Parking space management system and method based on sensor technology

Abstract

The present application relates to the technical field of urban parking resource management and traffic information service, in particular to a parking space management system and method based on sensor technology, the parking space management system based on sensor technology comprises sensing collection, transmission convergence, trusted situation assessment, induced risk quantification, output convergence evaluation, hierarchical convergence regulation and business output module, through the introduction of trusted situation assessment and induced risk quantification module, the arrival time, receipt information and time are collected, the parking space state crosstalk caused by time delay synchronization failure is inhibited, and the statistical caliber consistency is improved; adaptive output convergence is realized, real-time and induced safety are considered; through the output of link trusted evaluation and risk quantification results, the explainability and traceability of operation decision are enhanced, the management efficiency is improved, and the decision misleading risk is reduced.

Description

Technical Field

[0001] This invention relates to the field of urban parking resource management and traffic information service technology, and more specifically, to a parking space management system and method based on sensor technology. Background Technology

[0002] In sensor-based berth status acquisition and network transmission applications, berth events typically carry the acquisition time and are forwarded to the platform via a gateway to form arrival times. The platform then uses this information to generate berth queries, guidance information, and statistical results such as heat maps and occupancy rates. This type of architecture can support real-time services well when network conditions are relatively stable. However, in the complex wireless environment of urban roads and multi-hop aggregation scenarios, multiple acquisition points may experience inconsistent link delays and clock deviations. For example, within the same time window, one set of berth events arrives with a delay of about ten seconds, while another set arrives with a delay of tens of seconds due to backhaul congestion or batch retransmission. This causes the platform to aggregate events using arrival times or insufficiently corrected timestamps, misjudging that multiple berths switched simultaneously at the same time. This leads to statistical distortions such as abrupt changes in heat maps and amplified or diluted occupancy rates, which in turn misleads operational decisions regarding scheduling, inspection, and toll collection.

[0003] The problem essentially stems from latency asynchrony and time base inconsistency in distributed data acquisition scenarios. On one hand, time synchronization in IoT deployments faces adaptation and performance challenges under resource constraints and heterogeneous networks. Traditional network time protocols struggle to meet consistent time requirements in some scenarios, while precise time protocols, though improving accuracy, are still affected by link asymmetry, wireless characteristics, and device conditions, leading to fluctuations in synchronization quality during operation. On the other hand, statistical aggregation for event streams inherently faces issues of out-of-order and late data. Without a mechanism to manage the difference between event time and processing time, late events can cause result write-back and jumps in window statistics. Therefore, the stream processing field commonly uses mechanisms such as watermarks to define lateness tolerance and balance real-time performance with completeness. Existing parking applications often employ simple and low-cost methods such as fixed-delay publishing, simple debouncing, or updating according to arrival order, which can improve the user experience with slight jitter. However, when the link latency of different parking spaces exhibits directional bias, and batch backfilling and receipt anomalies occur simultaneously, these methods often fail to balance real-time performance and statistical consistency, easily leading to the aforementioned parking space status crosstalk phenomenon.

[0004] Based on this, it is necessary to introduce joint evaluation and hierarchical control of collection time, arrival time and receipt information on the basis of the existing event reporting framework, so that the output can adaptively converge with the reliability of the link, and to impose more robust constraints on statistical and induced calibers under high uncertainty conditions, so as to reduce the risk of heat map and occupancy statistics being amplified by asynchronous time delay. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a berth management system and method based on sensor technology.

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

[0007] A berth management system based on sensor technology includes: The sensing and acquisition module is used to collect occupancy status data of each berth and generate berth events. The berth events include at least the berth identifier and the collection time. The transmission aggregation module is used to receive berth events and forward them to the platform side. When a berth event arrives at the platform side, the arrival time is written and gateway receipt information is generated. The Trusted Situation Assessment Module is used to calculate the link timing feature component and the event consistency feature component based on the acquisition time, arrival time and gateway receipt information within a sliding time window, and to obtain the Trusted Situation Index by fusing the link timing feature component and the event consistency feature component. The induced risk quantification module is used to retrieve road risk parameters from the road risk parameter library based on berth identification, and to obtain the induced risk loss function value based on the road risk parameters; The output convergence evaluation module is used to fuse the credible situation index and the induced risk loss function value to obtain an adaptive output convergence index. The graded convergence control module is used to execute graded control strategies based on the level of the adaptive output convergence index and apply the graded control strategies to the berth status output. The business output module is used to output berth query and guidance information, as well as statistical analysis and management information, and adjust the output berth status according to the hierarchical control strategy executed by the hierarchical convergence control module.

[0008] Furthermore, the sensing and acquisition module performs threshold determination or classification identification on the berth occupancy status data to obtain the berth occupancy status, and generates a berth event when the occupancy status changes or the reporting trigger condition is met.

[0009] Furthermore, the transmission aggregation module includes a gateway node; when the gateway node receives a berth event, it records the platform-side timestamp as the arrival time and generates gateway receipt information. The gateway receipt information includes at least the berth identifier, collection time, arrival time, event sequence number, and check code or signature, and is then transmitted back to the sensing and acquisition module or edge node.

[0010] Furthermore, the link timing feature components include at least: delay spectrum feature components, asymmetric drift components, and out-of-order consistency entropy components; the event consistency feature components include at least: causal violation intensity components.

[0011] Furthermore, the time delay spectrum characteristic components are characterized by the single-trip time delay quantile, time delay jitter intensity, and time delay skewness; the asymmetric drift component is characterized by the statistical difference between uplink and downlink time delays and their rate of change over time; the out-of-order consistency entropy component is characterized by the out-of-order ratio, out-of-order span, and batch replenishment intensity; and the causal violation intensity component is characterized by the number of unreachable transfers, the number of reverse transfers, and the number of short-term back-and-forth jitters during berth status transitions.

[0012] Furthermore, the risk quantification module locates the road risk record corresponding to the berth in the road risk parameter database based on the berth identifier, and extracts road risk parameters from the road risk record; among which, the road risk parameters include at least road attribute parameters, traffic carrying capacity threshold, and operational risk level.

[0013] Furthermore, the tiered regulatory strategy includes at least the following: a) When the adaptive output convergence index is at the first level, the output is based on the berth status of the real-time value; b) When the adaptive output convergence index is at the second level, the berth status is output after performing asymmetric time delay compensation on the berth event; c) When the adaptive output convergence index is at the third level, the berth status is output after reordering the causal consistency constraints and arbitrating the conflicts for the berth events. d) When the adaptive output convergence index is at level four, the berth event reporting method is switched from reporting one by one to reporting status snapshots, and the berth status is output after edge merging of repeated jitter events.

