An intelligent management system for exhibition halls

Abstract

This invention relates to exhibition hall management technology and discloses an intelligent management system for exhibition halls, including at least one radio frequency (RF) module, a control module, and a client. The RF module collects exhibition hall information and transmits it to the control module. The control module stores and processes the received exhibition hall information and transmits the processed exhibition hall behavior information to the client. This exhibition hall management system does not disturb visitors, eliminates the need for frequent interaction with individual booths, and recommends suggested routes based on user behavior data, thereby effectively improving visitor efficiency.

Description

Technical Field

[0001] This invention relates to exhibition hall management technology, and more particularly to an intelligent management system for exhibition halls. Background Technology

[0002] The current exhibition industry pays insufficient attention to visitor behavior and lacks effective means to help visitors and exhibitors plan and record key points of their visits. Some ideas, even those not yet implemented, focus on frequent interaction with visitors to obtain information, which is highly intrusive and naturally encounters many obstacles in promotion. Due to space limitations, exhibitors in the current exhibition hall industry are mainly single companies or departments, resulting in a single business chain for visitors and a failure to retain valuable visitor information. Occasionally, camera solutions are promoted, but due to privacy concerns, they can only use infrared sensors to capture heat distribution, providing heat maps for a certain area, but cannot provide customized data services for individual visitors or identify more detailed behaviors.

[0003] For example, the exhibition hall system in the prior art, patent application number CN201610447843.5, only relies on some basic recommendation behaviors for user visiting behavior and cannot effectively analyze user behavior data, which results in users spending a lot of time visiting exhibition halls. Summary of the Invention

[0004] This invention addresses the problem that existing exhibition hall systems only rely on basic recommendations to analyze user behavior, resulting in users spending excessive time visiting exhibition halls. It provides an intelligent management system for exhibition halls.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] An intelligent management system for exhibition halls includes at least one radio frequency module, a control module, and a client;

[0007] The radio frequency module is used to collect exhibition hall information and transmit the collected exhibition hall information to the control module;

[0008] The control module is used to store and process the received exhibition hall information; and to transmit the processed exhibition hall behavior information to the client.

[0009] As a preferred option, the control module's information analysis and processing for the exhibition hall includes:

[0010] Information acquisition: Exhibition hall information is obtained through the radio frequency module;

[0011] The exhibition hall information is processed using clustering algorithms to obtain the similarity of visitor behavior characteristics.

[0012] Information on visitor behavior at exhibition halls is obtained by analyzing the similarity of visitor behavior characteristics.

[0013] As a preferred embodiment, the control module's information analysis and processing of the exhibition hall also includes: forming a behavioral self-portrait of the exhibition hall information; obtaining a behavioral self-portrait of the exhibition hall information through the exhibition hall behavioral information, and providing the behavioral self-portrait to the user terminal.

[0014] As a preferred approach, the formation of a self-portrait of information behavior at exhibition halls includes:

[0015] Select audience P x The exhibition hall information behavior feature matrix was used to calculate the audience P. x The mean distance d(P) from the exhibition hall information behavior feature matrix to the k cluster centers x ,a j ),

[0016]

[0017] Among them, a j Let A be a point mass of class Aj, i.e., the class center;

[0018] The mean distance d(P) of the k cluster centers is calculated. x ,a j Distance normalization,

[0019]

[0020] Where d j * Between [0, 1];

[0021] d max =max{d(P x ,a1),d(P x ,a2), …d(P x ,a k )}

[0022] d min =min{d(P x ,a1),d(P x ,a2), …d(P x ,a k )}

[0023] d j =d(P x ,a j );d(Px ,a j The smaller the distance, the higher the score for that category;

[0024] Using K categories as axes, the score d of each dimension j * Mark and connect the axes to form a radar chart, and use the radar chart as a self-portrait of the information behavior of the exhibition hall.

[0025] As a preferred option, output distance d(x) i ,a j );

[0026] Step 1: Composition of the feature matrix. A two-dimensional matrix is ​​formed by collecting information from each visitor Px at different booths. The columns of the two-dimensional matrix are the EPC list of each visitor Px; the rows are the collected information. The feature matrix is ​​obtained by associating the two-dimensional matrix with the information stored on the server.

