Intelligent control method and system of channel gate based on data analysis

Abstract

The invention relates to the technical field of gate control, and discloses an intelligent control method and system for a channel gate based on data analysis, and the method comprises the steps: obtaining historical and current user passing data and gate environment data, carrying out the time-frequency domain denoising of the historical data through wavelet transform, and obtaining the denoised data; performing correlation analysis on the denoised historical data, extracting a main environment item and a user behavior item which are remarkably correlated, and establishing an environment-behavior mapping model through linear regression; and generating portraits of different users by using a density clustering algorithm in combination with the mapping model. And further establishing a mapping relationship between the user portrait and the de-noised historical data through a deep learning network, and constructing a user portrait model. And finally, inputting the current data into the model, generating an optimal gate control strategy for different types of users, and realizing intelligent control of the channel gate. According to the method, the user portrait model is established to adapt to multi-user characteristics and complex environment changes, and personalized and accurate control of the gate system is realized.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an intelligent control method and system for a channel gate based on data analysis. Background Art

[0002] In the user usage scenarios of channel gates, user behaviors are highly dynamic and diverse. These behaviors include the user's travel time, frequency, direction, speed, and equipment carried; in addition, environmental factors such as lighting conditions, temperature, humidity, and surrounding crowd density will also affect user behavior.

[0003] Most of the existing gate systems process data based on fixed rules or simple statistical methods. The advantages of these traditional methods are their simplicity and ease of implementation, which can be quickly deployed and provide basic access control functions.

[0004] However, fixed rules lack flexibility and cannot dynamically adapt to changes in user behavior. Secondly, simple statistical methods can only capture the surface characteristics of user behavior and it is difficult to dig deep patterns and associations. In addition, traditional gate systems often perform poorly in complex environments. This not only affects traffic efficiency, but may also reduce users' trust in the gate system, thereby affecting the overall user experience. Therefore, traditional gate systems have obvious deficiencies in adapting to multi-user behavior characteristics and complex environmental changes, and it is difficult to meet the requirements of modern intelligent gate systems for personalized and precise control. Summary of the invention

[0005] The present invention provides an intelligent control method and system for a channel gate based on data analysis to solve the problem that the channel gate cannot dynamically adapt to changes in user behavior due to the dynamics and diversity of user behavior and the diversity of environmental factors in which the gate is located, thereby achieving efficient and precise control of the channel gate.

[0006] In the first aspect, in order to solve the above technical problems, the present invention provides an intelligent control method of a channel gate based on data analysis, comprising: Obtain historical user pass data, current user pass data, historical gate environment data and current gate environment data; Using wavelet transform to perform time-frequency domain denoising on the historical user passage data, the historical gate environment data, the current user passage data, and the current gate environment data to obtain denoised historical passage data, denoised historical environment data, denoised current passage data, and denoised current environment data; Perform correlation analysis on the denoised historical environmental data and the denoised historical passage data to obtain significantly correlated data items, extract the main environmental items and main user behavior items of the data items, and establish an environment-behavior mapping model using a linear regression algorithm; Use a density clustering algorithm to divide the main environmental items and the main user behavior items, and combine with the environment-behavior mapping model to obtain user portraits of different users in different environments; Establish a mapping relationship between the user portrait, the denoised historical passage data, and the denoised historical environmental data through a deep learning network model to obtain a user portrait model; Input the denoised current passage data and the denoised current environmental data into the user portrait model, and generate an optimal turnstile control strategy corresponding to the current environment for different types of users to control the access turnstiles.

[0007] In an alternative implementation, the obtaining of the historical user passage data, the current user passage data, the historical turnstile environmental data, and the current turnstile environmental data includes: Obtain the passage frequency, passage direction, passage speed, stay time, and user carried item characteristics of the user, the light intensity data, temperature and humidity data, and crowd density data in the turnstile area, and the corresponding time stamps during collection; Take the passage frequency, the passage direction, the passage speed, and the user carried item characteristics as behavior items, and divide them into historical user passage data and current user passage data according to the sequence of the time stamps; Take the light intensity data, the temperature and humidity data, and the crowd density data in the turnstile area as environmental items, and divide them into historical turnstile environmental data and current turnstile environmental data according to the sequence of the time stamps.

[0008] In an alternative implementation, the performing of time-frequency domain denoising on the historical user passage data and the historical turnstile environmental data using wavelet transform to obtain denoised historical passage data, denoised historical environmental data, denoised current passage data, and denoised current environmental data includes: Perform frequency domain decomposition on the historical user passage data, the historical turnstile environmental data, the current user passage data, and the current turnstile environmental data using wavelet transform to obtain high-frequency components and low-frequency components; Perform threshold filtering on the high-frequency components to obtain filtered high-frequency components; Reconstruct the filtered high-frequency components and the low-frequency components to generate denoised historical passage data, denoised historical environmental data, denoised current passage data, and denoised current environmental data.

[0009] In an alternative embodiment, the correlation analysis of the denoised historical environment data and the denoised historical passage data to obtain data items with a significant correlation between the denoised historical environment data and the denoised historical passage data, and extracting the main environmental items and main user behavior items of the data items, and establishing an environment-behavior mapping model using a linear regression algorithm includes: Calculating the correlation quantization value between the denoised historical environment data and the denoised historical passage data using the Pearson correlation coefficient; Judging whether the correlation quantization value is greater than a preset quantization threshold. If so, it is determined that there is a significant correlation between the denoised historical environment data and the denoised historical passage data. If not, it is determined that there is a weak correlation between the denoised historical environment data and the denoised historical passage data; Using the principal component analysis method to perform dimensionality reduction processing on the significantly correlated denoised historical environment data and denoised historical passage data, and extracting the main environmental items and main user behavior items; Constructing an environment-behavior mapping table according to the main environmental items and the main user behavior items; Using a linear regression algorithm to analyze the mapping table and establish an environment-behavior mapping model between the main environmental items and the main user behavior items.