[0014] Furthermore, when the adaptive output convergence index is at level four, the driver-side guidance information and the operation-side heat map output are switched from being based on real-time values ​​to being based on conservative values. The conservative values ​​are the output values ​​of available parking spaces after applying credible constraints based on the parking space status.

[0015] Furthermore, the berth management method based on sensor technology includes the following steps: Step 1: Collect occupancy status data for each berth and generate berth events; Step 2: Receive berth events and forward them to the platform side. When a berth event arrives at the platform side, record the arrival time and generate gateway receipt information. Step 3: Within the sliding time window, based on the acquisition time, arrival time, and gateway receipt information, calculate the link timing feature component and the event consistency feature component respectively, and obtain the trustworthy situation index by fusing the link timing feature component and the event consistency feature component. Step 4: Retrieve road risk parameters from the road risk parameter database based on berth signs, and calculate the induced risk loss function value based on the road risk parameters; Step 5: Obtain the adaptive output convergence index by fusing the credible situation index and the induced risk loss function value, and implement a graded control strategy according to the level to which the adaptive output convergence index belongs. Step 6: Output the berth status according to the graded control strategy and adjust the berth status according to the strategy.

[0016] Compared with the prior art, the present invention has the following beneficial effects: This invention introduces a reliable situation assessment module on the platform side, which jointly collects time, arrival time, and gateway feedback information within a sliding time window, forming link timing feature components and event consistency feature components respectively, and then fuses them to obtain a reliable situation index. The output convergence assessment module and the hierarchical convergence control module then implement hierarchical constraints and consistency processing on the berth status output. Therefore, when different sensors / links experience asynchronous delays (e.g., one group delays by 10 seconds, another by tens of seconds, or even retransmission), the probability of the platform misjudging "asynchronous arrival" as "multiple berths switching simultaneously at the same time" can be reduced. This significantly mitigates statistical distortions such as sudden changes in heatmaps and amplification / dilution of occupancy rates, improving the reliability and continuity of indicators such as occupancy and turnover rates on the operations side.

[0017] The system employs a risk quantification module to retrieve road risk parameters from the berth identification database and calculate the risk loss function value. This value is then fused with a credibility index to obtain an adaptive output convergence index. A tiered convergence control module then implements differentiated strategies (such as compensation, reordering, merging, conflict arbitration / more conservative release, etc.) based on tier level and applies them to the berth status output. This approach prioritizes real-time performance when link credibility is high. When uncertainties increase due to link disorder, batch replenishment, or abnormal receipts, the system automatically tends towards "converged output," reducing user confusion and unnecessary detours. Furthermore, in areas or scenarios with high road risk, a more cautious output approach mitigates potential safety and congestion risks from "erroneous guidance," achieving a dynamic balance between service experience and traffic safety.

[0018] By integrating the link credibility assessment results (credibility situation index), the induced risk quantification results (loss function value), and the execution effect of the hierarchical control strategy into the berth query / guidance and statistical analysis / management information output of the "business output module," operators can distinguish between "real supply changes" and "data fluctuations caused by link latency asynchrony / synchronization failure," avoiding over-response or incorrect scheduling when heatmaps are abnormal or occupancy rates suddenly jump. At the same time, gateway receipts and time-series characteristics provide a basis for review, facilitating the identification of problem sources (sensor jitter, link congestion, retransmission mechanism triggering, etc.), thereby supporting closed-loop improvement of management actions such as inspection strategy optimization, gateway parameter adjustment, zoning management, and toll auditing, reducing the risk of decision-making being misled by abnormal data. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the system structure of the present invention; Figure 2 This is a flowchart of the trusted situation assessment module in this invention; Figure 3 This is a schematic diagram illustrating the combination of the risk quantification module and the output convergence evaluation module. Detailed Implementation

[0020] Example 1: Refer to Figures 1 to 3 A berth management system based on sensor technology includes: The sensing and acquisition module collects occupancy status data for each berth and generates berth events. Each berth event includes at least a berth identifier and the acquisition time. Its main function is to collect occupancy status data for each berth and generate berth events. Berth events are the fundamental data unit of the berth management system and must include at least a berth identifier and the acquisition time. This data is acquired in real time through sensors or other detection devices, ensuring timely reflection of berth occupancy and providing accurate data support for subsequent scheduling, evaluation, and management. The efficiency of the sensing and acquisition module directly affects the response speed and accuracy of the entire system, playing a crucial role in management decision-making and resource optimization.

[0021] In one specific implementation, the sensing and acquisition module performs threshold determination or classification identification on the berth occupancy status data to obtain the berth occupancy status, and generates a berth event when the occupancy status changes or when the reporting trigger condition is met. This can be implemented according to the following steps: For each berth, the original sensor sampling sequence is preprocessed and features are constructed. The preprocessing includes at least removing abnormal spikes, smoothing filtering, and time alignment. Occupancy discrimination features are calculated within a unified sampling period. The occupancy discrimination features include one or more of the following: magnetic field disturbance amplitude, geomagnetic vector change rate, infrared obstruction duty cycle, millimeter wave echo energy, or ultrasonic ranging stability. Taking geomagnetic acquisition as an example, the background feature distribution of the berth can be established in the absence of a vehicle baseline, and a significant shift occurs when a vehicle enters, thereby making the occupancy discrimination features separable. Based on the obtained occupancy discrimination features, threshold determination or classification recognition is performed to obtain the berth occupancy status. The threshold determination can use dual thresholds and hysteresis logic to suppress critical jitter. For example, an entry threshold and an exit threshold are set, and the status is switched only after a preset number of consecutive samplings are met. The classification recognition can use supervised learning or rule fusion to map the occupancy discrimination features to the occupancy status. For example, the amplitude and rate of change of geomagnetic disturbance and the infrared occupancy duty cycle are jointly input into the classifier to output occupancy or vacancy, and confidence can be introduced for subsequent trigger control. Taking the fluctuation of ultrasonic ranging caused by rainwater accumulation as an example, the addition of stability features and confidence thresholds can avoid misjudgment. Based on the obtained berth occupancy status, event triggering and event generation are performed. When the occupancy status changes from idle to occupied or from occupied to idle, a status change report is triggered. Alternatively, a supplementary report is triggered when the occupancy status remains unchanged but the reporting trigger conditions are met. The reporting trigger conditions include, for example, periodic heartbeats, confidence levels below a preset threshold, consecutive jitter counts reaching a preset threshold, or reaching the maximum silent duration. After triggering, a berth event is generated. The berth event at least includes the berth identifier and collection time, and can further include the occupancy status and event sequence number for subsequent consistency verification and out-of-order processing. For example, when there is dense traffic near the entrance of a large commercial district, periodic heartbeats and status change triggering can be combined to enable the platform to promptly detect status changes and maintain status traceability even when there are no changes for a long time. This ensures real-time performance while reducing invalid reports and improving data reliability.

