Ticket safety verification method and system based on artificial intelligence

Abstract

The invention relates to the cross technical field of artificial intelligence and ticket business safety management, and discloses a ticket safety cancel-after-verification method based on artificial intelligence. The method comprises the following steps: firstly, acquiring ticket spectrum data, associating admission record data and batch admission queue data, and constructing a preliminary spectrum feature set; dimensionality reduction is carried out to determine a dominant frequency band, and a reflection mode and an absorption peak value are extracted to obtain a refined spectrum vector; the ultraviolet frequency band absorption peak offset of the vector is judged, and a potential forgery deviation frequency band subset is obtained; fusing environment illumination and ticket aging field variables to generate a corrected spectral feature vector, and comparing to obtain a consistency score; the authenticity labels of the tickets are obtained through classification of a support vector machine, and verification confirmation signals are obtained through judgment and matching in combination with admission records; and updating the bill library in real time, optimizing an admission sequence in combination with queue data, and realizing accurate authentic identification and efficient field scheduling. According to the method, multi-dimensional accurate authentic identification of the ticket can be realized, and the problems of difficulty in identifying false tickets and peak congestion in traditional verification are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of artificial intelligence and ticket security management, and in particular to a method and system for secure ticket verification based on artificial intelligence. Background Technology

[0002] Currently, with the rapid development of the cultural and entertainment industry, the scale of participation in offline activities such as concerts and sports events continues to expand. As the core credentials for users to enter the venue, the secure verification of tickets is directly related to the maintenance of order at the event site, the protection of the economic interests of the organizers, and is also closely related to the legitimate rights and interests of millions of users.

[0003] Existing traditional ticket verification technologies in the industry mainly rely on manual visual inspection of watermarks and patterns on tickets, or reading surface information such as barcodes and QR codes through simple scanning devices. They employ fixed verification processes and data processing logic, and while they still have some application in small-scale events, they are clearly insufficient in complex scenarios: high-quality counterfeit tickets can accurately imitate the appearance and basic identification information of tickets, while existing methods only focus on surface features, completely ignoring the multi-dimensional spectral characteristics differences inherent in the paper fibers and ink chemical composition of tickets; at the same time, changes in ambient lighting and ticket aging can lead to data distortion, and existing technologies lack dynamic adaptation capabilities, further reducing the accuracy of identification. During peak periods, slow verification speeds can also cause queue congestion.

[0004] In summary, existing technologies suffer from poor multimodal data collaboration, inability to eliminate interference and achieve accurate authenticity verification, lack of dynamic linkage between verification status and entry queues, resulting in scheduling delays during peak periods. These issues make it difficult to meet the dual requirements of ticketing security and management efficiency for large-scale offline events, and fail to provide efficient and secure ticketing management. Summary of the Invention

[0005] This invention provides an artificial intelligence-based secure ticket verification method to solve the problems in existing technologies, such as poor multimodal data coordination, inability to eliminate interference to achieve accurate authenticity identification, lack of dynamic linkage between verification status and entry queue, and resulting in scheduling delays during peak periods.

[0006] In a first aspect, to address the aforementioned technical problems, this invention provides an artificial intelligence-based secure ticket verification method, comprising:

[0007] Acquire ticket spectral data, associated ticket entry record data, and batch entry queue data to construct a preliminary spectral feature set;

[0008] The preliminary spectral feature set is dimensionality reduced to determine the dominant frequency band, the reflection mode and absorption peak of the dominant frequency band are extracted, and the refined spectral vector is determined.

[0009] If the absorption peak of the refined spectral vector in the ultraviolet band deviates from the absorption peak of the preset standard by more than the preset ultraviolet band absorption peak deviation threshold, it is judged as a potential counterfeiting deviation, and a subset of the deviation frequency band is obtained.

[0010] The system acquires on-site variables such as ambient light and ticket aging, fuses these on-site variables with the deviation frequency band subset data using a multilayer sensor to generate a corrected spectral feature vector, compares it with the spectral data of a preset standard, judges the consistency of the corrected spectrum, and obtains a consistency score.

[0011] Based on the consistency score, the tickets are classified as genuine or counterfeit using a support vector machine to obtain genuine / counterfeit labels.

[0012] Based on the ticket authenticity label and the associated ticket entry record data, determine whether the entry record data matches the already cancelled ticket and obtain a cancellation confirmation signal.

[0013] Based on the verification confirmation signal, a real-time database update mechanism is adopted, and the on-site scanning timestamp is integrated to determine the change in the invoice verification status and obtain the updated invoice database.

[0014] Based on the updated ticket database and the batch entry queue data, the entry order is adjusted through queue optimization to determine the risk of congestion during peak periods and obtain an optimized entry sequence.

[0015] Secondly, the present invention provides an artificial intelligence-based secure ticket verification system, comprising:

[0016] The data acquisition module is used to acquire ticket spectral data, associated ticket entry record data, and batch entry queue data to construct a preliminary spectral feature set.

[0017] The dimensionality reduction module is used to perform dimensionality reduction processing on the preliminary spectral feature set, determine the dominant frequency band, extract the reflection mode and absorption peak of the dominant frequency band, and determine the refined spectral vector.

[0018] The deviation judgment module is used to determine potential counterfeiting deviation if the absorption peak of the refined spectral vector in the ultraviolet band deviates from the absorption peak of the preset standard by more than a preset ultraviolet band absorption peak deviation threshold, and obtain a subset of the deviation frequency band.

[0019] The variable fusion module is used to acquire on-site variables such as ambient light and ticket aging degree. It fuses the on-site variables with the deviation frequency band subset data through a multilayer sensor to generate a corrected spectral feature vector and compares it with the spectral data of a preset standard to determine the consistency of the corrected spectrum and obtain a consistency score.

[0020] The authenticity classification module is used to classify tickets into authenticity categories using a support vector machine based on the consistency score, and obtain ticket authenticity labels; the verification confirmation module is used to determine whether the entry record data matches the verified tickets based on the ticket authenticity labels and the associated ticket entry record data, and obtain a verification confirmation signal.

[0021] The database update module is used to determine the change in the invoice cancellation status based on the cancellation confirmation signal, using a real-time database update mechanism and integrating the on-site scanning timestamp, and to obtain the updated invoice database.

[0022] The queue optimization module is used to adjust the entry order based on the updated ticket database and the batch entry queue data, determine the risk of congestion during peak periods, and obtain an optimized entry sequence.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] (1) This invention constructs a preliminary spectral feature set by acquiring ticket spectral data, entry record data and queue data, and extracts and refines spectral vectors by combining principal component analysis to reduce dimensionality. This breaks through the limitations of traditional identification that relies solely on the surface features of the ticket, explores the differences in spectral characteristics of ticket paper material and ink composition, provides multi-dimensional data support for authenticity identification, effectively improves the basic data accuracy of counterfeit ticket identification, and solves the problem that single surface features are easily counterfeited and imitated.

[0025] (2) This invention introduces on-site variables such as ambient light and ticket aging, and generates a corrected spectral feature vector by fusing multi-source data through a multi-layer sensor. It dynamically eliminates the influence of on-site environmental interference factors on spectral data, and then combines it with a support vector machine to achieve accurate classification of ticket authenticity, avoid identification errors caused by environmental interference, significantly improve the accuracy of authenticity identification in complex scenarios, and make up for the lack of dynamic adaptation capability in existing technologies.

[0026] (3) After generating the authenticity label of the ticket, the present invention links the entry record data to confirm the verification status, adopts a real-time database update mechanism to synchronize the ticket status, and combines batch entry queue data to optimize the entry sequence and judge the risk of congestion, so as to realize the full-process linkage of "authenticity identification-verification confirmation-queue scheduling", solve the problem of peak period scheduling delay caused by the disconnect between the verification status and the entry queue, take into account both ticket security and on-site management efficiency, and meet the practical needs of large-scale offline events. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of a ticket security verification method based on artificial intelligence provided in the first embodiment of the present invention;

[0028] Figure 2This is a schematic diagram of a ticket security verification system based on artificial intelligence provided in the second embodiment of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Reference Figure 1 The first embodiment of the present invention provides a ticket security verification method based on artificial intelligence, including the following steps:

[0031] S101, Obtain ticket spectral data, associated ticket entry record data, and batch entry queue data to construct a preliminary spectral feature set;

[0032] S102, Dimensionality reduction processing is performed on the preliminary spectral feature set to determine the dominant frequency band, the reflection mode and absorption peak of the dominant frequency band are extracted, and the refined spectral vector is determined;

[0033] S103, if the absorption peak of the refined spectral vector in the ultraviolet band deviates from the absorption peak of the preset standard by more than the preset ultraviolet band absorption peak deviation threshold, it is judged as a potential counterfeiting deviation, and a deviation frequency band subset is obtained.

[0034] S104. Acquire on-site variables of ambient light and ticket aging degree, fuse the on-site variables with the deviation frequency band subset data through a multilayer sensor to generate a corrected spectral feature vector, compare it with the preset standard spectral data, determine the consistency of the corrected spectrum, and obtain a consistency score.

[0035] S105, Based on the consistency score, the tickets are classified into genuine and counterfeit categories using a support vector machine to obtain genuine and counterfeit labels for the tickets;

[0036] S106. Based on the ticket authenticity label and the associated ticket entry record data, determine whether the entry record data matches the cancelled ticket and obtain a cancellation confirmation signal.

[0037] S107. Based on the cancellation confirmation signal, a real-time database update mechanism is adopted, and the on-site scanning timestamp is integrated to determine the change in the cancellation status of the invoice and obtain the updated invoice database.

[0038] S108, Based on the updated ticket database and the batch entry queue data, the entry order is adjusted through queue optimization to determine the risk of congestion during peak periods and obtain an optimized entry sequence.