[0010] In an alternative embodiment, the using the principal component analysis method to perform dimensionality reduction processing on the significantly correlated denoised historical environment data and denoised historical passage data, and extracting the main environmental items and main user behavior items includes: Performing data standardization, denoising, and normalization operations on the significantly correlated denoised historical environment data and denoised historical passage data to obtain preprocessed environmental features and preprocessed user behavior features; Calculating the covariance matrix of the preprocessed user behavior features and the preprocessed environmental features; Performing eigenvalue decomposition on the covariance matrix, and extracting the eigenvalues and their corresponding eigenvectors; Sorting the eigenvectors according to the magnitudes of the eigenvalues, and selecting the eigenvectors corresponding to the first d largest eigenvalues as the principal components, where d is the preset number of low-dimensional feature dimensions; Combining the eigenvectors to obtain the main environmental items and main user behavior items.

[0011] In an alternative embodiment, the using the density clustering algorithm to divide the main environmental items and the main user behavior items, and combining the environment-behavior mapping model to obtain user portraits of different users in different environments includes: Calculating the density value of the main user behavior items in the feature space using a Gaussian kernel function; Mark the area where the density value is greater than the preset density threshold as the high-density area, and divide the data points located in the same high-density area into the same user group; Combine the environment-behavior mapping model, the main environmental items, and the user group to generate user portraits of different users in different environments.

[0012] In an alternative embodiment, the step of combining the environment-behavior mapping model, the main environmental items, and the user group to generate user portraits of different users in different environments includes: Input the main environmental items into the environment-behavior mapping model to obtain the behavior characteristics of the user group under different main environmental items; According to the behavior characteristics, combine the user group to construct an initial framework of the user portrait library; Use a feature extraction algorithm to optimize the behavior characteristics in the user portrait library to obtain refined user portraits; Match the refined user portraits with the high-density areas in the feature space to generate the final user portraits of different users in different environments.

[0013] In an alternative embodiment, the step of establishing a mapping relationship between the user portrait, the denoised historical passage data, and the denoised historical environment data through a deep learning network model to obtain a user portrait model includes: Fuse the user portrait, the denoised historical passage data, and the denoised historical environment data to construct a comprehensive feature vector as the input of the deep learning network model; Use the comprehensive feature vector to train the deep learning network model, and optimize the performance of the model through cross-validation, early stopping mechanism, and regularization technology to obtain a trained network model; Based on the trained network model, introduce an attention mechanism to automatically learn the contribution differences of different features of the denoised historical passage data and the denoised historical environment data to user portrait generation, assign different weights to each modality, and generate a user portrait model.

[0014] In a second aspect, the present invention provides an intelligent control system for a channel gate based on data analysis, including: A data acquisition module for acquiring historical user passage data, current user passage data, historical gate environment data, and current gate environment data; A denoising processing module for performing time-frequency domain denoising on the historical user passage data and the historical gate environment data by using wavelet transform to obtain denoised historical passage data and denoised historical environment data; An association analysis module is used to perform association analysis on the denoised historical environment data and the denoised historical passage data, obtain data items with significant associations between the denoised historical environment data and the denoised historical passage data, extract the main environmental items and main user behavior items of the data items, and establish an environment-behavior mapping model using a linear regression algorithm; A clustering and partitioning module is used to partition the main environmental items and the main user behavior items using a density clustering algorithm, and combine with the environment-behavior mapping model to obtain user portraits of different users in different environments; A user portrait modeling module is used to establish a mapping relationship between the user portrait, the denoised historical passage data, and the denoised historical environment data through a deep learning network model to obtain a user portrait model; A turnstile control strategy generation module is used to input the current user passage data and the current turnstile environment data into the user portrait model, and generate an optimal turnstile control strategy corresponding to the current environment for different types of users to control the access turnstiles.

[0015] In a third aspect, the present invention also provides a computer-readable storage medium, which includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute any one of the above-mentioned intelligent control methods for a passage turnstile based on data analysis.

[0016] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: The present invention discloses an intelligent turnstile control method based on user portraits. This method obtains historical and current user passage data and turnstile environment data, uses wavelet transform for time-frequency domain denoising, and extracts main environmental items and user behavior items through association analysis. Subsequently, a linear regression algorithm is used to establish an environment-behavior mapping model, and a density clustering algorithm is used to partition user portraits. Finally, a deep learning network model is used to establish a mapping relationship between the user portrait and historical data to form a user portrait model. The present invention can generate optimal turnstile control strategies for different types of users according to the current environment and user characteristics, realizing personalized and precise control of the turnstile system, and improving the passage efficiency and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of an intelligent control method for a passage turnstile based on data analysis according to the present invention.

[0018] Figure 2 It is a schematic structural diagram of an intelligent control method and system for a passage turnstile based on data analysis according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] To enable those skilled in the art to better understand the technical solutions in this specification, the following will clearly and completely describe the technical solutions in the embodiments of this specification in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this specification.

[0020] As Figure 1 shown, an intelligent control method for a channel turnstile based on data analysis in this embodiment may specifically include: Step S101, obtaining historical user passing data, current user passing data, historical turnstile environment data, and current turnstile environment data.