[0022] The transmission aggregation module receives berth events and forwards them to the platform. Upon arrival at the platform, it writes the arrival time and generates gateway receipt information. Its task is to receive berth events generated by the sensing and acquisition module and forward them to the platform for further processing. During event transmission, the platform records the arrival time and generates gateway receipt information, including key information such as berth identifier, acquisition time, arrival time, and event sequence number. The generation and transmission of gateway receipt information helps the system track the data flow path and timeliness, thereby enabling monitoring and verification of the data transmission process. This module ensures accurate and timely data transmission from edge nodes to the core platform, preventing data loss or delay.

[0023] In one specific implementation, the transmission aggregation module includes a gateway node. When the gateway node receives a berth event, it records the platform-side timestamp as the arrival time and generates gateway receipt information. The gateway receipt information includes at least the berth identifier, collection time, arrival time, event sequence number, and check code or signature, and is then sent back to the sensing and acquisition module or edge node. This can be implemented according to the following steps: The gateway node performs access verification and clock anchoring processing on the received berth events. Access verification includes at least checking the integrity of the berth identifier, collection time, and event sequence number, as well as checking the sequence number continuity. After passing the verification, the arrival time is generated using the unified time source on the platform side. The arrival time and the original collection time are combined to form a double timestamp pair to ensure that the collection side time and the arrival side time can be distinguished in the future. During this process, the receiving channel identifier and the number of retransmissions can be further recorded as auxiliary diagnostic information to enhance the interpretability of link jitter and out-of-order events. The gateway node generates gateway receipt information based on the obtained dual timestamp pair. The gateway receipt information includes at least the berth identifier, collection time, arrival time, event sequence number, and check code or signature. The check code can be calculated by concatenating the berth identifier, collection time, arrival time, and event sequence number in a preset order. The signature can be obtained by signing the same concatenated result with the platform side key, thereby achieving joint constraints on the integrity of the receipt content and the credibility of the source. To improve robustness to batch out-of-order scenarios, an event sequence number window prompt can be added to the receipt, such as indicating the offset of the current event sequence number relative to the most recent valid sequence number, so that the collection side or edge nodes can perform bounded rearrangement and duplicate suppression. The gateway node sends the generated gateway receipt information back to the sensing and acquisition module or edge node, and performs receipt confirmation and anomaly closed-loop processing on the backhaul side. The receipt confirmation includes at least a backhaul success mark and a timeout retransmission strategy. When a backhaul failure or checksum inconsistency occurs, a retransmission request or state snapshot request for the corresponding parking space event is triggered to avoid long-term deviation of the platform side state due to single-point packet loss. For example, when the wireless coverage of an underground parking lot is unstable and the uplink is suddenly congested, the same parking space event may arrive with a delay or repeatedly. The gateway node records the arrival time and carries the event sequence number and signature in the receipt, so that the acquisition side can identify duplicate receipts and suppress duplicate reporting. At the same time, the platform side can use the difference between the acquisition time and the arrival time to characterize the link latency characteristics, so that the event link remains traceable, verifiable and has self-recovery capability even in high interference scenarios.

[0024] The Trusted Situation Assessment module calculates link timing characteristic components and event consistency characteristic components based on acquisition time, arrival time, and gateway feedback information within a sliding time window. It then fuses these components to obtain a Trusted Situation Index. The purpose is to analyze the acquired data and assess its reliability and consistency. Link timing characteristics include network performance metrics such as latency and jitter, reflecting the delay and fluctuation of data during transmission. Event consistency characteristics compare the consistency between actual and expected events to determine if there are errors or anomalies in the data. Through comprehensive analysis of these characteristics, the Trusted Situation Assessment module outputs a Trusted Situation Index, providing a basis for subsequent system decisions and ensuring the system can respond appropriately under unstable or unreliable conditions.

[0025] In one specific implementation, within a sliding time window, calculating the link timing feature component and the event consistency feature component based on the acquisition time, arrival time, and gateway receipt information can be carried out according to the following steps: Construct a sliding time window and complete event alignment. Sort berth events corresponding to the same berth identifier according to the collection time to form a collection sequence. At the same time, extract the corresponding arrival time, event number and check code or signature based on the gateway receipt information to form a pair of records of collection time and arrival time. Within the sliding time window, remove records that fail to verify the check code or signature and duplicate records with obvious backsliding event numbers that cannot be explained, so as to ensure that subsequent calculations are based on reliable events. The delay spectrum characteristic component and the asymmetric drift component are calculated. The delay spectrum characteristic component is obtained by statistically analyzing the single-trip delay of each record. The single-trip delay is obtained by subtracting the acquisition time from the arrival time. Within the sliding time window, the quantile value, fluctuation intensity, and distribution skewness of the single-trip delay are extracted to characterize short-term congestion, sudden jitter, and long-tail delay. The asymmetric drift component is obtained by comparing the statistical differences and their time change rates of transmission in different directions. Specifically, the arrival time of the receipt to the acquisition side can be recorded on the backhaul path of the gateway receipt information. Combined with the acquisition time reported by the acquisition side and the arrival time recorded by the platform side, the statistical characteristics of the uplink delay and downlink delay are estimated respectively. Then, the difference between the two and its rate of change with the sliding time window are calculated to reflect the drift and offset of the link in different directions. The out-of-order consistency entropy component is calculated. An arrival sequence is constructed based on the event sequence number and arrival time. The relative order of the arrival sequence and the collection sequence is compared. The out-of-order ratio, out-of-order span, and batch replenishment intensity are statistically analyzed. The out-of-order ratio describes the frequency of inconsistency between the arrival order and the collection order. The out-of-order span describes the maximum sequence number distance spanned by out-of-order events. The batch replenishment intensity describes the scale and density of the cluster of re-sending events that arrive in a short period of time. The above statistics are mapped to entropy measures to reflect the dispersion of the out-of-order state. The causal violation intensity component is calculated as the event consistency feature component. A reachability state graph is established based on berth occupancy state transition constraints under the same berth identifier. Within a sliding time window, the state evolution of berth events is checked according to the acquisition time. The number of unreachable transitions, reverse transitions, and short-term jitter are counted. The number of unreachable transitions characterizes abnormal jumps in the violation state graph; the number of reverse transitions characterizes state regressions opposite to common sense; and the number of short-term jitters characterizes jitters that repeatedly travel between occupied and idle states within a preset short-term threshold. For example, in densely built-up commercial areas, wireless links are prone to sudden congestion and... Batch retransmission results in long-tail delays and concentrated retransmissions for the same berth event on the platform side. At this time, the obtained delay spectrum feature components will show increased quantile values ​​and skewness. The obtained out-of-order consistency entropy component will increase with the increase of the retransmission cluster. At the same time, if the vehicle has not actually left but the sensor is misjudged for a short time due to strong electromagnetic interference, the obtained causal violation intensity component will increase due to the increase of short-term reciprocating jitter. Thus, the link timing feature component and the event consistency feature component can distinguish and quantify transmission anomalies and state anomalies within the same sliding time window, providing a fully public and reproducible calculation basis for the subsequent formation of the credibility situation index.

