Intelligent parking control system and method based on internet of things non-inductive payment

Abstract

The present application relates to the technical field of Internet of Things financial settlement and business management, and discloses a smart stop-charging control system and method based on Internet of Things non-inductive payment. The method comprises the following steps: acquiring a monitoring image sequence and extracting a pixel displacement trajectory, performing perspective transformation to obtain a multi-modal data set containing a video stream motion vector and a power pulse fingerprint; performing time domain cross-correlation mapping on the multi-modal data set to obtain a joint time sequence set, calculating an identity confidence and outputting an identity verification identification set; extracting a power supplement termination time point and calculating a residence time distribution, constructing a settlement judgment plane and dividing a functional partition, and performing settlement logic state conversion to obtain a logic switching pulse; using the pulse to drive an asynchronous shadow queue to perform time sequence reprogramming, performing amount offset through a time sequence deviation compensation factor, and generating a settlement instruction. The present application solves the problems of identity drift and billing shock caused by heterogeneous data asynchronization, and guarantees the reliability and fairness of financial-level settlement.

Description

Technical Field

[0001] This invention relates to the field of IoT financial settlement and business management technology, and more specifically, to a smart charging / stopping control system and method based on IoT-based contactless payment. Background Technology

[0002] With the promotion of integrated parking and charging pilot projects in smart city construction, IoT-based contactless payment technology has become key to improving parking and charging efficiency. However, in large underground parking garages or roadside scenarios, abnormal charging phenomena occur frequently. For example, high resource occupancy fees are still deducted after a vehicle finishes charging and leaves, or the settlement process cannot start normally due to discrepancies in the reporting timing of vehicle identification and sensing signals, resulting in risks such as evasion or double charging. The root cause of these problems lies in the independent operation of parking management and charging services at the underlying logic level. The communication mechanisms and data reporting cycles used by different systems differ significantly, leading to varying degrees of delay in information feedback. When the management platform attempts to correlate the spatial location status of the vehicle, the energy replenishment process, and the financial charging process, the information collected by IoT devices is discrete and asynchronous in the time dimension, often resulting in a mismatch between the received signal sequence and the actual order of events in the physical world. Traditional sequential triggering mechanisms are difficult to cope with logical misalignments caused by signal drift, network congestion, or sampling errors. Once precise spatiotemporal alignment cannot be achieved at the computational level, the trigger point of the settlement instruction will shift at the critical point of state transition, thus losing financial-grade reliability. Therefore, how to overcome the limitations of asynchronous transmission of heterogeneous signals and build a verification mechanism with physical causal correlation to ensure the accuracy of the settlement process is a technical problem that urgently needs to be solved in the current field.

[0003] In the prior art, Chinese Patent No. CN111402592B discloses a large-scale parking and charging station for electric vehicles and its parking and charging management method. This system includes parking lot hardware facilities, a vehicle type identification and decoy device, a vehicle flow detection mechanism, and an intelligent guidance screen, enabling accurate identification of vehicle types and flow guidance. Through the linkage between the vehicle exit detection mechanism and the first and second charging displays, it pushes vehicle parking information and charging status to the operation and maintenance platform and rest area in real time, optimizing the dynamic allocation of charging resources, improving equipment utilization, and effectively preventing fuel vehicles from occupying charging spaces. Chinese Patent No. CN117496496B discloses a license plate recognition system and method for intelligent charging piles. This system adopts a four-layer architecture of image acquisition, processing, vehicle sensing, and control modules. It acquires license plate images by deploying dual cameras, extracts the license plate area using coordinate and pixel grayscale value algorithms, accurately identifies the vehicle's energy type, and controls the opening and closing of the intelligent lock. It automatically fills the license plate information into the charging pile login module to trigger power-on, realizing the automation of the charging process and the intelligent access authentication, solving the recognition problem caused by image tilt and the problem of fuel vehicles occupying the charging pile incorrectly.

[0004] However, while the two existing technologies mentioned above have some value in terms of station resource scheduling, vehicle type recognition and authentication, and anti-occupancy guidance, they fail to address the core pain points of consistency conflicts and settlement atomicity guarantees in multi-dimensional state machines under distributed asynchronous environments. Specifically, the Chinese patent with authorization announcement number CN111402592B focuses on macro-level station traffic flow guidance and hardware linkage, without addressing the handling of temporal race conditions in the data aggregation process between the "stop" and "charge" business domains, thus failing to eliminate the risk of logical collapse during seamless payment settlement. The Chinese patent with authorization announcement number CN117496496B, while implementing visual tag-based entry control, relies on isolated license plate visual information and a one-way communication protocol, lacking a causal verification mechanism that strongly couples video stream motion vectors with power pulse fingerprints. Neither technology utilizes temporal cross-correlation mapping to construct a multimodal dataset to address identity drift, nor does it construct a settlement decision plane based on dwell time distribution to eliminate billing state oscillations. More importantly, neither of them built asynchronous shadow queues to align the physical occurrence sequence of heterogeneous data, could not use logical switching pulses to drive timing reprogramming, and could not use timing deviation compensation factors to perform amount offsetting. As a result, it was difficult to ensure the financial-grade reliability and absolute fairness of contactless payment settlement instructions in scenarios with multiple vehicles in concurrency and signal delay. Summary of the Invention

[0005] This invention is applicable to integrated smart parking and charging scenarios, meeting the collaborative management needs of vehicle parking and charging resources in different environments. By extracting pixel displacement trajectories and performing perspective transformation processing on monitoring image sequences, a multimodal dataset containing video stream motion vectors and power pulse fingerprints is constructed, transforming passive, fragmented, single-dimensional monitoring into proactive feature monitoring with physical causality. By using a joint time series set generated by temporal cross-correlation mapping, the physical mapping points between vehicle stillness and charging handshake are locked, solving the problem of incorrect matching between parking space resources and charging entities caused by isolated visual tags and asynchronous energy data. The settlement judgment plane constructed based on dwell time distribution executes settlement logic state transitions through functional partitioning, and the generated logic switching pulses eliminate billing state oscillations, ensuring the atomicity of billing mode switching. The asynchronous shadow queue, combined with the time series re-editing mechanism, corrects the logical misalignment caused by signal drift, and uses the time series deviation compensation factor to reduce the error amount, ensuring the fairness and reliability of financial-grade settlement in a distributed asynchronous environment.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A smart charging stop control method based on IoT-based contactless payment includes:

[0008] Acquire a sequence of monitoring images representing the parking process of a target vehicle in a smart parking and charging integrated scenario. Extract pixel displacement trajectories from the monitoring image sequence and perform perspective transformation on the pixel displacement trajectories to obtain a joint time series set representing the probability of physical causal matching. Perform spatiotemporal feature coupling calculation on the joint time series set to obtain the identity confidence score. Perform numerical comparison on the identity confidence score to obtain the identity verification identifier set.

[0009] Multidimensional feature parameter extraction is performed based on the identity verification identifier set to obtain the power replenishment termination time point. The duration of the power replenishment termination time point is accumulated to obtain the dwell time distribution. A settlement judgment plane is constructed based on the dwell time distribution, and functional partitions are divided in the settlement judgment plane. Settlement logic state transitions are performed according to the functional partitions to obtain logic switching pulses.

[0010] An asynchronous shadow queue is constructed, and a timing recompilation is performed on the asynchronous shadow queue using a logical switching pulse to obtain a timing-aligned sequence set. The amount is offset on the timing-aligned sequence set to obtain a settlement instruction with settlement atomicity.

[0011] Furthermore, the method for obtaining the joint time series set includes:

[0012] The monitoring image sequence is obtained by collecting two-dimensional image data of each sampling frame according to a preset unit sampling period and arranging them in chronological order;

[0013] A spatial motion monitoring coordinate system is constructed. An image coordinate system with the upper left corner as the origin is established in each frame of two-dimensional image data. Feature anchor points are set for the target vehicle. At the same time, the pixel position coordinates of the feature anchor points in the image coordinate system are obtained. The pixel position coordinates of all sampled frames are connected in time sequence to obtain the pixel displacement trajectory.

[0014] The perspective transformation of the pixel displacement trajectory is performed using the pre-calibrated homography matrix to obtain the physical position coordinate pairs. The instantaneous velocity amplitude and direction angle are calculated using the physical position coordinate pairs of the current sampling frame and the previous sampling frame. The instantaneous velocity amplitude and direction angle of all sampling frames are arranged in time sequence and combined into the video stream motion vector.

[0015] The current amplitude sequence output by the charging pile is obtained and defined as a power pulse fingerprint. The power pulse fingerprint and the motion vector of the video stream are encapsulated into a multimodal dataset. Temporal cross-correlation mapping is performed on the multimodal dataset to obtain a joint time series set.

[0016] Furthermore, the execution of the time-domain cross-correlation mapping specifically includes:

[0017] Instantaneous velocity amplitude sequence and current amplitude sequence are extracted from the multimodal dataset. The current amplitude sequence is resampled using the unit sampling period as the reference time axis to obtain the current response sequence. The instantaneous velocity amplitude sequence is a sequence composed of the instantaneous velocity amplitudes corresponding to all sampling frames arranged in time order.

