A multi-interface unified scheduling control method for rail transit self-service ticket checking

Abstract

The application discloses a kind of track traffic self-service ticket checking multi-interface unified scheduling control method, it is related to track traffic automatic ticket checking field;Solve the conflict arbitration when multiple ticket checking interfaces concurrent request on gate passage, transaction time sequence confusion and repeated charge problem.The application constructs environmental interference index by collecting environmental working condition data, dynamically adjusts conflict heat threshold and interface transaction time window;Based on interface space parameter and time window, construct space-time influence kernel function, get conflict heat value by weighted superposition and nonlinear entropy integral, trigger candidate scheduling set generation;Determine optimal activation sequence by minimizing scheduling loss function containing operation urgency and interface operation time interval index cost;When executing sequence, use global state snapshot and buffer queue to compare voucher and space-time continuity, complete unified scheduling control to multiple ticket checking interfaces.

Description

Technical Field

[0001] This invention belongs to the field of automatic fare collection in rail transit, specifically a multi-interface unified scheduling and control method for self-service fare collection in rail transit. Background Technology

[0002] The automated fare collection system for rail transit is a key component of urban public transportation operation and management. The turnstiles, as the core nodes for passenger entry and exit, play a crucial role in ticket verification and access control. With the widespread adoption of mobile payment and biometric technologies, a single turnstile typically houses multiple ticket verification interfaces. Typical configurations include a 13.56 MHz RFID reader, a QR code scanner, a near-field communication module, and a facial recognition camera. These various interfaces correspond to different types of ticket media, providing passengers with diverse ways to pass through the turnstiles.

[0003] In actual operation, rail transit stations are characterized by high passenger density and stringent timeliness requirements. During morning and evening rush hours, a large number of passengers pass through the turnstiles in a short period of time, and multiple passengers may use different ticket checking media at almost the same turnstile channel. For example, while one passenger holds a transit card close to the RFID reader, another passenger holds a mobile phone QR code to the scanner, and even the same passenger may trigger multiple interfaces at the same time due to unfamiliarity with the operation. Since the ticket checking interfaces are physically close to each other and each corresponds to independent reading and writing hardware and backend communication protocols, how to coordinate the execution order of these interfaces, avoid mutual interference, and prevent duplicate charges or wrongful passage when multiple interfaces generate trigger events in the same time period has become an urgent technical problem to be solved.

[0004] The following problems exist in the existing technology: Existing gate scheduling methods typically use fixed thresholds to determine whether to trigger multi-interface coordination, without considering the impact of ambient light flicker, electromagnetic noise interference, and changes in passenger flow density on sensor triggering accuracy. This results in frequent false triggers in noisy environments or sluggish responses in quiet environments. Existing methods for quantifying concurrent conflicts across multiple interfaces often employ simple event counting or linear weighting, which cannot sensitively reflect the true degree of conflict when multiple low-intensity events overlap in time and space, and are prone to missing or misjudging the timing of scheduling intervention. Existing methods typically determine the scheduling order of multiple interfaces based on fixed priorities or first-come-first-served rules, without considering the urgency of passage reflected by repeated passenger operations, or incorporating the physical constraints of safe intervals between interfaces into the decision model. This results in passengers with high urgency waiting for too long or the risk of accidental activation in the sequence arrangement. Existing methods, during sequential execution, typically handle sudden requests from interfaces not included in the current sequence by either locking them out or dropping them directly. They lack a mechanism to distinguish between requests triggered by the same passenger and independent new requests, which can easily lead to duplicate charges or missed detections. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a multi-interface unified scheduling and control method for self-service ticket checking in rail transit, which is used to solve the above-mentioned technical problem.

[0006] The first aspect of the present invention provides a multi-interface unified scheduling and control method for self-service ticket checking in rail transit, comprising the following steps: S1: Obtain the coordinates of the operation center and the effective sensing radius of each of the multiple ticket checking interfaces coexisting on the gate channel, allocate a benchmark transaction time window for each ticket checking interface, and initialize the benchmark threshold for conflict heat; collect real-time trigger events of multiple ticket checking interfaces and environmental condition data of the gate site; S2: Construct an environmental interference index using the environmental condition data; adjust the conflict heat benchmark threshold according to the environmental interference index to obtain an adjusted conflict heat threshold, and adjust the benchmark transaction time window of each ticket checking interface to obtain an adjusted transaction time window. S3: Filter the real-time triggered events that have not yet been verified and mark them as unprocessed events; calculate the urgency of the operation of the ticket checking interface corresponding to the unprocessed event based on the number of times each unprocessed event is repeatedly triggered within a preset time period, and determine its expected verification time based on the historical verification time record of the ticket checking interface; S4: Based on the operation center coordinates and effective sensing radius of each ticket checking interface, define the operation space in front of the gate, and construct a spatiotemporal influence kernel function for the unprocessed events in conjunction with the adjusted transaction time window of the ticket checking interface; perform weighted superposition of all spatiotemporal influence kernel functions within the operation space that are within the adjusted transaction time window and perform nonlinear entropy integration to obtain a conflict heat value; when the conflict heat value reaches the adjusted conflict heat threshold, include all ticket checking interfaces corresponding to the current unprocessed events into the candidate scheduling set; S5: For the ticket checking interfaces in the candidate scheduling set, a scheduling loss function is constructed based on the operation urgency, the expected verification time, and the nonlinear cost function with the operation time interval between the ticket checking interfaces as the variable for each ticket checking interface; by minimizing the scheduling loss function, the activation sequence of the ticket checking interfaces and their operation time intervals are determined. S6: Activate the ticket checking interfaces sequentially according to the activation sequence to perform ticket checking verification; save a global ticket checking status snapshot when activating the current ticket checking interface, and temporarily store the real-time trigger events newly generated by the inactive ticket checking interfaces from the time the global ticket checking status snapshot is saved in the buffer queue; after the current ticket checking verification is completed, extract the real-time trigger events one by one from the buffer queue, and compare them with the global ticket checking status snapshot for document consistency and spatiotemporal continuity. If the comparison results are consistent, the real-time trigger event is discarded; if they are inconsistent, the real-time trigger event is transferred to the next scheduling cycle for processing.

[0007] Preferably, step S2 includes the following steps: The environmental operating data includes ambient light flicker variance Lv, spatial electromagnetic noise power Rn, and passenger flow density in front of the gate Cd; Constructing the environmental disturbance index ,in, The preset electromagnetic noise reference zero point, , and These are the preset standardization coefficients; For the aforementioned conflict heat benchmark threshold Adjustments were made to obtain the adjusted conflict heat threshold. ; The baseline transaction time window for each of the ticket checking interfaces Adjustments are made to obtain the adjusted trading time window. ,in This is the preset time window adjustment coefficient.

[0008] Preferably, in step S3, calculating the operational urgency of the ticket checking interface corresponding to each unprocessed event based on the number of times each unprocessed event is repeatedly triggered within a preset time period includes the following steps: Starting from the trigger time of the unprocessed event, a preset time period is traced backward. The number of real-time triggered events generated by the same ticket checking interface within that preset time period, and whose voucher digest matches the voucher digest of the unprocessed event, is counted as the repeated trigger count of the unprocessed event. The voucher summary is generated by the ticket checking interface after reading the passenger's ticket card medium and is included in the real-time trigger event. Calculate the operational urgency of the ticket checking interface corresponding to the unprocessed event. ,in This is the preset urgency growth coefficient.

[0009] Preferably, in step S3, determining the expected verification time based on the historical verification time records of the ticket checking interface includes the following steps: Maintain a historical verification time queue for each ticket checking interface. The queue stores the actual time taken for the N most recent successful ticket checking verifications of the ticket checking interface. When the historical verification time queue is not empty, the arithmetic mean of all record values ​​in the queue is taken as the expected verification time of the ticket checking interface. When the historical verification time queue is empty, the baseline transaction time window allocated to the ticket checking interface is used as the expected verification time of the ticket checking interface.