[0027] Step 2: Select k samples from the feature matrix as initial cluster centers; a = a1, a2, ..., a k For each sample x in the feature matrix i The distance d(x) from the k cluster centers is calculated using Euclidean distance. i ,a j );

[0028]

[0029] Step 3, assign sample xi to d min In the corresponding class,

[0030] d min =min{d(x i ,a1),d(x i ,a2),…d(x i ,a k )}

[0031] And recalculate the cluster center a for each category Aj. j ,

[0032] a j That is, it belongs to category A. j The centroid of all samples x in the dataset;

[0033] Step 4, determine the centroid a of the sample. j ; when a j If the error of the change reaches the threshold, then the output distance d(x) is... i ,a j Otherwise, repeat steps 2 and 3.

[0034] Preferably, the information collected by the radio frequency module includes the equipment information of the exhibition hall, the tag number of the exhibition hall, the time information corresponding to the tag number of the exhibition hall, the signal strength value corresponding to the tag number of the exhibition hall, and the tag distance value corresponding to the tag number of the exhibition hall.

[0035] The tag signal strength value is RSSI. x101 RSSI x102 RSSI x103 ...RSSI x10n The tag distance value is inversely proportional to the tag signal strength value, and the tag distance values ​​are L... x101 L x102 L x103 ...L x10n ;

[0036] Preferably, the server obtains the EPC number list from the exhibition hall information by analyzing the time information corresponding to the exhibition hall's tag number. i According to the EPC number list i Thus, the time difference ΔT between adjacent moments in the EPC number list and the entry time t of the EPC number list are obtained. x Departure time t of adjacent EPC number list y The duration of stay in adjacent EPC number lists and the duration of visit to each EPC number list;

[0037] t i List of EPC numbers within a given time frame i The time difference between adjacent moments is ΔT, where i is a positive integer, and ΔT = t. i -t i-1 ;

[0038] EPC list entry time t x and departure time t y And the duration of stay is obtained through the EPC number list. i Get the entry time t of the adjacent EPC list. x The departure time t of the adjacent EPC list y and the duration t of the EPC list at adjacent times. y -t x .

[0039] As a preferred option, the booth visit duration information includes a list of each booth's visit duration, the total visit duration of all participating booths, the average visit duration per booth, and a list n of the top n booths by visit duration.

[0040] Through EPC xThe audience P x Booth U n The duration of the visit is δT m m≤n; a list of booths {1, 2, ..., k} and a list of booth durations {δT1, δT2, ..., δT} for each booth. k};

[0041] Total number of booths visited: M sum ,

[0042] Total exhibition duration sum ,

[0043] Average visit duration per booth (AveTime) x AveTime x =Time sum ÷M sum List of booths with the longest visit duration (listn(U)) a1 U a2 ,...U an );U an =rank{δT1, δT2,...δT k}

[0044] Preferably, the information collected by the radio frequency module also includes industry labels or product category labels for different booths; the server determines the visitor P based on the industry labels or product category labels for different booths collected by the video module. x A sorted list of industry types for visitors' preferences;

[0045] The industry label or product category label for different booths is H.

[0046]

[0047] Among them, the total number of industries x ≤ the total number of booths M sum H1(U1,U2...U n ) is the category label for industry 1, and similarly, H x (U o+1 U o+2 ...U p ) represents the category label for industry 1; U1, U2...U n List of booths for Industry 1; U n+1 U n+2 ...U n+m This is the list of booths for Industry 2; similarly for U. o+1 U o+2 ...U o+p List of booths for industry x;

[0048] The classification of different booth weight coefficients, the filtering rules for the duration of each booth, the weight coefficient S of each booth, and the weighting based on the visit duration δT. x Set up y categories, as follows:

[0049]

[0050] Among them, T level Based on the duration of the visit, T level0 This is the initial visit duration level;

[0051] Audience P x The visitor preferences are ranked by industry type; the industry category H is combined with the weight coefficient s to perform a weighted summation, resulting in visitor P. x A list sorted by industry type for visitor preferences.

[0052]

[0053] Among them, list sx Industry type sorted list; s U1 These are the weighting coefficients for U1.

[0054] This invention, by adopting the above technical solutions, has significant technical effects:

[0055] The exhibition hall system of this invention will not disturb visitors. By analyzing user behavior data, it recommends user tour routes. Visitors do not need to frequently interact with each booth, effectively improving visitor efficiency.