[0021] Obtain the passing frequency, passing direction, passing speed, staying time, user's carried item characteristics of the user, the light intensity data, temperature and humidity data, crowd density data in the turnstile area, and the corresponding timestamps at the time of collection; take the passing frequency, the passing direction, the passing speed, and the user's carried item characteristics as behavior items, and divide them into historical user passing data and current user passing data according to the sequence of the timestamps; take the light intensity data, the temperature and humidity data, and the crowd density data in the turnstile area as environmental items, and divide them into historical turnstile environment data and current turnstile environment data according to the sequence of the timestamps.

[0022] It should be noted that the passing frequency reflects the number of times a user passes through the turnstile within a specific time, and can be used to analyze the user's passing habits, patterns, and whether the user is a frequent user, etc. For example, some users pass through the turnstile at about 8:30 in the morning and 6:00 in the evening every day. Such regular data is of great value for analyzing the user's behavior characteristics.

[0023] It should be noted that the passing direction can be divided into two basic states: entering and leaving. By collecting the passing direction data at different time periods, the characteristics of the flow of people can be grasped. For example, in a shopping mall, the flow of people mainly shows an entering state at 10:00 in the morning, while it is mainly leaving before closing at night.

[0024] It should be noted that the passing speed refers to the moving speed of the user when passing through the turnstile, which is measured by the displacement within a unit time and reflects the urgency of the user when passing through the turnstile. For example, at a subway station, during the morning rush hour, the passing speed of users is generally relatively fast, and on average, each person only needs 0.8 seconds to pass through the turnstile, while during non-peak hours, this value may be extended to about 1.5 seconds.

[0025] It should be noted that the staying time refers to the time from when the user enters the turnstile channel to when the user completely passes through the turnstile.

[0026] It should be noted that the characteristics of the user's carried items refer to the types, sizes, shapes and other characteristics of the items carried by the user, which are often closely related to the passing speed. For example, the passing speed of passengers carrying large luggage will be significantly slowed down, and the average passing time can reach more than three seconds.

[0027] It should be noted that the light intensity will affect the recognition accuracy of the turnstile. When the light intensity is lower than 100 lux, supplementary lighting is required to ensure normal operation.

[0028] It should be noted that the temperature and humidity data are related to the stability of the equipment. When the temperature exceeds 35 degrees or the relative humidity is higher than 85%, the environmental conditioning equipment needs to be started.

[0029] It should be noted that the passenger flow density data refers to the number of people passing through the turnstile per unit time or per unit area, which can reflect the usage pressure of the turnstile. When the number of people per square meter area exceeds 2 - 3 people, it is necessary to consider opening the standby channel.

[0030] It should be noted that the time stamp refers to the specific time point of data collection, usually recorded in a format accurate to seconds. The behavior items and the environmental items are divided into the historical user passing data, the current user passing data, the historical turnstile environment data and the current turnstile environment data according to the chronological order of the time stamps.

[0031] In a feasible embodiment, the behavior items within one day are divided into the current user passing data, and the behavior items outside one day are divided into the historical user passing data. The division of the environmental items is the same and will not be elaborated here.

[0032] Step S102, perform time-frequency domain denoising on the historical user passing data and the historical turnstile environment data by using wavelet transform to obtain denoised historical passing data and denoised historical environment data.

[0033] Perform frequency domain decomposition on the historical user passing data, the historical turnstile environment data, the current user passing data and the current turnstile environment data by using wavelet transform to obtain high-frequency components and low-frequency components; perform threshold filtering processing on the high-frequency components to obtain filtered high-frequency components; reconstruct the filtered high-frequency components and the low-frequency components to generate denoised historical passing data, denoised historical environment data, denoised current passing data and denoised current environment data. Exemplarily, wavelet transform is a multi-resolution analysis method that realizes frequency-domain decomposition by decomposing a signal into sub-bands of different frequencies. Its core idea is to utilize the dilation and translation characteristics of wavelet basis functions to gradually extract the low-frequency components and high-frequency components of the signal. By decomposing the time-domain signal into different frequency components, it can effectively identify and process the noise in the data. To obtain the high-frequency components and low-frequency components, it is first necessary to select a wavelet basis function. Common wavelet basis functions include Daubechies wavelet, Symlet wavelet, Haar wavelet, etc. Each wavelet basis function has its corresponding scaling function (for low-frequency components) and wavelet function (for high-frequency components).

[0034] In the embodiment of the present invention, Daubechies wavelet is used as the wavelet basis function. The data is convolved with the scaling function of Daubechies wavelet and the wavelet function of Daubechies wavelet to obtain the approximation coefficients (low-frequency components) and detail coefficients (high-frequency components) of the first layer respectively. Then, the approximation coefficients of the first layer are continuously convolved to be further decomposed into the approximation coefficients and detail coefficients of the second layer, and so on. The approximation coefficients of the previous layer are continuously convolved by the next layer until the predetermined decomposition level is reached. The result of the decomposition is a set of coefficients, including the approximation coefficients of the last layer and the detail coefficients of each layer, which reflect the low-frequency components and high-frequency components of the data. Since the noise is mainly concentrated in the high-frequency components, the high-frequency components are processed by hard threshold filtering to remove the noise components.

[0035] It should be noted that the hard threshold filtering process is to directly set the coefficients whose absolute values are less than the threshold to zero and retain the coefficients greater than the threshold.

[0036] In a feasible embodiment, the selection of the threshold follows the following formula: where T represents the threshold, σ represents the standard deviation of the noise, and N represents the total length of the coefficients after decomposition. Exemplarily, the noise standard deviation can be obtained through the median absolute deviation.