[0010] Preferably, in step S4, the step of constructing a spatiotemporal impact kernel function for the unprocessed event based on the operation center coordinates and effective sensing radius of each ticket checking interface, and in conjunction with the adjusted transaction time window of that ticket checking interface, includes the following steps: For each of the unprocessed events, extract its trigger time. The location of the operation point, and the coordinates of the operation center of the ticket checking interface corresponding to the unprocessed event. Effective sensing radius and adjusted trading time window , where i is the identifier of the ticket checking interface, and j is the sequence number of the unprocessed event; Construct the spatiotemporal impact kernel function for this unprocessed event. Where p is any point in the operating space in front of the gate, and t is any moment within a sliding time window that ends at the current moment and traces back a preset integration time. From operation point p to operation center The Euclidean distance.

[0011] Preferably, in step S4, the step of weighting and superimposing all spatiotemporal influence kernel functions within the adjusted transaction time window in the operating space and performing a nonlinear entropy integral operation on them to obtain the conflict heat value includes the following steps: The spatiotemporal impact kernel function of all unprocessed events within the operating space in front of the gate is calculated according to the weight coefficient of its respective ticket checking interface. By performing weighted superposition, the spatiotemporal heat distribution is obtained. ; Calculated conflict heat value ,in, This refers to the operating space in front of the gate. Here, T is the preset normalized time constant, T is the sliding time window ending at the current time, and V is... The volume.

[0012] Preferably, step S5 includes the following steps: Perform a full permutation of all ticket checking interfaces in the candidate scheduling set, and treat each permutation result as a candidate activation sequence. , where n is the number of ticket checking interfaces in the candidate scheduling set, and the total number of candidate activation sequences is n!; For any of the candidate activation sequences, the k-th ticket checking interface in the sequence will be... The actual verification completion time under sequential scheduling is denoted as . The ideal verification completion time of the ticket checking interface under the condition of immediate activation is recorded as . The actual verification completion time is determined recursively based on the position of the ticket checking interface in the candidate activation sequence, its expected verification time, and the operation time interval between it and the previous ticket checking interface. Construct a nonlinear cost function with the time interval gap between the ticket checking interfaces as the variable. ,in The sensitivity coefficient is the preset time interval. Constructing the scheduling loss function ,in, The urgency of the operation at the k-th ticket checking interface. This is the time interval between the current ticket checking interface and the previous ticket checking interface. The preset interval cost weighting coefficient; For all candidate activation sequences and the combination of the operation time interval values, calculate the corresponding scheduling loss function value, determine the candidate activation sequence with the smallest scheduling loss function value as the activation sequence of the ticket checking interface, and determine the operation time interval between each adjacent ticket checking interface corresponding to the activation sequence.

[0013] Preferably, step S6 includes the following steps: When the current ticket checking interface is activated and ticket checking verification begins, a global ticket checking status snapshot is saved. The global ticket checking status snapshot includes the snapshot saving time, the operation point position of the current ticket checking interface and the read voucher summary, as well as the ticket checking verification status of each ticket checking interface. Create a buffer queue and store all real-time triggered events generated by other ticket checking interfaces that are not included in the activation sequence from the time the snapshot is saved to the time when the ticket checking verification of the current ticket checking interface is completed into the queue in sequence. After the ticket checking interface completes the ticket checking verification, each real-time triggered event is extracted one by one from the buffer queue. First, it is determined whether the ticket checking interface type corresponding to the event is the same as the currently activated ticket checking interface type. If the interface types are different, the event is directly marked to be transferred to the next scheduling cycle for processing, and no further credential consistency comparison is performed. If the interface types are the same, a credential consistency comparison is performed to determine whether the credential digest carried by the event is consistent with the credential digest of the current ticket checking interface in the global ticket checking status snapshot. For real-time trigger events that match the voucher summary, a spatiotemporal continuity comparison is performed to determine whether the time difference between the trigger time of the event and the snapshot saving time is less than the adjusted transaction time window of the current ticket checking interface, and whether the distance between the operation point position of the event and the operation point position of the current ticket checking interface in the global ticket checking status snapshot is less than the effective sensing radius of the current ticket checking interface. Real-time triggered events that have consistent credential summaries and meet the conditions for spatiotemporal continuity comparison are discarded, while real-time triggered events that have inconsistent credential summaries or do not meet the conditions for spatiotemporal continuity comparison are marked for processing in the next scheduling cycle.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention constructs an environmental interference index by collecting ambient light flicker variance, spatial electromagnetic noise power, and passenger flow density in front of the gate, and dynamically adjusts the benchmark threshold for conflict heat and the benchmark transaction time window, so that the scheduling sensitivity adapts to the on-site working conditions, thereby improving the stability and response accuracy of the system in complex environments. This invention constructs a spatiotemporal influence kernel function and performs nonlinear entropy integral operation to amplify the contribution of overlapping conflicts by including logarithmic terms in the integrand function. This causes the conflict heat value to generate a significant nonlinear increase when multiple events coincide in spatiotemporal time, which can sensitively capture the sudden changes in real operational conflicts and avoid missed or misjudged scheduling opportunities. This invention incorporates a weighted penalty for verification delay based on operational urgency into the scheduling loss function, along with a nonlinear cost function that uses the operational time interval as a variable. This integrates passenger urgency and physical safety constraints between interfaces into the optimization model. By minimizing the loss function, the optimal activation sequence and its operational safety interval are determined synchronously, achieving a scheduling decision that balances efficiency and safety. This invention saves a snapshot of the global ticket checking status when the current interface is activated, and temporarily stores newly generated events from inactive interfaces in a buffer queue. After the current verification is completed, the event is compared with the snapshot one by one for credential consistency and spatiotemporal continuity, thereby distinguishing between erroneous triggers and independent ticket checking requests from the same passenger, and realizing non-blocking safe handling of concurrent events. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0016] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0017] Please see Figure 1This invention relates to a multi-interface unified scheduling and control method for self-service ticket checking in rail transit, comprising the following steps: S1: Obtain the coordinates of the operation center and the effective sensing radius of each of the multiple ticket checking interfaces coexisting on the gate channel, allocate a benchmark transaction time window for each ticket checking interface, and initialize the benchmark threshold for conflict heat; collect real-time trigger events of multiple ticket checking interfaces and environmental condition data of the gate site; S2: Construct an environmental interference index using the environmental condition data; adjust the conflict heat benchmark threshold according to the environmental interference index to obtain an adjusted conflict heat threshold, and adjust the benchmark transaction time window of each ticket checking interface to obtain an adjusted transaction time window. S3: Filter the real-time triggered events that have not yet been verified and mark them as unprocessed events; calculate the urgency of the operation of the ticket checking interface corresponding to the unprocessed event based on the number of times each unprocessed event is repeatedly triggered within a preset time period, and determine its expected verification time based on the historical verification time record of the ticket checking interface; S4: Based on the operation center coordinates and effective sensing radius of each ticket checking interface, define the operation space in front of the gate, and construct a spatiotemporal influence kernel function for the unprocessed events in conjunction with the adjusted transaction time window of the ticket checking interface; perform weighted superposition of all spatiotemporal influence kernel functions within the operation space that are within the adjusted transaction time window and perform nonlinear entropy integration to obtain a conflict heat value; when the conflict heat value reaches the adjusted conflict heat threshold, include all ticket checking interfaces corresponding to the current unprocessed events into the candidate scheduling set; S5: For the ticket checking interfaces in the candidate scheduling set, a scheduling loss function is constructed based on the operation urgency, the expected verification time, and the nonlinear cost function with the operation time interval between the ticket checking interfaces as the variable for each ticket checking interface; by minimizing the scheduling loss function, the activation sequence of the ticket checking interfaces and their operation time intervals are determined. S6: Activate the ticket checking interfaces sequentially according to the activation sequence to perform ticket checking verification; save a global ticket checking status snapshot when activating the current ticket checking interface, and temporarily store the real-time trigger events newly generated by the inactive ticket checking interfaces from the time the global ticket checking status snapshot is saved in the buffer queue; after the current ticket checking verification is completed, extract the real-time trigger events one by one from the buffer queue, and compare them with the global ticket checking status snapshot for document consistency and spatiotemporal continuity. If the comparison results are consistent, the real-time trigger event is discarded; if they are inconsistent, the real-time trigger event is transferred to the next scheduling cycle for processing.

