos ai = <user a , uli, time, longtitude, latitude>

[0084] Wherein, Pos ai is the i-th point in the signaling trajectory Traj a of the signaling terminal user a, n is the number of points of the signaling trajectory, user a is the user name of the signaling terminal user a, uli is the base station serial number, time is the timestamp, longtitude is the longitude information, and latitude is the latitude information.

[0085] The user a heterogeneous trajectory Trck a is represented by a plurality of points Loc a , each point contains a heterogeneous user id, a heterogeneous device id, a timestamp, longitude, and latitude information, specifically:

[0086] Trck a = {Loc a1 , Loca2 ..., Loc am}

[0087] Loc i = <id a , device, time, longtitude, latitude>

[0088] wherein Loc i is the ith point in the hetero-track Trck a of the signaling terminal user a, m is the number of hetero-track points, id a is the hetero-user id, and device is the hetero-device id;

[0089] Based on the signaling track, the time difference between adjacent points of the same user can be calculated:

[0090] T i = time i - time i-1

[0091] Adjacent points of the same user construct the base station switching relationship:

[0092] <uli i-1 , uli i >

[0093] That is, the base station switching information:

[0094] <uli i-1 , uli i , T i >

[0095] According to the time sequence of the terminal signaling track, the time difference T i between base station switching is required to be within 100s, that is, a short-time switching relationship from one base station to another base station is obtained; the base station accumulates a sufficient number of terminal sequences in the calculation period, and the short-time switching base station and the short-time switching number of the base station in the calculation period are obtained by statistics. For base station i and its switching base station set J, the base station short-time switching information obtained by statistics is:

[0096] {<uli1, c i1 , d i1 >, …, <uli j , c ij , d ij >, …}, j∈J

[0097] d ij = Haversine(Lon i , Lat i , Lon j , Lat j ), j∈J

[0098] wherein c ij is the number of handovers between base station i and base station j, d ij is the handover distance, and J is the set of handover base stations;

[0099] Theoretically, the terminal-heterogeneous matching pair data of user a is:

[0100] <user a , id a >

[0101] In fact, the matching data is a one-to-many relationship, and the terminal-heterogeneous matching pair data of user a is the matching relationship constructed by the signaling terminal user id and a plurality of heterogeneous user ids:

[0102] <user a , id a1 , id a2 , …>.

[0103] In order to further implement the above technical solutions, in order to filter out high-confidence matching pair data from multiple matches, the specific content of the multi-level noise processing of the terminal-heterogeneous matching pair data A in step S2 includes:

[0104] Calculate the average longitude of all points of the heterogeneous trajectory of the heterogeneous user and the average latitude Calculate the average longitude of all points of the signaling trajectory of the signaling terminal user and the average latitude And calculate the matching pair daily distance D paira based on the full data;

[0105] Specifically:

[0106]

[0107] The matching pair daily distance D paira represents the distance between the center position of the heterogeneous trajectory and the center position of the signaling trajectory; in actual application, the heterogeneous trajectory has error data, such as error of the heterogeneous device position data or error in the data generation process, and the signaling trajectory also has base station position error and the like. This index calculation is biased due to the abnormal deviation of a small number of points;

[0108] Outlier analysis is performed on the signaling trajectory and the heterogeneous trajectory, and outliers are marked using the longitude and latitude standard score. Based on the matching pair daily distance based on the full data and the filtered outlier matching pair daily distance, the noise factor is calculated;

[0109] Specifically:

[0110]

[0111] A point is judged as an outlier by using a standard score Z-score, the Z-score is obtained by subtracting the average of the longitude / latitude of the day from the longitude / latitude of the point, and then dividing by the standard deviation of the longitude / latitude, and the point is marked as an abnormal outlier if the absolute value of the longitude Z-score or the absolute value of the latitude Z-score is greater than 2, and is not included in the calculation of the matching pair daily average distance;