[0056] The exhibition hall system designed by this invention is low in cost, the hardware equipment is reusable, and the layout of the exhibition hall can be optimized when combined with actual exhibition booths.

[0057] The exhibition hall system of this invention helps visitors automatically collect behavioral data without causing too much disturbance, further assisting business exchanges and information communication. Attached Figure Description

[0058] Figure 1 This is a flowchart of the present invention.

[0059] Figure 2 This is a schematic diagram of the visitor flow of this invention;

[0060] Figure 3 This is a schematic diagram of the suggested visitor route for this invention;

[0061] Figure 4 This is a schematic diagram of distance calculation according to the present invention;

[0062] Figure 5This is a schematic diagram of the feature matrix formation of the present invention. Detailed Implementation

[0063] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0064] Example 1

[0065] An intelligent management system for exhibition halls includes at least one radio frequency module, a control module, and a client; the radio frequency module is used to collect exhibition hall information and transmit the collected exhibition hall information to the control module.

[0066] The control module is used to store and process the received exhibition hall information; and to transmit the processed exhibition hall behavior information to the client.

[0067] For client input to the control module, it can be done via QR code scanning, mini-program, webpage, and app; for client signals received from the control module, the received information can be displayed via mini-program, app, and webpage.

[0068] The control module's information analysis and processing for the exhibition hall includes:

[0069] Information acquisition: Exhibition hall information is obtained through the radio frequency module;

[0070] The exhibition hall information is processed using clustering algorithms to obtain the similarity of visitor behavior characteristics.

[0071] Information on visitor behavior at exhibition halls is obtained by analyzing the similarity of visitor behavior characteristics.

[0072] The information collected by the radio frequency module includes equipment information of the exhibition hall, tag number of the exhibition hall, time information corresponding to the tag number of the exhibition hall, signal strength value corresponding to the tag number of the exhibition hall, and tag distance value corresponding to the tag number of the exhibition hall;

[0073] The tag signal strength value is RSSI. x101 RSSI x102 RSSI x103 ...RSSI x10n The tag distance value is inversely proportional to the tag signal strength value, and the tag distance values ​​are L... x101 L x102 L x103 ...L x10n ;

[0074] The server retrieves the list of EPC numbers from the exhibition hall information by analyzing the time information corresponding to the exhibition hall's tag number. iAccording to the EPC number list i Thus, the time difference ΔT between adjacent moments in the EPC number list and the entry time t of the EPC number list are obtained. x Departure time t of adjacent EPC number list y The duration of stay in adjacent EPC number lists and the duration of visit to each EPC number list;

[0075] t i List of EPC numbers within a given time frame i The time difference between adjacent moments is ΔT, where i is a positive integer, and ΔT = t. i -t i-1 ;

[0076] EPC list entry time t x and departure time t y And the duration of stay is obtained through the EPC number list. i Get the entry time t of the adjacent EPC list. x The departure time t of the adjacent EPC list y and the duration t of the EPC list at adjacent times. y -t x .

[0077] The booth visit duration information includes a list of each booth's duration, the total duration of all participating booths, the average visit duration per booth, and a list n of the top n booths by visit duration.

[0078] Through EPC x The audience P x Booth U n The duration of the visit is δT m m≤n; a list of booths {1, 2, ..., k} and a list of booth durations {δT1, δT2, ..., δT} for each booth. k};

[0079] Total number of booths visited: M sum ,

[0080] Total exhibition duration sum ,

[0081] Average visit duration per booth (AveTime) x AveTime x =Time sum ÷M sum List of booths with the longest visit duration (listn(U)) a1 Ua2 ,...U an );U an =rank{δT1, δT2,...δT k}

[0082] The information collected by the radio frequency module also includes industry labels or product category labels for different booths; the server determines the visitor P based on the industry labels or product category labels for different booths collected by the video module. x A sorted list of industry types for visitors' preferences;

[0083] The industry label or product category label for different booths is H.