[0018] Specifically, rail transit station turnstiles have multiple ticket checking interfaces, each corresponding to a ticket medium or biometric reader. A typical configuration includes a 13.56 MHz RFID reader, a QR code scanner, an NFC module, and a facial recognition camera, installed at different heights and angles within the turnstile housing.

[0019] Operation center coordinates of each ticket check interface A three-dimensional Cartesian coordinate system is established with the center of the turnstile channel as the origin, the passenger flow direction as the x-axis, the transverse direction of the turnstile housing as the y-axis, and the vertical direction as the z-axis. The unit is meters, where i is the identifier of the ticket checking interface. Taking a typical turnstile installation layout as an example, the RFID card reader... =(0.30,0.25,1.10), QR code scanner =(0.35,−0.15,1.25), NFC module =(0.30,0.20,1.15), the facial recognition camera's... =(0.40,0,1.55). The coordinate values ​​are derived from the gate hardware design drawings or installation calibration program, and are measured and configured according to the specific installation location during actual deployment.

[0020] Effective sensing radius of each ticket check interface The maximum spatial range within which the ticket checking interface can reliably read passenger tickets depends on the physical technology and operating parameters used by the interface, and is measured in meters. For example, an RFID reader... =0.08m, QR code scanner =0.12m, NFC module =0.05m, facial recognition camera =0.25m. The value is based on the stable reading distance measured by the interface under interference-free laboratory conditions, with a margin, and is stored in the gate configuration file.

[0021] For each ticket checking interface, allocate a baseline transaction time window. The unit is seconds. The baseline transaction time window represents the typical time required for this interface to complete a full ticket verification process—from credential reading and data verification to payment confirmation—under interference-free and queue-free conditions. This value is assigned based on the physical technology type and historical statistics corresponding to the interface. For example, RFID card readers... The time allocation is 0.30s, including approximately 0.05s for RFID card search, approximately 0.10s for data reading, approximately 0.12s for background verification, and approximately 0.03s for result return.

[0022] Simultaneously initialize a conflict heat baseline threshold. This is a basic decision parameter used to determine whether to initiate unified scheduling across multiple interfaces. Its initial value is 0.30, and it is dimensionless. When a single interface is in normal use, only one ticket check interface generates a trigger event. The integral value of its spatiotemporal heat distribution in the operating space, after normalization, usually does not exceed 0.25. Setting the baseline threshold to 0.30 can keep the threshold higher than the heat value when a single interface is in normal operation, avoiding triggering unnecessary scheduling decisions, while leaving a margin of about 0.05 to accommodate slight sensor noise.

[0023] Preset a weight coefficient for each ticket checking interface. The initial values ​​are all set to 1.0. During operation, the weight coefficients can be dynamically updated based on the cumulative usage frequency of each interface.

[0024] When a passenger brings their ticket (including a public transport card, mobile QR code, or NFC-enabled mobile device) close to the sensing area of ​​the ticket gate interface, or faces their face towards the facial recognition camera, the interface generates a real-time trigger event. Each real-time trigger event includes: the trigger time (recorded by the system clock with millisecond precision), the interface identifier, the operation point position p, and the credential digest. The credential digest is generated by the ticket gate interface after reading the passenger's ticket; for RFID readers and NFC modules, the credential digest is a fixed-length string obtained by performing a secure hash operation on the read card number, for example, truncating the first 8 bytes after performing a SHA-256 hash on a 16-byte card number; for QR code scanners, the credential digest is obtained by performing the same hash operation on the decoded user unique identifier string; for facial recognition cameras, the credential digest is obtained by performing a hash operation on the extracted facial feature vector. The location p of the operation point is determined in real time by the positioning module configured corresponding to the ticket checking interface. This positioning module uses sensing technology adapted to the interface type to obtain the three-dimensional spatial coordinates of the ticket medium or biometric features in the gate coordinate system. The gate coordinate system is set with the center of the channel ground as the origin. For example, a QR code scanner uses images captured by a camera to calculate the three-dimensional coordinates of the center of the handheld device screen in the gate coordinate system through a monocular vision positioning algorithm. Each real-time triggered event is recorded in the event status table after it occurs. In addition to storing the trigger time, interface identifier, operation point location, and voucher summary of the event, each record in the event status table also has a verification status flag, which is initially set to unverified. When an event is successfully checked and verified in S6, S6 updates the verification status flag of the event to verified.

[0025] Environmental condition data is collected at a fixed frequency by multiple environmental sensors installed on the turnstile. The environmental condition data includes at least: ambient light flicker variance Lv, in units of... The illuminance (lux squared) is calculated by taking the variance of 50 samples from an ambient light sensor installed on top of the turnstile facing the passenger operating area at a sampling frequency of 50Hz, with an update frequency of 10Hz; the spatial electromagnetic noise power Rn, in dBm (decibels per milliwatt), is calculated by taking the noise floor value from the received signal strength indicator register during idle periods when no card reading is performed by the RF card reader chip, and taking the arithmetic mean of the values ​​within the most recent second, with an update frequency of 10Hz; the passenger flow density Cd in front of the turnstile is calculated by taking the variance of 50 samples from the most recent 5 seconds, with an update frequency of 10Hz; (Per person per square meter) Depth images are captured by a depth camera installed on top of the turnstile facing the queuing area. A human target detection algorithm is used to count the number of human targets within a 1.2m wide and 0.8m deep area in front of the turnstile, and then the result is divided by the area of ​​that area, which is 0.96. The update frequency is 5Hz.

[0026] An environmental interference index is constructed using environmental condition data, and the conflict heat threshold and interface transaction time window are dynamically adjusted accordingly to adapt to on-site conditions. Unprocessed events are screened, and the urgency of operations is calculated based on the number of repeated triggers. The expected verification time is determined based on historical verification time records. A spatiotemporal impact kernel function is constructed, and the conflict heat value is obtained through weighted superposition and nonlinear entropy integration. When the conflict heat value reaches the adjusted conflict heat threshold, the corresponding interface is included in the candidate scheduling set. The interfaces in the candidate set are permuted to generate a candidate activation sequence. The optimal activation sequence and its operation time interval are determined synchronously by minimizing the scheduling loss function that includes the urgency-weighted delay and the cost of the operation time interval between interfaces. Interfaces are activated in sequence to perform ticket verification. By saving a global state snapshot and a temporary queue mechanism, the newly generated events of inactive interfaces are compared in terms of credentials and spatiotemporal continuity to achieve unified scheduling control of multiple interfaces.

[0027] In one embodiment of the present invention, step S2 includes the following steps: The environmental operating data includes ambient light flicker variance Lv, spatial electromagnetic noise power Rn, and passenger flow density in front of the gate Cd; Constructing the environmental disturbance index ,in, The preset electromagnetic noise reference zero point, , and These are the preset standardization coefficients; For the aforementioned conflict heat benchmark threshold Adjustments were made to obtain the adjusted conflict heat threshold. ; The baseline transaction time window for each of the ticket checking interfaces Adjustments are made to obtain the adjusted trading time window. ,in This is the preset time window adjustment coefficient.