[0112] For the convenience of subsequent discussion, for the matching A, the matching pair daily average distance calculated based on the full data is D, and the matching pair daily average distance calculated based on the filtered outliers is D clean The noise factor is specifically:

[0113]

[0114] The accompanying features of the heterogeneous trajectory and the signaling trajectory are calculated, including the accompanying space ratio and the accompanying time ratio, the accompanying space ratio = the number of accompanying spaces / the number of heterogeneous trajectory spaces, and the accompanying time ratio = the number of accompanying times / the number of heterogeneous trajectory times; in this embodiment, the accompanying features are used to judge whether the matching pair is in a normal accompanying state, the heterogeneous trajectory and the signaling trajectory are within 60s, and the distance is less than 500m, which is marked as accompanying, with 7-bit geohash as the space scale and 7-bit timestamp as the time scale.

[0115] In order to further implement the above technical solutions, the high-confidence matching pair data is specifically:

[0116] For the matching pair daily average distance D clean filtered out all the heterogeneous trajectories in the calculation period.

[0117] For the high-confidence matching pair whose matching pair daily average distance D based on full data is greater than the preset threshold, the accompanying space ratio <0.1, and the accompanying time ratio <0.1, all the heterogeneous trajectories in the calculation period are filtered out.

[0118] For the noise factor F noise greater than the preset threshold, and the heterogeneous trajectory point is an outlier, the corresponding heterogeneous point is filtered out.

[0119] In this embodiment, the high-confidence matching pair co-site information and the high-confidence matching pair allochthonous information are specifically:

[0120] In the calculation period, for the terminal signaling trajectory and its corresponding heterogeneous trajectory, the distance of the matching pair point within 60s is calculated, and when the distance is greater than 2km, it is marked as allochthonous, and when the distance is less than 500m, it is marked as co-site.

[0121] For the matching pair A: <user a , id aTerminal signaling Traj a and heterologous trajectory Trck a , High-confidence matching pairs with co-location information are:

[0122] U sam a = {<Pos sam k , Loc sam k >}, Haversine(Pos sam k , Loc sam k ) < 500

[0123] High-confidence matching pairs with allopatric information are:

[0124]

[0125] To further implement the above technical solutions, in step S3, the base station abnormal switching feature data includes a weighted base station switching distance mean, a high-frequency switching base station near-distance switching proportion, and a high-frequency switching base station far-distance switching proportion; and the base station abnormal allopatric feature data includes a co-location matching pair proportion, a co-location heterologous device proportion, an allopatric matching pair proportion, an allopatric heterologous device proportion, and a matching pair point distance mean.

[0126] To further implement the above technical solutions, the weighted base station switching distance mean is:

[0127]

[0128] wherein D i is the weighted base station switching distance mean of base station i, d ij is the distance between base station i and base station j, c ij is the number of switching times between base station i and base station j, w ij is the switching weight between base station i and base station j, and C i is the total number of switching times between base station i and all its switching base stations.

[0129] In this embodiment, the total number of switching base stations with a switching number greater than a set threshold value is calculated as the number of high-frequency switching base stations S i :

[0130]

[0131] First, the k base stations with the highest switching weight and a switching number greater than μ1 in the switching base station set of base station i are taken as the high-frequency switching base station set N i , and the total number of high-frequency switching base stations |N i | is counted.

[0132] N i = {uli i1 ,uli i2 …uli ik} c ij > μ1

[0133] The base station switching distance is required to be less than a set threshold μ2, and a near-distance high-frequency switching base station set Near i is obtained:

[0134] Near i = {n i1 ,n i2 …n im} m<=k,d ij < μ2

[0135] The base station switching distance is required to be greater than a set threshold μ3, and a far-distance high-frequency switching base station set Far i is obtained:

[0136] Far i = {n i1 ,n i2 …n ip} p<=k,d ij > μ3

[0137] The near-distance switching proportion is:

[0138]

[0139] The far-distance switching proportion is:

[0140]

[0141] Wherein, N i is a high-frequency switching base station set, and the k base stations with the highest switching weight and greater than a preset threshold μ1 in the base station i switching base station set are taken as the high-frequency switching base station set; Near i is a near-distance high-frequency switching base station set, and the near distance is a base station switching distance less than a preset threshold μ2; Far i is a far-distance high-frequency switching base station set, and the far distance is a base station switching distance greater than a preset threshold μ3.