[0037] It should be noted that based on the reversibility of wavelet transform, the approximation coefficients and detail coefficients obtained after decomposition are synthesized layer by layer to gradually reconstruct the denoised data. Specifically, starting from the approximation coefficients of the last layer, combined with the detail coefficients of the last layer, the inverse transform of the wavelet basis function is used to synthesize the approximation coefficients and detail coefficients of the penultimate layer, and so on. Finally, the filtered high-frequency components and low-frequency components are combined into denoised historical traffic data, denoised historical environmental data, denoised current traffic data, and denoised current environmental data.

[0038] Step S103: Perform correlation analysis on the denoised historical environment data and the denoised historical passage data to obtain the data items that show a significant correlation between the denoised historical environment data and the denoised historical passage data. Extract the main environmental items and the main user behavior items of these data items, and use the linear regression algorithm to establish an environment-behavior mapping model.

[0039] Calculate the correlation quantification value between the denoised historical environment data and the denoised historical passage data using the Pearson correlation coefficient. Determine whether the correlation quantification value is greater than a preset quantification threshold. If so, it is determined that there is a significant correlation between the denoised historical environment data and the denoised historical passage data; if not, it is determined that there is a weak correlation between the denoised historical environment data and the denoised historical passage data. Use the principal component analysis method to perform dimensionality reduction processing on the significantly correlated denoised historical environment data and the denoised historical passage data, and extract the main environmental items and the main user behavior items. According to the main environmental items and the main user behavior items, construct an environment-behavior mapping table. Use the linear regression algorithm to analyze the mapping table and establish an environment-behavior mapping model between the main environmental items and the main user behavior items. It should be noted that the preset quantification threshold is a key parameter, which determines the ability of the turnstile to discover the relationship between the environment data and the user passage data. If it is set too high, it is difficult for the turnstile to discover the relationship between the environment data and the user passage data; if it is set too low, the turnstile will associate irrelevant environment data and user passage data.

[0040] In the embodiment of the present invention, the preset quantification threshold is 0.6. When the absolute value of the calculated correlation coefficient is greater than 0.6, it indicates that the two data sequences have a significant correlation.

[0041] It should be noted that the Pearson correlation coefficient is a statistical index for measuring the linear correlation between variables, and its value range is from -1 to 1, where 1 represents a perfect positive correlation, -1 represents a perfect negative correlation, and 0 represents no correlation. For example: for the turnstile environment data and the user passage data, the relationship between temperature and passage speed can be considered. For example, when the temperature is too high, the user's passage speed through the turnstile becomes slower, showing a negative correlation.

[0042] In the embodiment of the present invention, the Pearson correlation coefficient is used to process the correlation quantification values between different data items of the denoised historical environment data and different data items of the denoised historical passage data. For example: A total of 12 groups of correlation quantification values are calculated respectively between the light intensity, temperature and humidity, and the crowd density and the passage frequency, passage direction, passage speed, and user carried object characteristics.

[0043] It should be noted that principal component analysis extracts the main features through dimensionality reduction, which can transform multi-dimensional data into a small number of important dimensions. Exemplarily, for multiple dimensions such as temperature and humidity, light intensity, and pedestrian flow density, it may be found through principal component analysis that temperature and light intensity have the greatest impact on user passage, and these are taken as the main environmental items; similarly, for passage speed, passage direction, passage frequency, etc., it is found through analysis that passage speed and residence time are the most representative and are taken as the main user behavior items.

[0044] It should be noted that the environment-behavior mapping table records the correspondence between the main environmental items and the main user behavior items. For example, when the temperature is 25 degrees, the average passage speed is 1.2 meters per second and the residence time is 2 seconds; when the temperature rises to 35 degrees, the passage speed drops to 0.8 meters per second and the residence time increases to 3 seconds; when the lighting intensity is 200 lux, the passage speed is 1.0 meters per second and the residence time is 2.5 seconds. Through these mapping relationships, the influence law of environmental factors on user behavior can be observed.

[0045] It should be noted that the linear regression model can quantitatively describe the relationship between the environment and behavior. Taking the influence of temperature as an example, it is assumed that for every 1-degree increase in temperature, the passage speed decreases by an average of 0.02 meters per second and the residence time increases by 0.05 seconds. For every 50-lux increase in lighting intensity, the passage speed increases by 0.1 meters per second and the residence time decreases by 0.2 seconds. This quantitative relationship helps to predict user behavior under different environmental conditions. In practical applications, environmental factors often interact with each other. For example, high temperature can cause discomfort to users and reduce passage efficiency; while appropriate lighting can improve user comfort and offset some of the negative impacts brought by high temperature. By establishing an environment-behavior mapping model, these complex interaction relationships can be better understood, providing a basis for optimizing the control strategy of the turnstile system. When the environmental conditions change, the system can predict the changes in user behavior and adjust the turnstile parameters accordingly, such as appropriately extending the verification timeout time in hot weather and increasing the display screen brightness in low light conditions, so as to improve the passage experience.

[0046] In step S103, the principal component analysis method is used to perform dimensionality reduction processing on the denoised historical environmental data and the denoised historical passage data with significant association, and the main environmental items and the main user behavior items are extracted.