[0028] Specifically, in constructing the environmental disturbance index In the calculation formula, the preset electromagnetic noise reference zero point The value is set to −90dBm, which corresponds to the noise floor level of the RFID reader measured in an electromagnetically shielded room free from external interference; since the spatial electromagnetic noise power Rn is negative under typical operating conditions, it is introduced... Electromagnetic noise values ​​can be mapped to non-negative differences while maintaining the relative proportionality between different interference levels. Light intensity scintillation normalization coefficient. The value is 0.01 (Reciprocal of the square of each lux); the ambient light flicker variance Lv under typical ground station daylight flicker scenarios is approximately 50. ,at this time =0.5, which is on the same order of magnitude as the constant 1 in the logarithmic term. This ensures that the weight is neither too large to suppress the contribution of other interference terms, nor too small to fail to reflect the effect of light flicker. Electromagnetic noise normalization factor The value is 0.1 (Reciprocal of milliwatts per decibel); under typical electromagnetic interference scenarios It is approximately 20 to 25 at this time. Approximately 2.0 to 2.5, on the same order of magnitude as the constant 1 in the logarithmic term, it effectively reflects the impact of electromagnetic interference on radio frequency communication quality. Passenger flow density normalization factor. The value is 1 (Reciprocal of the number of people per square meter); Under typical peak passenger flow scenarios, the passenger flow density Cd in front of the turnstile is approximately 3 to 5. ,at this time Approximately 4 to 6, this matches the magnitude of the passenger flow density term with that of the logarithmic term, reflecting the aggravating effect of multiple interface mis-touches in crowded scenarios. For example, during a peak period, the illuminance variance Lv collected by the ambient light sensor was 40. The electromagnetic noise floor of the RFID reader chip is Rn = −70dBm, and the passenger flow density measured by the depth camera is Cd = 3. Calculations yielded =4.896.

[0029] use Conflict heat benchmark threshold Adjustments are made to obtain the adjusted conflict heat threshold for practical application under the current operating conditions. The adjustment formula is , where tanh is the hyperbolic tangent function, whose graph is S-shaped. The function value approaches 1 as the independent variable approaches positive infinity, approaches -1 as the independent variable approaches negative infinity, and is 0 when the independent variable is zero. The output range is strictly limited to the interval (-1, 1). For example, when When =1, tanh(1−1)=0. = ×1=0.30, the threshold remains unchanged at the baseline value. At this time, the environmental interference is at the preset normal level, and there is no need to adjust the scheduling sensitivity; when When = 4.896 (peak scenario), tanh(4.896−1)≈0.999. With a threshold value of approximately 0.600, the threshold is steadily raised to about twice the baseline value, allowing the scheduler to intervene in unified scheduling earlier and preventing sensor-triggered events caused by environmental interference from being incorrectly identified as real multi-interface operation conflicts.

[0030] Using the same The baseline transaction time window allocated to each ticket checking interface. Adjustments are made to obtain the adjusted transaction time window for the ticket checking interface under the current operating conditions. The adjustment formula is Among them, the preset time window adjustment coefficient The value is 0.1, and the dimension is 1 (dimensionless); in extreme peak scenarios... Approximately 5, at this time =1.5, the adjusted trading time window is a maximum of 1.5 times the benchmark value; if If the value is too large, the window will over-expand in extreme scenarios, leading to excessively long passenger waiting times and a sharp decline in traffic efficiency; if... If the value is too small, the increase in window size will be insufficient to compensate for the extra time consumed by environmental interference, rendering the adjustment function ineffective. Taking an RFID card reader as an example, in a quiet environment… =0.876, ≈0.326s, the window size increased by approximately 0.026s compared to the baseline value; in peak scenarios =4.896, The window size is approximately 0.147s larger than the baseline value, representing a time increase of ≈0.447s. In environments with strong, flickering light, QR code scanners require longer image exposure times and multiple decoding attempts. Face recognition cameras, due to unstable lighting, need to capture multiple frames for comparison. Appropriately expanding the transaction time window can accommodate these additional time consumptions and prevent normal ticket verification from being mistakenly judged as failed due to timeouts. In environments with strong electromagnetic interference, RFID readers and NFC modules require more retries and longer response times. Expanding the transaction time window can improve communication success rates. In high-passenger-density environments, passengers' arms may sway due to crowding, causing the credential to remain in the sensing area for a shorter time or become unstable. Appropriately widening the transaction time window for effective judgment can prevent passengers from further exacerbating congestion due to repeated failed operations.

[0031] In one embodiment of the present invention, step S3, which calculates the operational urgency of the ticket checking interface corresponding to each unprocessed event based on the number of times each unprocessed event is repeatedly triggered within a preset time period, includes the following steps: Starting from the trigger time of the unprocessed event, a preset time period is traced backward. The number of real-time triggered events generated by the same ticket checking interface within that preset time period, and whose voucher digest matches the voucher digest of the unprocessed event, is counted as the repeated trigger count of the unprocessed event. The voucher summary is generated by the ticket checking interface after reading the passenger's ticket card medium and is included in the real-time trigger event. Calculate the operational urgency of the ticket checking interface corresponding to the unprocessed event. , where is the preset urgency growth coefficient.

[0032] Specifically, the event status table is traversed, and the verification status flag of each record (i.e., each real-time triggered event) is checked one by one. All records with an unverified status are extracted. The events represented by these records are those that have not yet completed ticket verification and are marked as unprocessed events. All unprocessed events are summarized into an unprocessed event list for the current scheduling cycle. Each element in the list contains the trigger time of the event. Interface identifier, operation point location p, and credential summary.

[0033] For each unprocessed event in the unprocessed event list, count the number of times it is triggered repeatedly. , where j is the index of the unprocessed event in the list. The unprocessed event... As the endpoint, a preset time period is traced back, with a length of 3.0 seconds. The typical retry interval for rail transit passengers after an initial ticket check failure is approximately 0.5 to 1.0 seconds. A 3.0-second time period is sufficient to cover 2 to 4 consecutive retries without introducing irrelevant historical triggers due to an excessively long window. Within the traceback period, the number of real-time triggered events that simultaneously meet two conditions is counted: Condition 1, the event is generated by the same ticket check interface as the unprocessed event, i.e., the interface identifier is the same; Condition 2, the credential digest carried by the event is completely identical to the credential digest of the unprocessed event. The count is subtracted by 1 (excluding the unprocessed event itself) to represent the number of times the unprocessed event was repeatedly triggered. If no other events meeting the conditions are found during the backtracking period, =0 indicates that this event is triggered for the first time and there are no repeated operations. For example, a certain RFID reader interface in An unprocessed event was generated at 12:00:00:00:00, with credential digest A1B2C3D4. Looking back 3.0 seconds, it was found that the same interface had generated two other events with credential digest A1B2C3D4 within that time period, occurring at 11:59:58:200 and 11:59:59:500 respectively. Including the current event, there were a total of three such events. What is the number of times this unprocessed event has been repeatedly triggered? =3−1=2.

[0034] Calculate the urgency of the ticket checking interface corresponding to this unprocessed event, denoted as . Among them, the urgency growth coefficient The value is 0.2, and the dimension is 1 (dimensionless); when =5、 =0.2 =2.0, which means the urgency level is twice the baseline value of 1.0. This gives sufficient priority to continuously retried events while avoiding excessive inflation of urgency due to excessive weight of a single behavior.