[0142] In order to further implement the above technical scheme, the calculation method of the base station abnormal ectopic feature data is:

[0143] For the matching pair set P i = {…, <user ik , id ikAccording to the high-confidence matching pair homologous information and the high-confidence matching pair heterologous information, the number of matching pairs and the number of alien devices can be calculated respectively; and

[0144] The homologous matching pair proportion is the number of homologous matching pairs / the total number of matching pairs.

[0145] The homologous alien device proportion is the number of homologous alien devices / the total number of alien devices.

[0146] The heterologous matching pair proportion is the number of heterologous matching pairs / the total number of matching pairs.

[0147] The heterologous alien device proportion is the number of heterologous alien devices / the total number of alien devices.

[0148] The matching pair point distance average is the average of the distances between all matching pair points within 60s.

[0149] In order to further implement the above technical solution, the combined features of the abnormal base station detection output are specifically:

[0150] The total number of high-frequency switching base stations is greater than a set threshold value, the total number of matching pairs is greater than a set threshold value, and the total number of heterologous alien devices is greater than a set threshold value.

[0151] The weighted base station switching distance average is greater than a set threshold value, the high-frequency switching base station near-distance switching proportion is less than a set threshold value, and the high-frequency switching base station far-distance switching proportion is greater than a set threshold value.

[0152] Or the matching pair point distance average is greater than a set threshold value, the heterologous matching pair proportion is greater than a set threshold value, the heterologous alien device proportion is greater than a set threshold value, the homologous matching pair proportion is less than a set threshold value, and the homologous alien device proportion is less than a set threshold value.

[0153] The combined features of the high-confidence base station detection output are:

[0154] The total number of high-frequency switching base stations is greater than a set threshold value, the total number of matching pairs is greater than a set threshold value, and the total number of heterologous alien devices is greater than a set threshold value.

[0155] The weighted base station switching distance average is less than a set threshold value, the high-frequency switching base station near-distance switching proportion is greater than a set threshold value, the high-frequency switching base station far-distance switching proportion is less than a set threshold value, the matching pair point distance average is less than a set threshold value, the heterologous matching pair proportion is less than a set threshold value, the heterologous alien device proportion is less than a set threshold value, the homologous matching pair proportion is greater than a set threshold value, and the homologous alien device proportion is greater than a set threshold value.

[0156] In order to further implement the above technical solution, the final correction position is specifically:

[0157] Final_lon = α * Cali_lon f +(1-α)Cali_lon p

[0158] Final_Lat = α * Cali_Lat f +(1-α)Cali_Lat p

[0159] Where (Final_lon, Final_Lat) represents the final correction position; (Cali_lon) represents the final correction position. f ,Cali_Lat f (Cali_lon) represents the correction position of the heterogeneous device, obtained by summing the products of the heterogeneous device position and its weight; p ,Cali_Lat p ) represents the base station correction position, which is obtained by summing the product of the base station position and the base station weight; α is the correction coefficient of the heterogeneous equipment, 0 < α < 1.

[0160] In this embodiment, the content of the correction unit cleaning is to screen out out-of-source devices with abnormal locations and high-confidence handover base stations to reduce correction noise; specifically: calculate the latitude and longitude Z-score values ​​of out-of-source devices under the jurisdiction of abnormal base stations, and the latitude and longitude Z-score values ​​of handover base stations of high-confidence handover base stations, screen out abnormal devices and base stations with an absolute Z-score value greater than 2, and obtain high-confidence neighboring out-of-source devices and high-confidence neighboring base stations.