[0039] In step S101, ticket spectral data, associated ticket entry record data, and batch entry queue data are acquired to construct a preliminary spectral feature set.

[0040] In one optional implementation, the step of acquiring ticket spectral data, associated ticket entry record data, and batch entry queue data to construct a preliminary spectral feature set includes:

[0041] Collect raw, multi-dimensional spectral data, including the characteristics of ticket paper material, reflection modes of ink spectral reflectance, and absorption peaks, to form an initial spectral dataset;

[0042] The initial spectral dataset is subjected to noise filtering, and feature vectors of forged identification patterns and reflectance peak absorption are extracted. The feature vectors are compared with preset data validity standards. If they meet the standards, a preliminary spectral feature set is determined.

[0043] Retrieve entry record data containing ticket number, timestamp, and user identity-related identifiers to obtain associated ticket entry record data; retrieve queue data containing pending entry ticket numbers and pre-query results of verification status to obtain batch entry queue data.

[0044] It should be noted that when collecting raw multi-dimensional spectral data, including the characteristics of ticket paper material, the reflection mode of ink spectral reflectance, and absorption peaks, the spectral analysis equipment used must meet the technical parameters of wavelength coverage of 200-1000nm, reflectance measurement accuracy of ±0.5%, and absorption peak wavelength accuracy of ±1nm. This equipment can accurately capture the core physical differences between genuine and counterfeit tickets—genuine ticket paper (cotton and linen fibers) typically has a reflectance of 0.7-0.75 in the 600-700nm band, while counterfeit ticket paper (ordinary wood pulp paper) has a reflectance of only 0.5-0.55, and the anti-counterfeiting ink on genuine tickets has an absorption peak of 0.4-0.45 in the 365nm ultraviolet band. During the scanning process, the equipment sampling frequency is set to 10 times / second and the average value is taken to eliminate random errors. At the same time, the scanning equipment number and the timestamp accurate to milliseconds are recorded to ensure the matching requirements of subsequent data traceability and historical ticket database verification. Finally, the similar data of all tickets to be verified are summarized to form the initial spectral dataset.

[0045] When filtering noise from the initial spectral dataset, a 3-layer db4 wavelet basis transform algorithm is used. The specific steps are as follows: First, the db4 wavelet basis is selected as the transform basis function because it has the best balance between smoothness and feature preservation of spectral data. Second, 3-layer wavelet decomposition is performed to decompose the initial spectral data into one approximation coefficient (low-frequency effective component) and three detail coefficients (high-frequency noise component). Third, "soft thresholding" is applied to the high-frequency detail coefficients. The threshold is set to 1.2 times the standard deviation of the detail coefficients. Noise coefficients with absolute values ​​less than the threshold are set to 0, and coefficients with values ​​greater than the threshold are recalculated as "coefficient value - threshold". Fourth, wavelet reconstruction is used to inversely transform the processed approximation coefficients and detail coefficients back to the original data dimension, completing the noise filtering. This process can eliminate high-frequency noise caused by venue lighting (such as interference from LED lights in the 550nm band), natural light (such as mid-infrared noise from sunlight), and equipment current fluctuations, improving the data signal-to-noise ratio from 20-25dB to 35-40dB. It can specifically eliminate spectral baseline drift caused by venue lighting and natural light, as well as high-frequency noise generated by equipment current fluctuations, improving the data signal-to-noise ratio from 20-25dB to 35-40dB. In the feature vector extraction stage, it is necessary to focus on identifying common counterfeit ticket marking patterns such as "absorption peak splitting caused by uneven ink layer thickness" and "abnormal reflectivity fluctuation caused by disordered paper fiber arrangement." At the same time, it extracts the absorption peak wavelength, intensity, and half-width (half-width at half-width of 365nm band for genuine anti-counterfeiting ink, while counterfeit tickets often exceed 15nm due to impure chemical composition) and other reflection peak absorption characteristics. When comparing feature vectors with preset data validity standards, three core indicators must be included: First, feature completeness, with a missing rate of less than 1% for core indicators such as reflectance, absorption peak value, and full width at half maximum (FWHM); second, numerical reasonableness, with reflectance controlled between 0 and 1, and the absorption peak wavelength deviating from the average of the historical genuine ticket database (storing no less than 10,000 verified genuine ticket data) by no more than 5 nm; and third, matching degree, with a matching degree of no less than 0.85 between the feature vectors calculated by the Euclidean distance algorithm and the historical genuine ticket database. Feature vectors that meet all the indicators can be included in the preliminary spectral feature set, while those that do not meet the indicators are marked as "feature vectors to be reviewed" and temporarily stored in a temporary database.

[0046] When retrieving entry record data containing ticket number, timestamp, and user identification information, the data originates from a ticketing management system built on a MySQL relational database. This system's RESTful API supports 50 concurrent queries per second with a single response time within 30ms, meeting the real-time data retrieval needs of large-scale events. During data extraction, the "ticket number" (formatted as "T + date + 6-digit serial number") is used as the unique key to ensure that the entry timestamp differs from the ticket scanning timestamp by no more than 30 minutes, preventing data confusion across sessions. Simultaneously, user identification information is anonymized (e.g., the middle 4 digits of the mobile phone number are hidden, and only the last 6 digits of the ID card are retained). A "reverse lookup of user identification information" function is used to complete missing ticket numbers, and duplicate records with the same ticket number are removed (keeping only the latest timestamp), ensuring data security and accuracy.

[0047] When retrieving queue data containing the ticket numbers to be entered and the pre-query results of the verification status, the data comes from the on-site entry queue management system, which uses the WebSocket real-time communication protocol. This system updates data once per second, enabling dynamic synchronization of the data to be entered. The pre-query stage connects to a MongoDB non-relational database. Through index optimization, the response time for "ticket number query" is reduced from 200ms to 30ms, quickly obtaining the pre-query results of the verification status (categorized into "not verified," "pre-verified," and "abnormal," with "abnormal" requiring specific reasons such as duplicate numbers, expired validity, or matching a blacklist). During queue data processing, invalid data with incorrect ticket number formats or pre-query response timeouts (exceeding 5 seconds) must be removed. Furthermore, the ticket numbers to be entered must correspond one-to-one with the ticket numbers in the preliminary spectral feature set. In addition, to ensure efficient real-time data retrieval, the preliminary spectral feature set and associated data are temporarily stored in a Redis cache (valid for 30 minutes) to reduce the time spent on repeated queries.

[0048] In summary, the process of collecting raw multi-dimensional spectral data relies on precise equipment to capture physical differences in tickets; noise filtering and feature verification ensure data quality through algorithm optimization and standard validation; and the acquisition of entry records and batch queue data utilizes a dedicated system for efficient correlation and real-time processing. The entire process not only meets the multi-dimensional data support required for genuine and counterfeit ticket identification but also balances security and efficiency through data anonymization and caching optimization. This process effectively addresses the shortcomings of traditional ticket verification, which relies solely on surface features and has a single data dimension. It provides a precise and complete data foundation for subsequent steps such as dimensionality reduction and forgery deviation detection. Furthermore, it can meet the verification requirements of 50-100 tickets per second for large-scale offline events, aligning with the core objectives of improving ticket security and management efficiency.

[0049] In step S102, the preliminary spectral feature set is subjected to dimensionality reduction processing to determine the dominant frequency band, the reflection mode and absorption peak of the dominant frequency band are extracted, and the refined spectral vector is determined.

[0050] In one optional implementation, the step of dimensionality reduction of the preliminary spectral feature set, determining the dominant frequency band, extracting the reflection mode and absorption peak of the dominant frequency band, and determining the refined spectral vector includes:

[0051] The reflectance and absorbance data in the preliminary spectral feature set are standardized, and the standardized data are then subjected to dimensionality reduction through principal component analysis to obtain the dominant frequency band data.

[0052] The reflection modes and absorption peaks of the dominant frequency band data are extracted and integrated into a low-dimensional feature vector to obtain a refined spectral vector.

[0053] It should be noted that when standardizing the reflectance and absorbance data in the preliminary spectral feature set, the Z-score standardization algorithm is used to eliminate dimensional differences between different indicators. The preliminary spectral feature set typically contains reflectance (range 0-1) and absorbance (range 0-2) data in the 200-1000nm band. After standardization, the mean of all data is uniformly 0, and the standard deviation is uniformly 1. For example, the original reflectance of genuine paper in the 600-700nm band is 0.7-0.75, which is transformed into 0.8-1.0 after standardization. The original reflectance of counterfeit paper in the same band is 0.5-0.55, which is standardized to -0.9 to -0.7. This avoids the subsequent dimensionality reduction bias towards a certain indicator due to differences in numerical ranges. During the standardization process, outliers also need to be handled: data exceeding the mean ± 3 times the standard deviation (such as a reflectance of 1.2 due to accidental equipment failure) are identified using the 3σ principle and replaced with linear interpolation of adjacent band data to ensure data continuity and provide high-quality input for subsequent principal component analysis.

[0054] When performing dimensionality reduction on standardized data using Principal Component Analysis (PCA), the covariance matrix of the standardized data is first calculated to quantify the linear correlation between reflectance and absorbance values ​​in each band. Then, the eigenvalues ​​and eigenvectors of the covariance matrix are solved. Principal components are selected based on the principle of "cumulative variance contribution rate ≥ 90%." For example, the original standardized data contains 1600 dimensions (2 indicators for each of 800 bands). After PCA calculation, the eigenvalues ​​of the first three principal components are 850, 620, and 310, respectively, with a cumulative variance contribution rate of 92%. The corresponding dominant frequency bands are 600-700nm (dominated by paper material characteristics), 365nm (dominated by ultraviolet characteristics of anti-counterfeiting ink), and 520nm (dominated by basic ink characteristics). The data in these three frequency bands constitute the dominant frequency band data, significantly reducing the data dimensionality (from 1600 dimensions to 3 dimensions) while fully preserving the core spectral information required for authenticating counterfeit tickets. To avoid overfitting during the dimensionality reduction process, 5-fold cross-validation is also required to adjust the number of principal components and ensure the consistency of dimensionality reduction effect for different batches of ticket data. For example, in the validation of 1,000 sample tickets (including 200 fake tickets), the retention rate of the first three principal components for fake ticket features remained stable at over 95%.

