WO2017202226A1 - 人群流量的确定方法及装置 - Google Patents

人群流量的确定方法及装置 Download PDF

Info

Publication number
WO2017202226A1
WO2017202226A1 PCT/CN2017/084388 CN2017084388W WO2017202226A1 WO 2017202226 A1 WO2017202226 A1 WO 2017202226A1 CN 2017084388 W CN2017084388 W CN 2017084388W WO 2017202226 A1 WO2017202226 A1 WO 2017202226A1
Authority
WO
WIPO (PCT)
Prior art keywords
base station
target base
training set
location information
crowd
Prior art date
Application number
PCT/CN2017/084388
Other languages
English (en)
French (fr)
Inventor
文韬
罗圣美
刘丽霞
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Publication of WO2017202226A1 publication Critical patent/WO2017202226A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/24Classification techniques
    • G06F18/241Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W24/00Supervisory, monitoring or testing arrangements
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/23Clustering techniques
    • G06F18/232Non-hierarchical techniques
    • G06F18/2321Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions

Definitions

  • the present invention relates to the field of communications, and in particular to a method and apparatus for determining a population flow.
  • the embodiment of the invention provides a method and a device for determining the flow rate of a crowd, so as to solve at least the problems of inaccurate detection results and high cost in the prediction method of the crowd flow in the related art.
  • a method for determining a population flow including:
  • the method before determining the crowd traffic of the target base station according to the processed result, the method further includes: filtering out the abnormal point from the merged result according to the discrete point discovery technology.
  • the abnormal point is selected from the combined processing result according to the discrete point discovery technology, including: dividing the result belonging to the normal value of the cluster into the first training result in the combined processing result The concentration is concentrated, and the results that are outside the normal values of the cluster are divided into the second training set.
  • the method further includes: constructing an initialization model according to the first training set; predicting the second training set by using the initialization model, and obtaining an updated second training set according to the prediction result;
  • the updated second training set and the first training set obtain a prediction model of the crowd flow, wherein the population flow determination value of the target base station at the current time is input into the prediction model, and the current time is obtained.
  • the crowd flow of the target base station at the specified time
  • the location information of the target base station includes at least one of the following: a user identifier ID, a timestamp, and a target base station identifier ID;
  • the historical location information of the user accessing the target base station includes at least one of the following: a target base station identifier ID, a target base station latitude, and a target base station longitude.
  • a device for determining crowd flow including:
  • An acquiring module configured to acquire location information of a target base station and historical location information of a user accessing the target base station
  • a processing module configured to combine the historical location information and the location information
  • the first determining module is configured to determine the crowd traffic of the target base station according to the result of the combining process.
  • the apparatus further includes: a screening module, configured to filter the abnormal point from the merged result according to the discrete point discovery technology.
  • the screening module is further configured to: in the result of the merge processing, divide the result that belongs to the normal value of the cluster into the first training set, and the result that is free from the normal value of the cluster Divided into the second training set.
  • the apparatus further includes: a building module configured to construct an initialization model according to the first training set; a prediction module configured to predict the second training set by using the initialization model; and a second determining module And being configured to obtain an updated second training set according to the prediction result; and the third determining module is configured to obtain the person according to the updated second training set and the first training set And a prediction model of the group traffic, wherein the population flow rate determination value of the target base station at the current time is input into the prediction model, and the crowd flow rate of the target base station at the specified time after the current time is obtained.
  • the location information of the target base station includes at least one of the following: a user identifier ID, a timestamp, and a target base station identifier ID;
  • the historical location information of the user accessing the target base station includes at least one of the following: a target base station identifier ID, a target base station latitude, and a target base station longitude.
  • a computer storage medium is further provided, and the computer storage medium may store an execution instruction for performing the implementation of the method for determining the crowd traffic in the foregoing embodiment.
  • the traffic of the target base station can be determined according to the location information of the target base station and the historical location information of the user accessing the target base station, and the foregoing technical solution is used to solve the traffic in the related art.
  • the prediction method has many problems such as inaccurate detection results and high cost, so that the intervention of video surveillance is not required to realize the determination of crowd flow and avoid excessive cost pressure.
  • FIG. 1 is a flow chart of a method for determining crowd flow according to an embodiment of the present invention
  • FIG. 2 is a structural block diagram of a device for determining a crowd flow according to an embodiment of the present invention
  • FIG. 3 is a block diagram showing another structure of a device for determining a flow rate of a crowd according to an embodiment of the present invention
  • FIG. 4 is a flow chart of a method for determining crowd traffic according to an embodiment of the present invention.
  • FIG. 1 is a flowchart of a method for determining a population flow according to an embodiment of the present invention. As shown in FIG. 1, the process includes the following steps:
  • Step S102 acquiring location information of the target base station and historical location information of the user of the access target base station;
  • Step S104 Combine the historical location information and the location information, and determine the crowd traffic of the target base station according to the result of the merge processing.
  • the population traffic of the target base station can be determined according to the location information of the target base station and the historical location information of the user of the access target base station, and the foregoing technical solution is used to solve the detection result of the prediction method for the crowd traffic in the related technology. Inaccurate, high cost and many other issues, in order to achieve the determination of crowd traffic without the intervention of video surveillance, avoiding excessive cost pressure.
  • the embodiment of the present invention further provides the following technical solution: before determining the traffic flow of the target base station according to the processed result.
  • the method further includes: selecting an abnormal point from the merged result according to the discrete point finding technique.