[0161] The content of weighting the correction units involves assigning different correction units with correction weights:

[0162]

[0163] Where is the correction weight of heterogeneous device j on base station i, and p ij Let j be the number of matching pairs between heterogeneous device j and base station i. The total number of matching pairs between base station i and each heterogeneous device;

[0164]

[0165] in, Let c be the correction weight of base station j for base station i. ij Let j be the number of handovers between base station j and base station i. This represents the total number of handovers between base station i and all other base stations;

[0166] Heterogeneous equipment correction position (Cali_lon) f ,Cali_Lat f )for:

[0167]

[0168] Cali_lon p Cali_Lat p ) are as follows:

[0169]

[0170] An abnormal base station detection and rectification system based on multi-source data, such as Figure 2 , an abnormal base station detection and rectification method based on multi-source data, comprising: a data acquisition and preprocessing module, a base station abnormal feature calculation module, an abnormal base station detection module and an abnormal base station rectification module;

[0171] The data acquisition and preprocessing module is configured to acquire signaling trajectory data and heterogeneous trajectory data and pre-process the same, construct signaling terminal short-time switching sequence data and statistically obtain base station short-time switching information, and construct terminal-heterogeneous matching pair data, perform multi-level noise processing on the terminal-heterogeneous matching pair data to screen out high-confidence matching pair data, and obtain high-confidence matching pair homotopic information and heterotopic information data.

[0172] The base station abnormal feature calculation module is configured to calculate base station abnormal switching feature data based on the base station short-time switching information, and calculate base station abnormal heterotopic feature data based on the high-confidence matching pair heterotopic information.

[0173] The abnormal base station detection module is configured to detect abnormal base stations and high-confidence base stations based on the combined features of the base station abnormal switching feature data and the base station abnormal heterotopic feature data, and output the same.

[0174] The abnormal base station rectification module is configured to utilize high-confidence neighboring area base stations and high-confidence neighboring area heterogeneous equipment to sequentially perform rectification unit cleaning, rectification unit weighting and joint weighted position rectification, and complete the position rectification of the abnormal base station.

[0175] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the related parts can be referred to the method part.

[0176] The foregoing description of the disclosed embodiments enables a person skilled in the art to make or use the application. Modifications of these embodiments will occur to persons of skill in the art, and that the appended claims are intended to cover all such modifications that do not depart from the true spirit and scope of the application. Therefore, the application is not limited to the embodiments shown but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting and rectifying an abnormal base station based on multi-source data, characterized in that, The method comprises the following steps: S1. Obtain signaling trajectory data and heterogeneous trajectory data, and preprocess, construct signaling terminal short-time switching sequence data, and obtain base station short-time switching information by statistics, and construct terminal-heterogeneous matching pair data; S2. Perform multi-level noise processing on the terminal-heterogeneous matching pair data to screen out high-confidence matching pair data, and obtain high-confidence matching pair homoplasia information and heteroplasia information data; S3. Calculate base station abnormal switching feature data based on the base station short-time switching information, and calculate base station abnormal heteroplasia feature data based on the high-confidence matching pair heteroplasia information; S4. Perform abnormal base station and high-confidence base station detection based on the combined features of the base station abnormal switching feature data and the base station abnormal heteroplasia feature data, and output; S5. Perform rectification unit cleaning, rectification unit weighting and joint weighted position rectification on the high-confidence neighboring area base stations and high-confidence neighboring area heterogeneous devices in turn to complete the position rectification of the abnormal base station.