[0018] The maximum instantaneous velocity amplitude in the instantaneous velocity amplitude sequence and the maximum current amplitude in the current response sequence are normalized and mapped to obtain the motion intensity sequence and energy intensity sequence, respectively.

[0019] Set a sliding time window and set a time offset with a discrete offset step size of unit sampling period within the sliding time window. For each time offset within the sliding time window, calculate the cumulative sum of the motion intensity sequence and the energy intensity sequence to obtain the cross-correlation intensity value. Arrange the cross-correlation intensity values ​​corresponding to all time offsets within the sliding time window in the order of time offset to form a joint time series set representing the probability of physical causal matching.

[0020] Furthermore, the identity confidence level includes:

[0021] The maximum cross-correlation strength value is searched from the joint time series set and defined as the time-domain matching peak.

[0022] Extract the direction angles corresponding to multiple consecutive sampled frames after the target vehicle comes to rest from the motion vector of the video stream, calculate the arithmetic mean, and obtain the end pointing angle that represents the final parking position of the target vehicle in the parking space.

[0023] In the spatial motion monitoring coordinate system, the straight line passing through the origin and pointing into the berth is defined as the berth's central axis, and the angle between the berth's central axis and the coordinate axis is defined as the reference angle.

[0024] By introducing a time-domain weight component, a multi-dimensional weighted fusion calculation is performed on the time-domain matching peak, end pointing angle, and reference reference angle to obtain the identity confidence degree, which represents the strength of the association between the target vehicle's identity and the physical resources of the charging pile.

[0025] Furthermore, the identity verification identifier set includes:

[0026] Set a logical settlement threshold to determine whether the target vehicle and the charging pile have a unique physical association, and compare the identity confidence level with the logical settlement threshold.

[0027] If the identity confidence level is greater than or equal to the logical settlement threshold, it is determined that the target vehicle and the charging pile belong to the same physical object, the physical association identifier is output, and the physical association identifier is used to initialize the financial deduction process;

[0028] If the identity confidence level is less than the logical settlement threshold, it is determined that the target vehicle and the charging pile do not belong to the same physical object, and an identity anomaly identifier is output; the physical association identifier and the identity anomaly identifier are combined to obtain the identity verification identifier set.

[0029] Furthermore, the residence time distribution includes:

[0030] The charging pile is located based on the physical association identifier in the identity verification identifier set, and the real-time charging power output by the charging pile to the target vehicle is obtained using the power monitoring module built into the charging pile.

[0031] A gradient observation window is set up, which consists of a preset number of sampling points arranged in chronological order, and a unique incremental sampling point index is assigned to each sampling point. The first-order numerical differentiation processing is performed on the real-time charging power within the gradient observation window to obtain the power consumption gradient characterizing the physical deceleration rate of the real-time charging power.

[0032] Set the energy compensation cutoff power for determining the end of the energy compensation state, and the gradient stability threshold for quantifying the quietness of power change. In a continuous sampling point index sequence with multiple consecutive sampling points, when the real-time charging power is less than the energy compensation cutoff power and the absolute value of the power consumption gradient is less than the gradient stability threshold, the physical time point associated with the first sampling point index in the continuous sampling point index sequence is recorded as the energy compensation termination time point.

[0033] Starting from the point when the refueling ends, the location occupancy signal representing the physical occupation of the berth by the target vehicle is obtained using a visual sensor. The cumulative physical duration between the point when the refueling ends and the physical time point corresponding to the current real-time sampling sequence index is calculated to obtain the dwell time distribution.

[0034] Furthermore, the logic switching pulse includes:

[0035] A two-dimensional coordinate system is constructed with the power consumption gradient as the horizontal axis and the residence time distribution as the vertical axis. This system is defined as the settlement judgment plane. Gradient feature values ​​are set to define the degree of energy consumption, and residence feature values ​​are set to define the duration of resource occupation.

[0036] Using gradient eigenvalues ​​and dwell eigenvalues, three logically mutually exclusive functional zones are divided in the settlement decision plane: the energy trading zone, the billing logic buffer zone, and the resource occupation zone.

[0037] Within the settlement determination plane, coordinate points determined based on real-time power consumption gradients and dwell time distributions are defined as state points. When a state point enters a resource occupancy area, a logic transfer identifier is generated. The logic transfer identifier is configured to have a low logic level representing the initial reset state and a high logic level representing the settlement trigger state, and is initially set to a low logic level.

[0038] The linear offset of the state point in the vertical direction beyond the dwell characteristic value is calculated and defined as the occupancy strength factor. When the occupancy strength factor is continuously greater than zero within the preset judgment confirmation window, the logic transfer flag is switched from low logic level to high logic level. The rising edge transition signal generated by the logic transfer flag at the instant of switching from low logic level to high logic level is defined as the logic switching pulse, and the physical moment when the rising edge transition signal is generated is defined as the transition moment.

[0039] Furthermore, the time-aligned sequence set includes:

[0040] A temporary storage area for data storage and verification is allocated in the computer memory and defined as an asynchronous shadow queue. The real-time charging power and the physical time point corresponding to the generation of the real-time charging power are combined to form an energy billing data packet. The location occupancy signal and the physical time point corresponding to the generation of the location occupancy signal are combined to form a resource occupancy billing data packet. The time when the energy billing data packet and the resource occupancy billing data packet are received is obtained and defined as the system receiving time. The energy billing data packet and the resource occupancy billing data packet are stored in the asynchronous shadow queue.

[0041] The transition time of the logic switching pulse is defined as the global synchronization reference point. The physical time point representing the energy state is extracted from the asynchronous shadow queue, and the physical time point representing the space state is extracted from the resource occupancy billing data packet. The physical time point representing the energy state and the physical time point representing the space state are uniformly identified as physical timestamps.

[0042] When the logic switching pulse is triggered as a rising edge signal, the system receiving time is discarded, and the energy billing data packet and the resource occupancy billing data packet are linearly rearranged according to the order of the physical timestamps to obtain a time-aligned sequence set.

[0043] Furthermore, the settlement instruction includes:

[0044] All energy billing data packets are filtered from the time-aligned sequence set, while maintaining the chronological order of the physical timestamps, to form an energy billing flow.

[0045] The physical timestamps corresponding to the energy billing data packets at the end of the energy billing flow are extracted from the time-aligned sequence set and defined as the true termination time. The system receiving time corresponding to the energy billing data packets at the end is extracted and defined as the system sensing time.

[0046] The time difference between the system's perceived time and the actual termination time is defined as the time deviation. A billing unit price factor is set, and the time deviation compensation factor is obtained by multiplying the time deviation by the billing unit price factor.

[0047] Obtain the initial total energy billing amount, subtract the timing deviation compensation factor from the initial total energy billing amount to obtain the atomic settlement amount, and encapsulate the atomic settlement amount, the physical association identifier corresponding to the target vehicle, and the logical switching pulse to obtain the settlement instruction.

[0048] A smart charging stop control system based on IoT-based contactless payment is used to implement the aforementioned smart charging stop control method based on IoT-based contactless payment. The system includes:

[0049] Identity verification module: used to acquire monitoring image sequences representing the parking process of target vehicles in the smart parking and charging integrated scenario, extract pixel displacement trajectories from the monitoring image sequences, perform perspective transformation processing on the pixel displacement trajectories to obtain a joint time series set representing the probability of physical causal matching, perform spatiotemporal feature coupling calculation on the joint time series set to obtain the identity confidence, and perform numerical comparison on the identity confidence to obtain the identity verification identifier set;

[0050] The billing decision module is used to extract multi-dimensional feature parameters based on the identity verification identifier set to obtain the power replenishment termination time point, perform cumulative calculation of the duration of the power replenishment termination time point to obtain the dwell time distribution, construct a settlement decision plane based on the dwell time distribution, divide functional areas in the settlement decision plane, and perform settlement logic state transitions according to the functional areas to obtain logic switching pulses.

[0051] Atomic Settlement Module: Used to construct asynchronous shadow queues, use logical switching pulses to drive asynchronous shadow queues to perform timing recompilation, obtain timing-aligned sequence sets, perform amount offsetting on timing-aligned sequence sets, and obtain settlement instructions with atomicity.

[0052] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0053] This invention solves the problems of decreased identity recognition accuracy and mismatch between parking space resources and charging entities caused by isolated visual tags, asynchronous energy data, and the need to lock physical causal mapping points by restoring monitoring image sequences into multimodal datasets and combining them with joint time series sets generated by time-domain cross-correlation mapping. It utilizes a settlement determination plane and its functional partitions constructed from the recharge termination point to orthogonally decouple the power consumption gradient and dwell time distribution of asynchronous operation from physical attributes, solving the problem of state oscillation during billing mode transitions. A logical switching pulse ensures the atomicity and financial-grade reliability of billing mode transitions. An asynchronous shadow queue, combined with a time-series reprogramming mechanism, maps discrete and asynchronous IoT data back to a continuous sequence in the physical world, correcting logical drift caused by communication link transmission delays. A time-series deviation compensation factor reduces the error amount caused by system perception delays, solving the problem of inflated billing amounts and ensuring absolute fairness in financial settlement under a distributed asynchronous environment. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 A flowchart illustrating the intelligent charging stop control method based on IoT-based contactless payment provided in this embodiment of the invention.