  • the abnormal point may be filtered by the following method, and the result of the merge processing is affiliated with The result of the clustering normal value is divided into the first training set, and the result that is outside the normal value of the cluster is divided into the second training set, wherein the element in the second training set can be understood as the abnormal point.
  • the base station signaling data is the result of many years of accumulation, there is no guarantee that the history records of each base station area cover the entire time period, resulting in a large number of data defects; if the broken data is directly discarded, the valuable data assets will be If the model is not fully utilized, the accuracy of the trained model will be reduced accordingly.
  • the incomplete and incomplete data can be effectively utilized.
  • the technical solution of the embodiment of the present invention can also predict the crowd flow value at a certain moment in the future, mainly by the following scheme: constructing an initialization model according to the first training set; predicting the second training set by using the initialization model, and predicting according to the prediction As a result, an updated second training set is obtained. According to the updated second training set and the first training set, a prediction model of the crowd flow is obtained, wherein the population flow determination value of the target base station at the current time is input into the prediction model, and the current The traffic of the target base station at a specified time after the moment, the technical solution of this part will be described below in conjunction with the preferred embodiment, and details are not described herein again.
  • the location information of the target base station includes at least one of the following: a user identifier ID, a timestamp, and a target base station identifier ID; and the historical location information of the user of the access target base station includes at least one of the following: the target base station identifier ID, the target base station. Latitude, target base station longitude.
  • the crowd situational awareness means has the basis of high accuracy prediction, and can also predict the population density of key locations in advance, so as to further advance the arrangement and prevent it from happening.
  • the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware, but in many cases, the former is A better implementation.
  • the technical solution of the present invention which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk,
  • the optical disc includes a number of instructions for causing a terminal device (which may be a cell phone, a computer, a server, or a network device, etc.) to perform the methods of various embodiments of the present invention.
  • a device for determining the flow rate of the crowd is also provided, and the device is used to implement the above-mentioned embodiments and preferred embodiments, and the description thereof has been omitted.
  • the term "module” may implement a combination of software and/or hardware of a predetermined function.
  • the described device is preferably implemented in software, but hardware, or a combination of software and hardware, is also possible and contemplated.
  • FIG. 2 is a structural block diagram of a device for determining a crowd flow according to an embodiment of the present invention. As shown in FIG. 2, the device includes:
  • the obtaining module 20 is configured to acquire location information of the target base station and historical location information of a user accessing the target base station;
  • the processing module 22 is configured to combine the historical location information and the location information
  • the first determining module 24 is configured to determine the crowd traffic of the target base station according to the result of the combining process.
  • the above-mentioned technical solutions can be used to determine the traffic of the target base station according to the location information of the target base station and the historical location information of the user accessing the target base station.
  • the method of predicting crowd flow has many problems such as inaccurate test results and high cost, so that the intervention of video surveillance is not needed to realize the determination of crowd flow and avoid excessive cost pressure.
  • FIG. 3 is another structural block diagram of a device for determining a crowd flow according to an embodiment of the present invention. As shown in FIG. 3, the device further includes:
  • the screening module 26 is configured to filter the abnormal points from the combined processing results according to the discrete point discovery technique.
  • the screening module 26 is further configured to: in the result of the merge processing, divide the result that belongs to the normal value of the cluster into the first training set, and divide the result that is free from the normal value of the cluster. Go to the second training set.
  • the apparatus further includes: a construction module 28 configured to construct an initialization model according to the first training set; and a prediction module 30 configured to predict the second training set by using the initialization model;
  • the second determining module 32 is configured to obtain the updated second training set according to the prediction result
  • the third determining module 34 is configured to obtain a prediction model of the crowd traffic according to the updated second training set and the first training set, where And inputting a population flow determination value of the target base station at the current moment into the prediction model, and obtaining a specified time after the current time The crowd flow of the target base station.
  • the location information of the target base station includes at least one of the following: a user identifier ID, a timestamp, and a target base station identifier ID; and the historical location information of the user accessing the target base station includes at least one of the following: the target base station identifier ID, target base station latitude, target base station longitude.
  • FIG. 4 is a flowchart of a method for determining crowd flow according to an embodiment of the present invention. As shown in FIG. 4, the following stages are included:
  • Data preparation phase the existing historical signaling data and the geographical location data of the base station are aggregated, specifically, the above two data are connected by using the base station number Cell ID, and the shape is further obtained as "a certain year - a certain month - some Day-a certain period of time", the flow statistics of people entering/staying within a certain base station.
  • the above aggregation result may not be completely in line with the actual situation, because the log records of the operator signaling data are affected by various situations, and there may be some defects in user switching and location timing reporting signaling. Specifically, because the historical data is incomplete, there are some "some people's flow statistics of some base stations are far lower than the statistical values of the neighboring base stations/previous statistics; and the statistics of the flow of some base stations are much lower than their statistics. The phenomenon of statistical values before and after the period, the above data can not reflect the real situation of the crowd flow, and the direct training will result in an inaccurate model. Directly discarding the statistics of these spatio-temporal slices will waste data resources. Second, the model will lack sufficient data support, which is also not accurate.
  • the preferred embodiment of the present invention utilizes outlier detection techniques to identify and identify these anomalies for use in the next phase.