[0035] In one embodiment of the present invention, step S3, determining the expected verification time based on the historical verification time record of the ticket checking interface, includes the following steps: Maintain a historical verification time queue for each ticket checking interface. The queue stores the actual time taken for the N most recent successful ticket checking verifications of the ticket checking interface. When the historical verification time queue is not empty, the arithmetic mean of all record values ​​in the queue is taken as the expected verification time of the ticket checking interface. When the historical verification time queue is empty, the baseline transaction time window allocated to the ticket checking interface is used as the expected verification time of the ticket checking interface.

[0036] Specifically, a historical verification time queue is maintained for each ticket checking interface. Each queue is a fixed-length queue with a length of N, where N is 10. Ten records can reflect the recent trend of verification time changes while avoiding drastic fluctuations in expected values ​​due to single outliers or excessively old historical data, and the storage overhead is negligible.

[0037] Each record in the queue stores the actual time taken for a successful ticket check at that interface, in seconds. Each time a ticket check interface successfully completes a check, the time difference between activation and completion is appended to the tail of the corresponding historical check time queue as the actual time taken. If the queue already contains N records before enqueueing, the oldest record at the head of the queue is removed first, and then the new record is added to maintain a constant queue length of N. If the interface has not yet successfully completed a ticket check, its historical check time queue is empty.

[0038] For each unprocessed event in the unprocessed event list, determine the expected verification time from the historical verification time queue of its corresponding ticket checking interface. , where j is the sequence number of the unprocessed event in the list. When the historical verification time queue of this interface is not empty, let the number of records currently stored in the queue be m (1≤m≤N), and the actual time consumption value of each record is as follows: The expected verification time is... For example, if a certain RFID reader interface has m=5 records in its historical verification time queue, with values ​​of 0.28s, 0.31s, 0.29s, 0.30s, and 0.32s respectively, then... =0.30s.

[0039] When the historical verification time queue of the interface is empty, that is, when the interface has no successful verification records, the baseline transaction time window initially allocated to it is used. As the expected verification time, i.e. .

[0040] In one embodiment of the present invention, step S4, which involves constructing a spatiotemporal impact kernel function for the unprocessed event based on the operation center coordinates and effective sensing radius of each ticket checking interface, and in conjunction with the adjusted transaction time window of that ticket checking interface, includes the following steps: For each of the unprocessed events, extract its trigger time. and operation point location And the operation center coordinates of the ticket checking interface corresponding to the unprocessed event. Effective sensing radius And the adjusted trading time window , where i is the identifier of the ticket checking interface, and j is the sequence number of the unprocessed event; Construct the spatiotemporal impact kernel function for this unprocessed event. Where p is any point in the operating space in front of the gate, and t is any moment within a sliding time window that ends at the current moment and traces back a preset integration time. From operation point p to operation center The Euclidean distance.

[0041] Specifically, a sliding time window T is defined, which extends backward from the current moment by a preset integration time, with a value of 2.0 seconds. t represents any moment within this sliding time window. The sliding time window is set based on the fact that the maximum adjusted transaction time window for a single ticket checking interface is approximately 1.2 seconds, and a window length of 2.0 seconds is sufficient to cover the time span from triggering to completion of any unprocessed event, without introducing too many irrelevant historical moments.

[0042] For each unprocessed event in the unprocessed event list, with index j, extract its trigger time. and operation point location And based on the ticket check interface identifier i corresponding to the event, obtain the coordinates of the operation center. Effective sensing radius And the adjusted trading time window Construct the spatiotemporal impact kernel function for this unprocessed event. , where max(0, The parentheses indicate that the larger value between 0 and the result within the parentheses is taken, ensuring the kernel function value is non-negative. The spatiotemporal influence of the kernel function's construction ensures that, in the spatial dimension, when the operation point p is exactly located at the interface operation center... Time, spatial item The value is 1, as p moves away from... As the distance squared term increases, the spatial term decreases linearly. When the distance reaches... The spatiotemporal term decays to 0, and once it exceeds its range, the spatial term no longer contributes; in the time dimension, when t is exactly equal to the trigger time... At that time, the time item The value is 1, and it deviates from t. The time term decays linearly, and when the time difference reaches... The time term decays to 0, and the kernel function value becomes 0 after the window expires. The decay terms in both dimensions are added together, meaning that the influence of the event decays simultaneously in both spacetime and time directions, only decreasing when p approaches zero. And t is close to When the kernel function value is close to 1, any deviation in either direction will cause the kernel function value to decrease until it drops to 0.

[0043] Taking an unprocessed event from an RFID reader interface as an example, the interface's =(0.30,0.25,1.10), =0.08m, =0.30s; the trigger time of this event. =12:00:00:00.000, Operation point location =(0.32,0.27,1.12). Take a point p=(0.32,0.27,1.12) within the operating space in front of the turnstile, and take the time t=12 hours 00 minutes 00.100 seconds; calculate... ≈0.0346m, the square of the distance is 0.0012, divided by =0.0064, which gives approximately 0.1875; time difference =0.100s, divided by =0.30s, yielding approximately 0.3333; kernel function value =0.4792. If the same location is taken, but t=12:00:00:300, the time term is 0.3 / 0.3=1.0, and the kernel function value is 0, indicating that the time window has expired and the event's influence has disappeared. If the trigger time is taken, but p=(0.50,0.25,1.10), the distance... =0.20m, the square of the distance is 0.04, divided by 0.0064, we get 6.25, the value in parentheses (i.e. If the value is negative, the kernel function value is 0, indicating that the location is not within the effective sensing radius.

[0044] In one embodiment of the present invention, step S4, which involves weighting and superimposing all spatiotemporal influence kernel functions within the adjusted transaction time window in the operating space and performing a nonlinear entropy integral operation on them to obtain a conflict heat value, includes the following steps: The spatiotemporal impact kernel function of all unprocessed events within the operating space in front of the gate is calculated according to the weight coefficient of its respective ticket checking interface. By performing weighted superposition, the spatiotemporal heat distribution is obtained. ; Calculated conflict heat value ,in, This refers to the operating space in front of the gate. Here, T is the preset normalized time constant, T is the sliding time window ending at the current time, and V is... The volume.

[0045] Specifically, based on the coordinates of the operation centers at each ticket checkpoint. and effective sensing radius Define the operating space in front of the turnstile. . Let it be the smallest circumscribed cuboid that can encompass the effective sensing range of all interfaces, and the boundaries of each dimension are defined by all... The coordinate extreme values ​​expand to both sides The result is as follows. In a typical installation layout, the cuboid is approximately 1.2m long, 0.8m wide, and 1.5m high, with a volume V of approximately 1.44. .

[0046] operating space The spatiotemporal impact kernel functions of all unprocessed events are weighted and superimposed according to the weight coefficients of their respective ticket checking interfaces to obtain the spatiotemporal heat distribution. , where p is Let t be any point within the sliding time window T, and t be any time within the sliding time window T. This represents the set of all unprocessed event sequence numbers belonging to the i-th interface. And on the sliding time window T, for By performing nonlinear entropy integration, the conflict heat value H is obtained through a formula.

[0047] In practical calculations, continuous integrals are transformed into discrete summations. With side length A cubic mesh of 0.05m was generated, resulting in a total of Spatial grid points ( T is measured in steps. Dividing the data at 0.02s, we obtained a total of [number missing] results. Time sampling points ( Volume element of each spatial grid. For each unprocessed event, the kernel function value is computed only within the local spatiotemporal region where its kernel function is non-zero. That is, for event j belonging to interface i, the kernel function value is computed only for events satisfying the condition... and The kernel function is calculated using grid points and time sampling points, while the kernel function values ​​for other sampling points are set to zero. At each valid sampling point, the weighted kernel function values ​​of all unprocessed events are summed to obtain the result. Then calculate Finally, the conflict heat value is obtained. The normalized time constant The value is set to 1 second, making H a dimensionless pure number. The above discrete summation operation is a conventional processing method in numerical integration, which can be completed within the specified response time under the current mainstream computing power of gate controllers.