[0055] When extracting the reflection modes and absorption peaks of the dominant frequency band data, a differentiated extraction strategy is adopted for different dominant frequency bands: For the 600-700nm band, the reflection mode is focused, and the smoothness of the reflectance curve is calculated using a 5nm sliding window (genuine tickets have a smoothness ≤0.02 due to the regular arrangement of cotton and linen fibers, while counterfeit tickets have a smoothness ≥0.05 due to the disordered fibers of ordinary wood pulp paper). Simultaneously, the reflection peak at 650nm is recorded (typical value 0.75 for genuine tickets, typical value 0.55 for counterfeit tickets); the 365nm ultraviolet band is further analyzed. The absorption peak value is extracted by finding the extreme points through differentiation to determine the peak wavelength (genuine tickets are stable at 365nm, while counterfeit tickets often shift to 370-375nm) and peak value (genuine tickets 0.4-0.45, counterfeit tickets 0.25-0.3). Simultaneously, the reflection mode (genuine tickets have no significant fluctuations in reflectivity curves, while counterfeit tickets have 1-2 small fluctuation peaks) and absorption peak value (genuine tickets have an absorption value of 0.35 at 520nm, while counterfeit tickets shift to 525nm and the absorption value drops to 0.28) are extracted in the 520nm band. During the extraction process, the feature range is compared with the historical genuine ticket database (storing ≥10,000 verified genuine ticket spectral data) to eliminate false peaks caused by ambient light interference (such as the slight absorption fluctuations caused by venue LED lights at 550nm), ensuring the effectiveness of the extracted features.

[0056] When integrating the reflection modes and absorption peaks of the dominant frequency bands to form a low-dimensional feature vector, two key parameters of each dominant frequency band are used as vector dimensions: for the 600-700nm band, the reflectivity at 650nm and the smoothness of the reflection curve are used; for the 365nm band, the absorption peak wavelength and absorption peak value are used; and for the 520nm band, the absorption peak wavelength and absorption peak value are used, ultimately forming a 6-dimensional refined spectral vector. For example, the refined spectral vector of a genuine ticket is [0.75, 0.01, 365, 0.42, 520, 0.35], and the refined spectral vector of a counterfeit ticket is [0.55, 0.06, 372, 0.29, 525, 0.28]. This low-dimensional vector not only significantly reduces the computational load of subsequent algorithms—the processing time for a single vector is reduced from 1.2 seconds for the original high-dimensional data to 0.3 seconds, meeting the real-time verification requirement of processing ≥3 tickets per second on-site—but also avoids interference from redundant data in subsequent forgery bias judgments by focusing on core identification features, ensuring the identification specificity of the feature vector.

[0057] In summary, the standardization of the preliminary spectral feature set provides a unified data benchmark for dimensionality reduction. Principal component analysis accurately identifies the dominant frequency band by screening high-contribution principal components. The extraction of reflection modes and absorption peaks focuses on the essential differences between genuine and counterfeit tickets. Finally, the refined spectral vector formed by integration compresses the data dimension while fully preserving the core identification information of ticket material and ink, thus solving the problem of low computational efficiency of the original high-dimensional spectral data.

[0058] In step S103, if the absorption peak of the refined spectral vector in the ultraviolet band deviates from the absorption peak of the preset standard by more than a preset ultraviolet band absorption peak deviation threshold, it is determined to be a potential counterfeiting deviation, and a subset of the deviation frequency band is obtained.

[0059] In one optional implementation, if the absorption peak of the refined spectral vector in the ultraviolet band deviates from the absorption peak of a preset standard by more than a preset ultraviolet band absorption peak deviation threshold, it is determined to be a potential counterfeiting deviation, and a subset of deviation frequency bands is obtained, including:

[0060] Calculate the offset between the ultraviolet absorption peak of the refined spectral vector and the absorption peak of the preset standard to obtain a set of offset values;

[0061] If at least one offset value in the set of offset values ​​exceeds the preset ultraviolet band absorption peak offset threshold, it is determined that there is a potential forgery deviation.

[0062] The ultraviolet bands corresponding to the offset value set are processed by frequency band segmentation, and the bands whose offset values ​​exceed the offset threshold of the ultraviolet band absorption peak are extracted. Then, the extracted bands are judged for potential forgery deviation category by combining cluster classification to determine the deviation frequency band subset.

[0063] It should be noted that when calculating the offset between the ultraviolet absorption peak of the refined spectral vector and the absorption peak of the preset standard, the ultraviolet frequency range is first defined as 200-400nm (this range covers the core spectral response range of ticket anti-counterfeiting ink and paper material). The preset standard data comes from a pre-constructed genuine ticket spectral database—this database stores no less than 10,000 verified genuine ticket ultraviolet frequency spectral data, including the absorption peak benchmark values ​​of genuine tickets from different production batches and with different usage durations. For example, the standard range of the absorption peak of genuine ticket anti-counterfeiting ink at 365nm is 0.40-0.45, and the standard range of the absorption peak of paper material at 280nm is 0.55-0.60. When calculating the offset values, a "two-dimensional offset calculation method" is adopted: first, wavelength offset, which is the difference between the wavelength corresponding to the absorption peak in the refined spectral vector and the peak wavelength of the standard (e.g., the peak at 365nm on the counterfeit ticket shifts to 372nm, with a wavelength offset of +7nm); second, numerical offset, which is the difference between the numerical value of the absorption peak in the refined spectral vector and the average numerical value of the peak in the standard (e.g., the peak value at 365nm on the counterfeit ticket is 0.32, with a numerical offset of -0.11 from the standard average of 0.43). The wavelength and numerical offset values ​​of all ultraviolet frequency band subdivisions (40 subdivisions in total, each 5nm) are summarized to form an offset value set containing 80 data points. At the same time, extreme outliers (such as wavelength offsets of more than 10nm caused by momentary equipment failure) are removed using the Grubbs test. The set is then completed using a weighted average of the offset values ​​of adjacent subdivisions to ensure the continuity and accuracy of the offset value set.

[0064] If at least one offset value in the set of offset values ​​exceeds the preset ultraviolet band absorption peak offset threshold, a potential counterfeiting deviation is determined. The threshold setting needs to be verified by multiple batches of samples: through statistical analysis of 5,000 genuine tickets (covering 10 production batches and 3 aging levels) and 2,000 known counterfeit tickets (including 5 types of counterfeiting such as high-quality counterfeits and ordinary counterfeits), the wavelength offset threshold is determined to be ±5nm (99.2% of the wavelength offsets in genuine tickets are within ±3nm), and the numerical offset threshold is ±0.08 (98.7% of the numerical offsets in genuine tickets are within ±0.05). During the judgment process, the principle of "single exceedance triggers" is adopted—if the wavelength deviation of a certain sub-band exceeds ±5nm, or the numerical deviation exceeds ±0.08, a potential counterfeiting deviation is initially judged. Simultaneously, secondary verification is performed based on the characteristic importance of that sub-band. For example, the weight of 365nm (the core band of anti-counterfeiting ink) is 0.8, and the weight of 280nm (the basic band of paper) is 0.5. If the 365nm band exceeds the limit, the confidence level of the potential counterfeiting deviation judgment is directly strengthened (confidence level increased to 90%). If only the 280nm band exceeds the limit, a comprehensive judgment is needed based on whether other adjacent bands are close to the threshold (confidence level adjusted to 60%), avoiding misjudgment due to accidental fluctuations in a single non-critical band. After the judgment, a preliminary report of potential counterfeiting deviation is generated, marking the location, direction, and amount of the exceeding band, providing a clear target for subsequent frequency band segmentation.

[0065] When processing the ultraviolet bands corresponding to the offset value set through frequency band segmentation, an "adaptive band merging algorithm" is adopted: taking the over-limit subdivision band as the center, it expands to the adjacent subdivision bands on both sides, calculates the average offset value of the expanded band, and if the average offset value still exceeds 70% of the threshold (i.e., wavelength offset ±3.5nm, numerical offset ±0.056), it continues to expand until the average offset value is lower than the threshold or reaches the ultraviolet band boundary (200nm or 400nm), forming a continuous "deviation candidate band". For example, if the three subdivision bands of 365nm, 370nm, and 375nm all exceed the limit, and the average wavelength offset after merging is +6nm and the average numerical offset is -0.10, exceeding the threshold of 70%, it continues to expand to the 380nm band. If the average offset value of the 380nm band drops to -0.05 (lower than the numerical offset threshold of 70%), then the deviation candidate band is determined to be 365-375nm. Subsequently, bands with offset values ​​exceeding the ultraviolet band absorption peak offset threshold are extracted. This involves removing marginal subdivision bands from the candidate bias bands whose average offset values ​​do not exceed the limit, ultimately obtaining the 365-375nm excess band. Then, clustering classification is used to determine the potential counterfeiting bias category of the extracted bands. The K-means clustering algorithm is employed (with K=3 clusters based on the clustering objective, corresponding to three common counterfeiting types: "ink composition counterfeiting," "paper material counterfeiting," and "missing anti-counterfeiting labels"). Before clustering, feature preprocessing is required: discrete offset directions (positive / negative) are encoded as 1 / -1; continuous features such as band position and offset amplitude distribution are standardized using Z-score along with the encoding results to eliminate dimensional differences; Euclidean distance is used as the distance metric to accurately calculate the spatial distance between samples and cluster centers for category classification. Clustering features include the position of the excess band (e.g., 360-380nm often corresponds to ink counterfeiting), offset direction (e.g., positive wavelength offset often indicates impure ink composition), and offset amplitude distribution (e.g., a significant drop in value often indicates inferior paper material). During the clustering process, a historical counterfeit case library (containing 1,000 band shift samples labeled with counterfeit types) is introduced as the initial value for cluster centers. Through iterative optimization, the clustering accuracy is stabilized at over 85%. For example, samples in the 365-375nm band with positive wavelength shift and significant value decrease are clustered into the "ink composition counterfeit" category. Finally, the out-of-limit bands that belong to the counterfeit-related categories after clustering are integrated and determined as a subset of deviation frequency bands (such as the 365-375nm ink counterfeit deviation frequency band subset). At the same time, the confidence level of the counterfeit type corresponding to this subset is output (such as 92% confidence level for ink composition counterfeit), providing category guidance for subsequent spectral correction.