  • the clustering technique is used to cluster the "personal flow statistics of a certain time period - a certain place", and the normal values belonging to each cluster are directly put into the TrainingSet training set (corresponding to the first training section of the above embodiment).
  • the outliers that are free from the respective clusters are placed in the AbnormalSet auxiliary training set (corresponding to the second training set of the above embodiment).
  • X is the vector form of the input feature
  • W is the weight vector of the corresponding feature
  • y is the expected value of the flow of the target area when the specific input X and the model parameter are W.
  • Model application stage Use Flow_Model to predict the newly entered data to be predicted, and calculate the population flow value at a certain time and target position in the future based on the current time and the statistical value of the flow around the target position.
  • the preferred embodiment of the present invention does not require video surveillance intervention, avoids excessive cost pressure; can process the missing signaling historical data, protect the data assets, and at the same time improve the accuracy of the prediction, and overcome the present
  • the visual blind spot of video surveillance traffic counting technology and the negative impact caused by the lack of historical signaling records of traditional operators' mobile phones have proposed a method for predicting crowd traffic in public places by using operator mobile phone signaling data. This technology can still work well in the absence of training data, helping city managers to predict the flow density of key areas in cities in advance. Provide data support for urban public safety, large-scale event organization, etc.
  • Definition point - cluster distance the Euclidean distance of a point from a centroid of a cluster in a multidimensional space
  • cluster radius the average Euclidean distance from the point to the centroid in a cluster
  • mapping function S(X) is defined, the input parameter X is an m-dimensional vector, and the output Y is also an m-dimensional vector.
  • W(1)j and W(2)j are the jth component values of the two vectors:
  • is a small floating point number defined externally.
  • each sample has n features in the form of:
  • x ij represents the jth feature value of the i-th historical training data sample
  • Y i represents the statistical value of the i-th training data sample
  • Model determination coefficient R used to measure the performance of the trained model. Let the number of rows in the training set be n; let Y be the actual dependent variable vector of the training set, then Y k corresponds to each component of it; let Z be the result variable predicted by the training set on the existing model, then Z The k corresponds to each of its components. make Then the formula for R is:
  • Step 101 The operator provides user location update data (including user ID, timestamp, base station ID) and base station location data (base station ID, base station longitude, base station latitude), and user historical location information and base station geographic location.
  • user location update data including user ID, timestamp, base station ID
  • base station location data base station ID, base station longitude, base station latitude
  • user historical location information and base station geographic location are used as the training sample set of the subsequent steps, and the format is as follows:
  • Step 201 After the work in step 101 is completed, abnormal data may exist due to the incompleteness of the operator data history signaling record and the inherent technical problem of the signaling base station drift. Therefore, the scheme uses outlier detection technology to find out these abnormal points and mark them for use in the next stage. specific:
  • each object selected represents the initial mean of the group or the initial group center value; for each of the remaining objects, assign them to the nearest based on their Euclidean distance from the initial mean of each group (most similar) group; then, recalculate the new mean for each group.
  • step 301 and step 201 the initialization model Flow_Model0 is first constructed by using the normal sample training set TrainingSet. Specifically:
  • the n+1-dimensional weight vector W is initialized: a random number between (0, 1) is generated for each weight component Wi and assigned, where W is W(0).
  • the loop jump condition is: the number of loop executions is greater than the external parameter t, or three consecutive loops, W(k-2), W(k-1), W(k) are "approximate" to each other:
  • Step 302 Using the internal distribution feature of the AbnormalSet's incomplete data to improve the accuracy of the Flow_Model0 model, specifically:
  • step 301 the method described in step 301 is used to train to obtain a new decision coefficient Rnew. If Rnew>Rtraining, ⁇ Xabnormal, ypredicti> is placed in the sample set NewSet.
  • the NewSet and the TrainingSet are recombined, and the method described in step 301 is used again to form the final model Flow_Model1.
  • Step 401 After the system is online, the new observation data (denoted as vector X) is substituted into the parameter vector W in the trained model Flow_Model1, and the expected flow value of the target base station region is obtained by using Equation 1.
  • the new observation data (denoted as vector X) is substituted into the parameter vector W in the trained model Flow_Model1, and the expected flow value of the target base station region is obtained by using Equation 1.
  • the embodiment of the present invention achieves the following technical effects: no video surveillance intervention is required, excessive cost pressure is avoided; and broken signaling history data can be processed, data assets are protected, and prediction accuracy can be improved at the same time. It overcomes the blind spots caused by the existing video surveillance traffic counting technology and the negative impact caused by the lack of the traditional operator's mobile phone historical signaling record.
  • Embodiments of the present invention also provide a storage medium.
  • the foregoing storage medium may be configured to store program code for performing the following steps:
  • the foregoing storage medium may include, but not limited to, a USB flash drive, a Read-Only Memory (ROM), a Random Access Memory (RAM), a mobile hard disk, and a magnetic memory.
  • ROM Read-Only Memory
  • RAM Random Access Memory
  • a mobile hard disk e.g., a hard disk
  • magnetic memory e.g., a hard disk
  • modules or steps of the present invention described above can be implemented by a general-purpose computing device that can be centralized on a single computing device or distributed across a network of multiple computing devices. Alternatively, they may be implemented by program code executable by the computing device such that they may be stored in the storage device by the computing device and, in some cases, may be different from the order herein. Execution shown or described The steps are either made into individual integrated circuit modules, or a plurality of modules or steps are made into a single integrated circuit module. Thus, the invention is not limited to any specific combination of hardware and software.
  • the foregoing technical solution provided by the embodiment of the present invention can be applied to determining a crowd flow of the target base station according to the location information of the target base station and the historical location information of the user accessing the target base station in the process of determining the crowd process.
  • the technical solution solves many problems in the related technology that the prediction method of the crowd flow has inaccurate detection results and high cost, and thus the intervention of the video monitoring is not required to realize the determination of the crowd flow, and the excessive cost pressure is avoided.