2. The method of claim 1, wherein, In step S1, the signaling trajectory data is: Traj a = {Pos a1 , Pos a2 , …, Pos an} Pos ai = <user a , uli, time, longtitude, latitude> Wherein, Pos ai is the i-th point in the signaling trajectory Traj a of the signaling terminal user a, n is the point number of the signaling trajectory, user a is the username of the signaling terminal user a, uli is the base station serial number, time is the timestamp, longtitude is the longitude information, and latitude is the latitude information. The heterogeneous trajectory data is: Trck a = {Loc a1 , Loc a2 , …, Loc am} Loc i = <id a , device, time, longtitude, latitude> Wherein, Loc i is the i-th point in the heterogeneous trajectory Trck a of the signaling terminal user a, m is the number of heterogeneous trajectory points, id a is the heterogeneous user id, and device is the heterogeneous device id; The obtained base station short-time switching information is: {<ul i1,c i1 ,d i1 >,…,<ul i j ,c ij ,d ij >,…},j∈J d ij = Haversine(Lon i ,Lat i , Lon j ,Lat j ), j e J wherein c ij is the number of handovers from base station i to base station j, d ij is the handover distance, and J is the set of handover base stations; The terminal-heterogeneous matching pair data is a matching relationship constructed by the signaling terminal user id and a plurality of heterogeneous user ids: <user a id a1 id a2 ,...>.

3. The method of claim 1, wherein, In step S2, the specific content of the multi-level noise processing on the terminal-heterogeneous matching pair data includes: The longitude mean and latitude mean of all point positions of the heterogeneous user heterogeneous trajectory are calculated, the longitude mean and latitude mean of all point positions of the signaling terminal user signaling trajectory are calculated, and the matching pair daily average distance based on the full data is calculated; Outlier points are analyzed for the signaling trajectory and the heterogeneous trajectory, and the outliers are marked by using the longitude and latitude standard scores, the noise factor is calculated based on the matching pair daily average distance based on the full data and the filtered outlier matching pair daily average distance; The accompanying features of the heterogeneous trajectory and the signaling trajectory are calculated, including the accompanying space ratio and the accompanying time ratio, the accompanying space ratio = the number of accompanying spaces / the number of heterogeneous trajectory spaces, and the accompanying time ratio = the number of accompanying times / the number of heterogeneous trajectory times.

4. The method of claim 3, wherein, The high-confidence matching pair data is: For the high-confidence matching pair with the filtered outlier matching pair daily average distance greater than the preset threshold, all the heterogeneous trajectories in the calculation period are screened out; For the high-confidence matching pair with the matching pair daily average distance based on the full data greater than the preset threshold, the accompanying space ratio <0.1, and the accompanying time ratio <0.1, all the heterogeneous trajectories in the calculation period are screened out; For the case where the noise factor is greater than the preset threshold and the heterogeneous trajectory point position is an outlier, the corresponding heterogeneous point position is screened out.

5. The method of claim 1, wherein, In step S3, the base station abnormal switching feature data includes the weighted base station switching distance mean, the high-frequency switching base station near-distance switching ratio, and the high-frequency switching base station far-distance switching ratio; and the base station abnormal heteroplasia feature data includes the homoplasia matching pair ratio, the homoplasia heterogeneous device ratio, the heteroplasia matching pair ratio, the heteroplasia heterogeneous device ratio, and the matching pair point position distance mean.

6. The method of claim 5, wherein, The weighted base station switching distance mean is: where D i is the weighted base station handover distance mean for base station i, d ij is the distance between base station i and base station j, c ij is the number of handovers between base station i and base station j, w ij is the handover weight between base station i and base station j, C i is the sum of the number of handovers between base station i and all of its handover base stations; The near-distance switching ratio is: The far-distance switching ratio is: N i is a high-frequency handover base station set, taking the k base stations with the highest handover weights and the handover times greater than a preset threshold μ1 in the handover base station set of base station i as the high-frequency handover base station set; Near i is a near-distance high-frequency handover base station set, the near distance being a base station handover distance less than a preset threshold μ2; Far i is a far-distance high-frequency handover base station set, the far distance being a base station handover distance greater than a preset threshold μ3.