[0066] In summary, the precise calculation of the offset value set provides a data foundation for determining potential forgery deviations. The threshold determination mechanism combined with feature importance ensures the accuracy and anti-interference capability of the determination. Frequency band segmentation and cluster classification realize in-depth processing from "the existence of deviation" to "locating the deviation frequency band and identifying the deviation type". It not only accurately locks the core interval of the spectral difference between genuine and counterfeit tickets, but also provides a clear direction for deviation correction for subsequent on-site variable fusion. It effectively solves the problem in traditional technologies that can only determine "whether it is abnormal" but cannot locate "where the abnormality is and why it is abnormal", laying a key foundation for improving the accuracy and interpretability of ticket authenticity identification.

[0067] In step S104, the on-site variables of ambient light and ticket aging degree are acquired. The on-site variables are fused with the deviation frequency band subset data through a multilayer sensor to generate a corrected spectral feature vector, which is then compared with the spectral data of a preset standard to determine the consistency of the corrected spectrum and obtain a consistency score.

[0068] In one optional implementation, the acquisition of ambient light and ticket aging degree field variables, the fusion of the field variables with the deviation frequency band subset data through a multilayer perceptron to generate a corrected spectral feature vector, and the comparison with preset standard sample spectral data to determine the consistency of the corrected spectrum and obtain a consistency score, includes:

[0069] Data on ambient light intensity and ticket aging were obtained and used as on-site variables.

[0070] The field variables are processed using a data standardization method to obtain a standardized set of field variables;

[0071] The biased frequency band subset data and the standardized field variable set are input into the multilayer sensor. The multilayer sensor performs hierarchical calculations to fuse the multi-source data and generate a corrected spectral feature vector.

[0072] The matching degree between the modified spectral feature vector and the preset standard spectral data is calculated. If the matching degree meets the preset matching degree judgment criteria, the consistency score is calculated by the preset scoring method.

[0073] It should be noted that when acquiring data on ambient light intensity and ticket aging as the on-site variables, precise data collection is required using specialized sensing equipment: Ambient light intensity is collected in real time using a high-precision light sensor (measurement range 0-10000 lux, accuracy ±5%), with a sampling frequency of 1 time / second. The average value is taken after three consecutive samplings to eliminate instantaneous light fluctuations (such as sudden strong light caused by venue flashlights or moving light sources). For example, the ambient light intensity values ​​collected at a concert were 800 lux, 820 lux, and 790 lux, ultimately determined to be 803 lux. Ticket aging data is captured using a high-definition industrial camera (12 megapixel resolution, spectral response range 400-700nm) to capture ticket surface features, which are then quantified using image analysis algorithms. —Focusing on extracting paper yellowing (calculated by the difference between the blue channel grayscale value in the RGB image and the standard new ticket, with a difference range of 0-255, the larger the difference, the more severe the aging) and ink fading degree (comparing the deviation of the ink grayscale value at the key anti-counterfeiting mark on the ticket from the standard value, with a deviation range of 0-1, and a deviation of 0.3 or more is considered obvious fading). When calculating the aging degree value, first perform Min-Max standardization on the paper yellowing (standardized value is yellowing difference / 255) to eliminate the dimensional difference with the ink fading degree; then perform weighted summation according to preset weights (such as ink fading degree 0.6, standardized yellowing degree 0.4, set based on the intuitive appearance of ticket aging, generally the ink fading of aged tickets is more intuitive, in practice, this weight setting can be adjusted according to the characteristics of the ticket). For example, if the yellowing difference of a ticket paper is 45 (approximately 0.176 after standardization) and the ink fading deviation is 0.2, the calculated value is 0.176×0.4+0.2×0.6≈0.19. After taking the approximate value, the overall judgment is 0.2 (0 for brand new, 1 for severely aged). At the same time, it needs to be compared and calibrated with a preset ticket aging sample library (which stores 500 sets of ticket images and quantitative data with different aging degrees) to ensure the consistency and accuracy of the aging degree data.

[0074] When processing the field variables using data standardization, an appropriate method is selected based on the characteristics of different variables: the original ambient light intensity data ranges from 0-10000 lux, and Min-Max standardization is used to map it to the 0-1 interval. The Min-Max standardization calculation formula is as follows:

[0075]

[0076] in, This is the original value of ambient light intensity. This is the minimum illumination threshold (value 0). The maximum illumination threshold (value 10000), for example, 803 lux after standardization is calculated. =0.0803; The original data for ticket aging is already in the range of 0-1. Z-score standardization is used to further eliminate differences in data distribution. The Z-score standardization formula is:

[0077]

[0078] in, This represents the original value for the ticket's aging condition. The value represents the average degree of aging (calibrated statistically from 500 aging samples, with a value of 0.2). The standard deviation of aging degree (statistically calibrated using 500 aging samples, with a value of 0.15), for example, is calculated after standardization to an aging degree of 0.3. During the standardization process, outliers need to be identified using the 3σ principle. For example, an outlier of 15,000 lux caused by a light sensor malfunction (exceeding the 3σ range) is replaced by a moving average of five adjacent sampling points to avoid the outlier affecting the subsequent fusion effect. Finally, the standardized light value and aging value are integrated to form a standardized set of field variables with two dimensions, such as [0.0803, 0.666], to ensure that the data meets the requirement of the multilayer perceptron for the uniformity of input feature dimensions.

[0079] When the biased frequency band subset data and the standardized field variable set are input into the multilayer perceptron, and the multi-source data is fused through hierarchical calculation to generate the corrected spectral feature vector, the network structure of the multilayer perceptron must first be determined: the input layer dimension is the sum of the number of bands in the biased frequency band subset and the dimension of the field variables—assuming the biased frequency band subset is the 365-375nm ultraviolet band (subdivided into 11 bands in 1nm increments), then the input layer has a total of 11+2=13 neurons; the hidden layer has 2 layers, the first layer has 32 neurons (using the ReLU activation function to alleviate the gradient vanishing problem), and the second layer has 16 neurons (also using the ReLU activation function); the output layer dimension is consistent with the number of bands in the biased frequency band subset (11 neurons, using a linear activation function to ensure the continuity of the output corrected value). The network training relies on a historical sample dataset (containing 10,000 samples, each containing a subset of biased frequency bands, field variables, and manually labeled true correction vectors). Mean squared error is used as the loss function, and the Adam optimizer (learning rate 0.001, batch size 32) is used for iterative training until the loss function converges (validation set loss is below 0.001). In actual fusion computation, the input layer first receives the absorption peak data of the biased frequency band subset (e.g., 0.32 at 365nm, 0.31 at 366nm, etc.) and standardized field variables [0.0803, 0.666]. After weighted summation (weights are the optimal parameters obtained during training) and ReLU activation in the first hidden layer, 32 feature values ​​are output. These are then passed to the second hidden layer for secondary feature transformation. Finally, the output layer generates a corrected spectral feature vector—for example, to address the low absorption peak caused by illumination, the absorption peak at 365nm is corrected to 0.38, and at 366nm to 0.37, making the corrected data closer to the true spectral characteristics without environmental interference.

[0080] When calculating the matching degree between the modified spectral feature vector and the preset standard spectral data, a cosine similarity algorithm is used to quantify the similarity between the two. The cosine similarity formula for calculating the matching degree is:

[0081]

[0082] in, To correct the spectral eigenvector, The preset standard sample spectral vector is taken from the preset standard sample database, which stores the ultraviolet frequency spectrum data of different batches of genuine tickets under standard conditions, such as the standard value of the absorption peak in the 365-375nm band being 0.40-0.42. For vectors and dot product, For vectors L2 norm, For vectors The L2 norm, for example, the modified vector =[0.38,0.37,...,0.39], standard vector =[0.41,0.40,...,0.42], calculated as follows The preset matching score criterion is "matching score ≥ 0.85 meets the consistency requirements". If it is lower than 0.85, it is marked as "pending review" and requires further manual verification. If it meets the standard, the consistency score is calculated using a preset linear scoring method. The formula for calculating the linear consistency score is:

[0083]

[0084] in, For consistency score, The matching degree is defined as follows: 0.85 is the lower limit of the matching degree judgment threshold (verified by 10,000 sets of genuine and fake ticket samples, this threshold can balance the false positive rate and the false negative rate), and 1 is the perfect matching degree (the matching degree when the corrected vector is completely consistent with the standard vector). For example, a matching degree of 0.96 is calculated as follows: The final score is rounded down to 95 points; however, the score needs to be adjusted based on the band weights. The formula for adjusting the consistency score based on the band weights is as follows:

[0085]

[0086] in, The final consistency score after weight adjustment. The matching score for the core wavelength band (around 365nm) of the anti-counterfeiting ink. For the matching score of other bands, 0.6 and 0.4 are the weighting coefficients of the core band and other bands, respectively (calibrated by 5000+ sets of genuine and counterfeit ticket identification experiments, the core band contributes more to the identification of genuine and counterfeit tickets), to ensure that the score can accurately reflect the consistency of key features and avoid the excessive influence of non-core band deviations on the overall score.