[0026] In one specific implementation, the calculation of the time delay spectrum characteristic component, asymmetric drift component, out-of-order consistency entropy component, and causal violation intensity component can be performed according to the following steps: Calculate the characteristic components of the delay spectrum. Obtain the one-way delay for each berth event, which is obtained by subtracting the acquisition time from the arrival time. Perform statistical analysis on the one-way delay of all events within a sliding time window. Specifically, by calculating the quantile values ​​of the delay (e.g., 50th and 90th percentiles), the delay situation of berth events within a given time window can be reflected. The intensity of delay jitter can be represented by calculating the standard deviation or variance of the one-way delay, which helps to assess the delay fluctuations in the network. Delay skewness measures the asymmetry of the data by calculating the skewness of the delay distribution. The greater the skewness, the more uneven the delay. This usually occurs when there are sudden changes in network load or bottlenecks, such as during peak business hours when delay skewness increases. Calculate the asymmetric drift component. By comparing the statistical differences between uplink and downlink delays, calculate the average, standard deviation, and trends of the uplink and downlink delays respectively. Represent the difference between uplink and downlink delays as a drift value, reflecting the performance differences of the network link. Furthermore, by calculating the rate of change of this difference over time, the rate of change of delay drift can be obtained, thus revealing potential problems in the link. For example, in intercity communication links, significant differences in uplink and downlink delays may occur due to imbalances between operator networks, thereby affecting the stability of data transmission. Calculate the out-of-order consistency entropy component. Sort the received berth events according to their arrival time and acquisition time, and calculate the out-of-order ratio, i.e., measure the proportion of events that are out of order among all arriving events. Out-of-order span is described by calculating the span of the largest out-of-order sequence; a larger out-of-order span usually indicates a longer delay and backtracking in data transmission. Batch backtracking intensity measures the degree to which multiple events arrive in a short period of time. These events may be caused by network latency or retransmissions. The greater the intensity, the more batch backtracking events exist in the link, which may lead to increased system response latency. Calculate the causal violation intensity component. Based on the berth state transition diagram, determine the effective state transition path for each berth and count the number of unreachable transitions (i.e., the event flow cannot occur according to the predetermined path), the number of reverse transitions (i.e., the state transitioned in the opposite direction to the expectation), and the number of short-term back-and-forth jitters (i.e., the berth state changes frequently in a short period of time). The count values ​​of these abnormal events reflect the inconsistencies in berth state transitions. The higher the value of the causal violation intensity component, the more frequent the abnormal state transitions in the system. For example, during peak traffic periods, berth states may change frequently due to sensor errors or external interference, leading to a significant increase in the causal violation intensity component. Through the above calculations, the latency fluctuations, link instability, data out-of-order issues, and state consistency problems in the berth state management system can be comprehensively and accurately reflected, thus providing detailed quantitative basis for system reliability assessment and optimization.

[0027] For the same berth marker within the sliding time window The berth incident was recorded at the time of data collection. The platform-side arrival time is The event number is Define one-way delay as: If the verification code or signature verification fails in the gateway receipt, or if the event sequence number shows an inexplicable and obvious rollback, the record will be marked as untrustworthy and removed from subsequent statistics.

[0028] The sliding time window is denoted as The window length is Step size is Without limiting the possible unique values, a set of enforceable deterministic rules is given: Let To ensure a uniform sampling period on the acquisition side, then ;in Pick , Pick ,and Desirable Seconds or longer; thus ensuring stable estimation of short-term statistics even in congested, retransmission, and jitter scenarios.

[0029] Time delay spectrum characteristic components: within the time window Internal collection of one-way delay sets Define quantile values ​​(e.g., median and high quantile): ; Define the delay jitter intensity (standard deviation): ; Define the time delay skewness (skewness): ;in To prevent division by zero of extremely small positive numbers, the above three factors are combined to obtain the time delay spectrum characteristic components: ;in This represents normalization, where the weights satisfy... The weights The median of the corresponding time delay spectral features, Corresponding to the high quantile, Corresponding to latency jitter, This corresponds to the latency skew. In practical applications, if the network environment is relatively stable and latency fluctuations are small, the skewness can be appropriately reduced. and Increase the weight and The weights are adjusted to focus more on the median and high quantile of latency, which more accurately reflects the synchronicity of berth status and normal fluctuations in link latency. For example, suppose... , , , This allocation ensures that, in most cases, the delay spectrum component has a higher weight, enabling the system to effectively assess the timing consistency of berth events under low latency conditions; while the lower weights for delay jitter and skew are suitable when the network is in good condition, reducing sensitivity to delay fluctuations.

[0030] Asymmetric drift component: If the arrival time of the acknowledgment can be obtained on the acquisition side. $, then the approximate delays for uplink and downlink are defined as follows: ; within the time window Calculate the mean for each part. And define directional differences: To reflect the rate of change of drift as the window slides, let the directional differences between two adjacent windows be respectively... , Define the rate of change of drift: Based on this, the asymmetric drift component is defined as follows: ;in Example weight assignment: Assigning a high weight to the mean difference in latency statistics indicates that in most scenarios, the balance of the link (i.e., the difference between uplink and downlink) is crucial to system stability, especially when traffic flow is relatively stable. Assigning a lower weight to the rate of change of drift indicates that the impact of the rate of change of drift is smaller in a relatively stable network environment, but it is still a key measure of system flexibility, especially in response to emergencies.

[0031] Out-of-order consistent entropy component: within the time window Within, the collection sequence is obtained according to the collection time. Arrival sequence order obtained according to arrival time Define the disorder ratio:

[0032] Define out-of-order span: ; Define batch replenishment strength: ; in For the maximum bucket count, Let's count the average buckets. Based on the above statistics, define entropy: ; in These are the normalized values ​​of the out-of-order ratio, out-of-order span, and batch replenishment strength, respectively. Example weight assignment: Assigning a higher weight to the out-of-order ratio indicates that the system pays more attention to the consistency of the order of events during normal operation. A high out-of-order ratio may indicate that the system is unstable or abnormal. Assigning a moderate weight to the out-of-order span indicates that when events are out of order, the span also has a certain impact, but it is not as important as the out-of-order ratio. Assigning a lower weight to batch replenishment strength means that the concentrated arrival of events in a short period of time has a smaller impact on system stability, but it still needs to be considered to avoid excessively concentrated load.