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)
  • Alarm Systems (AREA)

Abstract

一种人群流量的确定方法及装置。其中,该方法包括:获取目标基站的位置信息和接入所述目标基站的用户的历史位置信息(S102);将所述历史位置信息和所述位置信息合并处理,并根据合并处理后的结果确定所述目标基站的人群流量(S104)。采用上述技术方案,解决了相关技术中,对人群流量的预测方法存在检测结果不准确、成本高等诸多问题,进而不需要视频监控的介入就能实现对人群流量的确定,避免了过高的成本压力。

Description

人群流量的确定方法及装置 技术领域
本发明涉及通信领域,具体而言,涉及一种人群流量的确定方法及装置。
背景技术
公共安全在当前日益得到政府和人民群众的重视和关注,尤其是针对公共场所的密集人群来说,当密度到达一定阈值时,就存在非常大的安全隐患。当前的公共场所人流密度检测的主流方法,还停留在利用视频监控设备对人群进行拍摄,然后对实时视频做识别、分析。以上方法成本较高、对摄像设备的布置位置也有要求、而且存在监控盲区导致人流检测不准确,对后续预测造成不利影响。
针对相关技术中,对人群流量的预测方法存在检测结果不准确、成本高等诸多问题,尚未提出有效的解决方案。
发明内容
本发明实施例提供了一种人群流量的确定方法及装置,以至少解决相关技术中对人群流量的预测方法存在检测结果不准确、成本高等诸多问题。
根据本发明的一个实施例,提供了一种人群流量的确定方法,包括:
获取目标基站的位置信息和接入所述目标基站的用户的历史位置信息;将所述历史位置信息和所述位置信息合并处理,并根据合并处理后的结果确定所述目标基站的人群流量。
可选地,根据处理后的结果确定所述目标基站的人群流量之前,所述方法还包括:根据离散点发现技术从合并处理后的结果中筛选出异常点。
可选地,根据离散点发现技术从合并处理后的结果中筛选出异常点,包括:在合并处理后的结果中,将隶属于聚类正常值的结果划分到第一训 练集中,以及将游离于所述聚类正常值之外的结果划分到第二训练集中。
可选地,所述方法还包括:根据所述第一训练集构建初始化模型;利用所述初始化模型对所述第二训练集进行预测,并根据预测结果得到更新后的第二训练集;根据更新后的第二训练集和所述第一训练集得到人群流量的预测模型,其中,将当前时刻所述目标基站的人群流量确定值输入到所述预测模型中,得到所述当前时刻之后的指定时刻的目标基站的人群流量。
可选地,所述目标基站的位置信息至少包括以下之一:用户标识ID、时间戳、目标基站标识ID;
接入所述目标基站的用户的所述历史位置信息至少包括以下之一:目标基站标识ID、目标基站纬度、目标基站经度。
根据本发明的另一个实施例,还提供了一种人群流量的确定装置,包括:
获取模块,设置为获取目标基站的位置信息和接入所述目标基站的用户的历史位置信息;
处理模块,设置为将所述历史位置信息和所述位置信息合并处理;
第一确定模块,设置为根据合并处理后的结果确定所述目标基站的人群流量。
可选地,所述装置还包括:筛选模块,设置为根据离散点发现技术从合并处理后的结果中筛选出异常点。
可选地,所述筛选模块,还设置为在合并处理后的结果中,将隶属于聚类正常值的结果划分到第一训练集中,以及将游离于所述聚类正常值之外的结果划分到第二训练集中。
可选地,所述装置还包括:构建模块,设置为根据所述第一训练集构建初始化模型;预测模块,设置为利用所述初始化模型对所述第二训练集进行预测;第二确定模块,设置为根据预测结果得到更新后的第二训练集;第三确定模块,设置为根据更新后的第二训练集和所述第一训练集得到人 群流量的预测模型,其中,将当前时刻所述目标基站的人群流量确定值输入到所述预测模型中,得到所述当前时刻之后的指定时刻的目标基站的人群流量。
可选地,所述目标基站的位置信息至少包括以下之一:用户标识ID、时间戳、目标基站标识ID;
接入所述目标基站的用户的所述历史位置信息至少包括以下之一:目标基站标识ID、目标基站纬度、目标基站经度。
在本发明实施例中,还提供了一种计算机存储介质,该计算机存储介质可以存储有执行指令,该执行指令用于执行上述实施例中的人群流量的确定方法的实现。
通过本发明实施例,能够根据目标基站的位置信息和接入所述目标基站的用户的历史位置信息确定所述目标基站的人群流量,采用上述技术方案,解决了相关技术中,对人群流量的预测方法存在检测结果不准确、成本高等诸多问题,进而不需要视频监控的介入才能实现对人群流量的确定,避免了过高的成本压力。
附图说明
此处所说明的附图用来提供对本发明的进一步理解,构成本申请的一部分,本发明的示意性实施例及其说明用于解释本发明,并不构成对本发明的不当限定。在附图中:
图1是根据本发明实施例的人群流量的确定方法的流程图;
图2是根据本发明实施例的人群流量的确定装置的结构框图;
图3是根据本发明实施例的人群流量的确定装置的另一结构框图;
图4为根据本发明实施例的人群流量的确定方法的流程图。
具体实施方式
下文中将参考附图并结合实施例来详细说明本发明。需要说明的是, 在不冲突的情况下,本申请中的实施例及实施例中的特征可以相互组合。
需要说明的是,本发明的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象,而不必用于描述特定的顺序或先后次序。
在本实施例中提供了一种人群流量的确定方法,图1是根据本发明实施例的人群流量的确定方法的流程图,如图1所示,该流程包括如下步骤:
步骤S102,获取目标基站的位置信息和接入目标基站的用户的历史位置信息;
步骤S104,将历史位置信息和位置信息合并处理,并根据合并处理后的结果确定目标基站的人群流量。
通过上述各个步骤,能够根据目标基站的位置信息和接入目标基站的用户的历史位置信息确定目标基站的人群流量,采用上述技术方案,解决了相关技术中,对人群流量的预测方法存在检测结果不准确、成本高等诸多问题,进而不需要视频监控的介入才能实现对人群流量的确定,避免了过高的成本压力。
实际上,不可能所有的历史数据都是正确完整的,当上述位置信息和历史位置信息不完整时,本发明实施例还提供了以下技术方案:根据处理后的结果确定目标基站的人群流量之前,上述方法还包括:根据离散点发现技术从合并处理后的结果中筛选出异常点,在一个可选示例中,可以通过以下方式筛选出异常点,在合并处理后的结果中,将隶属于聚类正常值的结果划分到第一训练集中,以及将游离于聚类正常值之外的结果划分到第二训练集中,其中,第二训练集中的元素可以理解为是上述异常点。
也就是说,由于基站信令数据是多年积累的结果,并不能保证每个基站区域的历史记录都覆盖全时间段,导致数据存在大量残缺;如果将残缺数据直接丢弃,则宝贵的数据资产将不能得到充分利用,同时训练出的模型准确率也会相应降低,采用本发明上述实施例的提供的技术方案,能够有效利用不完整残缺的数据。
本发明实施例的技术方案还可以对未来某一时刻的人群流量值进行预测,主要通过以下方案实现:根据第一训练集构建初始化模型;利用初始化模型对第二训练集进行预测,并根据预测结果得到更新后的第二训练集;根据更新后的第二训练集和第一训练集得到人群流量的预测模型,其中,将当前时刻目标基站的人群流量确定值输入到预测模型中,得到当前时刻之后的指定时刻的目标基站的人群流量,该部分的技术方案以下将结合优选实施例进行说明,此处不再赘述。