[0087] In summary, accurate collection of on-site variables provides data support for eliminating environmental and aging interference, standardized processing ensures the compatibility of multi-source data fusion, hierarchical calculation of multilayer sensors enables deep collaborative correction of deviation data and on-site variables, and matching degree calculation and scoring quantify the authenticity of the corrected spectrum. The entire process not only solves the identification error problem caused by ignoring on-site interference in traditional technologies, but also significantly improves the accuracy and reliability of ticket authenticity identification in complex on-site environments.

[0088] In step S105, based on the consistency score, the tickets are classified into genuine and counterfeit categories using a support vector machine to obtain genuine and counterfeit ticket labels.

[0089] In one optional implementation, the step of classifying tickets into authenticity categories using a support vector machine based on the consistency score to obtain ticket authenticity labels includes:

[0090] The consistency scores are standardized to obtain standardized consistency scores; the standardized consistency scores are then classified into true and false categories using a support vector machine to obtain preliminary labeling results.

[0091] The initial marking results are converted into a unified format of identification information through a label mapping operation, which is then used to determine the authenticity of the ticket.

[0092] It should be noted that when standardizing the consistency score, the range of the consistency score must first be defined—this score originates from the linear scoring in step S104, ranging from 80 to 100 points (80 points corresponds to the minimum passing score of 0.85 matching degree, and 100 points corresponds to perfect consistency of 1.0 matching degree; verified by 10,000 sets of genuine and fake ticket samples, this range can effectively distinguish between genuine and fake features). Therefore, the Min-Max standardization method is used to map it to the 0-1 interval to ensure that the data adapts to the input requirements of the support vector machine. The standardization calculation formula is as follows:

[0093]

[0094] in, The original consistency score, This is the minimum consistency score (80, which corresponds to the critical value where the matching degree does not reach the qualified line). This represents the maximum consistency score (taken as 100, corresponding to the perfect match). For example, if a ticket's original consistency score is 95, substituting it into the formula yields... =0.75; If the original score is below 80 (e.g., 78, corresponding to a matching degree of 0.83, which does not meet the passing standard), the standardized score will be uniformly set to 0 and marked as a low-confidence sample to avoid abnormal data interfering with the classification results. After standardization, a data integrity check is required. If the standardized score of a sample is missing (e.g., the original score is abnormal during the calculation), the S104 step of that sample will be retrieved from the temporary database and recalculated to ensure the integrity and accuracy of the standardized consistency score set.

[0095] When classifying true and false categories using the standardized consistency scores via Support Vector Machine (SVM), model training and parameter calibration must be completed first. The SVM model uses Radial Basis Function (RBF) as the kernel function, which effectively handles the non-linear mapping between standardized scores and true / false categories. The model training dataset is taken from a pre-set ticketing sample database (containing standardized consistency scores of 10,000 labeled genuine tickets and 2,000 labeled fake tickets). The training process uses 5-fold cross-validation to optimize model parameters, determining the penalty parameter C=1.2 (balancing the risk of overfitting and underfitting; verification shows that the model achieves 98.2% accuracy on the validation set under this parameter), and the kernel function parameters... =0.8 controls the influence range of the kernel function on the samples, ensuring robustness in classifying ambiguous samples. During classification, the standardized consistency score is input into the trained support vector machine model. The model outputs the class probability by calculating the distance between the sample and the classification hyperplane. If the probability of a genuine ticket is ≥0.9, it is initially labeled as "genuine ticket"; if the probability of a fake ticket is ≥0.9, it is initially labeled as "fake ticket"; if both probabilities are between 0.4 and 0.6 (e.g., standardized score 0.52, genuine ticket probability 0.55, fake ticket probability 0.45), it is initially labeled as "awaiting verification," forming a preliminary labeling result set. After classification, the results need to be sampled and verified. 1% of the samples are randomly selected and compared with the results of manual verification. If the error rate exceeds 1%, the model parameters are readjusted and the model is retrained to ensure that the classification accuracy meets the requirements of on-site verification.

[0096] When converting the initial marking results into a unified format of identification information through label mapping, the mapping rules must be adapted to the database field specifications of the ticketing management system: the initial marking "genuine ticket" is mapped to the number "1", "fake ticket" is mapped to the number "0", and "pending verification" is mapped to the number "2", and associated with the ticket's unique identifier "ticket number" (formatted as "T + date + 6-digit serial number", such as T20240927000001) to generate a unified format ticket authenticity label. For example, for a ticket with ticket number T20240927000001, if the initial marking is "genuine ticket", the mapped label will be "T20240927000001_1"; if the initial marking is "pending verification", the label will be "T20240927000001_2". During the mapping process, a format verification mechanism needs to be triggered. If the label has problems such as "missing ticket number" or "incorrect identification number" (such as "3" or other undefined numbers), the previous steps will be automatically returned to regenerate the initial labeling result to prevent invalid labels from entering the subsequent verification process. At the same time, the mapped ticket authenticity label will be stored in the Redis cache.

[0097] In summary, the standardization of consistency scores provides support vector machines with highly adaptable input features, avoiding classification bias caused by differences in the original score ranges. Through nonlinear kernel functions and parameter optimization, support vector machines achieve high-precision differentiation between genuine and counterfeit categories, especially effectively reducing the risk of misjudgment for the "to be verified" label of ambiguous samples. The label mapping operation, through a unified format and verification mechanism, ensures that labels can be directly connected to the ticketing management system, further enhancing the practicality and system compatibility of ticket authenticity verification.

[0098] In step S106, based on the ticket authenticity label and the associated ticket entry record data, it is determined whether the entry record data matches the already verified ticket, and a verification confirmation signal is obtained.

[0099] In one optional implementation, the step of determining whether the entry record data matches the already verified ticket based on the ticket authenticity label and the associated ticket entry record data, and obtaining a verification confirmation signal, includes:

[0100] Obtain a set of entry records containing ticket numbers and timestamps, and filter the set of entry records according to identity identifiers to obtain a filtered set of entry records;

[0101] Based on the ticket numbers in the filtered set of entry records, query the pre-established cancellation status database to determine whether the ticket number exists in the list of cancelled tickets, and obtain a set of matching results.

[0102] If the matching result set shows that the ticket number has not been cancelled and the ticket authenticity label is genuine, then the cancellation status of the ticket number in the cancellation status database is updated to cancelled, and a status update confirmation is obtained.

[0103] Based on the status update confirmation, a cancellation signal is generated. If the generation timestamp of the cancellation signal is within a preset valid period after the entry timestamp of the corresponding ticket in the entry record set, it is determined to be the cancellation confirmation signal.

[0104] It should be noted that the process involves obtaining a set of entry records containing ticket numbers and timestamps, and then filtering the set of entry records according to the user's identity. The resulting set of entry records originates from a ticketing management system built on a MySQL relational database. This system provides data retrieval services through a RESTful API interface, supporting 50 concurrent queries per second with a single response time controlled within 30ms, meeting the real-time data requirements of large-scale events. The core fields of the entry record set include "Ticket Number" (formatted as "T + Date + 6-digit Serial Number", e.g., "T20240927000001"), "Entry Timestamp" (accurate to milliseconds, e.g., "2024-09-27 14:05:30.123"), and "User Identity Identifier" (including anonymized mobile phone number, the last 6 digits of ID card, etc., e.g., mobile phone number "138****5678"). When filtering by identity, a mapping relationship is established based on the user's unique identity (such as a de-identified user ID). Only the entry records corresponding to the user currently awaiting verification are retained, while other user records and expired records (such as those past 24:00 on the day of the event) are removed. At the same time, the latest entry timestamp of the same ticket number is retained through the "duplicate record deduplication rule" (if there are multiple entry applications). Finally, a set of filtered entry records is formed, with the following format example: "[{Ticket Number: T20240927000001, Entry Timestamp: 2024-09-27 14:05:30.123, User Identity: U001},{...}]".

[0105] Based on the ticket numbers in the filtered set of entry records, a pre-established verification status database is queried to determine whether the ticket number exists in the list of verified tickets. When obtaining the matching result set, the verification status database uses a MongoDB non-relational database. Core fields include "ticket number," "verification status" (0 indicates not verified, 1 indicates verified), "verification timestamp," and "operation log." An index is created using the ticket number to optimize query efficiency, reducing the single query response time from 200ms to less than 30ms. During the query, each ticket number in the filtered set of entry records is used as a query keyword to perform an "exact match query"—for example, for the ticket number "T20240927000001," the "verification status" field corresponding to that number is queried in the database. If the field value is 1, it is determined that "it exists in the list of verified tickets"; if it is 0 or there is no matching record, it is determined that "it does not exist in the list of verified tickets." The query results for all invoice numbers are summarized into key-value pairs in the form of "invoice number - matching result" to form a matching result set. An example format is "[{invoice number: T20240927000001, matching result: does not exist},{invoice number: T20240927000002, matching result: exists},...]". At the same time, the response time and operation log of each query are recorded to facilitate subsequent data traceability and problem investigation. If the matching result set shows that the ticket number has not been cancelled and the ticket authenticity label is genuine, then the cancellation status of the ticket number in the cancellation status database will be updated to cancelled. When the status update confirmation is obtained, the transaction processing logic of "verify first, then update" must be strictly followed. First, a second verification is performed on ticket numbers that are "not cancelled" (matching result is "does not exist") and "authenticity label is genuine" (label format is "ticket number_1", such as "T20240927000001_1") to confirm that their entry record timestamp has not exceeded the event validity period (such as 14:00-22:00 on the day of the event) to avoid accidental operation on expired tickets. After the verification is passed, the cancellation status update operation is performed: through the database transaction management mechanism, the "cancellation status" corresponding to the ticket number is updated from 0 to 1, and the "cancellation timestamp" (accurate to milliseconds, consistent with the current system time) and "operator identifier" (such as scanning device number "SCAN001") are recorded to ensure the traceability of the update operation. After the update is complete, the system will automatically return a "Status Update Confirmation" message. This message must include the invoice number, the status before the update, the status after the update, the update timestamp, and the verification result. An example of the format is: "Status Update Confirmation: Invoice Number T20240927000001, Status before the update 0 (not cancelled), Status after the update 1 (cancelled), Update Timestamp 2024-09-27 14:06:15.789, Verification Result: Passed".If a concurrent conflict occurs during the update process (such as other devices initiating an update for the ticket number at the same time), a retry mechanism will be triggered (up to 3 retries, with an interval of 100ms between each retry). If the retry fails, it will be marked as "update error" and an alarm message will be pushed to ensure data consistency.