[0056] Figure 2 This is a schematic diagram of the spatial motion monitoring coordinate system and the target vehicle's reversing into the parking space, provided in an embodiment of the present invention.

[0057] Figure 3This is a schematic diagram of the partition structure of the settlement determination plane provided in an embodiment of the present invention;

[0058] Figure 4 This is a functional block diagram of a smart charging and stopping control system based on IoT-based contactless payment provided in an embodiment of the present invention. Detailed Implementation

[0059] 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.

[0060] Example 1

[0061] Please see Figure 1 As shown, this embodiment provides a smart charging stop control method based on IoT-based contactless payment, including:

[0062] Step S10: Obtain a sequence of monitoring images representing the parking process of the target vehicle in the smart parking and charging integrated scenario; extract pixel displacement trajectories from the monitoring image sequence; perform perspective transformation on the pixel displacement trajectories to obtain a joint time series set representing the probability of physical causal matching; perform spatiotemporal feature coupling calculation on the joint time series set to obtain the identity confidence score; perform numerical comparison on the identity confidence score to obtain the identity verification identifier set.

[0063] Further, step S10 includes:

[0064] Step S11: Obtain a sequence of monitoring images representing the parking process of the target vehicle in the smart parking and charging integrated scenario, and construct a spatial motion monitoring coordinate system to extract the trajectory of the monitoring image sequence to obtain the pixel displacement trajectory. Perform perspective transformation processing on the pixel displacement trajectory to obtain a multimodal dataset representing the initial state of the parking and charging integrated system.

[0065] In integrated smart parking and smart charging scenarios, the identification of target vehicles and their association with physical resources—namely, parking spaces and charging piles—usually rely on isolated visual tags or one-way communication protocols. However, in confined spaces such as commercial parking lots or underground garages with high traffic volume, license plate information identified solely by video streams is prone to accuracy degradation due to light occlusion, license plate damage, or multipath reflection interference. Furthermore, the energy data acquired by the charging pile only reflects the energy replenishment status and cannot directly prove that the entity consuming energy belongs to the same physical object as the target vehicle in the video surveillance. The parking management logic and charging management logic operate asynchronously at the underlying level, lacking a causal verification mechanism that strongly couples spatial displacement features with energy pulse features. This leads to a high risk of incorrect matching between vehicle owner identities and charging resources under race conditions where multiple vehicles simultaneously enter and perform charging / plugging operations. To address the challenge of identity drift at the logical level in multi-source heterogeneous data streams, an initial dataset capable of fusing spatial displacement vectors and power feature codes is constructed. The goal is to transform passive, fragmented, single-dimensional monitoring into proactive, multimodal feature monitoring with physical causal relationships.

[0066] Specifically, to quantify the spatial topological trajectory of a target vehicle during parking, a fixed two-dimensional coordinate system is established with the geometric center point of the parking space entrance as the origin, defined as the spatial motion monitoring coordinate system. The Y-axis of the spatial motion monitoring coordinate system is defined as a straight line parallel to the direction the vehicle enters the parking space, pointing towards the geometric center point between the inner barriers of the parking space. The X-axis is defined as a straight line perpendicular to the Y-axis and passing through the origin in the parking space plane. A vision sensor deployed directly above the parking space continuously collects two-dimensional image data for each sampling frame at a frequency of one sampling period. The collection of two-dimensional image data arranged in chronological order of the sampling frames is defined as the monitoring image sequence. The unit sampling period is set based on the rate of change of the motion vector of the car during the parking maneuver; for example, the unit sampling period is set to 30 milliseconds. In each frame of the monitoring image sequence, a two-dimensional coordinate system is established with the upper left corner of the two-dimensional image data as the origin, defined as the image coordinate system. The axis parallel to the bottom edge of the two-dimensional image data in the image coordinate system is defined as the horizontal pixel axis. The vertical pixel axis is defined as the axis perpendicular to the horizontal pixel axis and passing through the origin in the image coordinate system. Feature anchor points are defined as geometric reference points on the surface of the target vehicle that have significant visual features and relatively fixed positions; for example, the geometric center point of the license plate area of ​​the target vehicle. The spatial orientation of the feature anchor point in the image coordinate system is quantized using pixel position coordinates, where the pixel position coordinates refer to the coordinates (U, V) of the feature anchor point in the image coordinate system. The horizontal pixel value U in the pixel position coordinates is defined as the pixel offset of the feature anchor point relative to the origin of the image coordinate system in the horizontal pixel axis direction. The vertical pixel value V in the pixel position coordinates is defined as the pixel offset of the feature anchor point relative to the origin of the image coordinate system in the vertical pixel axis direction. The pixel position coordinates corresponding to the feature anchor point in all sampling frames of the monitoring image sequence are vector-connected according to the temporal order of the sampling frames, and the resulting continuous two-dimensional coordinate point set is defined as the pixel displacement trajectory. See also... Figure 2This is a schematic diagram of the spatial motion monitoring coordinate system and the reversing parking state of a target vehicle provided in an embodiment of the present invention. The diagram uses a rectangular frame to represent parking spaces in a parking lot, and the parking space entrance is represented by a thickened line at the bottom edge of the parking space. The geometric center point of the parking space entrance is the origin of the spatial motion monitoring coordinate system, denoted by O. A dashed line pointing vertically upwards from point O and towards the center point between the parking barriers 1 and 2 inside the parking space represents the Y-axis of the spatial motion monitoring coordinate system; a dashed line intersecting the Y-axis perpendicularly at point O and extending laterally along the parking space entrance line represents the X-axis of the spatial motion monitoring coordinate system. The diagram illustrates a target vehicle in the process of reversing into a parking space. The left and right rear wheels of the target vehicle are aligned with parking barriers 1 and 2 on the longitudinal axis, respectively. At the end of the target vehicle closest to the inside of the parking space, i.e., the rear of the target vehicle, the rear license plate of the target vehicle is indicated. The geometric center of the rear license plate serves as the aforementioned feature anchor point, represented by a solid black dot in the diagram. Figure 2 It intuitively presents the physical working environment in which the feature anchor points are tracked within the spatial motion monitoring coordinate system.

[0067] The pre-calibration homography matrix of the visual sensor is obtained. This pre-calibration homography matrix is ​​a mapping matrix obtained by homogeneous coordinate projection transformation of the intrinsic and extrinsic parameter matrices of the visual sensor. Perspective transformation processing is performed on the pixel displacement trajectory using the pre-calibration homography matrix to obtain physical position coordinate pairs. These physical position coordinate pairs are the coordinates of the feature anchor points in the spatial motion monitoring coordinate system, used to quantify the geometric displacement of the target vehicle in physical space. The purpose is to restore the pixel displacement in the monitoring image sequence to motion coordinates with real physical scale units, thereby eliminating perspective distortion caused by the tilt of the visual sensor's shooting angle. The physical position coordinate pairs corresponding to the current sampling frame t are extracted from the monitoring image sequence. and the physical location coordinates corresponding to the previous sampling frame t-1 Then, the changes in horizontal and vertical displacements of the feature anchor points in the X and Y axes of the spatial motion monitoring coordinate system are respectively... and , , The instantaneous velocity amplitude S and direction angle A are calculated using the changes in horizontal and vertical axis displacement. The instantaneous velocity amplitude is used to quantify the real-time movement speed of the target vehicle within the parking space plane. ,in, This indicates the unit sampling period. The direction angle is the positive deflection phase of the feature anchor point relative to the X-axis in the spatial motion monitoring coordinate system. ,in, The arctangent function is used to arrange the instantaneous velocity amplitude and direction angle corresponding to all sampling frame times in chronological order, resulting in instantaneous velocity amplitude sequences and direction angle sequences. These sequences are then combined to form a video stream motion vector. Simultaneously, the power pulse fingerprint of the charging pile is acquired. This fingerprint is a sequence of current amplitude values ​​directly collected and output by the charging pile at each sampling moment during the charging handshake protocol execution with the target vehicle. The power pulse fingerprint and the video stream motion vector are then encapsulated to generate a multimodal dataset representing the initial state of the integrated charging and stopping process.

[0068] Step S12: Perform temporal cross-correlation mapping on the multimodal dataset to obtain a joint time series set representing the probability of physical causal matching.

[0069] After obtaining the multimodal dataset, correlation quantization in the temporal dimension is performed on the multimodal dataset to align the video stream motion vector, which represents spatial displacement characteristics, with the power pulse fingerprint, which represents energy state characteristics, to a unified time reference. In the distributed acquisition architecture of intelligent charging and stopping integration, the video stream motion vector records the dynamic evolution of the target vehicle's geometric displacement in physical space, while the power pulse fingerprint records the transient waveform of electrical energy injection at the target vehicle's charging port. By calculating the cross-correlation matching degree between the instantaneous velocity amplitude contained in the video stream motion vector and the power pulse fingerprint in a discrete time offset sequence with a unit sampling period as the step, the physical causal mapping point between the instant the target vehicle is completely stationary and the instant the charging handshake begins is locked.