[0084]

[0085] Among them, the total number of industries x ≤ the total number of booths M sum H1(U1,U2...U n ) is the category label for industry 1, and similarly, H x (U o+1 U o+2 ...U p ) represents the category label for industry 1; U1, U2...U n List of booths for Industry 1; U n+1 U n+2 ...U n+m This is the list of booths for Industry 2; similarly for U. o+1 U o+2 ...U o+p List of booths for industry x;

[0086] The classification of different booth weight coefficients, the filtering rules for the duration of each booth, the weight coefficient S of each booth, and the weighting based on the visit duration δT. x Set up y categories, as follows:

[0087]

[0088] Among them, T level Based on the duration of the visit, T level0 This is the initial visit duration level;

[0089] Audience P x The visitor preferences are ranked by industry type; the industry category H is combined with the weight coefficient s to perform a weighted summation, resulting in visitor P. x A list sorted by industry type for visitor preferences.

[0090]

[0091] Among them, listsx Industry type sorted list; s U1 These are the weighting coefficients for U1.

[0092] Example 2

[0093] Based on Example 1, the control module of this example further includes the following for analyzing and processing exhibition hall information: forming a behavioral self-portrait of exhibition hall information; obtaining a behavioral self-portrait of exhibition hall information through exhibition hall behavioral information, and providing the behavioral self-portrait to the user terminal.

[0094] The formation of a self-portrait of information behavior at exhibition halls includes:

[0095] Select audience P x The exhibition hall information behavior feature matrix was used to calculate the audience P. x The mean distance d(P) from the exhibition hall information behavior feature matrix to the k cluster centers x ,a j ),

[0096]

[0097] Among them, a j Let A be a point mass of class Aj, i.e., the class center;

[0098] The mean distance d(P) of the k cluster centers is calculated. x ,a j Distance normalization,

[0099]

[0100] Where d j * Between [0, 1];

[0101] d max =max{d(P x ,a1),d(P x ,a2), …d(P x ,a k )}

[0102] d min =min{d(P x ,a1),d(P x ,a2), …d(P x ,a k )}

[0103] d j =d(P x ,a j );d(P x ,a jThe smaller the distance, the higher the score for that category;

[0104] Using K categories as axes, the score d of each dimension j * Mark and connect the axes to form a radar chart, and use the radar chart as a self-portrait of the information behavior of the exhibition hall.

[0105] Output distance d(x) i ,a j );

[0106] Step 1: Composition of the feature matrix. A two-dimensional matrix is ​​formed by collecting information from each visitor Px at different booths. The columns of the two-dimensional matrix are the EPC list of each visitor Px; the rows are the collected information. The feature matrix is ​​obtained by associating the two-dimensional matrix with the information stored on the server.

[0107] Step 2: Select k samples from the feature matrix as initial cluster centers; a = a1, a2, ..., a k For each sample x in the feature matrix i The distance d(x) from the k cluster centers is calculated using Euclidean distance. i ,a j );

[0108]

[0109] Step 3, assign sample xi to d min In the corresponding class,

[0110] d min =min{d(x i ,a1),d(x i ,a2),…d(x i ,a k )}

[0111] And recalculate the cluster center a for each category Aj. j ,

[0112] a j That is, it belongs to category A. j The centroid of all samples x in the dataset;

[0113] Step 4, determine the centroid a of the sample. j ; when a j If the error of the change reaches the threshold, then the output distance d(x) is... i ,a j Otherwise, repeat steps 2 and 3.

[0114] Example 3

[0115] Based on the above embodiments, this embodiment includes a transmitter and a reader for the radio frequency module, where the transmitter is an RFID transmitter.

[0116] The card reader is always resident within the automatic discovery range for tags and uploads the tag list; the card reader deployment methods include:

[0117] The receiver is placed in a corner, which is suitable for irregularly shaped booths. By adjusting the receiving angle of the antenna, it can cover the entire exhibition area as much as possible. At the same time, the sensitivity of the receiver can be adjusted so that it only receives signals within the booth area. Different booth shapes can be set with different angles and effective distances to achieve the best receiving sensitivity and avoid false triggering of tags outside the booth.

[0118] The receiver is positioned in the center, which is suitable for relatively regular booth shapes. It uses an open antenna, and the receiving distance can be adjusted by adjusting the receiving sensitivity to adapt to different booth sizes.

[0119] For certain special booth applications, receiver reliability can be ensured by combining receivers, such as placing them diagonally, to avoid weak signal strength at a distance.