[0106] Based on the status update confirmation, a verification signal is generated. If the generation timestamp of the verification signal is within a preset valid period after the entry timestamp of the corresponding ticket in the entry record set, then it is determined to be the verification confirmation signal. The verification signal format must include core identification information, for example, "Verification Signal_Ticket Number T20240927000001_Generation Timestamp 2024-09-27 14:06:15.789_Status 1". The preset valid period is set to 30 minutes, that is, the time difference between the verification signal generation timestamp and the entry timestamp must be within 30 minutes to avoid invalid verification due to delayed operation. If the time difference is within 30 minutes, it is determined to be "within the preset valid time period," and the cancellation signal is identified as a cancellation confirmation signal. An example format is "Cancellation Confirmation Signal_Ticket Number T20240927000001_Within Valid Time Period_Status: Cancelled." If the time difference exceeds 30 minutes, it is marked as a "Time-Expired Signal," and the entry record review process is automatically triggered to check for time synchronization anomalies or human error delays. After the cancellation confirmation signal is generated, it must be synchronously pushed to a real-time database (such as a Redis cache) and the on-site cancellation terminal to ensure that subsequent ticketing data retrieval and on-site entry guidance can obtain the signal in real time.

[0107] In summary, the accurate screening of entry records provides a targeted data foundation for subsequent matching, the query and update of the verification status ensures the consistency and security of ticketing data, and the valid time period verification of the timestamp avoids business risks caused by delayed operations. The entire process takes over the generation of ticket authenticity labels, realizing a business closed loop of "authenticity verification - verification status confirmation", effectively solving problems such as "duplicate verification", "fake ticket verification" and "loss of timeliness" in traditional verification, and improving the rigor and real-time nature of ticketing management for large-scale events.

[0108] The system acquires scan data uploaded by the on-site scanning device, including the ticket number and on-site scanning timestamp. The ticket number is then matched with the verification confirmation signal to filter out valid scan data. The on-site scanning device must meet the requirements of real-time transmission and high-precision timestamp acquisition—the device sampling frequency is set to 1 time / second to ensure that only one record is generated for each ticket scan, avoiding duplicate data; the on-site scanning timestamp is accurate to milliseconds (format: “YYYY-MM-DDHH:MM:SS.fff”, e.g., “2024-09-27 14:10:25.345”) and synchronized with the device's local clock (clock error ≤ 1 second, periodically calibrated via NTP protocol). The core fields of the scan data include the “ticket number” (following the format “T + date + 6-digit serial number”, e.g., “T20240927000003”), the “on-site scanning timestamp”, and the “scanning device number” (e.g., “SCAN005”). The data is uploaded to the backend server in real-time via the WebSocket protocol, with transmission latency controlled within 50ms. During the matching process, the "invoice number" is used as the unique association key. The verification confirmation signal contains a verified invoice number (e.g., "T20240927000003"). The invoice numbers in the scanned data are compared one by one with those in the verification confirmation signal. If a corresponding record can be found in the verification confirmation signal and the confirmation signal status is "confirmed," the scanned data is considered valid. If no corresponding verification confirmation signal is found, or the confirmation signal status is "expired," it is marked as invalid data and temporarily stored in the exception database (retained for 72 hours for traceability). For example, the scanned data might be "{Invoice Number:T20240927000003, On-site Scan Timestamp:2024-09-27 14:10:25.345, Scanning Device Number:SCAN005}".

[0109] In step S107, based on the cancellation confirmation signal, a real-time database update mechanism is adopted, and the on-site scanning timestamp is integrated to determine the change in the cancellation status of the invoice, thereby obtaining the updated invoice database.

[0110] In one optional implementation, the step of determining the change in invoice verification status and obtaining the updated invoice database based on the verification confirmation signal, using a real-time database update mechanism and integrating the on-site scan timestamp, includes:

[0111] The system acquires scanned data uploaded by the on-site scanning device, which includes the invoice number and the on-site scanning timestamp. It then associates and matches the invoice number with the verification confirmation signal to filter out valid scanned data.

[0112] Call the preset invoice database to query the current status of the invoice number in the valid scanned data. If the current status is not cleared, trigger the real-time status update operation.

[0113] Update the status of the corresponding ticket in the ticket database from unverified to verified, and record the update timestamp that merges the on-site scanning timestamp;

[0114] According to the preset data consistency verification rules, the updated invoice database data is verified, and if the verification passes, it is determined to be the updated invoice database.

[0115] It should be noted that if the invoice number exists in the verification confirmation signal and its status is valid, then the scanned data is filtered as valid scanned data, forming a valid scanned dataset.

[0116] The system calls a pre-defined invoice database to query the current status of invoice numbers in the valid scanned data. If the current status is "not yet verified," a real-time status update operation is triggered. The pre-defined invoice database is built using a MySQL relational database (supporting transaction processing and high-concurrency queries). Core fields include "Invoice Number," "Current Verification Status" (0 indicates not verified, 1 indicates verified), "First Scan Time," "Last Update Time," and "Associated User Identifier." A unique index is created for "Invoice Number," optimizing the single query response time to within 20ms. When calling the invoice database, the current status of each invoice number in the valid scanned data is obtained through an SQL query statement (such as "SELECT Current Verification Status FROM Invoice Database WHERE Invoice Number='T20240927000003'"). During the query process, a database read / write separation mechanism is enabled (read operations are distributed to the slave database to avoid pressure on the master database). If the query result shows "Current verification status = 0" (not verified), a real-time status update operation is immediately triggered, generating an update task queue (using a FIFO (First-In-First-Out) mechanism to ensure sequential task execution). If the query result is "Current verification status = 1" (verified), a "duplicate scan" log is recorded (including the invoice number, scan timestamp, and device number), but no update operation is triggered to avoid duplicate data modification. For example, if the current status of invoice number "T20240927000003" is 0, an update task is triggered, and the task is added to the queue for execution.

[0117] When updating the status of the corresponding ticket in the ticket database from unverified to verified, and recording the update timestamp that merges the on-site scan timestamp, a database transaction mechanism is used to ensure the atomicity of the update operation (either all succeed or all rollback) – the update SQL statement contains two core operations: first, updating the "current verification status" from 0 to 1, and second, setting the "last update time" to a composite timestamp that merges the on-site scan timestamp (format: "system update time_on-site scan timestamp", such as "2024-09-27 14:10:30.123_2024-09-27 14:10:25.345"). At the same time, an "update operation log" is also recorded (fields include operation serial number, ticket number, operation device, and status before and after the update). During the update process, row-level locking (locking the record corresponding to the current ticket number) is enabled to prevent state chaos caused by concurrent updates from multiple threads. If concurrent update requests exist (such as the same ticket being scanned by two devices simultaneously), a retry mechanism is triggered (up to 3 retries, each with a 150ms interval). If a retry fails, it is marked as an "update conflict" and an alarm is pushed to the operation and maintenance system for manual verification. For example, for ticket number "T20240927000003", after the update, its "current reconciliation status" becomes 1, and the "last update time" is recorded as "2024-09-27 14:10:30.123_2024-09-27 14:10:25.345". A corresponding transaction record is generated in the operation log to ensure that the update process is traceable.

[0118] According to the preset data consistency verification rules, the updated invoice database data is verified. Once the verification passes, it is determined to be the updated invoice database. The preset verification rules include three core dimensions: First, **field integrity verification**, checking whether there are any missing core fields such as "invoice number", "current cancellation status", "last update time", and "update operation log" in the updated records. If any field is empty, the verification is deemed to have failed. Second, **logical consistency verification**, verifying that "current cancellation status" can only be 1 (cancelled), and the system update time in "last update time" must be later than the on-site scan timestamp (error ≤ 10 seconds, ensuring that the update operation is completed within a reasonable time after scanning). If the status is abnormal or the time logic is incorrect, the verification is deemed to have failed. Third, **redundant data verification**, querying whether there are duplicate records with the same invoice number in the invoice database (verified through the unique index of "invoice number"). If duplicates exist, the verification is deemed to have failed. The verification process combines batch verification and sampling verification. First, batch field and logic verification is performed on all updated records. Then, 5% of the records are randomly selected for redundancy verification. If the batch verification pass rate is 100% and the sampling verification shows no anomalies, the verification is considered successful. If the verification fails, a data rollback operation is immediately triggered (restoring the updated records in that batch to their pre-update state). Simultaneously, a verification failure report (including the invoice number of the failed record and the reason for failure) is generated and pushed to the data operation and maintenance module for manual review. For example, in the updated invoice database, if all core fields of invoice number "T20240927000003" are complete, its status is 1, its time logic is reasonable, and there are no duplicate records, and the sampling verification shows no anomalies, then this invoice database passes the verification and is confirmed as the updated invoice database.