可选地,目标基站的位置信息至少包括以下之一:用户标识ID、时间戳、目标基站标识ID;接入目标基站的用户的历史位置信息至少包括以下之一:目标基站标识ID、目标基站纬度、目标基站经度。
通过上述技术方案,利用运营商手机信令数据对重点地段人群进行定位,则摆脱了视频监控的高成本,且无需采用高性能的计算设备进行视频分析、更没有监控死角,是当前最为有效的人群态势感知手段,具备了高准确率预测的基础,同时还能够对重点地段的人群密度能够提前预测、从而进一步提前布置,防范于未然。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到根据上述实施例的方法可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。基于这样的理解,本发明的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中,包括若干指令用以使得一台终端设备(可以是手机,计算机,服务器,或者网络设备等)执行本发明各个实施例的方法。
实施例2
在本实施例中还提供了一种人群流量的确定装置,该装置用于实现上述实施例及优选实施方式,已经进行过说明的不再赘述。如以下所使用的,术语“模块”可以实现预定功能的软件和/或硬件的组合。尽管以下实施例 所描述的装置较佳地以软件来实现,但是硬件,或者软件和硬件的组合的实现也是可能并被构想的。
图2是根据本发明实施例的人群流量的确定装置的结构框图,如图2所示,该装置包括:
获取模块20,设置为获取目标基站的位置信息和接入所述目标基站的用户的历史位置信息;
处理模块22,设置为将所述历史位置信息和所述位置信息合并处理;
第一确定模块24,设置为根据合并处理后的结果确定所述目标基站的人群流量。
通过本发明上述各个模块的综合作用,能够根据目标基站的位置信息和接入所述目标基站的用户的历史位置信息确定所述目标基站的人群流量,采用上述技术方案,解决了相关技术中,对人群流量的预测方法存在检测结果不准确、成本高等诸多问题,进而不需要视频监控的介入才能实现对人群流量的确定,避免了过高的成本压力。
图3是根据本发明实施例的人群流量的确定装置的另一结构框图,如图3所示,上述装置还包括:
筛选模块26,设置为根据离散点发现技术从合并处理后的结果中筛选出异常点。
可选地,筛选模块26,还设置为在合并处理后的结果中,将隶属于聚类正常值的结果划分到第一训练集中,以及将游离于所述聚类正常值之外的结果划分到第二训练集中。
如图3所示,上述装置还包括:构建模块28,设置为根据所述第一训练集构建初始化模型;预测模块30,设置为利用所述初始化模型对所述第二训练集进行预测;第二确定模块32,设置为根据预测结果得到更新后的第二训练集;第三确定模块34,设置为根据更新后的第二训练集和所述第一训练集得到人群流量的预测模型,其中,将当前时刻所述目标基站的人群流量确定值输入到所述预测模型中,得到所述当前时刻之后的指定时刻 的目标基站的人群流量。
可选地,目标基站的位置信息至少包括以下之一:用户标识ID、时间戳、目标基站标识ID;接入所述目标基站的用户的所述历史位置信息至少包括以下之一:目标基站标识ID、目标基站纬度、目标基站经度。
为了更好的理解上述人群流量的确定方法,以下结合优选实施例进行说明,但不用于限定本发明实施例。
优选实施例1
本发明所阐述的技术方案分为4个步骤阶段,图4为根据本发明实施例的人群流量的确定方法的流程图,如图4所示,包括以下阶段:
数据准备阶段:将现有的历史信令数据和基站地理位置数据进行汇聚,具体来说利用基站编号Cell ID,将以上两种数据进行连接,并进一步获取形如“某年-某月-某日-某时间段”,进入/停留在某基站范围内的人流统计值。
数据处理阶段:以上汇聚结果并不一定完全符合实际情况,因为运营商信令数据的日志记录受到各种情况影响,可能存在某些用户切换、位置定时上报信令的残缺。具体来说,就是由于历史数据不完整,存在部分“某些基站点的人流统计值远远低于其周边基站同期统计值/前期统计值;以及某些基站点的人流统计值远低于其前后时段的统计值”的现象,以上数据不能反应人群流动的真实情况,直接进行训练得出的将是不准确的模型。而直接抛弃这些时空切片的统计数据,一来浪费了数据资源,二来模型将缺少足够数据支撑,同样也不准确。本发明优选实施例利用离群点发现技术将这些异常点找出,进行标示,以备下一阶段有选择性地利用。具体来说,利用聚类技术将“某时间段-某地点的人流统计值”进行聚类,隶属于各个聚类的正常值直接放入TrainingSet训练集(相当于上述实施例的第一训练节)、游离于各个聚类之外的异常值被放入AbnormalSet辅助训练集(相当于上述实施例的第二训练集)。
模型构建阶段:
1)利用TrainingSet训练集的样本,通过机器学习方法构建初始化预测模型Flow_Model0,并计算模型的拟合指标R;Flow_Model0的形式为:
y=XTW    (公式1)
其中,X是输入特征的向量形式,W是对应特征的权重向量,y指的是当特定输入X、模型参数为W的时候,目标区域人流量的期望值。
2)迭代执行如下循环,直到R指标不再提升:
利用当前模型Flow_Model0对AbnormalSet做出预测,用预测值代替AbnormalSet中原有因变量y,与原有TrainingSet合并构造最新训练集TrainingAllSet.
利用TrainingAllSet训练集的样本,通过机器学习方法构建预测模型,更新到Flow_Model0,并计算拟合系数R.
将最后得到的模型Flow_Model0命名为Flow_Model1。
模型应用阶段:利用Flow_Model对新进入的待预测数据进行预测,根据当前时刻、目标位置周边的人流统计值,求出未来某时刻、目标位置的人群流量值。
与现有技术相比,本发明优选实施例不需要视频监控介入,避免了过高的成本压力;能够处理残缺的信令历史数据,保护了数据资产、同时能够提高预测的精度,克服了现有视频监控流量计数技术的视觉盲点、以及传统运营商手机历史信令记录部分缺失导致的负面影响,提出了一种利用运营商手机信令数据对公共场所人群流量进行预测的方法。该技术可以在训练数据有残缺的情况下依旧良好工作,帮助城市管理者对城市重点区域的人流密度实现提前预判。为城市公共安全、大型活动组织等提供数据支撑。
优选实施例2
在介绍本发明优选实施例2的技术方案之前,先对以下技术术语进行说明:
定义聚簇质心C-Center:某批多元数据集合的几何中心点
定义点-聚簇距离:在多维空间中,某点与某聚簇质心的欧氏距离
定义聚簇半径:某聚簇中各点到质心的平均欧式距离
定义某远离某聚簇:某点与该聚簇的距离大于其3倍半径
定义离群点:某点与所有聚簇都远离
定义映射函数S(X),输入参数X是m维向量,输出量Y也是m维向量。
令Xi为向量X的第i分量,Yi为向量Y的第i分量,则存在映射关系:Yi=Xi
定义两个k维向量W(1),W(2)的距离distance(W(1),W(2)),