[0033] Causal violation intensity component: Let the set of berth occupancy states be... The set of allowed state transitions is Define the number of unreachable transitions: ; Define the number of reverse transitions: ; Define the number of short-term reciprocating jitters: ; Based on this definition, the causal violation intensity component is: ;in Example weight assignment: Assign a higher weight to unreachable transitions because disallowed state transitions usually indicate a serious logical error in the system, which has a significant impact. Assigning appropriate weights to backtransfers can affect the normal operation of the system, but this can usually be avoided by adjusting the strategy. Assigning lower weights to short-term cyclical jitter is because such short-term fluctuations may be caused by temporary changes in network conditions and do not necessarily reflect the overall stability of the system.

[0034] In one specific implementation, the process of calculating the trustworthiness index based on the fusion of link time-series feature components and event consistency feature components can be carried out according to the following steps: The link-time characteristic components and event consistency characteristic components are standardized. Specifically, the link-time characteristic components include delay spectrum characteristic components, asymmetric drift components, and out-of-order consistency entropy components. Each characteristic component is converted into a standardized value by calculating its mean and standard deviation within a sliding time window, ensuring that all characteristic components have consistent dimensions and eliminating unit differences. The causal violation intensity component in the event consistency characteristic components is also standardized using the Z-score standardization method, so that all characteristic components can be weighted and synthesized on the same quantization scale, avoiding the excessive influence of some features due to their large dimensions on the final result. Through this standardization step, all characteristic components are ensured to have the same importance and comparability during fusion. Based on the standardized feature components, a weighted model is designed to fuse the various feature components. The weighting factors are optimized using machine learning algorithms or expert knowledge, and adjusted according to the contribution of different features to the credibility index. For example, the latency spectrum feature component can be learned in a data-driven manner to determine its weight under specific environments; if network latency fluctuations are large and significantly impact system stability, its weight is higher. Meanwhile, the causal violation intensity component in the event consistency feature component is given a higher weight based on the actual abnormal state transitions, especially in high-risk scenarios where the contribution of the causal violation intensity component is greater. After weighting, each standardized feature component is multiplied by its corresponding weight, and then summed to obtain the fused credibility index. This index, by considering the combined performance of link stability and event consistency, can reflect the credibility of the system in real time. Based on the fused Trust Situation Index, subsequent anomaly detection and decision support are performed. The Trust Situation Index is a dynamically changing value; a higher value indicates a more reliable current link and event status, while a lower value indicates a greater risk to the system. By setting threshold ranges, the Trust Situation Index can be classified. When the index falls below a certain preset threshold, a risk warning is triggered, and corresponding error correction or control measures are taken. For example, in a busy city parking lot, if the calculated Trust Situation Index is lower than a preset safety threshold, the system can automatically adjust the parking space allocation strategy to reduce data transmission pressure or switch to a backup link to avoid system failures or data loss caused by link instability. Through this fusion process, the Trust Situation Index not only dynamically reflects the current operating status of the system but also provides real-time risk assessment basis for decision-making, effectively improving the stability and reliability of the system.

[0035] To avoid the non-reproducible nature of the fusion due to differences in dimensions, "in-window robust normalization" or "historical baseline normalization" is uniformly applied to all the above statistics. A set of feasible implementation methods is given: for any original value of a feature... Take the median of the historical baseline segment. Interquartile range Define robust normalization:

[0036] in As a truncation function, C takes The historical baseline segment can be taken as "data from the same period of the most recent several hours to several days", and is updated on a daily / weekly cycle, thus ensuring that the parameter acquisition method is clear and feasible.

[0037] Trustworthy Situation Index: This index fuses link timing characteristic components with event consistency characteristic components. Definition of Link Timing Composite Quantity: Example weight allocation: : Assign high weight to spectral features because spectral features can reflect the overall frequency distribution of the system. For stable links, frequency characteristics are crucial to system performance. Assigning appropriate weights to drift features, which reflect the trend of link latency changes, especially in long-term evaluations, drift features can reveal potential system instability factors. Assign a low weight to the entropy feature, which mainly measures the time delay uncertainty of the system. In short-term or relatively stable network environments, its impact is relatively small.

[0038] Define the credibility situation index: The method for obtaining weights provides at least one feasible rule: the initial weights can be given by expert experience (e.g., equal distribution), and the loss function can be constructed offline or online during operation based on "in-window receipt anomaly rate, conflict arbitration rate, and manual sampling consistency rate"; if learning is not enabled, the default weights can also be used.

[0039] The induced risk quantification module retrieves road risk parameters from a road risk parameter database based on berth markers and then calculates the induced risk loss function value based on these parameters. Road risk parameters typically include road attribute parameters (such as road width and traffic flow), traffic carrying capacity thresholds (such as the road's maximum carrying capacity), and operational risk levels (such as assessing road risk levels based on historical data). Through the calculation of these parameters, the induced risk quantification module can assess potential risks caused by changes in road conditions, traffic flow fluctuations, and other factors, providing a quantitative basis for subsequent risk management and scheduling decisions.

[0040] In one specific implementation, the process of locating the road risk record corresponding to the berth in the road risk parameter database based on the berth identifier and extracting the road risk parameters can be carried out according to the following steps: A mechanism for linking berth markers with road risk records is established. Specifically, berth markers are mapped to road segment markers and spatial index keys. The spatial index key is composed of at least the road name, road direction, road segment start and end mileage, and the lateral position of the berth relative to the road centerline. Based on this, a unique or highest-priority road risk record is retrieved from the road risk parameter database. When multiple candidate road risk records are retrieved, disambiguation is performed according to the rules of road level priority, minimum spatial distance, and latest time validity period to ensure that the positioning results are stable and reproducible. Road risk parameters are extracted from the located road risk records. These parameters include at least road attribute parameters, traffic capacity thresholds, and operational risk levels. Road attribute parameters describe the inherent and variable properties of the road and include at least one or more of the following: number of lanes, lane width, speed limit, intersection density, proportion of bus lanes, sight-limiting factor, and congestion-prone time period label. Traffic capacity thresholds describe the upper limit of the road's capacity under different operating conditions and include at least one or more of the following: unit time capacity threshold, queue length threshold, saturation threshold, and bottleneck triggering threshold. Operational risk levels describe the strength of the road's risk under current or typical conditions and can be multi-level discrete levels with confidence levels and applicable scenario labels to distinguish between normal and sudden risks in subsequent calculations. Consistency verification and adaptive completion are performed on the extracted road risk parameters. Specifically, the verification includes at least the physical rationality verification of road attribute parameters, the interval constraint verification of traffic carrying capacity thresholds, and the time-period consistency verification of operational risk levels. When parameters are missing or conflicting, statistical interpolation and conservative correction strategies for adjacent road segments of the same type are used for completion, and the source and weight of the completion are recorded for traceability. For example, if a parking space in front of a large hospital corresponds to a road with temporary parking and pedestrian traffic during morning and evening peak hours, after the parking space identifier is mapped to the main road segment in front of the hospital, road attribute parameters with fewer lanes and higher intersection density can be extracted from the road risk record. Traffic carrying capacity thresholds with lower capacity thresholds and smaller queue length thresholds can also be extracted, and high-risk operational risk levels applicable during peak hours can be extracted and marked. This allows subsequent induced risk assessments to quantify the actual constraints of the road where the parking space is located and to be interpretable.