[0047] Perform data standardization, denoising, and normalization operations on the significantly correlated denoised historical environmental data and the denoised historical passage data to obtain preprocessed environmental features and preprocessed user behavior features; calculate the covariance matrix of the preprocessed user behavior features and the preprocessed environmental features; perform eigenvalue decomposition on the covariance matrix to extract eigenvalues and their corresponding eigenvectors; sort the eigenvectors according to the magnitudes of the eigenvalues, and select the eigenvectors corresponding to the top d largest eigenvalues as the principal components, where d is the preset number of low-dimensional feature dimensions; combine the eigenvectors to obtain the main environmental items and the main user behavior items.

[0048] It should be noted that data standardization refers to transforming data into a distribution with zero mean and unit variance, eliminating the differences in dimension and magnitude between different features. For example, in the turnstile environment data, the temperature unit is Celsius, the light intensity unit is lux, and the humidity is a percentage. These data have inconsistent dimensions. Through standardization, the data can be converted into a standard normal distribution with a mean of zero and a variance of one.

[0049] It should be noted that denoising is achieved by smoothing the data through the median filtering method; normalization refers to scaling the data to a specific range (such as [0, 1]), so that different features have the same scale; for example, mapping the temperature from ten degrees to forty degrees to zero to one, and mapping the light intensity from zero to one thousand lux to zero to one, so that data in different dimensions have the same scale.

[0050] It should be noted that the covariance matrix is used to measure the strength and direction of the linear relationship between different features. A positive value indicates a positive correlation, a negative value indicates a negative correlation, and the larger the absolute value, the stronger the correlation. For example, temperature may be negatively correlated with the passage speed, and light is positively correlated with the passage efficiency. By calculating the covariance matrix, the association pattern between environmental factors and user behavior can be discovered. The eigenvalue represents the degree of dispersion of the data in the direction of the corresponding eigenvector, and the eigenvector indicates the main direction of data change. By selecting the eigenvectors corresponding to the larger eigenvalues, the main components in the data can be extracted to achieve dimensionality reduction. For example, the first eigenvector reflects the influence of temperature on user behavior, and the second eigenvector reflects the influence of light on user behavior. Selecting the eigenvectors with a larger contribution rate as the principal components can reduce the data dimension and retain key information. This dimensionality reduction process not only retains key information but also simplifies subsequent modeling analysis. By establishing the mapping relationship between the main environmental items and the main behavior items, the operating state of the turnstile system can be predicted and controlled more accurately.

[0051] It should be noted that the eigenvectors are arranged and combined as column vectors in the order of the magnitudes of the corresponding eigenvalues to obtain the main environmental items and the main user behavior items, where the eigenvalue represents the main environmental item and the eigenvector represents the main user behavior item.

[0052] Step S104: Use the density clustering algorithm to divide the main environmental items and the main user behavior items, and combine the environment-behavior mapping model to obtain user portraits of different users in different environments.

[0053] Calculate the density values of the main user behavior items in the feature space using the Gaussian kernel function; mark the regions where the density values are greater than the preset density threshold as high-density regions, and divide the data points located in the same high-density region into the same user group; combine the environment-behavior mapping model, the main environmental items, and the user group to generate user portraits of different users in different environments.

[0054] Exemplarily, it should be noted that the Gaussian kernel function is a classic kernel function that can map data into a high-dimensional feature space to better reflect the distribution characteristics of the data.

[0055] It should be noted that the selection of the preset density threshold is crucial for the division of user groups. If the density threshold is preset too high, it is difficult to divide the same user group together. If the density threshold is preset too low, different user groups will also be divided into the same class.

[0056] In the embodiment of the present invention, the preset density threshold is 0.5, and the regions where the density values are greater than this threshold are marked as high-density regions. This threshold can also be adaptively determined by the silhouette coefficient or data distribution.

[0057] It should be noted that the user portrait can accurately describe the behavioral characteristics and environmental preferences of user groups. For example, within the temperature range of 22 degrees to 28 degrees Celsius and the illumination intensity range of 150 to 250 lux, the user passing speed is relatively concentrated, forming a high-density area. These data points represent the normal passing group in a comfortable environment. The division of user groups is based on the distribution characteristics of high-density areas. Through analysis, it is found that there are multiple different high-density areas. For example, during the period from 7 am to 9 am, when the light is sufficient, a high-density area is formed, corresponding to the fast-passing group during the morning rush hour; while from 12 pm to 2 pm, another high-density area is formed in a high-temperature environment, corresponding to the leisurely-passing group during the lunch break. Combining the above-mentioned environment-behavior mapping model, a richer user portrait can be constructed. Taking the shopping mall scenario as an example, the portrait of the elderly shopping group shows that they tend to pass during the period from 9 am to 11 am, are relatively sensitive to temperature and light. When the temperature exceeds 30 degrees Celsius, their passing speed will decrease by 10 - 20%, and the staying time will double; while the user portrait of the young shopping group indicates that they mostly enter and exit in the afternoon and evening. Their passing speed is relatively less affected by the environment, but they will significantly slow down when the crowd density is large. During the generation process of the user portrait, for example, in the scenario of the subway turnstile, a certain group maintains a relatively stable passing speed even in the face of high temperature and large crowd density during the morning rush hour, which reflects the characteristic that this group is more sensitive to time. Through this multi-dimensional analysis, the behavioral patterns of different groups under various environmental conditions can be predicted more accurately, providing data support for the precise control strategy of the turnstile system.

[0058] In step S104, combine the environment-behavior mapping model, the main environmental items, and the user groups to generate user portraits of different users in different environments.

[0059] Input the main environmental items into the environment-behavior mapping model to obtain the behavioral characteristics of user groups under different main environmental items. According to the behavioral characteristics, combine the user groups to construct the initial framework of the user portrait library. Use a feature extraction algorithm to optimize the behavioral characteristics in the user portrait library to obtain refined user portraits. Match the refined user portraits with the high-density areas in the feature space to generate the final user portraits of different users in different environments.