其中W(1)j和W(2)j分别是两个向量的第j分量值:
Figure PCTCN2017084388-appb-000001
定义两个k维向量W(1),W(2)“近似相等”条件:
distance(W(1),W(2))<ε
其中ε是外部定义的一个很小的浮点数。
令历史数据有m条训练样本,每条样本有n个特征,其形式为:
Figure PCTCN2017084388-appb-000002
其中,xij表示第i历史训练数据样本的第j特征值,Yi表示第i训练数据样本的统计值。
模型判定系数R:用来衡量训练出的模型性能。令训练集的行数为n;令Y为训练集的实际因变量向量,则Yk对应就是它的每个分量;令Z为训练集作用在现有模型上预测出的结果变量,则Zk对应就是它的每个分 量。令
Figure PCTCN2017084388-appb-000003
则R的计算公式为:
Figure PCTCN2017084388-appb-000004
具体实施步骤如下:
步骤101、将运营商提供了用户位置更新数据(包含用户ID,时间戳,基站ID)和基站的位置数据(基站ID,基站经度,基站纬度)进行关联,将用户历史位置信息与基站地理位置结合,按时间段统计,形成人流空间分布的时间切片,汇总信息作为后续步骤的训练样本集,格式如下:
Figure PCTCN2017084388-appb-000005
Figure PCTCN2017084388-appb-000006
步骤201、在步骤101工作完成后,由于运营商数据历史信令记录不定完备,以及存在信令基站漂移的固有技术问题,可能存在异常数据。因此本方案利用离群点发现技术将这些异常点找出,进行标示,以备下一阶段有选择性地利用。具体的:
利用外部业务知识设置参数k,这里将k设置为5。
随机选择k个对象,并且所选择的每个对象都代表一个组的初始均值或初始的组中心值;对剩余的每个对象,根据其与各个组初始均值的欧式距离,将它们分配给最近的(最相似)小组;然后,重新计算每个小组新的均值。
以上过程不断重复,质心逐渐向最合适的位置靠拢;直到各组质心变化趋向为0,此时到达稳态:所有的对象在K组分布中都找到离自己最近的组,完成聚类;求出每个聚簇的质心和半径。
重新遍历历史数据集,如果其与所有聚簇都远离(距离质心超过3倍半径),则认为是离群点,构造出数据集AbnormalSet1;非离群点构造出训练数据集TrainingSet.
将海量时空统计数据去除因变量属性“人流计数值”,重复以上前述过程,构造数据集AbnormalSet2。
最后将数据集AbnormalSet1和AbnormalSet2集合求交,获得离群数据集AbnormalSet,这部分数据可以供下一阶段训练训练使用。
步骤301、步骤201完成后,首先利用正常样本训练集TrainingSet构建初始化模型Flow_Model0,具体的:
初始化循环计数器k=0;记W(k)为第k轮循环后的权重向量W。
初始化训练步长参数α=0.1;记α(k)为第k轮循环后的步长参数。
初始化训练衰减λ参数(=0.95)。
对n+1维权重向量W进行初始化:对每个权重分量Wi生成(0,1)之间的随机数并赋值,此时W记为W(0)。
进入循环,循环跳出条件为:循环执行次数大于外部参数t,或者之前连续3次循环,W(k-2)、W(k-1)、W(k)都彼此“近似相等”:
通过如下公式迭代计算W(k+1):
Wn+1,1(k+1)=Wn+1,1(k)+α(k)*Xn+1,mT*(Ym,1-(Xm,n+1*Wn+1,1(k)))(公式2)
通过如下公式迭代计算α(k+1):
α(k+1)=α(k)×λk          (公式3)
跳出循环后的W(k)作为训练结果Whistory,α(k)作为最后步长参数αhistory。
在TrainingSet上,基于当前模型(参数即为Whistory)和公式4,计算判定系数Rtraining.。
步骤302、利用AbnormalSet的未完整数据的内部分布特征提升Flow_Model0模型的准确性,具体的:
对AbnormalSet中的每一条训练数据Xabnormal,利用现有模型给出预测值ypredicti,将<Xabnormal,ypredicti>作为新的训练样本,将该样本放入TrainingSet.
在新的TrainingSet,利用步骤301描述的方法进行训练,得出新的判定系数Rnew,如果Rnew>Rtraining,将<Xabnormal,ypredicti>放入样本集NewSet.
在遍历全部AbnormalSet后,将NewSet与TrainingSet重新合并,再次利用步骤301描述的方法,形成最后的模型Flow_Model1。
步骤401、系统上线后,对于新观测数据(记为向量X),代入训练好的模型Flow_Model1中的参数向量W,利用公式1求出目标基站区域的人流期望值。对较大区域(面积大于某一基站覆盖区域)进行人流预测时,只要将该区域所在的多个基站人流期望值进行相加即可。
综上所述,本发明实施例达到了以下技术效果:不需要视频监控介入,避免了过高的成本压力;能够处理残缺的信令历史数据,保护了数据资产、同时能够提高预测的精度,克服了现有视频监控流量计数技术的视觉盲点、以及传统运营商手机历史信令记录部分缺失导致的负面影响。
本发明的实施例还提供了一种存储介质。可选地,在本实施例中,上述存储介质可以被设置为存储用于执行以下步骤的程序代码:
S1,获取目标基站的位置信息和接入所述目标基站的用户的历史位置信息;
S2,将所述历史位置信息和所述位置信息合并处理;
S3,根据合并处理后的结果确定所述目标基站的人群流量。
可选地,在本实施例中,上述存储介质可以包括但不限于:U盘、只读存储器(ROM,Read-Only Memory)、随机存取存储器(RAM,Random Access Memory)、移动硬盘、磁碟或者光盘等各种可以存储程序代码的介质。
可选地,本实施例中的具体示例可以参考上述实施例及可选实施方式中所描述的示例,本实施例在此不再赘述。
显然,本领域的技术人员应该明白,上述的本发明的各模块或各步骤可以用通用的计算装置来实现,它们可以集中在单个的计算装置上,或者分布在多个计算装置所组成的网络上,可选地,它们可以用计算装置可执行的程序代码来实现,从而,可以将它们存储在存储装置中由计算装置来执行,并且在某些情况下,可以以不同于此处的顺序执行所示出或描述的 步骤,或者将它们分别制作成各个集成电路模块,或者将它们中的多个模块或步骤制作成单个集成电路模块来实现。这样,本发明不限制于任何特定的硬件和软件结合。
以上所述仅为本发明的优选实施例而已,并不用于限制本发明,对于本领域的技术人员来说,本发明可以有各种更改和变化。凡在本发明的精神和原则之内,所作的任何修改、等同替换、改进等,均应包含在本发明的保护范围之内。
工业实用性
本发明实施例提供的上述技术方案,能够应用于人群流程的确定过程中,根据目标基站的位置信息和接入所述目标基站的用户的历史位置信息确定所述目标基站的人群流量,采用上述技术方案,解决了相关技术中,对人群流量的预测方法存在检测结果不准确、成本高等诸多问题,进而不需要视频监控的介入才能实现对人群流量的确定,避免了过高的成本压力。