[0119] In summary, the accurate acquisition and correlation matching of on-site scanning data ensures the "legitimacy" of the update operation, the ticket database status query and real-time update mechanism guarantees the "timeliness" of the data, and multi-dimensional consistency verification safeguards the "accuracy" of the data. The entire process inherits the verification confirmation signal from the preceding stage, and through technical means such as equipment calibration, transaction control, and verification rules, it realizes real-time and accurate changes in the ticket verification status, builds a traceable and highly reliable ticket database update system, and effectively solves the problems of "data delay", "state chaos" and "missing verification" in traditional ticket database updates, providing accurate and real-time basic data support for ticket management.

[0120] In step S108, based on the updated ticket database and the batch entry queue data, the entry order is adjusted through queue optimization to determine the risk of congestion during peak periods and obtain an optimized entry sequence.

[0121] In one optional implementation, the step of adjusting the entry order through queue optimization based on the updated ticket database and the batch entry queue data, and determining the risk of congestion during peak periods to obtain an optimized entry sequence includes:

[0122] The updated ticket database is used to filter out ticket numbers that have been cancelled and are within their validity period, and these numbers are then merged with the batch entry queue data to obtain an entry queue to be sorted.

[0123] The system determines whether the current period is a peak period based on the preset peak period determination rules. If it is not a peak period, the entry sequence is sorted according to the priority of the bill type. If it is a peak period, the entry sequence is sorted according to the bill clearing time to obtain the preliminary entry sequence.

[0124] The number of people in the initial entry sequence is calculated and combined with the preset risk assessment criteria. If the number of people meets the congestion risk assessment conditions, it is determined that there is a congestion risk, triggering a dynamic adjustment operation to rearrange the entry order and obtain the optimized entry sequence.

[0125] It should be noted that when filtering out the cancelled and valid invoice numbers from the updated invoice database and merging them with the batch entry queue data to obtain the entry queue to be sorted, the core fields of the updated invoice database must first be defined—including "Invoice Number" (formatted as "T + Date + 6-digit serial number", such as "T20240927000005"), "Cancellation Status" (1 indicates cancelled), and "Validity Expiry Date" (formatted as "YYYY-MM-D"). DHH:MM:SS”, such as “2024-09-27 22:00:00”, must meet the following conditions during the screening: “Verification status = 1” and “Validity expiration time > current system time” (current time such as “2024-09-27 15:00:00”). Expired (such as validity expiration time of “2024-09-27 14:00:00”) or unverified ticket numbers will be removed to ensure that the screening results are all valid tickets that are “accessible and verified”. The batch entry queue data includes "ticket number to be entered" and "pre-query result of verification status" (such as "not verified and pending verification" or "pre-verified"). When merging, "ticket number" is used as the unique association key, and the "duplicate priority rule" is adopted. If the same ticket number exists in both the filtered ticket database result and the batch entry queue data, only the record in the ticket database is retained (because the ticket database data has completed verification and confirmation, its priority is higher than the pre-query result). Finally, a queue to be sorted is formed. The format example is "[{ticket number:T20240927000005,ticket type:VIP,verification time:2024-09-2714:55:30},{ticket number:T20240927000006,ticket type:ordinary ticket,verification time:2024-09-2714:56:10},...]".

[0126] The system determines whether the current time period is a peak period based on preset peak period judgment rules. If it is not a peak period, the entry sequence is sorted according to the priority of ticket type. If it is a peak period, the entry sequence is sorted according to the ticket redemption time. When obtaining the preliminary entry sequence, the preset peak period judgment rules need to be formulated in combination with historical entry data and on-site capacity thresholds. By statistically analyzing the entry flow curves of similar past events, the system sets "the proportion of people entering per unit time to the total venue capacity ≥ 60%" or "the number of people entering per minute ≥ 80" as the criteria for determining peak periods. For example, if the total venue capacity is 5,000 people, and the number of people entering during the current time period (15:00-15:05) reaches 320 people (64 people per minute), which is less than 80 people per minute, it is judged as not a peak period. If the number of people entering from 15:30 to 15:35 reaches 420 people (84 people per minute), it is judged as a peak period. During off-peak periods, when sorting tickets by type, the priority rule is preset as "VIP tickets > Group tickets > Regular tickets > Discount tickets". For example, if the queue to be sorted contains VIP tickets (T20240927000005), regular tickets (T20240927000006), and group tickets (T20240927000007), the initial entry sequence after sorting would be "T20240927000005 (VIP) → T20240927000007 (Group) → T20240927000007 (Group) → T20240927000005 (VIP) → T20240927000007 (Group) → T20240927000007 (Group) → T20240927000006 ...6 (Group) → T20240927000006 (Group) → T20240927000006 (Group) → T20240927000006 (Group) → T20240927000006 (Group) → T20240927000006 (Group) (Body) → T20240927000006 (Ordinary)”; During peak periods, when sorting by invoice verification time, the “verification time” (accurate to the second) in the invoice database is used as the basis, and the invoices are arranged according to the “first-to-enter” principle. For example, T20240927000005 with a verification time of 14:55:30 is placed before T20240927000006 with a verification time of 14:56:10, to ensure that the entry order is consistent with the verification rhythm and to avoid queue congestion caused by priority sorting during peak periods.

[0127] The number of people in the initial entry sequence is calculated and combined with preset risk assessment criteria. If the number of people meets the congestion risk assessment conditions, a congestion risk is determined, triggering a dynamic adjustment operation to rearrange the entry order and obtain the optimized entry sequence. The number of people in the initial entry sequence is the total number of ticket numbers in the queue (each ticket corresponds to 1 person). For example, if the initial sequence contains 120 ticket numbers, the number of people is 120. The preset risk assessment criteria need to be related to the instantaneous carrying capacity of the entrance—based on the entrance channel width (e.g., a 1.5-meter channel) and the processing efficiency of the security equipment (e.g., each security machine processes 15 people per minute), the congestion risk assessment conditions are set as "initial sequence number of people at a single entrance > 100 people" or "estimated queue length > 30 meters" (calculated based on an average queue spacing of 0.3 meters per person, 100 people correspond to 30 meters). If the number of people in the initial sequence is 120, exceeding the 100-person threshold, a congestion risk is determined. When dynamic adjustment is triggered, a "diversion + batching" strategy is adopted: First, **entrance diversion**: the real-time queue length of each entrance is queried, and the number of people exceeding the threshold (e.g., 20 people) is allocated to entrances with queue lengths of less than 50 people. For example, tickets T20240927000025 to T20240927000044 (20 tickets in total) are diverted to entrance No. 2. Second, **batch entry**: the 100 people remaining at the current entrance are split into batches of 20 people each, with a 2-minute interval between batches. The batch number and entry time window are marked in the sequence (e.g., "T20240927000005_batch"). The optimized entry sequence, which includes both the diversion of entry information and the batch and time markings, is formatted as "[{ticket number:T20240927000005,Entrance 1,Batch 1,15:10-15:12},{ticket number:T20240927000007,Entrance 1,Batch 1,15:10-15:12},...,{ticket number:T20240927000025,Entrance 2,Batch 1,15:10-15:12}]". This ensures a balanced flow of people at each entrance and avoids instantaneous congestion.

[0128] In summary, the construction of the queue for entry, through screening and merging, ensures the "validity" of entry data; peak period determination and differentiated sorting achieve the "adaptability" of entry order; and congestion risk assessment and dynamic adjustment guarantee the "smoothness" of the entry process. The entire process inherits the updated ticket database data from the previous stage, formulates rules based on historical data, controls risks with on-site capacity as the core, and optimizes the entry sequence through diversion, batching, and other means. This effectively solves the problems of "peak congestion," "chaotic order," and "uneven load at the entrance" in traditional entry management, providing a feasible technical solution for efficient entry scheduling of large-scale offline events.

[0129] Reference Figure 2 The second embodiment of the present invention provides an artificial intelligence-based ticket security verification system, comprising:

[0130] The data acquisition module is used to acquire ticket spectral data, associated ticket entry record data, and batch entry queue data to construct a preliminary spectral feature set.

[0131] The dimensionality reduction module is used to perform dimensionality reduction processing on the preliminary spectral feature set, determine the dominant frequency band, extract the reflection mode and absorption peak of the dominant frequency band, and determine the refined spectral vector.

[0132] The deviation judgment module is used to determine potential counterfeiting deviation if the absorption peak of the refined spectral vector in the ultraviolet band deviates from the absorption peak of the preset standard by more than a preset ultraviolet band absorption peak deviation threshold, and obtain a subset of the deviation frequency band.

[0133] The variable fusion module is used to acquire on-site variables such as ambient light and ticket aging degree. It fuses the on-site variables with the deviation frequency band subset data through a multilayer sensor to generate a corrected spectral feature vector and compares it with the spectral data of a preset standard to determine the consistency of the corrected spectrum and obtain a consistency score.

[0134] The authenticity classification module is used to classify tickets into authenticity categories using a support vector machine based on the consistency score, and obtain ticket authenticity labels; the verification confirmation module is used to determine whether the entry record data matches the verified tickets based on the ticket authenticity labels and the associated ticket entry record data, and obtain a verification confirmation signal.

[0135] The database update module is used to determine the change in the invoice cancellation status based on the cancellation confirmation signal, using a real-time database update mechanism and integrating the on-site scanning timestamp, and to obtain the updated invoice database.

[0136] The queue optimization module is used to adjust the entry order based on the updated ticket database and the batch entry queue data, determine the risk of congestion during peak periods, and obtain an optimized entry sequence.

[0137] The present invention provides an artificial intelligence-based ticket security verification system for executing all process steps of the artificial intelligence-based ticket security verification method described in the above embodiments. The working principles and beneficial effects of the two are one-to-one, and therefore will not be described in detail again.