[0060] It should be noted that behavioral characteristics refer to the behavioral performance of users under different environmental conditions. The environment-behavior mapping model establishes the correlation between environmental factors and behavioral characteristics by collecting and analyzing user behavior data under different environmental conditions. Taking the subway station as an example, when the temperature is between 20 and 25 degrees Celsius and the lighting is moderate, the passenger speed is relatively stable, with an average of 0.8 meters per second; when the temperature rises to above 30 degrees Celsius, the speed will drop to 0.6 meters per second, and the stay time will increase.

[0061] It should be noted that the initial framework of the user portrait library is a preliminary user portrait collection based on user groups and behavioral characteristics, which is used to describe the behavioral patterns of different user groups in different environments. Feature extraction can identify the behavioral patterns of different user groups. For example, in the shopping mall scene, analysis shows that the elderly group prefers to shop in the morning and is highly sensitive to temperature and lighting; the young group tends to shop in the evening and at night and has strong adaptability to the environment. These behavioral characteristics constitute the basic framework of the user portrait library. Feature extraction algorithms can discover deeper behavioral characteristics from raw data. For example, in the office building scene, analysis shows that different occupational groups have unique traffic patterns: management personnel arrive before 9 a.m. and have high requirements for the environment; technical personnel have flexible working characteristics and strong adaptability to the environment. This refined analysis helps to optimize the accuracy of user portraits. In the process of matching high-density areas with user portraits, the dynamic changes of environmental factors need to be considered. Taking the dining area in the mall as an example, during the peak dining period, even if the ambient temperature is high and the flow density is large, customers are still willing to stay, which shows that the influence of time factors in some scenarios is more important. Through this matching analysis, user needs can be predicted and responded to more accurately. In the process of generating the final user portrait, the impact of unexpected situations also needs to be considered. For example, in public places, when the density of people suddenly increases, some user groups will show obvious avoidance behavior, while other groups will show strong adaptability. The identification of such differentiated features is of great significance for scene management.

[0062] Step S105, establishing a mapping relationship between the user portrait and the denoised historical traffic data and the denoised historical environment data through a deep learning network model to obtain a user portrait model.

[0063] Fuse the user portrait, the denoised historical passing data, and the denoised historical environmental data to construct a comprehensive feature vector as the input of the deep learning network model; use the comprehensive feature vector to train the deep learning network model, and optimize the performance of the model through cross-validation, early stopping mechanism, and regularization techniques to obtain the trained network model; based on the trained network model, introduce the attention mechanism to automatically learn the contribution differences of different features of the denoised historical passing data and the denoised historical environmental data to the generation of the user portrait, assign different weights to each modality, and generate the user portrait model.

[0064] Exemplarily, the construction of the comprehensive feature vector is a process of organically integrating multi-source data. In the turnstile system, the user passing data includes behavioral features such as passing speed and residence time, the environmental data includes physical parameters such as temperature and light, and the user portrait reflects the behavioral preferences of different groups. Taking the turnstile system of a certain shopping mall as an example, for the elderly shopping group, there is a significant correlation between the passing speed and the temperature. When the room temperature exceeds 28 degrees Celsius, the passing speed will decrease by 10–20%. These features can be constructed into a multi-dimensional vector, where the behavioral data, environmental data, and user portrait respectively occupy different dimensions.

[0065] It should be noted that during the training process of the deep learning network model, the cross-validation technique evaluates the generalization ability of the model by dividing the dataset into a training set and a validation set. Taking a certain subway station as an example, the data for one month is divided into five parts in chronological order. Each time, four parts are used as the training set, and the remaining one part is used as the validation set, which can better evaluate the performance of the model in different time periods. The early stopping mechanism monitors the performance metrics on the validation set and stops training in a timely manner when the model shows a tendency of overfitting. When the accuracy on the validation set has not improved for five consecutive cycles, the training process is stopped. The introduction of the attention mechanism enables the model to automatically learn the importance of different features. In the turnstile scenario of an office building, the influence weight of the passenger flow density on the passing speed is relatively large during the morning and evening rush hours, while during non-peak hours, the weights of environmental factors such as temperature and light are relatively higher. By introducing the attention layer, the model can dynamically adjust the feature weights according to different time periods and scenarios. For example, on a rainy day, the weight of the feature of slipperiness will be automatically increased, and the model will pay more attention to the impact of this factor on the passing behavior. In the practice of the outpatient hall of a certain hospital, the model found that for the elderly patient group, when the temperature is relatively high, the correlation weight between their passing speed and the temperature reaches 60%, while for the young group, this weight is only 20%. This difference is automatically captured by the model through the attention mechanism, so as to generate a more accurate user portrait in different situations. A more accurate user portrait provides an important basis for optimizing the turnstile control strategy, enabling the system to make more intelligent adjustments according to the characteristics of different groups and environmental conditions.

[0066] Step S106: Input the current user access data and the current turnstile environment data into the user portrait model, and generate an optimal turnstile control strategy corresponding to the current environment for different types of users to control the access turnstiles.

[0067] It should be noted that the optimal turnstile control strategy refers to the turnstile response strategy formulated by the access turnstiles for specific users, which can help improve the efficiency and experience of users passing through the access turnstiles.