[0070] Specifically, the instantaneous velocity amplitude sequence from the video stream motion vector and the current amplitude sequence from the power pulse fingerprint are extracted from the multimodal dataset. To eliminate the discrepancy in the number of data points caused by the inconsistency between the video stream acquisition frequency and the charging data acquisition frequency, the current amplitude sequence is resampled using the unit sampling period as the reference time axis to obtain a resampled current amplitude sequence. Specifically, the total number of samples in the instantaneous velocity amplitude sequence is obtained, and the current amplitude sequence is resampled using a first-order Lagrange polynomial interpolation algorithm based on linear time-base mapping, so that the number of current amplitudes in the resampled current amplitude sequence equals the total number of samples, and the time step between two adjacent current sampling points in the resampled current amplitude sequence equals the unit sampling period. The resampled current amplitude sequence is defined as the current response sequence. Since the instantaneous velocity amplitude and the current response sequence have different dimensions, a normalization mapping process is performed on the instantaneous velocity amplitude sequence and the current response sequence. The specific process is as follows: The largest instantaneous velocity amplitude in the instantaneous velocity amplitude sequence is obtained and defined as the maximum velocity amplitude. The instantaneous velocity amplitude corresponding to each sampling frame time is divided by the maximum velocity amplitude to obtain the motion intensity component. The motion intensity components obtained from each sampling frame time are arranged in time sequence to obtain the motion intensity sequence. The motion intensity sequence characterizes the activity level of the target vehicle relative to a completely stationary state. Simultaneously, the largest current amplitude in the current response sequence is obtained and defined as the maximum current amplitude. The current amplitude corresponding to each sampling moment is divided by the maximum current amplitude to obtain the energy intensity component. The motion intensity components obtained from each sampling moment are arranged in time sequence to obtain the energy intensity sequence. The energy intensity sequence represents the activity level of the charging pile interface's electrical energy response relative to the charging handshake state.

[0071] A time-domain cross-correlation mapping calculation is performed on the energy intensity sequence and the motion intensity sequence. Specifically, a sliding time window is set, which is based on the average physical interaction time from when the target vehicle is completely stationary to when it completes the charging gun insertion / removal action. For example, the sliding time window is set to 10 seconds. The purpose is to ensure that the characteristic waveforms of the video stream motion vector and the characteristic waveforms of the power pulse fingerprint can be completely included in the calculation area. Within the sliding time window, a time offset is set with a discrete offset step size of unit sampling period. The time offset is the unit offset of the energy intensity sequence relative to the motion intensity sequence on the time axis. The time offset is set based on the maximum difference range between the streaming media transmission delay of the monitoring image sequence and the communication protocol response delay of the power pulse fingerprint. The purpose is to search for the optimal alignment position of motion features and electrical features in physical causal logic by simulating signal delays of different dimensions within the sliding time window. For example, the value range of the time offset is set to [-100, 100]. For each time offset within the sliding time window, the cumulative sum of the motion intensity component and the energy intensity component is calculated to obtain the cross-correlation strength value R. This cross-correlation strength value reflects the statistical correlation strength between parking behavior and charging behavior at a specific time offset. For the time offset... The corresponding cross-correlation strength value The calculation formula is: In this formula, M represents the total number of sampling points participating in the calculation within the sliding time window, determined by dividing the duration of the sliding time window by the unit sampling period; j represents the common time base index within the sliding time window, ranging from 1 to M. By traversing all common time base indices from 1 to M, it can be ensured that the motion intensity component and energy intensity component involved in the calculation are completely traversed throughout the entire duration of the sliding time window, guaranteeing that the calculation results can truly reflect the overall correlation of the two sets of sequences in the local time domain. Since the sampling frequency of the current response sequence has been aligned to the unit sampling period before performing the cross-correlation operation, the letter j in the formula uniformly represents the time sequence number of the j-th discrete time point within the sliding time window, and B represents the motion intensity sequence. This represents the element in the motion intensity sequence corresponding to the common time base index j, i.e., the value of the motion intensity component of the target vehicle at the j-th unit sampling period within the sliding time window, where I represents the energy intensity sequence. This indicates that the common time base index in the energy intensity sequence is... The element is the current response value of the charging pile within the sliding time window, which is the result of a time offset relative to the motion intensity component. The logic behind the calculation formula for the cross-correlation intensity value is as follows: when the physical moment when the target vehicle changes from a driving state to a completely stationary state in a parking space coincides with the starting moment when the charging pile injects electrical energy into the target vehicle and generates current feedback, the decreasing characteristic of the target vehicle's deceleration and stopping action in the motion intensity sequence and the increasing characteristic of the charging handshake starting action in the energy intensity sequence will achieve phase overlap at the same common time base index position. The feature vector formed by arranging the cross-correlation intensity values ​​corresponding to all time offsets within the sliding time window in time offset order is defined as the joint time series set. The joint time series set is used to quantify the physical causal matching probability of parking behavior and charging behavior under different time offsets. The physical causal matching probability reflects the degree of phase overlap between the vehicle stopping action recorded by the video stream motion vector and the charging response action recorded by the power pulse fingerprint on the physical time axis.

[0072] Step S13: Perform spatiotemporal feature coupling calculation on the joint time series set to obtain the identity confidence level representing the target vehicle and the charging pile. Perform numerical comparison based on the identity confidence level to obtain the identity verification identifier set.

[0073] After obtaining the joint time series set, to address the temporal feature confusion caused by vehicles in adjacent parking spaces performing parking actions within similar timeframes, and to ensure the physical uniqueness of the energy consumption generated by the charging pile and the target vehicle's identity, feature peak localization is performed on the cross-correlation strength values ​​included in the joint time series set. This aims to calculate the identity confidence between the target vehicle and the charging pile. The joint time series set reflects the probability of overlap between the target vehicle's parking characteristics and the charging pile's response characteristics in the temporal dimension, while the direction angle sequence records the final geometric orientation of the target vehicle within the parking space. By fusing the sequence correlation in the temporal dimension and the trajectory consistency in the spatial dimension, logical entanglement caused by multiple vehicles entering the parking lot concurrently is eliminated, thereby ensuring the atomicity of the seamless payment process.

[0074] Specifically, the cross-correlation intensity values ​​corresponding to all time biases are searched from the joint time series set, and the cross-correlation intensity value with the largest value in the joint time series set is defined as the time-domain matching peak. The temporal matching peak value is used to quantify the envelope similarity between the motion characteristics of the target vehicle evolving from a driving state to a stationary state and the electrical energy characteristics of the charging pile entering the charging handshake state at the optimal alignment point. A sequence of directional angles corresponding to C consecutive sampling frames is extracted from the motion vector of the video stream simultaneously. Here, C represents the number of stationary sampling frames at the end, set based on the smooth observation period required for the vehicle suspension system to complete mechanical damping vibration after the target vehicle enters its parking space and reaches a physically stationary state. The purpose is to eliminate the instantaneous interference caused by the vehicle's forward and backward swaying during braking on the determination of the feature anchor point angle, thereby obtaining the parking orientation in a stable state. For example, the value of C is set to 10, and the arithmetic mean of the C directional angles is calculated, defined as the final pointing angle, which represents the final parking orientation of the target vehicle in its parking space within the parking lot. In the spatial motion monitoring coordinate system, a straight line passing through the origin and pointing towards the center point between parking space barrier 1 and barrier 2 is defined as the parking space center axis. The angle between the parking space center axis and the positive X-axis is defined as the reference angle. Since the parking space center axis coincides with the Y-axis of the spatial motion monitoring coordinate system and is perpendicular to the X-axis, the reference angle is always 90 degrees. Using the time-domain matching peak value, end-pointing angle, and reference angle, a multi-dimensional weighted fusion calculation is performed to obtain the identity confidence score D. The identity confidence score represents the correlation strength score between the target vehicle's identity and the charging pile physical resource currently outputting current amplitude. Its calculation formula is as follows: ,in, This represents the temporal weight component, set based on the ambient light contrast level when the visual sensor captures feature anchor points and the electromagnetic pulse interference frequency around parking spaces in the parking lot. In scenarios where parking spaces are under intense ambient light, the temporal weight component is reduced to strengthen the weight provided by the end-pointing angle; in scenarios with relatively small fluctuations in network signal reporting delay, the value of the temporal weight component is increased to strengthen the weight of the temporal matching peak. Its range is limited to 0 to 1. e is a natural constant. and These represent the end-pointing angle and the reference angle, respectively. The logic behind the identity confidence calculation formula is as follows: by nonlinearly coupling the temporal feature similarity reflected by the joint time series set with the spatial trajectory consistency reflected by the end-pointing angle after smoothing based on C sampling frames, a dual identity verification model with spatial exclusivity and temporal causality is established. Only when the target vehicle's parking time and charging start time highly coincide on the time axis, and the target vehicle's final pose conforms to the physical orientation of the parking space in the parking lot, will the identity confidence reach the admission interval that meets the recognition requirements. This eliminates the risk of identity drift caused by heterogeneous data asynchrony from the algorithm's underlying layer.