7. The method of claim 5, wherein, The calculation method of the base station abnormal heteroplasia feature data is: For the matching pair set P attached to the base station i i = {…, <user ik , id ik >, …}, according to the high-confidence matching pair homologous information and high-confidence matching pair heterologous information, the matching pair number and the alien device number can be calculated respectively; then: The homoplasia matching pair ratio = the number of homoplasia matching pairs / the total number of matching pairs; The homoplasia heterogeneous device ratio = the number of homoplasia heterogeneous devices / the total number of heterogeneous devices; The proportion of ectopic matching pairs = the number of matching pairs with ectopia / the total number of matching pairs; The proportion of ectopic foreign devices = the number of foreign devices with ectopia / the total number of foreign devices; The average distance of matching pair points = the average distance of all matching pair points within 60s.

8. The method of claim 1, wherein, The combined features of the abnormal base station detection output are specifically: The total number of high-frequency switching base stations is greater than a set threshold, the total number of matching pairs is greater than a set threshold, and the total number of ectopic foreign devices is greater than a set threshold; The weighted average base station switching distance is greater than a set threshold, the proportion of high-frequency switching base stations with short-distance switching is less than a set threshold, and the proportion of high-frequency switching base stations with long-distance switching is greater than a set threshold; Or the average distance of matching pair points is greater than a set threshold, the proportion of ectopic matching pairs is greater than a set threshold, the proportion of ectopic foreign devices is greater than a set threshold, the proportion of co-located matching pairs is less than a set threshold, and the proportion of co-located foreign devices is less than a set threshold; The combined features of the high-confidence base station detection output are: The total number of high-frequency switching base stations is greater than a set threshold, the total number of matching pairs is greater than a set threshold, and the total number of ectopic foreign devices is greater than a set threshold; The weighted average base station switching distance is less than a set threshold, the proportion of high-frequency switching base stations with short-distance switching is greater than a set threshold, the proportion of high-frequency switching base stations with long-distance switching is less than a set threshold, the average distance of matching pair points is less than a set threshold, the proportion of ectopic matching pairs is less than a set threshold, the proportion of ectopic foreign devices is less than a set threshold, the proportion of co-located matching pairs is greater than a set threshold, and the proportion of co-located foreign devices is greater than a set threshold.

9. The method of claim 1, wherein, The final correction position is specifically: Final_lon = a * Cali_lon f + (1 - a) Cali_lon p Final_Lat = a * Cali_Lat f + (1 - a) Cali_Lat p Wherein, (Final_lon, Final_Lat) is the final rectification position; (Cali_lon f ,Cali_Lat f ) is the rectification position of the heterogeneous device, obtained by summing the product of the heterogeneous device position and the heterogeneous device weight; (Cali_lon p ,Cali_Lat p ) is the rectification position of the base station, obtained by summing the product of the base station position and the base station weight; and α is the rectification coefficient of the heterogeneous device, 0<α<1. 10.A system for detecting and rectifying abnormal base stations based on multi-source data, characterized in that, The method comprises a data acquisition and preprocessing module, a base station abnormal feature calculation module, an abnormal base station detection module, and an abnormal base station correction module. The data acquisition and preprocessing module is used to acquire signaling trajectory data and foreign trajectory data and preprocess them, construct signaling terminal short-time switching sequence data, and obtain base station short-time switching information. The terminal-foreign matching pair data is constructed, and the high-confidence matching pair data is screened out by performing multi-level noise processing on the terminal-foreign matching pair data to obtain co-located information and ectopic information data of the high-confidence matching pairs. The base station abnormal feature calculation module is used to calculate base station abnormal switching feature data based on the base station short-time switching information and calculate base station abnormal ectopic feature data based on the high-confidence matching pair ectopic information. The abnormal base station detection module is used to perform abnormal base station and high-confidence base station detection output based on the combined features of the base station abnormal switching feature data and the base station abnormal ectopic feature data. The abnormal base station correction module is used to perform unit cleaning, unit weighting, and joint weighted position correction on the high-confidence neighboring base station and the high-confidence neighboring foreign device to complete the position correction of the abnormal base station.

Images