[0068] It should be noted that when the current user access data and the current turnstile environment data are input into the user portrait model, the system will, based on these real-time data, combine the user behavior characteristics and environmental preferences stored in the user portrait model, and generate an optimal turnstile control strategy that matches the current environment for different types of users. For example, for users carrying large luggage, the turnstile will automatically adjust to a wider channel mode; during peak hours, the turnstile enters a high-power consumption mode to speed up the response to users, and at the same time opens more turnstile channels to reduce congestion; while when there are few people, the turnstile enters a low-power consumption mode, appropriately reducing the response speed to users, and at the same time closing some turnstile channels. In this way, the access turnstiles can achieve intelligent and personalized control, not only improving the access efficiency, but also enhancing the user experience and security.

[0069] Refer to Figure 2 , the present invention provides an intelligent control system for access turnstiles based on data analysis, mainly including: A data acquisition module, which is used to acquire historical user access data, current user access data, historical turnstile environment data, and current turnstile environment data; A denoising processing module, which is used to perform time-frequency domain denoising on the historical user access data and the historical turnstile environment data by using wavelet transform to obtain denoised historical access data and denoised historical environment data; An association analysis module, which is used to perform association analysis on the denoised historical environment data and the denoised historical access data to obtain data items with significant associations between the denoised historical environment data and the denoised historical access data, and extract the main environmental items and main user behavior items of the data items, and establish an environment-behavior mapping model by using a linear regression algorithm; A clustering and partitioning module, which is used to partition the main environmental items and the main user behavior items by using a density clustering algorithm, and combine the environment-behavior mapping model to obtain user portraits of different users in different environments; A user portrait modeling module, which is used to establish a mapping relationship between the user portrait and the denoised historical access data and denoised historical environment data through a deep learning network model to obtain a user portrait model; A turnstile control strategy generation module is configured to input the current user access data and the current turnstile environment data into the user portrait model, and generate an optimal turnstile control strategy corresponding to the current environment for different types of users to control the access turnstile.

[0070] The above content is only an example and illustration of the structure of the present invention. Those skilled in the art of this technology can make various modifications or supplements to the described specific embodiments or use similar methods to replace them. As long as they do not deviate from the structure of the invention or exceed the scope defined by this claim book, they should fall within the protection scope of the present invention.

[0071] An embodiment of the present invention also provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as an intelligent control program for an access turnstile based on data analysis. When the processor executes the computer program, it implements the steps in the above-mentioned embodiments of the intelligent control method for an access turnstile based on data analysis, such as Figure 1 the step S101 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module in the above-mentioned device embodiments, such as the user portrait modeling module.

[0072] Exemplarily, the computer program can be divided into one or more modules. The one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more modules can be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device.

[0073] The electronic device can be a computing device such as a desktop computer, a notebook, a palm computer, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the above components are only examples of the electronic device and do not constitute a limitation to the electronic device. It may include more or fewer components than the above, or combine certain components, or different components. For example, the electronic device may further include input / output devices, network access devices, a bus, etc.

[0074] The so-called processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the electronic device and connects all parts of the entire electronic device using various interfaces and circuits.

[0075] The memory can be used to store the computer programs and modules. By running or executing the computer programs and modules stored in the memory and calling the data stored in the memory, the processor realizes various functions of the electronic device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the mobile phone (such as audio data, phone book, etc.). In addition, the memory may include high-speed random access memory and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, FlashCard, at least one magnetic disk storage device, flash device, or other volatile solid-state storage devices.

[0076] Among them, if the modules integrated in the electronic device are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-described embodiment methods of the present invention, it can also be completed by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-described various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0077] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that they have a communication connection, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0078] The above-described specific embodiments have further elaborated on the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. It is particularly pointed out that for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent control method for a channel turnstile based on data analysis, characterized in that, The method includes: Obtaining historical user passage data, current user passage data, historical turnstile environment data, and current turnstile environment data; Performing time-frequency domain denoising on the historical user passage data, the historical turnstile environment data, the current user passage data, and the current turnstile environment data by using wavelet transform to obtain denoised historical passage data, denoised historical environment data, denoised current passage data, and denoised current environment data; Performing correlation analysis on the denoised historical environment data and the denoised historical passage data to obtain significantly correlated data items, extracting the main environmental items and main user behavior items of the data items, and establishing an environment-behavior mapping model by using a linear regression algorithm; Dividing the main environmental items and the main user behavior items by using a density clustering algorithm, and combining with the environment-behavior mapping model to obtain user portraits of different users in different environments; Establishing a mapping relationship between the user portrait, the denoised historical passage data, and the denoised historical environment data through a deep learning network model to obtain a user portrait model; Inputting the denoised current passage data and the denoised current environment data into the user portrait model, and generating an optimal turnstile control strategy corresponding to the current environment for different types of users to control the access turnstile.

2. The method according to claim 1, wherein The obtaining of the historical user passage data, the current user passage data, the historical turnstile environment data, and the current turnstile environment data includes: Obtaining the passage frequency, passage direction, passage speed, stay time, and user carried item characteristics of the user, the light intensity data, temperature and humidity data, and pedestrian flow density data in the turnstile area, and the corresponding timestamps at the time of collection; Taking the passage frequency, the passage direction, the passage speed, and the user carried item characteristics as behavior items, and dividing them into historical user passage data and current user passage data according to the sequence of the timestamps; Taking the light intensity data, the temperature and humidity data, and the pedestrian flow density data in the turnstile area as environmental items, and dividing them into historical turnstile environment data and current turnstile environment data according to the sequence of the timestamps.