[0120] The receiver is in the form of an access control system. It does not need to read the list of tags within the range in real time. When a tag enters the access control system at time t1, the EPC code in the tag and time t1 are stored locally or reported to the cloud. When a tag leaves the access control system at time t2, the EPC code in the tag and time t2 are stored or reported.

[0121] By using the different receiver arrangement methods described above, a list of EPC numbers within the time range of t1 can be automatically discovered, list1(EPC a1 EPC a2 EPC a3 ...EPC an ); Stored in a local or cloud-based database; Simultaneously, at time t2 after a fixed interval ΔT, the EPC number list list2(EPC) within the range is automatically detected again. b1 EPC b2 EPC b3 ...EPC bm At the same time, list2 is stored locally or sent to a cloud database; too short an interval will result in higher perception accuracy, but will result in a large amount of data and high transmission requirements; while too long an interval will reduce the pressure on the device, but will also result in low accuracy of audience data recognition; therefore, an appropriate interval can be selected according to the application's accuracy requirements for audience data and the actual load capacity of the device; in this embodiment, ΔT = 30s;

[0122] Data processing can be performed locally at the edge or on servers in a data center or cloud service; by cyclically comparing the EPC lists of adjacent times, the entry time t of each EPC number at a certain booth can be obtained. x and departure time t y And the length of time spent there; based on the above data, we can obtain the data at booth U. x Visitor P1, carrying RFID number EPC1, has a visit duration of δ1; visitor P2, carrying RFID number EPC2, has a visit duration of δT2... and so on. i The audience P i The duration of the visit is δT i Based on the list above, we can determine the location of booth U. x Total number of visitors Average visitor time at the booth List of the top ten visitors by visit duration (list3(P1,P2,...P)) 10 Based on this information, the booth U will be automatically generated. x Visitor profiles will be sent to exhibitors via mini-program, app, SMS, or the official website in the suggested manner. x This is to optimize customer experience and enhance customer loyalty.

[0123] like Figure 2 As shown, the EPC carries an RFID number. x The audience P x The visit duration at booth U1 is δT1, the visit duration at booth U3 is δT2, ..., the visit duration at booth Un is δT. m m≤n; a group of viewers P can be formed. x The tour route map, booth list, and duration list for each booth are provided.

[0124] Based on the above list, we can obtain the EPC carrying the RFID number. x The audience P x Total number of booths visited in this exhibition Total exhibition duration Average visit duration per booth (AveTime) x =Time x ÷M x List of the top ten booths by visit duration (list4) a1 U a2 ,...U a10 Information such as audience demographics is used to automatically generate audience profiles (P). x A summary of the exhibition will be sent to attendees via mini-program, app, SMS, or official website in a suggested manner.x This is to optimize the audience experience and enhance user engagement;

[0125] For certain important booths with a large scope, three receiving devices (101, 102, and 103) can be set up, each corresponding to a list of tag strengths within the specified range (list101, list102, and list103). The devices can then be clicked on the same tag (EPC). x The intensity values ​​RSSI in list 101, list 102, list 103 x101 RSSI x102 RSSI x103 The corresponding EPC carrying the tag can be calculated. x The audience P x The distance L between each of the three receiving devices 101, 102, and 103 x101 ,L x102 ,L x103 This allows us to locate the audience P. x The location within the overall exhibition booth is a specific sub-area, preferably, in this embodiment, area 2 as shown in the figure. Combining the aforementioned time calculation method, the total value of each visitor P can be obtained. x A set of two-dimensional data matrices representing the duration of stay in different exhibition areas;

[0126] According to each audience member P x A set of two-dimensional data matrices representing the duration of time spent in different exhibition areas, combined with specific information about each exhibition area, allows for the analysis of visitor behavior characteristics to create user profiles. There are several ways to obtain specific exhibition area data. For example, exhibitors can fill in exhibition area information when applying for a booth; or the curatorial unit can collect and fill in the information. Specific exhibition area data includes, but is not limited to, the location of the exhibition area (e.g., distance from the entrance), the area of ​​the exhibition area, the number of graphic display panels, the number of multimedia devices, the number of physical models, and the number of interactive installations. This data, combined with the visitor's specific exhibition experience (from entering the booth to leaving), can be analyzed to create user profiles. Figure 4 (As shown in the visitor flow diagram), feature extraction is performed on the visitor data to obtain the exhibitor behavior feature matrix;

[0127] After obtaining the exhibitor behavior feature matrix, the features in the matrix are filtered to obtain the final features for cluster analysis. The similarity of visitor behavior characteristics for each visit is measured based on the final features, and the K-means clustering algorithm is used to cluster the visitors to obtain the clustering results. K-means clustering is a typical distance-based clustering algorithm that uses distance as a similarity evaluation index; that is, the closer two objects are, the greater their similarity.