[0075] A logical settlement threshold is set, representing the minimum confidence level required to determine whether a target vehicle and the charging pile currently consuming electricity have a unique physical association. This threshold is set based on the balance requirement between the false recognition rate and the missed settlement rate in a densely populated parking environment. For example, the logical settlement threshold is set to 0.85. The logical settlement threshold is compared with the identity confidence level. If the identity confidence level is greater than or equal to the logical settlement threshold, it is determined that the target vehicle in the monitoring image sequence and the charging pile currently consuming electricity belong to the same physical object. A physical association identifier representing the physical association between the target vehicle and the charging pile is output, and this identifier is used to initialize the financial deduction process based on the IoT-based contactless payment account. Conversely, if the identity confidence level is less than the logical settlement threshold, it is determined that the target vehicle in the monitoring image sequence and the charging pile currently consuming electricity do not belong to the same physical object. The financial settlement initialization process for the target vehicle is terminated, and an identity anomaly identifier is output to prevent erroneous deduction of resource occupancy fees. The identity anomaly identifier and the physical association identifier are combined to obtain an identity verification identifier set. The identity verification identifier set refers to a set of logical tags generated based on the numerical comparison results of identity confidence and preset logical settlement threshold, which are used to characterize the unique physical association between the target vehicle and the charging pile.

[0076] Step S10 addresses the challenges of decreased identity recognition accuracy and mismatched parking space resources with charging entities in smart parking scenarios by using a multimodal dataset, performing temporal cross-correlation mapping, and performing spatiotemporal feature coupling calculation. This achieves a strong causal relationship between the target vehicle's identity and physical resources. Specifically, the multimodal dataset transforms passive, fragmented, single-dimensional monitoring into proactive feature monitoring with physical causality; the temporal cross-correlation mapping aligns discrete, asynchronous video streams and current pulses to a unified time reference, accurately pinpointing the physical mapping point between vehicle stillness and charging handshake; and the spatiotemporal feature coupling calculation eliminates the logical entanglement caused by multiple vehicles entering the field concurrently, ensuring the physical uniqueness of identity binding.

[0077] Step S20: Perform multi-dimensional feature parameter extraction based on the identity verification identifier set to obtain the power replenishment termination time point, perform time accumulation calculation on the power replenishment termination time point to obtain the dwell time distribution, construct a settlement judgment plane based on the dwell time distribution, divide the settlement judgment plane into functional partitions, and perform settlement logic state transitions according to the functional partitions to obtain logic switching pulses.

[0078] Further, step S20 includes:

[0079] Step S21: Based on the identity verification identifier set, perform multi-dimensional feature parameter extraction on the target vehicle to obtain the recharge termination time point, and perform cumulative calculation on the recharge termination time point to obtain the residence time distribution representing the resource occupancy intensity.

[0080] After identifying the physical association identifier in the identity verification identifier set, in order to solve the problem of insufficient accuracy in switching between energy billing and resource occupancy billing caused by the slight power fluctuation of the target vehicle at the end of the charging period, and to ensure the atomicity of billing logic conversion, a set of billing conversion criteria is established by performing multi-dimensional parameterized extraction on the energy consumption characteristics and spatial residence characteristics of the target vehicle in a single charging cycle. This aims to eliminate the billing state switching oscillation caused by discrete signal jumps, thereby ensuring the financial reliability of the seamless payment settlement process at the state transition critical point.

[0081] Specifically, based on the physical association identifier in the identity verification identifier set, the charging pile corresponding to the target vehicle is located. The charging pile's built-in power monitoring module continuously acquires the real-time charging power output from the charging pile to the target vehicle. This real-time charging power is the amplitude of electrical energy injected into the target vehicle by the charging pile at the current sampling time. To quantify the decay slope and energy fluctuation characteristics of the real-time charging power over time, a gradient observation window is set to calculate the energy change rate. The gradient observation window is set based on the electrical signal sampling period of the charging pile's built-in power monitoring module when performing constant-voltage charging on the target vehicle. For example, the gradient observation window is set to 60 seconds. N represents the total number of sampling points in the gradient observation window, and H represents the sampling point index. To accurately reflect the decay rate of the real-time charging power over the entire time axis, the sampling point index is set to a range of 2 to N. Within the gradient observation window, a first-order numerical differentiation is performed on the real-time charging power to obtain the power consumption gradient G. This power consumption gradient characterizes the physical decay rate of the real-time charging power per unit time. The formula for calculating the power consumption gradient is: Where ∆H represents the time difference between two adjacent acquisition points in the gradient observation sliding window. K represents the real-time charging power. This indicates the real-time charging power at the collection point index H. This represents the real-time charging power at collection point index H-1. A power threshold value is set to determine whether the target vehicle transitions from an energy-compensated state to a resource-occupied state, defined as the energy replenishment cutoff power. The energy replenishment cutoff power is set based on the sum of the reference current loss of the charging pile after energy output stops and the static energy consumption amplitude of the target vehicle after energy replenishment is completed. For example, the energy replenishment cutoff power is set to 100. Simultaneously, a gradient stability threshold is set to quantify the quietness of power changes. The gradient stability threshold is set based on the random thermal noise power fluctuation range of the power monitoring module under no-load conditions. For example, the gradient stability threshold is set to 5. Only when the real-time charging power is less than the energy replenishment cutoff power and the absolute value of the power consumption gradient is less than the gradient stability threshold within the time points corresponding to L consecutive collection point indices is the target vehicle considered to have completed a single energy replenishment task. The physical time point corresponding to the first collection point index among the L consecutive collection point indices is recorded as the energy replenishment termination time point. The value of L is set based on the minimum steady-state observation period required for the charging pile to complete the charging termination determination; for example, L is set to 5. From the point of recharging termination, a visual sensor deployed directly above the parking space in the parking lot is used to acquire the target vehicle's position occupancy signal in real time. This position occupancy signal is the monitoring feedback from the visual sensor recognizing that the target vehicle is continuously within the spatial motion monitoring coordinate system, used to characterize whether the parking space in the parking lot is physically occupied by the target vehicle. After recognizing the position occupancy signal, the cumulative physical duration between the point of recharging termination and the physical time point corresponding to the current real-time acquisition point index is calculated, defined as the dwell time distribution. The dwell time distribution characterizes the intensity of the target vehicle's continuous occupancy of the parking space's physical resources after completing the energy transaction.

[0082] Step S22: Construct a settlement decision plane based on the residence time distribution, divide the settlement decision plane into functional partitions, and perform settlement logic state transitions according to the functional partitions to obtain logic switching pulses used to drive atomic switching of billing modes.

[0083] After obtaining the power consumption gradient and dwell time distribution, in order to solve the billing oscillation problem caused by the physical noise generated during the energy replenishment interaction process at the end of charging, which leads to the switching from energy trading to resource occupation trading, and to ensure the atomicity of billing logic transfer, a settlement determination plane is constructed and a nonlinear envelope boundary is defined to eliminate false settlements caused by fluctuations in a single sensor signal and to ensure the settlement accuracy of the contactless payment process at the state transition critical point.