3. The method according to claim 1, wherein The performing of time-frequency domain denoising on the historical user passage data and the historical turnstile environment data by using wavelet transform to obtain the denoised historical passage data, the denoised historical environment data, the denoised current passage data, and the denoised current environment data includes: Performing frequency domain decomposition on the historical user passage data, the historical turnstile environment data, the current user passage data, and the current turnstile environment data by using wavelet transform to obtain high-frequency components and low-frequency components; Performing threshold filtering on the high-frequency components to obtain filtered high-frequency components; Reconstructing the filtered high-frequency components and the low-frequency components to generate the denoised historical passage data, the denoised historical environment data, the denoised current passage data, and the denoised current environment data.

4. The method according to claim 1, wherein The performing of correlation analysis on the denoised historical environment data and the denoised historical passage data to obtain significantly correlated data items, extracting the main environmental items and main user behavior items of the data items, and establishing an environment-behavior mapping model by using a linear regression algorithm includes: Calculate the correlation quantification value between the denoised historical environmental data and the denoised historical traffic data using the Pearson correlation coefficient; Judge whether the correlation quantification value is greater than a preset quantification threshold. If so, determine that there is a significant correlation between the denoised historical environmental data and the denoised historical traffic data. If not, determine that there is a weak correlation between the denoised historical environmental data and the denoised historical traffic data; Use the principal component analysis method to perform dimensionality reduction on the significantly correlated denoised historical environmental data and the denoised historical traffic data, and extract the main environmental items and the main user behavior items; Construct an environment-behavior mapping table based on the main environmental items and the main user behavior items; Use the linear regression algorithm to analyze the mapping table and establish an environment-behavior mapping model between the main environmental items and the main user behavior items.

5. The method according to claim 4, characterized in that, The step of using the principal component analysis method to perform dimensionality reduction on the significantly correlated denoised historical environmental data and the denoised historical traffic data, and extract the main environmental items and the main user behavior items includes: Perform data standardization, denoising, and normalization operations on the significantly correlated denoised historical environmental data and the denoised historical traffic data to obtain preprocessed environmental features and preprocessed user behavior features; Calculate the covariance matrix of the preprocessed user behavior features and the preprocessed environmental features; Perform eigenvalue decomposition on the covariance matrix to extract the eigenvalues and their corresponding eigenvectors; Sort the eigenvectors according to the magnitudes of the eigenvalues, and select the eigenvectors corresponding to the first d largest eigenvalues as the principal components, where d is the preset low-dimensional feature dimension number; Combine the eigenvectors to obtain the main environmental items and the main user behavior items.

6. The method according to claim 1, wherein The step of using the density clustering algorithm to partition the main environmental items and the main user behavior items, and combine the environment-behavior mapping model to obtain user portraits of different users in different environments includes: Use the Gaussian kernel function to calculate the density values of the main user behavior items in the feature space; Mark the regions where the density values are greater than the preset density threshold as high-density regions, and divide the data points located in the same high-density region into the same user group; Combine the environment-behavior mapping model, the main environmental items, and the user group to generate user portraits of different users in different environments.

7. The method according to claim 6, wherein The step of combining the environment-behavior mapping model, the main environmental items, and the user group to generate user portraits of different users in different environments includes: Input the main environmental items into the environment-behavior mapping model to obtain the behavior features of the user group under different main environmental items; Construct an initial framework of the user portrait library according to the behavior features and the user group; Use the feature extraction algorithm to optimize the behavior features in the user portrait library to obtain refined user portraits; Match the refined user portraits with the high-density regions in the feature space to generate the final user portraits of different users in different environments.

8. The method according to claim 1, wherein Establishing a mapping relationship between the user profile, the denoised historical passage data, and the denoised historical environment data through a deep learning network model to obtain a user profile model, including: Fusing the user profile, the denoised historical passage data, and the denoised historical environment data to construct a comprehensive feature vector as the input of the deep learning network model; Training the deep learning network model using the comprehensive feature vector, and optimizing the performance of the model through cross-validation, early stopping mechanism, and regularization techniques to obtain a trained network model; Based on the trained network model, introducing an attention mechanism to automatically learn the contribution differences of different features of the denoised historical passage data and the denoised historical environment data to user profile generation, assigning different weights to each modality, and generating a user profile model.

9. An intelligent control system for a channel gate based on data analysis, characterized in that, The system includes: A data acquisition module for acquiring historical user passage data, current user passage data, historical turnstile environment data, and current turnstile environment data; A denoising processing module for performing time-frequency domain denoising on the historical user passage data, the historical turnstile environment data, the current user passage data, and the current turnstile environment data using wavelet transform to obtain denoised historical passage data, denoised historical environment data, denoised current passage data, and denoised current environment data; An association analysis module for performing association analysis on the denoised historical environment data and the denoised historical passage data to obtain data items with significant associations, extracting the main environmental items and the main user behavior items of the data items, and establishing an environment-behavior mapping model using a linear regression algorithm; A clustering and partitioning module for partitioning the main environmental items and the main user behavior items using a density clustering algorithm, and combining the environment-behavior mapping model to obtain user profiles of different users in different environments; A user profile modeling module for establishing a mapping relationship between the user profile, the denoised historical passage data, and the denoised historical environment data through a deep learning network model to obtain a user profile model; A turnstile control strategy generation module for inputting the current user passage data and the current turnstile environment data into the user profile model, and generating an optimal turnstile control strategy corresponding to the current environment for different types of users to control the access turnstile.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein when the computer program runs, it controls the device where the computer-readable storage medium is located to execute an intelligent control method for an access turnstile based on data analysis as described in any one of claims 1 to 8.

Images