[0048] For example, there are two events in the current unprocessed event list. Event 1 belongs to the RFID reader interface (i=1). =1.0, Spatiotemporal influence kernel function At a certain spatiotemporal sampling point The value is 0.6; Event 2 belongs to the QR code scanner interface (i=2). =1.0, Spatiotemporal influence kernel function If the value at the same sampling point is 0.5, then the spatiotemporal heat distribution at that sampling point is... =1.1, the integrand value is ≈0.816. If the two events are completely separated in space and time and do not overlap, then they contribute independently. In this case, the integrand values ​​in their respective non-zero regions are 0.6×ln(1.6)=0.282 and 0.5×ln(1.5)=0.202, respectively, with a sum of 0.484. It can be seen that the integral contribution of 0.816 when they coincide is significantly greater than that of 0.484 when they separate, demonstrating the sensitive ability of nonlinear entropy integrals to capture conflict and overlap.

[0049] The calculated conflict intensity value H is compared with the adjusted conflict intensity threshold. Comparison. When When a significant multi-interface operation conflict is identified in front of the current gate, a unified scheduling decision is initiated. At this time, all ticket checking interfaces corresponding to unprocessed events in the unprocessed event list are included in the candidate scheduling set D. The interfaces in this set are the objects for sequence optimization by S5. If the current conflict level is not serious, unified scheduling will not be triggered. If there are no unprocessed events at this time, the process will continue to wait. If there is only one unprocessed event, the interface corresponding to that event will be allowed to pass the ticket check directly without performing candidate scheduling set construction and sequence optimization, and will directly jump to S6 to perform the check. If there are two or more unprocessed events, the interfaces will be activated sequentially according to the order of their triggering time, with a preset safety time interval (0.05s). The scheduling cycle ends after all interfaces have been checked.

[0050] In one embodiment of the present invention, step S5 includes the following steps: Perform a full permutation of all ticket checking interfaces in the candidate scheduling set, and treat each permutation result as a candidate activation sequence. , where n is the number of ticket checking interfaces in the candidate scheduling set, and the total number of candidate activation sequences is n!; For any of the candidate activation sequences, the actual completion time of the verification of the k-th ticket checking interface in the sequence under sequential scheduling is denoted as . The ideal verification completion time of the ticket checking interface under the condition of immediate activation is recorded as . The actual verification completion time is determined recursively based on the position of the ticket checking interface in the candidate activation sequence, its expected verification time, and the operation time interval between it and the previous ticket checking interface. Construct a nonlinear cost function with the time interval gap between the ticket checking interfaces as the variable. ,in The sensitivity coefficient is the preset time interval. Constructing the scheduling loss function ,in, The urgency of the operation at the k-th ticket checking interface. This is the time interval between the current ticket checking interface and the previous ticket checking interface. The preset interval cost weighting coefficient; For all candidate activation sequences and the combination of the operation time interval values, calculate the corresponding scheduling loss function value, determine the candidate activation sequence with the smallest scheduling loss function value as the activation sequence of the ticket checking interface, and determine the operation time interval between each adjacent ticket checking interface corresponding to the activation sequence.

[0051] Specifically, all ticket checking interfaces in the candidate scheduling set D are permuted, and each permutation result is taken as a candidate activation sequence, denoted as . Let n be the number of ticket checking interfaces in D, and k represent the k-th activated position in the sequence (k=1,2,…,n). The total number of candidate activation sequences generated by all permutations is n!. In a typical turnstile scenario, the number of interfaces n in D is usually no more than 4, so the maximum number of permutations is 24.

[0052] For any candidate activation sequence S, the k-th ticket check interface in the sequence The actual verification completion time under sequential scheduling is denoted as . The ideal verification completion time under the condition of immediate activation is denoted as . The ideal verification completion time is defined as the time from the current scheduling start time if there is no interference from any other interface. The corresponding completion time is activated immediately, therefore ,in This represents the expected verification time for the interface. The actual verification completion time is determined recursively based on the interface's position in the sequence. For the first interface in the sequence (k=1), its activation time is the scheduling start time. For the k-th interface in the sequence (k≥2), its activation time depends on the completion time of the previous interface and the time interval between the two operations. ,in Let be the operation time interval (k≥2) between the k-th interface and its preceding interface. This interval is one of the variables to be solved, and its unit is seconds.

[0053] For each pair of adjacent interfaces, the allowable range of the operation time interval gap is determined based on the physical type of the adjacent interfaces. If both adjacent interfaces are radio frequency (RF) interfaces in the 13.56MHz band (including RF card readers and NFC modules), to avoid residual RF field interference from the preceding interface on the card search process of subsequent interfaces, the lower limit of the gap is 0.15s, and the upper limit is 0.5s. If at least one of the two adjacent interfaces is a non-RF interface (including QR code scanners and face recognition cameras), then the lower limit of the gap is 0.05s, and the upper limit is 0.5s. A nonlinear cost function with the operation time interval gap as the variable is constructed. The time interval sensitivity coefficient Values (Reciprocal per second); the function value approaches 0 when gap approaches 0, but the derivative is γ, indicating that in the region of minimal gap, the cost increases sharply as the gap decreases; as gap increases, the function value approaches 1 at an exponential rate, and the growth rate slows down rapidly, making the marginal benefit of continuing to increase the gap diminish.

[0054] For a candidate activation sequence S, construct a scheduling loss function L(S). In the function formula, Let $k$ be the operation time interval between the $k$-th interface and its preceding interface. For the first interface (k=1), there is no preceding interval, so we define $k$ as $k$. Then the corresponding cost term 1. This item does not affect sequence comparisons; it serves only as a formal standardization. Preset interval cost weighting coefficient. It takes a value of 0.5 and has a dimensionless value of 1. It is used to balance the relative importance between the squared deviation term and the interval cost term.

[0055] The first term of L(S) The measurement considers the verification delay loss caused by queuing. The squared deviation ensures that the longer the delay, the faster the loss increases. Multiplying by the operation urgency assigns higher weight to the delay of high-urgency interfaces. (Second term) The physical safety risks caused by excessively small operation intervals were measured. By imposing a nonlinear cost on overly compact intervals through exponential saturation, the optimization process was forced to automatically avoid sequence arrangements that could cause electromagnetic interference or accidental triggering.

[0056] Under the condition that n≤4, L(S) is minimized by combining full permutation enumeration with interval grid search. For each candidate activation sequence S generated by the full permutation, the operation time interval between all adjacent interfaces is used. (k=2,…,n) are variables. A grid search is performed within the allowed range [0.05s, 0.5s] with a step size of 0.01s. Consider a set of interval values. Calculate the corresponding scheduling loss function value for each value combination. Select the sequence that minimizes L(S) among all sequences and all interval combinations as the final determined activation sequence. And determine the operation time interval between each adjacent interface corresponding to the sequence. In the most complex case (n=4), 24× ≈2.19× Each evaluation involves only a few multiplication-addition and exponentiation lookup operations, and takes no more than 20ms under the typical computing power of the gate equipment.

[0057] For example, the candidate scheduling set D contains two ticket checking interfaces, and the operational urgency of the RFID card reader interface is... =1.4, expected verification time =0.30s; urgency of operation of the QR code scanner interface =1.0, expected verification time =0.50s; There are two candidate activation sequences, sequence and sequence The search was performed within a gap range of 0.05s to 0.5s. Calculations showed that... The minimum loss value occurs around gap = 0.15s. The minimum loss value occurs around gap = 0.20s, and Less than Therefore, the activation sequence was ultimately determined to be... That is, the RFID reader is activated first, followed by the QR code scanner, with the interval between the two determined to be 0.15s.