[0128] The specific implementation steps of the K-means algorithm are as follows:

[0129] Randomly select k samples from the feature matrix as the initial cluster centers a = a1, a2, ..., a k 2. For each sample x in the feature matrix i Calculate the distance from each of the k cluster centers using the Euclidean distance formula, i.e.

[0130] Calculate the centroid a of all samples j ,; Divide sample xi into d min (x i ,a k In the class corresponding to );

[0131] d min (x i ,a k )=min{d(x i ,a1),d(x i ,a2),…d(x i ,a k )}

[0132] And recalculate each category A j Cluster centroid a j ,

[0133] a j Belongs to category A j The centroid of all samples x in the dataset;

[0134] Determine the centroid a of the sample j ; when a j If the error of the change reaches the threshold, then the classification result and the centroids of each class a1, a2…a1 are output. j Otherwise, repeat steps 2 and 3.

[0135] Based on the clustering results, typical audience participation behaviors can be categorized and user profiles can be constructed in the form of radar charts.

[0136] Each category in the clustering results is defined as a dimension, and each dimension is scored. Multiple dimensions are formed based on multiple categories, and the scores for each dimension are presented in a radar chart to create a user profile. Creating user profiles helps to discover visitor habits and provides a basis for further optimizing visitor route recommendations, thereby improving the visitor experience.

[0137] The specific steps for building a user profile are as follows:

[0138] Select a certain audience member P xThe exhibitor behavior feature matrix is ​​used to calculate the mean distance from each exhibitor to the k cluster centers.

[0139] Normalizing the above distances, the smaller the distance, the higher the score for that category; therefore, the normalization formula can be chosen. Where d j * Between [0, 1];

[0140] Using K categories as axes, the score d of each dimension j * Mark and connect the axes to form a radar chart as a user profile.

[0141] Example 4

[0142] Based on the above embodiments, the appendix Figure 5 In this process, a feature matrix is ​​formed by merging samples using a two-dimensional matrix and specific booth information. k samples from the feature matrix are selected as initial cluster centers; a = a1, a2, ..., a k For each sample x in the feature matrix i The distance d(x) from the k cluster centers is calculated using Euclidean distance. i ,a j );

[0143]

[0144] Divide sample xi into d min In the corresponding class,

[0145] d min =min{d(x i ,a1),d(x i ,a2),…d(x i ,a k )}

[0146] And recalculate the cluster center a for each category Aj. j ,

[0147] a j That is, it belongs to category A. j The centroid of all samples x in the dataset;

[0148] Determine the centroid a of the sample j ; when a j If the error of the change reaches the threshold, then the output distance d(x) is... i ,a j ).