[0041] In one specific implementation, the process of calculating the induced risk loss function value based on road risk parameters can be performed according to the following steps: An induced risk loss function is defined and a quantitative model is established by combining road risk parameters. The induced risk loss function is used to assess the potential losses to traffic flow, parking space utilization efficiency, and driving safety caused by road risk factors. This loss function combines multiple risk factors, such as road attribute parameters, traffic capacity thresholds, and operational risk levels, to reflect the relationship between road usage status and potential risks. Specifically, the loss function can be in the form of a weighted sum model, where each risk parameter is weighted according to its contribution to the overall risk. For example, parameters such as the number of lanes, intersection density, and capacity threshold can be assigned different weights based on their impact on traffic efficiency and flow control. In the model, certain parameters can be set with critical values ​​or thresholds under specific conditions. When these thresholds are exceeded, the loss function value increases sharply, indicating that the road capacity is exceeded or there is an extremely high risk. Calculate the induced risk loss function value. Based on the defined induced risk loss function and combined with actual road risk parameters, calculate the contribution of each risk factor to the total risk loss value. For example, when there is high intersection density or temporary road closures, traffic flow is restricted, leading to an increase in the loss function value. Simultaneously, when the road's traffic capacity threshold reaches a critical value, increased vehicle queuing time may lead to driver anxiety and an increased probability of accidents; these factors are also reflected in the loss function. For each parking space and its corresponding road segment, calculate the combined impact of its road attributes, traffic capacity threshold, and operational risk level to obtain the induced risk loss function value for that road segment. This loss function value can represent the potential loss caused by risk factors such as traffic flow, congestion, or accidents around a specific parking space under given traffic conditions. Dynamic assessment and decision support can be achieved using induced risk loss function values. Based on the calculated induced risk loss function values, the allocation of traffic flow and parking resources can be optimized. For example, on high-risk roads, a high induced risk loss function value may prompt the parking management system to implement temporary traffic control or adjust parking strategies to reduce unnecessary vehicle entry, thereby mitigating potential risk losses. Conversely, a low induced risk loss function value indicates lower traffic risk on that road segment, allowing for a moderate increase in parking resource allocation or traffic flow. Dynamic monitoring and calculation of induced risk loss function values ​​provide timely and effective decision support for traffic management departments, ensuring the efficient and safe operation of the traffic system even under high-risk conditions.

[0042] Induced risk loss function value: Construct a risk factor vector from the road risk parameters (road attribute parameters, traffic capacity threshold, operational risk level, etc.) extracted from the road segment corresponding to the berth. For each risk factor Definition of normalization: ;in The statistical range or standard limit range can be taken from the road risk parameter database and updated as the database is updated, thus ensuring traceability.

[0043] Define the induced risk loss function as:

[0044] in , .function This is a piecewise gain function used to represent a reproducible form of "loss increasing sharply after exceeding the load threshold," for example:

[0045] in The critical point, This represents the penalty coefficient for exceeding the limit. The enforceable, definitive rules are given as follows: The high quantile of the historical distribution of this factor can be taken (e.g. (quantile) or engineering limit mapping point. Desirable The range value is adjusted up or down according to the sensitivity of accident / congestion cost.

[0046] The output convergence evaluation module fuses the credible situation index and the induced risk loss function value to obtain an adaptive output convergence index. This module acts as a bridge in the berth management system, comprehensively considering the system's credibility and potential risks. By adjusting the system output through the adaptive convergence index, it ensures that the system can dynamically adjust the output method of berth information based on real-time network conditions and road risks. The adaptive output convergence index can dynamically respond to different environmental changes, achieving an optimal balance between accuracy and efficiency in the system's output.

[0047] In one specific implementation, the process of obtaining the adaptive output convergence index by fusing the credible situation index and the induced risk loss function value can be carried out according to the following steps: A computational framework for the adaptive output convergence index is defined. The adaptive output convergence index aims to comprehensively evaluate the overall reliability and risk of a traffic system, and through dynamic adjustment of response strategies, enable the system to maintain optimal stability and safety under different conditions. The reliability situation index is calculated by fusing link temporal characteristic components and event consistency characteristic components; it reflects the overall performance of the current traffic system in terms of stability and reliability. The induced risk loss function value quantifies the potential risks and losses faced by the road system through analysis of road risk parameters. The fusion of these two indicators can adopt a weighted fusion model, assigning different weights to the reliability situation index and the induced risk loss function value, and adjusting them according to the needs of the actual application scenario. For example, during peak hours, the induced risk loss function value has a higher weight, while during normal operation, the influence of the reliability situation index is emphasized. Based on the fused credible situation index and the induced risk loss function value, an adaptive output convergence index is calculated. This index calculation requires a sliding update mechanism based on a time window, combining the changing trends of the credible situation index and the induced risk loss function value over a past period, and obtaining the fused value through a weighted summation method. Specifically, a dynamic weight factor is set, and the system automatically adjusts this factor over time to adapt to different traffic conditions. For example, in the event of a sudden surge in traffic flow, the induced risk loss function value may suddenly increase; in this case, the dynamic weight factor will increase the weight of this value, thus reflecting the current high-risk state of the system. Under normal conditions, the weight of the credible situation index may dominate, thereby ensuring system stability. Real-time monitoring and decision optimization are achieved through an adaptive output convergence index. The calculated adaptive output convergence index serves as a health indicator for the traffic system, assessing the effectiveness of current traffic management strategies. A low index indicates system stability and low risk; a high index suggests potential higher risks or instability, requiring timely intervention. For example, in a busy commercial area, when the adaptive output convergence index reaches a preset threshold, the system can trigger an emergency response mechanism, such as adjusting traffic light cycles, restricting vehicle access to certain road sections, or adjusting parking space allocation strategies to reduce potential traffic accidents and congestion.