Claims (10)

  1. 一种人群流量的确定方法,包括:
    获取目标基站的位置信息和接入所述目标基站的用户的历史位置信息;
    将所述历史位置信息和所述位置信息合并处理,并根据合并处理后的结果确定所述目标基站的人群流量。
  2. 根据权利要求1所述的方法,其中,根据处理后的结果确定所述目标基站的人群流量之前,所述方法还包括:
    根据离散点发现技术从合并处理后的结果中筛选出异常点。
  3. 根据权利要求2所述的方法,其中,根据离散点发现技术从合并处理后的结果中筛选出异常点,包括:
    在合并处理后的结果中,将隶属于聚类正常值的结果划分到第一训练集中,以及将游离于所述聚类正常值之外的结果划分到第二训练集中。
  4. 根据权利要求3所述的方法,其中,所述方法还包括:
    根据所述第一训练集构建初始化模型;
    利用所述初始化模型对所述第二训练集进行预测,并根据预测结果得到更新后的第二训练集;
    根据更新后的第二训练集和所述第一训练集得到人群流量的预测模型,其中,将当前时刻所述目标基站的人群流量确定值输入到所述预测模型中,得到所述当前时刻之后的指定时刻的目标基站的人群流量。
  5. 根据权利要求1-4任一项所述的方法,其中,所述目标基站 的位置信息至少包括以下之一:用户标识ID、时间戳、目标基站标识ID;
    接入所述目标基站的用户的所述历史位置信息至少包括以下之一:目标基站标识ID、目标基站纬度、目标基站经度。
  6. 一种人群流量的确定装置,包括:
    获取模块,设置为获取目标基站的位置信息和接入所述目标基站的用户的历史位置信息;
    处理模块,设置为将所述历史位置信息和所述位置信息合并处理;
    第一确定模块,设置为根据合并处理后的结果确定所述目标基站的人群流量。
  7. 根据权利要求6所述的装置,其中,所述装置还包括:
    筛选模块,设置为根据离散点发现技术从合并处理后的结果中筛选出异常点。
  8. 根据权利要求7所述的装置,其中,所述筛选模块,还设置为在合并处理后的结果中,将隶属于聚类正常值的结果划分到第一训练集中,以及将游离于所述聚类正常值之外的结果划分到第二训练集中。
  9. 根据权利要求8所述的装置,其中,所述装置还包括:
    构建模块,设置为根据所述第一训练集构建初始化模型;
    预测模块,设置为利用所述初始化模型对所述第二训练集进行预测;
    第二确定模块,设置为根据预测结果得到更新后的第二训练集;
    第三确定模块,设置为根据更新后的第二训练集和所述第一训练集得到人群流量的预测模型,其中,将当前时刻所述目标基站的人群流量确定值输入到所述预测模型中,得到所述当前时刻之后的指定时刻的目标基站的人群流量。
  10. 根据权利要求6-9任一项所述的装置,其中,所述目标基站的位置信息至少包括以下之一:用户标识ID、时间戳、目标基站标识ID;
    接入所述目标基站的用户的所述历史位置信息至少包括以下之一:目标基站标识ID、目标基站纬度、目标基站经度。
PCT/CN2017/084388 2016-05-23 2017-05-15 人群流量的确定方法及装置 WO2017202226A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610344118.5 2016-05-23
CN201610344118.5A CN107423742A (zh) 2016-05-23 2016-05-23 人群流量的确定方法及装置

Publications (1)

Publication Number Publication Date
WO2017202226A1 true WO2017202226A1 (zh) 2017-11-30

Family

ID=60411042

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/084388 WO2017202226A1 (zh) 2016-05-23 2017-05-15 人群流量的确定方法及装置

Country Status (2)

Country Link
CN (1) CN107423742A (zh)
WO (1) WO2017202226A1 (zh)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108647735A (zh) * 2018-05-15 2018-10-12 广州杰赛科技股份有限公司 用户出行规律分析方法、装置、设备及存储介质
CN110621026A (zh) * 2019-02-18 2019-12-27 北京航空航天大学 一种基站流量多时刻预测方法
CN110968341A (zh) * 2018-09-30 2020-04-07 上海汽车集团股份有限公司 一种设置空调参数的方法和装置
CN111709778A (zh) * 2020-05-29 2020-09-25 北京百度网讯科技有限公司 出行流量预测方法、装置、电子设备和存储介质
CN111985568A (zh) * 2020-08-21 2020-11-24 浪潮天元通信信息系统有限公司 数据处理方法及装置、电子设备
CN112000755A (zh) * 2020-08-14 2020-11-27 青岛市城市规划设计研究院 一种基于手机信令数据的区域出行廊道识别方法
CN112446406A (zh) * 2019-09-05 2021-03-05 中移信息技术有限公司 应急点确定方法、装置以及存储介质
CN112600728A (zh) * 2020-12-07 2021-04-02 华东交通大学理工学院 一种基于大数据的5g移动基站流量预测分析系统
CN113052327A (zh) * 2021-03-30 2021-06-29 北京骑胜科技有限公司 数据处理方法、装置、可读存储介质和电子设备
CN114358375A (zh) * 2021-11-29 2022-04-15 重庆邮电大学 一种基于大数据的人群密度预测方法及系统

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108876033A (zh) * 2018-06-13 2018-11-23 华南理工大学 一种基于多源数据融合的客流爆点人流量预测方法
CN109640249B (zh) * 2018-11-27 2020-08-11 佛山科学技术学院 一种基于大数据的商场人流量预测系统
CN109816175B (zh) * 2019-02-19 2022-02-25 迪爱斯信息技术股份有限公司 基于可变分量和结构风险最小的区域人数预测方法及系统
CN112381559A (zh) * 2020-10-14 2021-02-19 浪潮软件股份有限公司 一种基于非监督机器学习算法的烟草零售商细分方法
CN114283373A (zh) * 2021-11-16 2022-04-05 阿里巴巴达摩院(杭州)科技有限公司 人群态势获取方法、装置、存储介质及系统
CN117332886B (zh) * 2023-08-29 2024-08-09 齐鲁工业大学(山东省科学院) 基于对比学习的城市流量预测方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1852353A (zh) * 2005-10-18 2006-10-25 华为技术有限公司 一种处理用户数据的方法、装置以及系统
JP2008242556A (ja) * 2007-03-26 2008-10-09 Ntt Comware Corp 移動体監視システム
CN101394645A (zh) * 2008-10-30 2009-03-25 中国移动通信集团北京有限公司 一种目标区域移动通信终端用户的流量统计方法及系统
CN102982596A (zh) * 2012-11-07 2013-03-20 中兴通讯股份有限公司 人流量的统计方法及装置
CN103906086A (zh) * 2012-12-26 2014-07-02 中国移动通信集团江苏有限公司 一种基于移动网络的人员流向监测方法和装置