[0058] In one embodiment of the present invention, step S6 includes the following steps: When the current ticket checking interface is activated and ticket checking verification begins, a global ticket checking status snapshot is saved. The global ticket checking status snapshot includes the snapshot saving time, the operation point position of the current ticket checking interface and the read voucher summary, as well as the ticket checking verification status of each ticket checking interface. Create a buffer queue and store all real-time triggered events generated by other ticket checking interfaces that are not included in the activation sequence from the time the snapshot is saved to the time when the ticket checking verification of the current ticket checking interface is completed into the queue in sequence. After the ticket checking interface completes the ticket checking verification, each real-time triggered event is extracted one by one from the buffer queue. First, it is determined whether the ticket checking interface type corresponding to the event is the same as the currently activated ticket checking interface type. If the interface types are different, the event is directly marked to be transferred to the next scheduling cycle for processing, and no further credential consistency comparison is performed. If the interface types are the same, a credential consistency comparison is performed to determine whether the credential digest carried by the event is consistent with the credential digest of the current ticket checking interface in the global ticket checking status snapshot. For real-time trigger events that match the voucher summary, a spatiotemporal continuity comparison is performed to determine whether the time difference between the trigger time of the event and the snapshot saving time is less than the adjusted transaction time window of the current ticket checking interface, and whether the distance between the operation point position of the event and the operation point position of the current ticket checking interface in the global ticket checking status snapshot is less than the effective sensing radius of the current ticket checking interface. Real-time triggered events that have consistent credential summaries and meet the conditions for spatiotemporal continuity comparison are discarded, while real-time triggered events that have inconsistent credential summaries or do not meet the conditions for spatiotemporal continuity comparison are marked for processing in the next scheduling cycle.

[0059] Specifically, for activation sequences The k-th ticket check interface At the same time, determine Operation time interval between adjacent interfaces If k=1, the interface is immediately activated to begin ticket verification; if k≥2, the interface waits for the previous interface. From the time the ticket check is completed, then... Activate in seconds Ticket checking and verification begins.

[0060] Activation And confirm the moment when the ticket checking and verification process has begun. Save a global ticket check status snapshot, which includes the following information: the snapshot save time. Current ticket checking interface Read voucher summary, current ticket check interface The operation point location and ticket verification status of each ticket checking interface are specified. The operation point location of the current ticket checking interface is taken from the unprocessed event record corresponding to that interface in the unprocessed event list; the ticket verification status of each ticket checking interface is the value of the verification status flag bit of the current event record corresponding to each ticket checking interface in the event status table. The status of the active interface is "Verification in progress," while the status of the other inactive interfaces is "Idle" or "Verified."

[0061] While saving a snapshot of the global ticket check status, an empty temporary event buffer queue is created. From, to The ticket check verification is completed by the time the check is completed. During this period, other ticket check interfaces not included in the activation sequence (i.e., except for...) For all new real-time triggered events (excluding those from ticket checking interfaces), the system does not respond immediately. Instead, it sequentially stores the complete data of each event (trigger time, interface identifier, operation point location, and voucher summary) into the temporary event buffer queue. If no new events are generated by inactive interfaces during this period, the queue will be empty.

[0062] treat After the ticket verification process is completed, each real-time triggered event is retrieved from the temporary event buffer queue in the order it was enqueued and compared. First, the ticket verification interface type corresponding to the event is determined to be different from the currently active interface. If the interface types are different, the event is marked for processing in the next scheduling cycle without further credential consistency comparison. If the interface types are the same, credential consistency comparison is performed to determine if the credential digest carried by the event matches the one in the global ticket check status snapshot. The event is checked against the credential digests. If they do not match, the event is marked as being transferred to the next scheduling cycle. For events with matching credential digests, a spatiotemporal continuity comparison is performed, which includes two parallel conditions: Condition 1, temporal continuity, determining the trigger time of the event. and Does the time difference satisfy? ,in for Condition 1: Adjusted transaction time window; Condition 2: Spatial continuity, determining the operation point location p of the event and its relationship to the snapshot. Does the distance between the operation point locations meet the requirements? ,in For interface The effective sensing radius. If both conditions are met simultaneously, the event is determined to be the same passenger in [the location / location]. Accidental triggers during ticket verification are discarded without any response. If at least one condition is not met, the event is considered an independent ticket verification request and marked for processing in the next scheduling cycle. When making this mark, if the ticket verification interface corresponding to the event is a subsequent interface in the current active sequence that has not yet been executed, then the subsequent interface is deleted from the current active sequence, and subsequent interfaces are moved forward sequentially. If the interface corresponding to the event is not a subsequent interface in the current active sequence, it is only marked, and processing is carried out in the next cycle after all interfaces in the current active sequence have been executed. For events marked for processing in the next scheduling cycle, the verification status flag in the corresponding event status table remains unchanged as unverified. After all interfaces in the current active sequence have been executed, these events are captured and processed again starting from S3 in the next scheduling cycle.

[0063] Complete the After ticket verification and queue processing, the sequence position pointer k is incremented by 1. If k ≤ n, the next interface is activated according to the above interval rule and the above process is repeated; if k > n, the activation sequence is completed.

[0064] Current interface After the ticket checking verification and queue comparison processing are completed, it is determined whether there is a next ticket checking interface to be activated in the activation sequence. If k is less than n, according to the operation time interval determined in S5 , after waiting for the interval duration on the basis of the completion time of the previous interface verification, the k+1-th ticket checking interface is activated , and the snapshot saving, queue filling and comparison processing procedures are repeatedly executed. If k equals n, all ticket checking interfaces in the activation sequence have been executed, and the sequence execution of this scheduling cycle ends.

[0065] After the execution of the activation sequence is completed, the actual time consumption for each interface to successfully complete the ticket checking verification in the current cycle is appended to the historical verification time consumption queue of the corresponding interface, for use in determining the expected verification time consumption in S3 in subsequent scheduling cycles. For the real-time trigger event marked as transferred to the next scheduling cycle for processing, its verification status remains unverified, and it is recaptured by S3 in the next scheduling cycle.

[0066] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that modifications or equivalent substitutions can be made to the technical method of the present invention without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A multi-interface unified scheduling and control method for self-service ticket checking in rail transit, characterized in that, Includes the following steps: S1: Obtain the coordinates of the operation center and the effective sensing radius of each of the multiple ticket checking interfaces coexisting on the gate channel, allocate a benchmark transaction time window for each ticket checking interface, and initialize the benchmark threshold for conflict heat; collect real-time trigger events of multiple ticket checking interfaces and environmental condition data of the gate site; S2: Construct an environmental disturbance index using the environmental condition data; Based on the environmental interference index, the conflict heat baseline threshold is adjusted to obtain an adjusted conflict heat threshold, and the baseline transaction time window for each ticket checking interface is adjusted to obtain an adjusted transaction time window. S3: Filter the real-time triggered events that have not yet been verified and mark them as unprocessed events; Based on the number of times each unprocessed event is repeatedly triggered within a preset time period, the operational urgency of the ticket checking interface corresponding to the unprocessed event is calculated, and based on the historical verification time record of the ticket checking interface, its expected verification time is determined. S4: Based on the operation center coordinates and effective sensing radius of each ticket checking interface, define the operation space in front of the gate, and construct a spatiotemporal influence kernel function for the unprocessed events in conjunction with the adjusted transaction time window of the ticket checking interface; perform weighted superposition of all spatiotemporal influence kernel functions within the operation space that are within the adjusted transaction time window and perform nonlinear entropy integration to obtain a conflict heat value; when the conflict heat value reaches the adjusted conflict heat threshold, include all ticket checking interfaces corresponding to the current unprocessed events into the candidate scheduling set; S5: For the ticket checking interfaces in the candidate scheduling set, a scheduling loss function is constructed based on the operation urgency, the expected verification time, and the nonlinear cost function with the operation time interval between the ticket checking interfaces as the variable for each ticket checking interface; by minimizing the scheduling loss function, the activation sequence of the ticket checking interfaces and their operation time intervals are determined. S6: Activate the ticket checking interfaces sequentially according to the activation sequence to perform ticket checking verification; save a global ticket checking status snapshot when activating the current ticket checking interface, and temporarily store the real-time trigger events newly generated by the inactive ticket checking interfaces from the time the global ticket checking status snapshot is saved in the buffer queue; after the current ticket checking verification is completed, extract the real-time trigger events one by one from the buffer queue, and compare them with the global ticket checking status snapshot for document consistency and spatiotemporal continuity. If the comparison results are consistent, the real-time trigger event is discarded; if they are inconsistent, the real-time trigger event is transferred to the next scheduling cycle for processing.