[0084] Specifically, a two-dimensional coordinate system is constructed with the power consumption gradient as the horizontal axis and the dwell time distribution as the vertical axis, defined as the settlement judgment plane. The coordinates of any point in the settlement judgment plane are defined as a state point. To quantify the calculation process, variables are defined to represent the numerical values ​​of the dwell time distribution. A gradient feature value P is set to define the degree of energy dissipation. This gradient feature value corresponds to a physical judgment point in the settlement judgment plane with a vertical coordinate of 0 and located on the horizontal axis. The setting is based on the gradient stability threshold and the sampling step size of the power monitoring module built into the charging pile. For example, the gradient feature value is set to 8. Simultaneously, a dwell time feature value Q is set to define the resource occupancy duration. This dwell time feature value corresponds to a physical judgment point in the settlement judgment plane with a horizontal coordinate of 0 and located on the vertical axis. The setting is based on the free parking buffer time defined by the commercial scenario to which the parking space in the parking lot belongs. For example, the dwell time buffer time is set to 8. The dwell characteristic value is set to 180. Using the dwell characteristic value and gradient characteristic value, three logically mutually exclusive functional zones are divided within the first quadrant of the settlement determination plane: an energy trading zone, a billing logic buffer zone, and a resource occupation zone. The energy trading zone, within the first quadrant of the settlement determination plane, is represented as a semi-open geometric region with an abscissa greater than or equal to the gradient characteristic value, i.e., a semi-open geometric region where G≥P. When the state point falls within the energy trading zone, the target vehicle is determined to be in an energy-compensated state, and energy billing for the target vehicle is maintained, corresponding to the traditional charging billing based on charging power or charging duration. The logic buffer zone, within the first quadrant of the settlement determination plane, is represented as a straight line perpendicular to the abscissa axis where the power consumption gradient is always less than the gradient characteristic value, and a straight line perpendicular to the ordinate axis where the dwell time distribution is less than or equal to the dwell characteristic value, together with the abscissa and ordinate axes of the settlement determination plane, forming a rectangular closed region. When the state point is within the billing logic buffer zone, the target vehicle is determined to have met the conditions for a single energy replenishment task, and the current energy billing mode is locked and maintained, while parking billing is temporarily suspended. The resource occupancy area, within the first quadrant of the settlement determination plane, is represented as a semi-open geometric region enclosed by a dwell time distribution greater than the dwell characteristic value and a power consumption gradient less than the gradient characteristic value. When a state point enters the resource occupancy area, it is determined that the physical resource occupancy of the target vehicle in the parking lot has reached the billing trigger condition. Energy billing for the target vehicle is stopped, and physical resource occupancy billing for the parking space is started, which corresponds to the traditional parking billing based on the duration of parking space occupancy. The construction logic of the functional partition is as follows: based on the temporal race condition principle in a distributed Internet of Things environment, the energy replenishment characteristic (power consumption gradient) and the spatial resource occupancy characteristic (dwell time distribution) in an asynchronous operating state are orthogonally decoupled in terms of physical attributes within the settlement determination plane. The gradient characteristic value defines the physical critical line in the energy dimension where the power injection behavior transitions from a continuous fluctuating state to a steady-state silent state, aiming to shield the electromagnetic noise and sampling error of the charging pile's built-in power monitoring module at the critical point of completing a single energy replenishment task.The dwell feature value defines the time threshold for the transition of berth occupancy behavior from physical parking to billable occupancy in the time domain, aiming to offset the reporting delay caused by the inconsistency of communication heartbeat cycles between the visual sensor and the charging management platform.

[0085] Based on the functional area into which the state point falls within the settlement decision plane, a logical transition identifier is generated. This logical transition identifier controls the logical state variable of the settlement process jump and has a low logic level representing the initial reset state and a high logic level representing the settlement trigger state. Specifically, the linear offset of the state point in the vertical direction beyond the dwell characteristic value is calculated and defined as the occupancy intensity factor. The occupancy intensity factor is obtained by subtracting the dwell characteristic value from the dwell time distribution value acquired in real time. Simultaneously, a decision confirmation window is set to evaluate the stability of state switching. The decision confirmation window is set based on the average parking space turnover redundancy time defined according to the commercial scenario to which the parking lot belongs; for example, the decision confirmation window is set to 180. When the occupancy intensity factor is continuously greater than zero and the duration of the continuous value is greater than the duration of the decision confirmation window, the logical transition identifier is switched from the reset state corresponding to the low logic level to the trigger state corresponding to the high logic level. The rising edge transition signal generated by the logical transition flag at the moment of state switching is defined as a logical switching pulse. This logical switching pulse, as a globally unique settlement driving vector, represents the instantaneous action command for billing mode conversion in the time domain. It is used at the command execution level to forcibly stop energy billing for the target vehicle and simultaneously start physical resource occupancy billing for the parking space. See also... Figure 3 This is a schematic diagram of the partition structure of the settlement determination plane provided in an embodiment of the present invention, as shown below. Figure 3 As shown, in the first quadrant of the settlement decision plane, the power consumption gradient is used as the horizontal axis and the dwell time distribution as the vertical axis, dividing the first quadrant into three logically mutually exclusive functional zones. The energy trading zone is defined by a semi-open geometric region whose horizontal axis coordinate is greater than or equal to the gradient characteristic value P. The energy trading zone has no closed boundary line in the positive direction of the horizontal axis. The billing logic buffer is defined by a rectangular closed region enclosed by the horizontal axis, the vertical axis, and decision boundary lines passing through the gradient characteristic value P and the dwell time characteristic value Q respectively, and perpendicular to the corresponding coordinate axes. The resource occupancy zone is defined by a semi-open geometric region whose vertical axis coordinate is greater than the dwell time characteristic value Q and whose horizontal axis coordinate is less than the gradient characteristic value P. The resource occupancy zone also has no closed boundary line in the positive direction of the vertical axis. An example of a solid black circle located inside the resource occupancy zone, i.e., a state point, is provided to exemplarily represent the current real-time position of the target vehicle.

[0086] Step S20, through multi-dimensional feature parameters, constructing a settlement determination plane, and dividing functional zones, solves the problem of insufficient accuracy in switching between energy billing and resource occupancy billing and billing state oscillation caused by slight power fluctuations at the end of the charging period for the target vehicle. This achieves atomicity and financial-grade reliability in billing mode conversion. Specifically, the multi-dimensional feature parameters quantify the physical indicators of the evolution from energy-paid state to resource-occupancy state; the settlement determination plane eliminates false settlements caused by fluctuations in single sensor signals; and the settlement logic state transition drives the instantaneous action commands for stopping energy billing and starting resource occupancy billing.

[0087] Step S30: Construct an asynchronous shadow queue, use a logic switching pulse to drive the asynchronous shadow queue to perform timing reprogramming, obtain a timing-aligned sequence set, perform amount offsetting on the timing-aligned sequence set, and obtain a settlement instruction with settlement atomicity.

[0088] Further, step S30 includes:

[0089] Step S31: Construct an asynchronous shadow queue, and use a logic switching pulse to drive the asynchronous shadow queue to perform timing reprogramming to obtain a timing-aligned sequence set.

[0090] After receiving the logic switching pulse, in order to resolve the timing race condition caused by the different unit sampling period and communication protocol response delay between the monitoring image sequence collected by the visual sensor and the real-time charging power collected by the power monitoring module, and to ensure the strong coupling consistency between energy billing and resource occupancy billing at the moment of contactless payment settlement, an asynchronous shadow queue with non-blocking characteristics is introduced. This aims to establish a virtual timing alignment buffer for the billing flow to be processed, thereby providing fault-tolerant support for eliminating settlement logic misalignment caused by signal drift.

[0091] Specifically, a temporary storage area, defined as an asynchronous shadow queue, is allocated in the computer memory of the integrated charging and stopping management platform for data storage and verification. This asynchronous shadow queue is configured to uniformly receive and store heterogeneous data streams from different sources with varying transmission delays. It combines the real-time charging power provided by the power monitoring module with the physical timestamp corresponding to the real-time charging power to form an energy billing data packet, where the physical timestamp representing the energy status is a timestamp. Similarly, it combines the location occupancy signal provided by the visual sensor with the physical timestamp corresponding to the location occupancy signal to form a resource occupancy billing data packet, where the physical timestamp representing the location occupancy signal represents the spatial status. The asynchronous shadow queue is capable of performing time-sequence alignment and reordering calculations on the physical timestamps carried in the energy billing data packet and the resource occupancy billing data packet, aiming to recreate the real event sequence in the physical world within the virtual logical environment. When a logic switching pulse is triggered as a rising edge signal, the physical moment of generating the rising edge transition signal is defined as the transition moment, and the management platform defines the transition moment of the logic switching pulse as the global synchronization reference point. Using the global synchronization reference point as time zero, timestamps representing spatial state are extracted from the monitoring image sequence provided by the visual sensor, and timestamps representing energy state are extracted from the real-time charging power provided by the power monitoring module.

[0092] The energy billing data packets and resource occupancy billing data packets stored in the asynchronous shadow queue undergo time-series rearrangement. Specifically: the timestamps representing energy status contained in the energy billing data packets are extracted from the asynchronous shadow queue, and the timestamps representing spatial status are extracted from the resource occupancy billing data packets. The extracted timestamps representing energy status and spatial status are uniformly identified as physical timestamps. The set of energy billing data packets arranged in chronological order in the asynchronous shadow queue is defined as the energy billing stream, and the set of resource occupancy billing data packets arranged in chronological order in the asynchronous shadow queue is defined as the resource occupancy billing stream. The physical time at which the energy billing data packets and resource occupancy billing data packets arrive at the management platform through the communication link is defined as the system reception time. The management platform discards the system reception time and instead performs linear rearrangement of the energy billing stream and resource occupancy billing stream according to the order of the physical timestamps. By mapping discrete, asynchronous IoT-collected data, consisting of monitoring image sequences and real-time charging power, back to a continuous sequence of events in the physical world, logical drift caused by communication link transmission delays is corrected. This ensures that the termination action of energy billing and the initiation action of resource occupancy billing are logically aligned in the asynchronous shadow queue, thus providing deterministic timing evidence for achieving settlement atomicity in a distributed asynchronous environment. This yields a time-aligned sequence set characterizing the timing alignment relationship between energy billing and resource occupancy billing.

[0093] Step S32: Calculate the timing deviation compensation factor based on the timing alignment sequence set, and use the timing deviation compensation factor to perform amount offsetting to obtain a settlement instruction with settlement atomicity.