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8121618B2 (en) * 2009-10-28 2012-02-21 Digimarc Corporation Intuitive computing methods and systems
CN103793613B (zh) * 2013-12-13 2017-01-11 北京航空航天大学 基于回归分析和rbf神经网络的退化数据缺失插补方法
CN104484993B (zh) * 2014-11-27 2017-01-18 北京交通大学 用于交通小区划分的手机信令信息的处理方法
CN104464291B (zh) * 2014-12-08 2017-02-01 杭州智诚惠通科技有限公司 一种交通流量预测方法和系统
CN105336097B (zh) * 2015-11-17 2017-11-14 中国联合网络通信集团有限公司 人口流动轨迹的交通预警方法及装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1852353A (zh) * 2005-10-18 2006-10-25 华为技术有限公司 一种处理用户数据的方法、装置以及系统
JP2008242556A (ja) * 2007-03-26 2008-10-09 Ntt Comware Corp 移動体監視システム
CN101394645A (zh) * 2008-10-30 2009-03-25 中国移动通信集团北京有限公司 一种目标区域移动通信终端用户的流量统计方法及系统
CN102982596A (zh) * 2012-11-07 2013-03-20 中兴通讯股份有限公司 人流量的统计方法及装置
CN103906086A (zh) * 2012-12-26 2014-07-02 中国移动通信集团江苏有限公司 一种基于移动网络的人员流向监测方法和装置

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108647735A (zh) * 2018-05-15 2018-10-12 广州杰赛科技股份有限公司 用户出行规律分析方法、装置、设备及存储介质
CN108647735B (zh) * 2018-05-15 2021-01-12 广州杰赛科技股份有限公司 用户出行规律分析方法、装置、设备及存储介质
CN110968341A (zh) * 2018-09-30 2020-04-07 上海汽车集团股份有限公司 一种设置空调参数的方法和装置
CN110968341B (zh) * 2018-09-30 2024-01-19 上海汽车集团股份有限公司 一种设置空调参数的方法和装置
CN110621026A (zh) * 2019-02-18 2019-12-27 北京航空航天大学 一种基站流量多时刻预测方法
CN110621026B (zh) * 2019-02-18 2023-09-05 北京航空航天大学 一种基站流量多时刻预测方法
CN112446406A (zh) * 2019-09-05 2021-03-05 中移信息技术有限公司 应急点确定方法、装置以及存储介质
CN111709778B (zh) * 2020-05-29 2023-04-07 北京百度网讯科技有限公司 出行流量预测方法、装置、电子设备和存储介质
CN111709778A (zh) * 2020-05-29 2020-09-25 北京百度网讯科技有限公司 出行流量预测方法、装置、电子设备和存储介质
CN112000755A (zh) * 2020-08-14 2020-11-27 青岛市城市规划设计研究院 一种基于手机信令数据的区域出行廊道识别方法
CN112000755B (zh) * 2020-08-14 2024-03-12 青岛市城市规划设计研究院 一种基于手机信令数据的区域出行廊道识别方法
CN111985568A (zh) * 2020-08-21 2020-11-24 浪潮天元通信信息系统有限公司 数据处理方法及装置、电子设备
CN111985568B (zh) * 2020-08-21 2024-01-30 浪潮通信信息系统有限公司 数据处理方法及装置、电子设备
CN112600728A (zh) * 2020-12-07 2021-04-02 华东交通大学理工学院 一种基于大数据的5g移动基站流量预测分析系统
CN113052327A (zh) * 2021-03-30 2021-06-29 北京骑胜科技有限公司 数据处理方法、装置、可读存储介质和电子设备
CN113052327B (zh) * 2021-03-30 2024-04-19 北京骑胜科技有限公司 数据处理方法、装置、可读存储介质和电子设备
CN114358375A (zh) * 2021-11-29 2022-04-15 重庆邮电大学 一种基于大数据的人群密度预测方法及系统
CN114358375B (zh) * 2021-11-29 2024-05-24 重庆邮电大学 一种基于大数据的人群密度预测方法及系统

Also Published As

Publication number Publication date
CN107423742A (zh) 2017-12-01

Similar Documents

Publication Publication Date Title
WO2017202226A1 (zh) 人群流量的确定方法及装置
Singh et al. Crowd forecasting based on WiFi sensors and LSTM neural networks
EP3132592B1 (en) Method and system for identifying significant locations through data obtainable from a telecommunication network
KR101825023B1 (ko) 리스크 조기 경보 방법 및 장치
CN113276874B (zh) 一种车辆行驶轨迹处理方法及相关装置
She et al. Weighted network Voronoi Diagrams for local spatial analysis
KR101937940B1 (ko) 빅 데이터를 활용한 단계별 방범 cctv 위치 선정 방법
CN110837582A (zh) 数据关联方法、装置、电子设备和计算机可读存储介质
CN110335399B (zh) 蓝牙门禁控制方法、计算机终端和计算机可读存储介质
Kulkarni et al. Mobidict: A mobility prediction system leveraging realtime location data streams
CN109873790A (zh) 网络安全检测方法、装置以及计算机可读存储介质
CN114970621A (zh) 异常聚集事件的检测方法、装置、电子设备和存储介质
CN114037140A (zh) 预测模型训练、数据预测方法、装置、设备及存储介质
CN116337093A (zh) 一种路径规划方法、装置、设备、存储介质及产品
CN117010537A (zh) 目标区域预测方法、装置、计算机设备和存储介质
CN110399537B (zh) 一种基于人工智能技术的警情时空预测方法
WO2022267455A1 (zh) 欺诈用户集中区域的预测方法、装置、设备及存储介质
CN108416293A (zh) 一种特殊群体聚集行为早期检测及聚集地预测方法和系统
CN111949840B (zh) 一种基于物联网数据的拓扑图结构的构建方法、装置
CN115345324A (zh) 故障定位方法、装置、设备、存储介质及产品
CN116227929B (zh) 通信数据的分析方法、装置、设备及存储介质
CN115412851A (zh) 室内定位方法、装置、服务器及存储介质
CN116266946A (zh) 一种基站位置的确定方法、装置及设备
CN119694590B (zh) 一种基于大数据的流行性病原体预测方法
TWI844202B (zh) 用於計算起訖路徑車流量之系統、方法以及其電腦程式產品

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17802072

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 17802072

Country of ref document: EP

Kind code of ref document: A1