[0048] Adaptive output convergence index Trustworthiness index The adaptive output convergence index C is obtained by fusing the risk loss L. To ensure directional consistency, the two are first mapped to... : Then define fusion: ;in Let the weights be dynamic. A set of implementable dynamic update rules is given: Let... For the magnitude of risk change in adjacent windows, then

[0049] in Pick , Pick , Sensitivity coefficient (e.g.) ).

[0050] The tiered convergence control module executes tiered control strategies based on the level of the adaptive output convergence index and applies these strategies to the berth status output. Different convergence index levels correspond to different system response strategies. For example, level one may only require outputting real-time berth status, while level four may require more complex data processing, such as status snapshot reporting or edge merging of duplicate events. The module's function is to adjust the system's operating state based on the current credibility index and induced risk quantification results, thereby ensuring that the system provides the most suitable control and response under different operating environments.

[0051] In one specific implementation, the hierarchical control strategy is executed based on the level of the adaptive output convergence index and applied to the berth status output, which can be carried out in the following steps: The system completes the level determination and strategy binding, mapping the adaptive output convergence index corresponding to the same berth identifier within the sliding time window to level 1, level 2, level 3, or level 4, and binding a unique set of output rules to each level. Level 1 indicates that both link timing and event consistency are within a directly acceptable range; Level 2 indicates that there is a directional delay bias but event consistency can still be maintained; Level 3 indicates that out-of-order and conflict have affected causal evolution and require consistency constraint processing; Level 4 indicates that both links and events exhibit strong instability and require switching the reporting mode and transitioning to conservative output. To avoid frequent switching caused by level jitter, a hysteresis band and a minimum hold duration are introduced during level determination. That is, when the index crosses the level boundary, the hold conditions must be continuously met before upgrading or downgrading, and the triggering reason is recorded for subsequent traceability. The system executes a graded control strategy corresponding to the berth level and generates berth status output: When the adaptive output convergence index is at the first level, the berth status based on real-time values ​​is directly output, that is, the latest berth event is mapped to occupied or idle according to the collection time sequence and published immediately; when the adaptive output convergence index is at the second level, the berth status is output after performing asymmetric delay compensation on the berth events. Specifically, the statistical difference and time change rate of the uplink and downlink delays are estimated respectively, and a directional compensation amount is constructed to correct the arrival time, so that events under the same berth identifier return to a comparable scale on the time axis, and then the status is output according to the corrected time sequence to reduce the status lag caused by unidirectional congestion; when the adaptive output convergence index is at the third level, the berth events are reordered and conflict arbitration is performed after performing causal consistency constraint reordering on the berth events and outputting the berth status. Specifically, a reachable transition set is established based on the berth occupancy status transition constraint, and the events are jointly sorted according to the collection time and the corrected arrival time, and unreachable transitions and reverse transitions are handled accordingly. Conflicts caused by transfers and short-term repetitive jitter are resolved using arbitration rules that prioritize verification by check code or signature, monotonicity of event sequence number, and minimum cost repair. If necessary, conflicting events are discarded, merged, or delayed for confirmation, thereby outputting a state sequence that satisfies causal evolution. When the adaptive output convergence index is at level four, the reporting method for parking space events is switched from reporting one by one to reporting a state snapshot. Repeated jitter events are edge-merged and the parking space status is output. Specifically, edge merging is performed on multiple state flips of the same parking space identifier within a preset snapshot period, and only stable states that continuously exceed the threshold are retained as snapshot results. At the same time, the output of driver-side guidance information and operation-side heat map is switched from being based on real-time values ​​to being based on conservative values. The conservative value is the output value of the available parking space after applying credible constraints on the parking space status. For example, parking spaces that repeatedly change from idle to occupied and back to idle within a short period of time under level four are considered uncertain and treated as occupied to avoid overly optimistic guidance that causes vehicles to detour ineffectively. This system generates interpretable examples of output delivery and linkage, simultaneously including level identifiers, compensation parameters, reordering counts, and arbitration statistics during output for operational review. For instance, during evening rush hour in a densely populated commercial area with high-rise buildings, if long-tail latency and batch replenishment occur in the link, and the adaptive output convergence index of a certain parking space enters the third level, then directional latency compensation is first performed on the events within the replenishment cluster, followed by reordering and conflict arbitration based on the reachable transition set. The final output is a continuous and consistent occupancy status, avoiding false alarms of idle space caused by out-of-order delivery. If the same area experiences strong interference leading to widespread jitter and entering the fourth level, then snapshot-based output is used, and jitter edges are merged. Simultaneously, conservative values ​​are used for driver guidance and operational heatmaps, ensuring that the number of available parking spaces converges under credible constraints, thus maintaining controllability and safety of the output even under unstable link conditions.

[0052] Level threshold, hysteresis band, and minimum hold time (graded control can be implemented): The convergence index C is divided into multiple levels. Three level boundary thresholds are defined. For example, determined by the quantile of historical C: ;in Desirable Alternatively, adjustments may be made based on business risk appetite. To avoid rating fluctuations, a hysteresis band may be introduced. With minimum holding time : Upgrade requirements: ; Downgrade conditions: ; in A certain proportion of the threshold span can be taken (e.g.) ), One or more step sizes can be selected (e.g.) This ensures that the level switching rules are clear and implementable.

[0053] Level 4 Snapshot Period and Stable Persistence Threshold (Edge Merging Available): When entering the highest level, snapshot reporting is adopted. Let the snapshot period be... Edge merging is performed on the same berth state sequence within a snapshot period: if the cumulative stable duration of a certain state \(x\) satisfies ; If the current state is not specified, output this state as a snapshot result; otherwise, mark the berth as uncertain and output a conservative value. Implementable parameter acquisition rules are given: ; ; The business output module outputs berth query and guidance information, as well as statistical analysis and management information. It adjusts the output berth status according to the hierarchical control strategy executed by the hierarchical convergence control module. Its function is to output berth query results, guidance information, and statistical analysis and management information, and further adjust these outputs based on the hierarchical control strategy executed by the hierarchical convergence control module. For example, when the system is at a high convergence index level, it may need to output more conservative data or take more cautious business decisions. This module not only supports business queries but also involves optimizing data output to facilitate decision analysis and subsequent optimization by management personnel. The business output module ensures the transparency and effectiveness of the system output, enabling the system to maintain high flexibility and adaptability while providing services.