[0094] After obtaining the time-aligned sequence set, in order to eliminate the error amount caused by the communication protocol response delay, where the management platform still maintains the energy billing even though the target vehicle has physically finished recharging, and to ensure the absolute fairness of financial settlement, the state switching deviation under physical time and space is calculated, aiming to output a final settlement instruction with a compensation mechanism.

[0095] Specifically, the physical timestamp corresponding to the energy billing data packet at the end of the energy billing flow is extracted from the time-aligned sequence set and defined as the true termination time; the system reception time corresponding to the energy billing data packet at the end of the energy billing flow is extracted synchronously and defined as the system perception time. The time difference between the system perception time and the true termination time is calculated and defined as the time deviation. The time deviation quantifies the artificial increase in billing time caused by IoT signal drift. The management platform sets a coefficient to quantify the billing weight per unit time and defines it as the billing unit price factor. The billing unit price factor is set based on the tiered electricity price corresponding to the charging pile associated with the target vehicle at the settlement time and the energy service fee standard of the management platform. The purpose is to provide a quantitative benchmark for the time-series correction logic. For example, the billing unit price factor is set to 0.03. The time deviation is multiplied by the billing unit price factor to obtain the time deviation compensation factor. The time deviation compensation factor is used to characterize the energy billing amount that should be offset due to signal delay. The management platform obtains the original deduction value accumulated by the power monitoring module within a single energy replenishment cycle, defined as the initial energy billing total. The atomic settlement amount is obtained by subtracting the time-series deviation compensation factor from the initial energy billing total. The process of obtaining the atomic settlement amount by subtracting the time-series deviation compensation factor calculated from the time-series deviation and the billing unit price factor from the initial energy billing total is defined as amount offsetting. This aims to reduce the inflated energy billing amount caused by communication link transmission delays or system perception delays through physical timestamp restoration. The atomic settlement amount, the physical association identifier corresponding to the target vehicle, and the logical switching pulse are encapsulated to generate a settlement instruction. The management platform asynchronously pushes the settlement instruction to a third-party financial settlement gateway with financial-grade security verification capabilities. The settlement instruction forcibly stops the energy billing operation and simultaneously submits the resource occupancy billing deduction transaction. The settlement instruction ensures the instantaneous and indivisible nature of the billing mode switching at the instruction level, solving the billing oscillation and abnormal deduction problems caused by discrete and asynchronous data composed of monitoring image sequences and real-time charging power.

[0096] Step S30, by constructing an asynchronous shadow queue, performing timing reprogramming, and performing amount offsetting, solves the problems of settlement logic misalignment and inflated billing amounts caused by inconsistent data collection delays of heterogeneous nodes and communication protocol response delays in a distributed asynchronous environment. This achieves final payment instruction generation with settlement atomicity and financial-grade security verification. Specifically, the asynchronous shadow queue establishes a non-blocking virtual alignment buffer for heterogeneous data streams, providing fault-tolerant support to eliminate logic misalignments caused by signal drift; timing reprogramming maps asynchronous IoT data collection back to a continuous sequence of events in the physical world, ensuring that billing termination and activation actions are sequentially linked; and amount offsetting uses timing deviation to calculate a timing deviation compensation factor, reducing the error amount caused by system-perceived delays, ensuring absolute fairness in financial settlement.

[0097] Example 2

[0098] This embodiment, based on Embodiment 1, provides a smart charging stop control system based on IoT-based contactless payment, such as... Figure 4 As shown, it includes:

[0099] Identity verification module: used to acquire monitoring image sequences representing the parking process of target vehicles in the smart parking and charging integrated scenario, extract pixel displacement trajectories from the monitoring image sequences, perform perspective transformation processing on the pixel displacement trajectories to obtain a joint time series set representing the probability of physical causal matching, perform spatiotemporal feature coupling calculation on the joint time series set to obtain the identity confidence, and perform numerical comparison on the identity confidence to obtain the identity verification identifier set;

[0100] The billing decision module is used to extract multi-dimensional feature parameters based on the identity verification identifier set to obtain the power replenishment termination time point, perform cumulative calculation of the duration of the power replenishment termination time point to obtain the dwell time distribution, construct a settlement decision plane based on the dwell time distribution, divide functional areas in the settlement decision plane, and perform settlement logic state transitions according to the functional areas to obtain logic switching pulses.

[0101] Atomic Settlement Module: Used to construct asynchronous shadow queues, use logical switching pulses to drive asynchronous shadow queues to perform timing recompilation, obtain timing-aligned sequence sets, perform amount offsetting on timing-aligned sequence sets, and obtain settlement instructions with atomicity.

[0102] In the identity verification module, the process involves acquiring a sequence of monitoring images representing the parking process of a target vehicle in a smart charging and parking integrated scenario, extracting pixel displacement trajectories from the monitoring image sequence, performing perspective transformation on the pixel displacement trajectories to obtain a joint temporal set representing the probability of physical causal matching, performing spatiotemporal feature coupling calculation on the joint temporal set to obtain the identity confidence score, and performing numerical comparison on the identity confidence score to obtain an identity verification identifier set, including:

[0103] Step S11: Obtain a sequence of monitoring images representing the parking process of the target vehicle in the smart parking and charging integrated scenario, and construct a spatial motion monitoring coordinate system to extract the trajectory of the monitoring image sequence to obtain the pixel displacement trajectory. Perform perspective transformation processing on the pixel displacement trajectory to obtain a multimodal dataset representing the initial state of the parking and charging integrated system.

[0104] Step S12: Perform temporal cross-correlation mapping on the multimodal dataset to obtain a joint time series set representing the probability of physical causal matching;

[0105] Step S13: Perform spatiotemporal feature coupling calculation on the joint time series set to obtain the identity confidence level representing the target vehicle and the charging pile. Perform numerical comparison based on the identity confidence level to obtain the identity verification identifier set.

[0106] In the billing decision module, multi-dimensional feature parameter extraction is performed based on the identity verification identifier set to obtain the power replenishment termination time point. The duration of the power replenishment termination time point is accumulated to obtain the dwell time distribution. A settlement decision plane is constructed based on the dwell time distribution, and functional zones are divided within the settlement decision plane. Settlement logic state transitions are performed according to the functional zones to obtain logic switching pulses, including:

[0107] Step S21: Based on the identity verification identifier set, perform multi-dimensional feature parameter extraction on the target vehicle to obtain the recharge termination time point, and perform cumulative calculation on the recharge termination time point to obtain the residence time distribution representing the resource occupancy intensity.

[0108] Step S22: Construct a settlement decision plane based on the residence time distribution, divide the settlement decision plane into functional partitions, and perform settlement logic state transitions according to the functional partitions to obtain logic switching pulses used to drive atomic switching of billing modes.

[0109] In the atomic settlement module, the construction of an asynchronous shadow queue, the use of a logic switching pulse to drive the asynchronous shadow queue to perform timing reprogramming to obtain a timing-aligned sequence set, and the execution of amount offsetting on the timing-aligned sequence set to obtain a settlement instruction with atomicity, including:

[0110] Step S31: Construct an asynchronous shadow queue, and use a logic switching pulse to drive the asynchronous shadow queue to perform timing reprogramming to obtain a timing-aligned sequence set;

[0111] Step S32: Calculate the timing deviation compensation factor based on the timing alignment sequence set, and use the timing deviation compensation factor to perform amount offsetting to obtain a settlement instruction with settlement atomicity.

[0112] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0113] 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 invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart charging stop control method based on IoT-based contactless payment, characterized in that, The method includes: Acquire a sequence of monitoring images representing the parking process of a target vehicle in a smart parking and charging integrated scenario. Extract pixel displacement trajectories from the monitoring image sequence and perform perspective transformation on the pixel displacement trajectories to obtain a joint time series set representing the probability of physical causal matching. Perform spatiotemporal feature coupling calculation on the joint time series set to obtain the identity confidence score. Perform numerical comparison on the identity confidence score to obtain the identity verification identifier set. Multidimensional feature parameter extraction is performed based on the identity verification identifier set to obtain the power replenishment termination time point. The duration of the power replenishment termination time point is accumulated to obtain the dwell time distribution. A settlement judgment plane is constructed based on the dwell time distribution, and functional partitions are divided in the settlement judgment plane. Settlement logic state transitions are performed according to the functional partitions to obtain logic switching pulses. A temporary storage area for data storage and verification is allocated in the computer memory and defined as an asynchronous shadow queue. The real-time charging power and the physical time point corresponding to the generation of the real-time charging power are combined to form an energy billing data packet. The location occupancy signal and the physical time point corresponding to the generation of the location occupancy signal are combined to form a resource occupancy billing data packet. The time when the energy billing data packet and the resource occupancy billing data packet are received is obtained and defined as the system receiving time. The energy billing data packet and the resource occupancy billing data packet are stored in the asynchronous shadow queue. The transition time of the logic switching pulse is defined as the global synchronization reference point. The physical time point representing the energy state is extracted from the asynchronous shadow queue, and the physical time point representing the space state is extracted from the resource occupancy billing data packet. The physical time point representing the energy state and the physical time point representing the space state are uniformly identified as physical timestamps. When the logic switching pulse is triggered as a rising edge signal, the system receiving time is discarded, and the energy billing data packet and the resource occupancy billing data packet are linearly rearranged according to the order of the physical timestamps to obtain a time-aligned sequence set. All energy billing data packets are filtered from the time-aligned sequence set, while maintaining the chronological order of the physical timestamps, to form an energy billing flow. The physical timestamps corresponding to the energy billing data packets at the end of the energy billing flow are extracted from the time-aligned sequence set and defined as the true termination time. The system receiving time corresponding to the energy billing data packets at the end is extracted and defined as the system sensing time. The time difference between the system's perceived time and the actual termination time is defined as the time deviation. A billing unit price factor is set, and the time deviation compensation factor is obtained by multiplying the time deviation by the billing unit price factor. Obtain the initial total energy billing amount, subtract the timing deviation compensation factor from the initial total energy billing amount to obtain the atomic settlement amount, and encapsulate the atomic settlement amount, the physical association identifier corresponding to the target vehicle, and the logical switching pulse to obtain the settlement instruction.