Claims

1. An intelligent management system for exhibition halls, comprising at least one radio frequency module, a control module, and a client; characterized in that: The radio frequency module is used to collect exhibition hall information and transmit the collected exhibition hall information to the control module; The control module is used to store and process the received exhibition hall information; The processed exhibition hall behavior information will then be transmitted to the client. The control module's information analysis and processing for the exhibition hall includes: Information acquisition: Exhibition hall information is obtained through the radio frequency module; The exhibition hall information is processed using clustering algorithms to obtain the similarity of visitor behavior characteristics. The acquisition of exhibition hall behavior information involves obtaining such information through the similarity of visitor behavior characteristics. The control module's information analysis and processing also includes: the formation of a self-portrait of exhibition hall information behavior; obtaining this self-portrait based on the exhibition hall behavior information and providing it to the user; the formation of the self-portrait of exhibition hall information behavior includes: Select audience P x The exhibition hall information behavior feature matrix was used to calculate the audience P. x The mean distance from the exhibition hall information behavior feature matrix to the k cluster centers , ； Among them, a j Let A be a point mass of class Aj, i.e., the class center; The average distance between the k cluster centers Distance normalization, ； in, Between [0, 1]; ； ； ; The smaller the distance, the higher the score for that category; Using K categories as axes, the scores for each dimension are... Plot and connect the axes to form a radar chart, and use the radar chart as a self-portrait of the information behavior of the exhibition hall; output distance. ; Step 1, the composition of the feature matrix, through each viewer P x Information collected from different booths is used to form a two-dimensional matrix, with each column representing a visitor's profile (P). x The EPC list; the row information is the collected information; the feature matrix is ​​formed by merging the two-dimensional matrix and the specific information of the booth; the specific information of the booth includes the location of the exhibition area, the area of ​​the exhibition area, the number of graphic display boards, the number of multimedia devices, the number of physical models and the number of interactive devices; Step 2: Select k samples from the feature matrix as the initial cluster centers; ； For each sample in the feature matrix The distance from the node to the k cluster centers is calculated using Euclidean distance. ; Step 3, take sample X i Assigned In the corresponding class, ； And recalculate the cluster center a for each category Aj. j , a j That is, the centroid of all samples x belonging to category Aj; Step 4, determine the centroid a of the sample. j ; when a j If the error of the change reaches the threshold, then the output distance will be... Otherwise, repeat steps 2 and 3.

2. The intelligent management system for exhibition halls according to claim 1, characterized in that, The information collected by the radio frequency module includes equipment information of the exhibition hall, tag number of the exhibition hall, time information corresponding to the tag number of the exhibition hall, signal strength value corresponding to the tag number of the exhibition hall, and tag distance value corresponding to the tag number of the exhibition hall; The tag signal strength value is ; The tag distance value is inversely proportional to the tag signal strength value. The label distance values ​​are respectively .

3. The intelligent management system for exhibition halls according to claim 1, characterized in that, The server retrieves the list of EPC numbers from the exhibition hall information by analyzing the time information corresponding to the exhibition hall's tag number. i According to the EPC number list i Thus, the time difference between adjacent moments in the EPC number list is obtained. Entry time t of the EPC number list x Departure time t of adjacent EPC number list y The duration of stay in adjacent EPC number lists and the duration of visit to each EPC number list; t i List of EPC numbers within a given time frame i and the time difference between adjacent moments are Where i is a positive integer, and the time difference between adjacent times is . , ; EPC list entry time t x and departure time t y And the duration of stay is obtained through the EPC number list. i Get the entry time t of the adjacent EPC list. x The departure time t of the adjacent EPC list y And the duration of the EPC list at adjacent times. .

4. The intelligent management system for exhibition halls according to claim 3, characterized in that, The booth visit duration information includes a list of each booth's duration, the total duration of all participating booths, the average visit duration per booth, and a list n of the top n booths by visit duration. Through EPC x The audience P x The duration of a visit to booth Un is , Booth List And a list of booth durations for each booth: ; Total number of booths visited ; Total exhibition duration ; Average visit duration per booth , ， List of booths with the most n visit durations ; 。 5. The intelligent management system for exhibition halls according to claim 1, characterized in that, The information collected by the radio frequency module also includes industry labels or product category labels for different booths; the server determines the visitor P based on the industry labels or product category labels collected by the radio frequency module. x A sorted list of industry types for visitors' preferences; The industry label or product category label for different booths is H. ； Among them, the total number of industries x ≤ the total number of booths M sum ; For the category label of industry 1, similarly... Category labels for industry x; U1, U2…U n List of booths for Industry 1; U n+1 U n+2 …U n+m This is the list of booths for Industry 2; similarly for U. o+1 U o+2 …U o+p A list of booths for industry x; The classification of different booth weight coefficients, the filtering rules for the duration of each booth, and the weight coefficient S of each booth, based on the duration of the visit. Set up y categories, as follows: ； Among them, T level Based on the duration of the visit, T level0 This is the initial visit duration level; Audience P x The visitor preferences are ranked by industry type; the industry category H is combined with the weight coefficient s to perform a weighted summation, resulting in visitor P. x A list sorted by industry type for visitor preferences. in, Sort the list by industry type; s U1 These are the weighting coefficients for U1.

Images