[0054] In one specific implementation, the process of outputting berth query and guidance information, as well as statistical analysis and management information, can be carried out according to the following steps: The system receives and processes user query requests, and based on real-time berth status data, outputs the required berth status information to the user, including berth occupancy and vacancy status, estimated availability time, and guidance information. When the query involves a specific area, the system will provide corresponding berth information based on real-time data for that area, and perform data correction and supplementation to ensure the accuracy and timeliness of the information. The system dynamically adjusts the output berth status according to a hierarchical convergence control strategy. At lower levels, the berth status is directly output based on real-time values; while at higher levels, the system compensates, reorders, or merges the output results according to the hierarchical control strategy to ensure the reliability of the output berth status. For example, when the system is at level three, the output berth status is processed by causal consistency constraints to avoid information distortion caused by out-of-order events. The system generates and outputs statistical analysis and management information, and provides periodic management reports based on dynamic adjustments to berth status, including berth utilization rates, the number of vacant berths, and statistics on high-risk areas. This information can be used for management decision-making and operational adjustments to ensure the system maintains a balance between efficiency and safety. For example, if the berth status in a certain area fluctuates repeatedly, the management report will identify potential problems in that area and recommend appropriate risk control measures.

[0055] Example 2: A berth management method based on sensor technology, comprising the following steps: Step 1: Collect occupancy status data for each berth and generate berth events; Step 2: Receive berth events and forward them to the platform side. When a berth event arrives at the platform side, record the arrival time and generate gateway receipt information. Step 3: Within the sliding time window, based on the acquisition time, arrival time, and gateway receipt information, calculate the link timing feature component and the event consistency feature component respectively, and obtain the trustworthy situation index by fusing the link timing feature component and the event consistency feature component. Step 4: Retrieve road risk parameters from the road risk parameter database based on berth signs, and calculate the induced risk loss function value based on the road risk parameters; Step 5: Obtain the adaptive output convergence index by fusing the credible situation index and the induced risk loss function value, and implement a graded control strategy according to the level to which the adaptive output convergence index belongs. Step 6: Output the berth status according to the graded control strategy and adjust the berth status according to the strategy.

[0056] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A berth management system based on sensor technology, characterized in that include: The sensing and acquisition module is used to collect occupancy status data of each berth and generate berth events. The berth events include at least the berth identifier and the collection time. The transmission aggregation module is used to receive berth events and forward them to the platform side. When a berth event arrives at the platform side, the arrival time is written and gateway receipt information is generated. The Trusted Situation Assessment Module is used to calculate the link timing feature component and the event consistency feature component based on the acquisition time, arrival time and gateway receipt information within a sliding time window, and to obtain the Trusted Situation Index by fusing the link timing feature component and the event consistency feature component. The induced risk quantification module is used to retrieve road risk parameters from the road risk parameter library based on berth identification, and to obtain the induced risk loss function value based on the road risk parameters; The output convergence evaluation module is used to fuse the credible situation index and the induced risk loss function value to obtain an adaptive output convergence index. The graded convergence control module is used to execute graded control strategies based on the level of the adaptive output convergence index and apply the graded control strategies to the berth status output. The business output module is used to output berth query and guidance information, as well as statistical analysis and management information, and adjust the output berth status according to the hierarchical control strategy executed by the hierarchical convergence control module.

2. The berth management system based on sensor technology according to claim 1, characterized in that, The sensing and acquisition module performs threshold determination or classification on the berth occupancy status data to obtain the berth occupancy status, and generates a berth event when the occupancy status changes or the reporting trigger condition is met.

3. The berth management system based on sensor technology according to claim 1, characterized in that, The transmission aggregation module includes a gateway node. When the gateway node receives a berth event, it records the platform-side timestamp as the arrival time and generates gateway receipt information. The gateway receipt information includes at least the berth identifier, collection time, arrival time, event sequence number, and check code or signature, and is then sent back to the sensing and acquisition module or edge node.

4. The berth management system based on sensor technology according to claim 1, characterized in that, The link timing feature components include at least: delay spectrum feature components, asymmetric drift components, and out-of-order consistency entropy components; the event consistency feature components include at least: causal violation intensity components.

5. The berth management system based on sensor technology according to claim 4, characterized in that, The time delay spectrum characteristic components are characterized by the single-trip time delay quantile, time delay jitter intensity, and time delay skewness; the asymmetric drift component is characterized by the statistical difference between uplink and downlink time delay and their time change rate; the out-of-order consistency entropy component is characterized by the out-of-order ratio, out-of-order span, and batch replenishment intensity; the causal violation intensity component is characterized by the number of unreachable transfers, the number of reverse transfers, and the number of short-term back-and-forth jitters during berth status transitions.

6. The berth management system based on sensor technology according to claim 1, characterized in that, The induced risk quantification module locates the road risk record corresponding to the berth in the road risk parameter database based on the berth identifier, and extracts road risk parameters from the road risk record; among which, the road risk parameters include at least road attribute parameters, traffic carrying capacity threshold, and operational risk level.

7. The berth management system based on sensor technology according to claim 1, characterized in that, A tiered regulatory strategy should include at least the following: a) When the adaptive output convergence index is at the first level, the output is based on the berth status of the real-time value; b) When the adaptive output convergence index is at the second level, the berth status is output after performing asymmetric time delay compensation on the berth event; c) When the adaptive output convergence index is at the third level, the berth status is output after reordering the causal consistency constraints and arbitrating the conflicts for the berth events. d) When the adaptive output convergence index is at level four, the berth event reporting method is switched from reporting one by one to reporting status snapshots, and the berth status is output after edge merging of repeated jitter events.

8. The berth management system based on sensor technology according to claim 7, characterized in that, When the adaptive output convergence index is at level four, the driver-side guidance information and the operation-side heat map output are switched from being based on real-time values ​​to being based on conservative values. The conservative values ​​are the output values ​​of available parking spaces after applying credible constraints based on the parking space status.

9. A berth management method based on sensor technology, applied to the berth management system based on sensor technology as described in claims 1-8, characterized in that, Includes the following steps: Step 1: Collect occupancy status data for each berth and generate berth events; Step 2: Receive berth events and forward them to the platform side. When a berth event arrives at the platform side, record the arrival time and generate gateway receipt information. Step 3: Within the sliding time window, based on the acquisition time, arrival time, and gateway receipt information, calculate the link timing feature component and the event consistency feature component respectively, and obtain the trustworthy situation index by fusing the link timing feature component and the event consistency feature component. Step 4: Retrieve road risk parameters from the road risk parameter database based on berth signs, and calculate the induced risk loss function value based on the road risk parameters; Step 5: Obtain the adaptive output convergence index by fusing the credible situation index and the induced risk loss function value, and implement a graded control strategy according to the level to which the adaptive output convergence index belongs; Step 6: Output the berth status according to the graded control strategy and adjust the berth status according to the strategy.

Images