2. The intelligent charging stop control method based on IoT contactless payment according to claim 1, characterized in that, The method for obtaining the joint time series set includes: The monitoring image sequence is obtained by collecting two-dimensional image data of each sampling frame according to a preset unit sampling period and arranging them in chronological order; A spatial motion monitoring coordinate system is constructed. An image coordinate system with the upper left corner as the origin is established in each frame of two-dimensional image data. Feature anchor points are set for the target vehicle. At the same time, the pixel position coordinates of the feature anchor points in the image coordinate system are obtained. The pixel position coordinates of all sampled frames are connected in time sequence to obtain the pixel displacement trajectory. The perspective transformation of the pixel displacement trajectory is performed using the pre-calibrated homography matrix to obtain the physical position coordinate pairs. The instantaneous velocity amplitude and direction angle are calculated using the physical position coordinate pairs of the current sampling frame and the previous sampling frame. The instantaneous velocity amplitude and direction angle of all sampling frames are arranged in time sequence and combined into the video stream motion vector. The current amplitude sequence output by the charging pile is obtained and defined as a power pulse fingerprint. The power pulse fingerprint and the motion vector of the video stream are encapsulated into a multimodal dataset. Temporal cross-correlation mapping is performed on the multimodal dataset to obtain a joint time series set.

3. The intelligent charging stop control method based on IoT contactless payment according to claim 2, characterized in that, The execution of the time-domain cross-correlation mapping specifically includes: Instantaneous velocity amplitude sequence and current amplitude sequence are extracted from the multimodal dataset. The current amplitude sequence is resampled using the unit sampling period as the reference time axis to obtain the current response sequence. The instantaneous velocity amplitude sequence is a sequence composed of the instantaneous velocity amplitudes corresponding to all sampling frames arranged in time order. The maximum instantaneous velocity amplitude in the instantaneous velocity amplitude sequence and the maximum current amplitude in the current response sequence are normalized and mapped to obtain the motion intensity sequence and energy intensity sequence, respectively. Set a sliding time window and set a time offset with a discrete offset step size of unit sampling period within the sliding time window. For each time offset within the sliding time window, calculate the cumulative sum of the motion intensity sequence and the energy intensity sequence to obtain the cross-correlation intensity value. Arrange the cross-correlation intensity values ​​corresponding to all time offsets within the sliding time window in the order of time offset to form a joint time series set representing the probability of physical causal matching.

4. The intelligent charging stop control method based on IoT contactless payment according to claim 3, characterized in that, The identity confidence level includes: The maximum cross-correlation strength value is searched from the joint time series set and defined as the time-domain matching peak. Extract the direction angles corresponding to multiple consecutive sampled frames after the target vehicle comes to rest from the motion vector of the video stream, calculate the arithmetic mean, and obtain the end pointing angle that represents the final parking position of the target vehicle in the parking space. In the spatial motion monitoring coordinate system, the straight line passing through the origin and pointing into the berth is defined as the berth's central axis, and the angle between the berth's central axis and the coordinate axis is defined as the reference angle. By introducing a time-domain weight component, a multi-dimensional weighted fusion calculation is performed on the time-domain matching peak, end pointing angle, and reference reference angle to obtain the identity confidence degree, which represents the strength of the association between the target vehicle's identity and the physical resources of the charging pile.

5. The intelligent charging stop control method based on IoT contactless payment according to claim 4, characterized in that, The identity verification identifier set includes: Set a logical settlement threshold to determine whether the target vehicle and the charging pile have a unique physical association, and compare the identity confidence level with the logical settlement threshold. If the identity confidence level is greater than or equal to the logical settlement threshold, it is determined that the target vehicle and the charging pile belong to the same physical object, the physical association identifier is output, and the physical association identifier is used to initialize the financial deduction process; If the identity confidence level is less than the logical settlement threshold, it is determined that the target vehicle and the charging pile do not belong to the same physical object, and an identity anomaly identifier is output; the physical association identifier and the identity anomaly identifier are combined to obtain the identity verification identifier set.

6. The intelligent charging stop control method based on IoT contactless payment according to claim 5, characterized in that, The residence time distribution includes: The charging pile is located based on the physical association identifier in the identity verification identifier set, and the real-time charging power output by the charging pile to the target vehicle is obtained using the power monitoring module built into the charging pile. A gradient observation window is set up, which consists of a preset number of sampling points arranged in chronological order, and a unique incremental sampling point index is assigned to each sampling point. The first-order numerical differentiation processing is performed on the real-time charging power within the gradient observation window to obtain the power consumption gradient characterizing the physical deceleration rate of the real-time charging power. Set the energy compensation cutoff power for determining the end of the energy compensation state, and the gradient stability threshold for quantifying the quietness of power change. In a continuous sampling point index sequence with multiple consecutive sampling points, when the real-time charging power is less than the energy compensation cutoff power and the absolute value of the power consumption gradient is less than the gradient stability threshold, the physical time point associated with the first sampling point index in the continuous sampling point index sequence is recorded as the energy compensation termination time point. Starting from the point when the refueling ends, the location occupancy signal representing the physical occupation of the berth by the target vehicle is obtained using a visual sensor. The cumulative physical duration between the point when the refueling ends and the physical time point corresponding to the current real-time sampling sequence index is calculated to obtain the dwell time distribution.

7. The intelligent charging stop control method based on IoT contactless payment according to claim 6, characterized in that, The logic switching pulse includes: A two-dimensional coordinate system is constructed with the power consumption gradient as the horizontal axis and the residence time distribution as the vertical axis. This system is defined as the settlement judgment plane. Gradient feature values ​​are set to define the degree of energy consumption, and residence feature values ​​are set to define the duration of resource occupation. Using gradient eigenvalues ​​and dwell eigenvalues, three logically mutually exclusive functional zones are divided in the settlement decision plane: the energy trading zone, the billing logic buffer zone, and the resource occupation zone. Within the settlement determination plane, coordinate points determined based on real-time power consumption gradients and dwell time distributions are defined as state points. When a state point enters a resource occupancy area, a logic transfer identifier is generated. The logic transfer identifier is configured to have a low logic level representing the initial reset state and a high logic level representing the settlement trigger state, and is initially set to a low logic level. The linear offset of the state point in the vertical direction beyond the dwell characteristic value is calculated and defined as the occupancy strength factor. When the occupancy strength factor is continuously greater than zero within the preset judgment confirmation window, the logic transfer flag is switched from low logic level to high logic level. The rising edge transition signal generated by the logic transfer flag at the instant of switching from low logic level to high logic level is defined as the logic switching pulse, and the physical moment when the rising edge transition signal is generated is defined as the transition moment.

8. A smart charging stop control system based on IoT-based contactless payment, used to implement the smart charging stop control method based on IoT-based contactless payment as described in any one of claims 1-7, characterized in that, The system includes: Identity verification module: used to acquire monitoring image sequences representing the parking process of target vehicles in the smart parking and charging integrated scenario, extract pixel displacement trajectories from the monitoring image sequences, perform perspective transformation processing on the pixel displacement trajectories to obtain a joint time series set representing the probability of physical causal matching, perform spatiotemporal feature coupling calculation on the joint time series set to obtain the identity confidence, and perform numerical comparison on the identity confidence to obtain the identity verification identifier set; The billing decision module is used to extract multi-dimensional feature parameters based on the identity verification identifier set to obtain the power replenishment termination time point, perform cumulative calculation of the duration of the power replenishment termination time point to obtain the dwell time distribution, construct a settlement decision plane based on the dwell time distribution, divide functional areas in the settlement decision plane, and perform settlement logic state transitions according to the functional areas to obtain logic switching pulses. Atomic Settlement Module: Used to construct asynchronous shadow queues, use logical switching pulses to drive asynchronous shadow queues to perform timing recompilation, obtain timing-aligned sequence sets, perform amount offsetting on timing-aligned sequence sets, and obtain settlement instructions with atomicity.

Images