2. The multi-interface unified scheduling and control method for self-service ticket checking in rail transit according to claim 1, characterized in that, S2 includes the following steps: The environmental operating data includes ambient light flicker variance Lv, spatial electromagnetic noise power Rn, and passenger flow density in front of the gate Cd; Constructing the environmental disturbance index ,in, The preset electromagnetic noise reference zero point, , and These are the preset standardization coefficients; For the aforementioned conflict heat benchmark threshold Adjustments were made to obtain the adjusted conflict heat threshold. ; The baseline transaction time window for each of the ticket checking interfaces Adjustments are made to obtain the adjusted trading time window. ,in This is the preset time window adjustment coefficient.

3. The multi-interface unified scheduling and control method for self-service ticket checking in rail transit according to claim 1, characterized in that, In step S3, calculating the operational urgency of the ticket checking interface corresponding to each unprocessed event based on the number of times each unprocessed event is repeatedly triggered within a preset time period includes the following steps: Starting from the trigger time of the unprocessed event, a preset time period is traced backward. The number of real-time triggered events generated by the same ticket checking interface within that preset time period, and whose voucher digest matches the voucher digest of the unprocessed event, is counted as the repeated trigger count of the unprocessed event. The voucher summary is generated by the ticket checking interface after reading the passenger's ticket card medium and is included in the real-time trigger event. Calculate the operational urgency of the ticket checking interface corresponding to the unprocessed event. ,in This is the preset urgency growth coefficient.

4. The multi-interface unified scheduling and control method for self-service ticket checking in rail transit according to claim 1, characterized in that, In step S3, determining the expected verification time based on the historical verification time record of the ticket checking interface includes the following steps: Maintain a historical verification time queue for each ticket checking interface. The queue stores the actual time taken for the N most recent successful ticket checking verifications of the ticket checking interface. When the historical verification time queue is not empty, the arithmetic mean of all record values ​​in the queue is taken as the expected verification time of the ticket checking interface. When the historical verification time queue is empty, the baseline transaction time window allocated to the ticket checking interface is used as the expected verification time of the ticket checking interface.

5. A multi-interface unified scheduling and control method for self-service ticket checking in rail transit according to claim 2, characterized in that, In step S4, the step of constructing a spatiotemporal impact kernel function for the unprocessed event based on the operation center coordinates and effective sensing radius of each ticket checking interface, and in conjunction with the adjusted transaction time window of that ticket checking interface, includes the following steps: For each of the unprocessed events, extract its trigger time. and operation point location And the operation center coordinates of the ticket checking interface corresponding to the unprocessed event. Effective sensing radius And the adjusted trading time window , where i is the identifier of the ticket checking interface, and j is the sequence number of the unprocessed event; Construct the spatiotemporal impact kernel function for this unprocessed event. Where p is any point in the operating space in front of the gate, and t is any moment within a sliding time window that ends at the current moment and traces back a preset integration time. From operation point p to operation center The Euclidean distance.

6. A multi-interface unified scheduling and control method for self-service ticket checking in rail transit according to claim 5, characterized in that, In step S4, the step of weighting and superimposing all spatiotemporal influence kernel functions within the adjusted transaction time window in the operation space and performing a nonlinear entropy integral operation on them to obtain the conflict heat value includes the following steps: The spatiotemporal impact kernel function of all unprocessed events within the operating space in front of the gate is calculated according to the weight coefficient of its respective ticket checking interface. By performing weighted superposition, the spatiotemporal heat distribution is obtained. ; Calculated conflict heat value ,in, This refers to the operating space in front of the gate. Here, T is the preset normalized time constant, T is the sliding time window ending at the current time, and V is... The volume.

7. A multi-interface unified scheduling and control method for self-service ticket checking in rail transit according to claim 1, characterized in that, S5 includes the following steps: Perform a full permutation of all ticket checking interfaces in the candidate scheduling set, and treat each permutation result as a candidate activation sequence. , where n is the number of ticket checking interfaces in the candidate scheduling set, and the total number of candidate activation sequences is n!; For any of the candidate activation sequences, the k-th ticket checking interface in the sequence will be... The actual verification completion time under sequential scheduling is denoted as . The ideal verification completion time of the ticket checking interface under the condition of immediate activation is recorded as . The actual verification completion time is determined recursively based on the position of the ticket checking interface in the candidate activation sequence, its expected verification time, and the operation time interval between it and the previous ticket checking interface. Construct a nonlinear cost function with the time interval gap between the ticket checking interfaces as the variable. ,in The sensitivity coefficient is the preset time interval. Constructing the scheduling loss function ,in, The urgency of the operation at the k-th ticket checking interface. This is the time interval between the current ticket checking interface and the previous ticket checking interface. The preset interval cost weighting coefficient; For all candidate activation sequences and the combination of the operation time interval values, calculate the corresponding scheduling loss function value, determine the candidate activation sequence with the smallest scheduling loss function value as the activation sequence of the ticket checking interface, and determine the operation time interval between each adjacent ticket checking interface corresponding to the activation sequence.

8. A multi-interface unified scheduling and control method for self-service ticket checking in rail transit according to claim 1, characterized in that, S6 includes the following steps: When the current ticket checking interface is activated and ticket checking verification begins, a global ticket checking status snapshot is saved. The global ticket checking status snapshot includes the snapshot saving time, the operation point position of the current ticket checking interface and the read voucher summary, as well as the ticket checking verification status of each ticket checking interface. Create a buffer queue and store all real-time triggered events generated by other ticket checking interfaces that are not included in the activation sequence from the time the snapshot is saved to the time when the ticket checking verification of the current ticket checking interface is completed into the queue in sequence. After the ticket checking interface completes the ticket checking verification, each real-time triggered event is extracted one by one from the buffer queue. First, it is determined whether the ticket checking interface type corresponding to the event is the same as the currently activated ticket checking interface type. If the interface types are different, the event is directly marked to be transferred to the next scheduling cycle for processing, and no further credential consistency comparison is performed. If the interface types are the same, a credential consistency comparison is performed to determine whether the credential digest carried by the event is consistent with the credential digest of the current ticket checking interface in the global ticket checking status snapshot. For real-time trigger events that match the voucher summary, a spatiotemporal continuity comparison is performed to determine whether the time difference between the trigger time of the event and the snapshot saving time is less than the adjusted transaction time window of the current ticket checking interface, and whether the distance between the operation point position of the event and the operation point position of the current ticket checking interface in the global ticket checking status snapshot is less than the effective sensing radius of the current ticket checking interface. Real-time triggered events that have consistent credential summaries and meet the conditions for spatiotemporal continuity comparison are discarded, while real-time triggered events that have inconsistent credential summaries or do not meet the conditions for spatiotemporal continuity comparison are marked for processing in the next scheduling cycle.

Images