[0138] This 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 AI-based secure ticket verification program. When the processor executes the computer program, it implements the steps described in the various AI-based secure ticket verification method embodiments above, for example... Figure 1 The step S101 shown is (acquiring ticket spectral data, associated ticket entry record data, and batch entry queue data to construct a preliminary spectral feature set). Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as a data acquisition module (used to acquire ticket spectral data, associated ticket entry record data, and batch entry queue data to construct a preliminary spectral feature set).

[0139] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0140] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0141] The processor can be a Central Processing Unit (CPU), or 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. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0142] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0143] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0144] 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 separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0145] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. An artificial intelligence-based ticket security verification method, characterized in that, The computer is executed, comprising: Obtain ticket spectrum data, associated ticket entry record data, batch entry queue data, and construct a preliminary spectrum feature set; Perform dimensionality reduction processing on the preliminary spectrum feature set, determine the dominant frequency band, extract the reflection mode and absorption peak value of the dominant frequency band, and determine the refined spectrum vector; If the absorption peak value of the refined spectrum vector in the ultraviolet frequency band deviates from the absorption peak value of the preset standard product by more than a preset ultraviolet frequency band absorption peak value deviation threshold, it is judged as a potential forgery deviation, and a deviation frequency band subset is obtained; Obtain the on-site variables of environmental illumination and ticket aging degree, fuse the on-site variables and the deviation frequency band subset data through a multilayer perceptron to generate a corrected spectrum feature vector and compare it with the preset standard product spectrum data, judge the consistency of the corrected spectrum, and obtain a consistency score; According to the consistency score, classify the tickets into true and false categories through a support vector machine to obtain a ticket true and false label; According to the ticket true and false label, combined with the associated ticket entry record data, judge whether the entry record data matches the already cancelled ticket, and obtain a cancellation confirmation signal; According to the cancellation confirmation signal, adopt a real-time database update mechanism, fuse the on-site scanning time stamp, determine the ticket cancellation state change, and obtain an updated ticket library; According to the updated ticket library and the batch entry queue data, adjust the entry order through queue optimization, judge the peak period congestion risk, and obtain an optimized entry sequence. 2.The AI-based ticket security validation method of claim 1, wherein, The obtaining of the ticket spectrum data, the associated ticket entry record data, and the batch entry queue data, and the construction of the preliminary spectrum feature set, comprises: Collecting original multi-dimensional spectrum data containing ticket paper material characteristics, ink spectrum reflection, reflection mode and absorption peak value, forming an initial spectrum data set; Filtering noise from the initial spectrum data set, extracting feature vectors of forgery identification mode and reflection peak absorption, comparing the feature vectors with the preset data validity standard, and determining the preliminary spectrum feature set if it meets the standard; Obtain entry record data containing ticket number, time stamp and user identity related identification, and obtain associated ticket entry record data; Obtain queue data containing to-be-entered ticket number, cancellation state pre-query result, and obtain batch entry queue data. 3.The AI-based ticket security validation method of claim 1, wherein, The dimensionality reduction processing on the preliminary spectrum feature set, the determination of the dominant frequency band, the extraction of the reflection mode and the absorption peak value of the dominant frequency band, and the determination of the refined spectrum vector, comprise: Standardizing the reflectance and absorption value data in the preliminary spectrum feature set, and performing dimensionality reduction processing on the standardized data through principal component analysis to obtain dominant frequency band data; Extract the reflection mode and absorption peak value of the dominant frequency band data and integrate them into a low-dimensional feature vector to obtain a refined spectrum vector. 4.The AI-based ticket security validation method of claim 1, wherein, If the absorption peak value of the refined spectrum vector in the ultraviolet frequency band deviates from the absorption peak value of the preset standard product by more than a preset ultraviolet frequency band absorption peak value deviation threshold, it is judged as a potential forgery deviation, and a deviation frequency band subset is obtained, comprising: Calculate the deviation value of the ultraviolet frequency band absorption peak value of the refined spectrum vector and the absorption peak value of the preset standard product to obtain a deviation value set; If at least one offset value in the offset value set exceeds a preset ultraviolet frequency band absorption peak offset threshold, it is determined that there is a potential counterfeit bias; The offset value set is processed by frequency band segmentation to obtain a deviation frequency band subset. 5.The AI-based ticket security validation method of claim 1, wherein, Obtain on-site variables of ambient light and ticket aging degree, fuse the on-site variables and the deviation frequency band subset data through a multilayer perceptron to generate a corrected spectral feature vector and compare it with preset standard sample spectral data to determine the consistency of the corrected spectrum and obtain a consistency score, including: Obtain on-site ambient light intensity and ticket aging degree data as the on-site variables; Process the on-site variables using a data standardization processing method to obtain a standardized on-site variable set; Input the deviation frequency band subset data and the standardized on-site variable set into a multilayer perceptron, fuse multi-source data through hierarchical calculation of the multilayer perceptron, and generate a corrected spectral feature vector; Calculate the matching degree of the corrected spectral feature vector and the preset standard sample spectral data, and if the matching degree meets the preset matching degree determination standard, calculate the consistency score through a preset scoring method. 6.The AI-based ticket security validation method of claim 1, wherein, According to the consistency score, classify the tickets into true and false categories through a support vector machine to obtain a ticket true and false label, including: Standardize the consistency score to obtain a standardized consistency score; Classify the consistency score into true and false categories through a support vector machine to obtain a preliminary labeling result; Convert the preliminary labeling result into identification information in a unified format through a label mapping operation to determine the ticket true and false label. 7.The AI-based ticket security validation method of claim 1, wherein, According to the ticket true and false label, combine the associated ticket entry record data to determine whether the entry record data matches the already canceled ticket, and obtain a cancellation confirmation signal, including: Obtain an entry record set containing ticket numbers and timestamps, filter the entry record set according to identity identifiers to obtain a filtered entry record set; According to the ticket numbers in the filtered entry record set, query a pre-established cancellation status database to determine whether the ticket numbers exist in an already canceled ticket list to obtain a matching result set; If the matching result set shows that the ticket numbers have not been canceled and the ticket true and false label is a true ticket, update the cancellation status of the ticket numbers in the cancellation status database to already canceled to obtain a status update confirmation; According to the status update confirmation, generate a cancellation signal, and if the generation timestamp of the cancellation signal is within a preset valid time period after the entry timestamp of the corresponding ticket in the entry record set, determine the cancellation confirmation signal. 8.The AI-based ticket security validation method of claim 1, wherein, According to the cancellation confirmation signal, use a real-time database update mechanism to fuse the on-site scanning timestamp to determine the ticket cancellation status change and obtain an updated ticket library, including: Obtain scanning data uploaded by the on-site scanning device, the scanning data containing a ticket number and an on-site scanning timestamp, and perform association matching between the ticket number and the cancellation confirmation signal to screen valid scanning data; Call a preset ticket library, query a current state of the ticket number in the valid scanning data, and if the current state is not cancelled, trigger a real-time state updating operation; update the state of the corresponding ticket in the ticket library from not cancelled to cancelled, and record an updating timestamp that integrates the on-site scanning timestamp; According to a preset data consistency checking rule, check the updated ticket library data, and if the checking is passed, determine the updated ticket library. 9.The AI-based ticket security validation method of claim 1, wherein, According to the updated ticket library and the batch entry queue data, adjust an entry order through queue optimization, judge a peak period congestion risk, and obtain an optimized entry sequence, including: Screen ticket numbers that are cancelled and within a valid period from the updated ticket library, and merge the ticket numbers with the batch entry queue data to obtain an entry queue to be sorted; According to a preset peak period judgment rule, judge whether a current period is a peak period, if the current period is not a peak period, sort the entry sequence according to a ticket type priority, if the current period is a peak period, sort the entry sequence according to a ticket cancellation time sequence, and obtain a preliminary entry sequence; Calculate a number of people in the preliminary entry sequence, and if the number of people meets a congestion risk judgment condition according to a preset risk judgment standard, judge that there is a congestion risk, trigger a dynamic adjustment operation, and rearrange the entry order to obtain the optimized entry sequence.

10. An artificial intelligence based ticket security validation system characterized in that, including: A data acquisition module configured to acquire ticket spectrum data, associated ticket entry record data, and batch entry queue data, and construct a preliminary spectrum feature set; A dimension reduction processing module configured to perform dimension reduction processing on the preliminary spectrum feature set, determine a dominant frequency band, extract a reflection mode and an absorption peak value of the dominant frequency band, and determine a refined spectrum vector; A deviation judgment module configured to, if the absorption peak value of the refined spectrum vector in an ultraviolet frequency band deviates from an absorption peak value of a preset standard product by more than a preset ultraviolet frequency band absorption peak value deviation threshold, judge that there is a potential forgery deviation, and obtain a deviation frequency band subset; A variable fusion module configured to acquire on-site variables of environmental illumination and ticket aging degree, generate a corrected spectrum feature vector by fusing the on-site variables and the deviation frequency band subset data through a multilayer perceptron, and compare the corrected spectrum feature vector with preset standard product spectrum data to judge spectrum consistency and obtain a consistency score; A true-false classification module configured to classify tickets into true-false categories according to the consistency score through a support vector machine to obtain a ticket true-false label; A cancellation confirmation module configured to, according to the ticket true-false label, judge whether the entry record data matches a cancelled ticket in combination with the associated ticket entry record data to obtain a cancellation confirmation signal; A database updating module configured to, according to the cancellation confirmation signal, use a real-time database updating mechanism, integrate an on-site scanning timestamp, determine a ticket cancellation state change, and obtain an updated ticket library. A queue optimization module is configured to adjust the entry order through queue optimization, judge the risk of congestion during rush hours, and obtain an optimized entry sequence according to the updated ticket pool and the batch entry queue data.

Images