Campus dormitory intelligent door lock gateway system and method based on edge processing

Abstract

The invention relates to the field of edge processing, and discloses a campus dormitory intelligent door lock gateway system and method based on edge processing, and the method comprises the steps: obtaining a door lock processing queue length, a passage trigger count and a door body occupation duration, mapping the door lock processing queue length, the passage trigger count and the door body occupation duration into a door area operation load vector, and determining a state evolution sequence of a door area operation load; identifying a transition interval in which the gate area operation load is converted from a steady state to a congestion state, and forming a corresponding processing period; based on the time adjacency relation of the passing request and the door body occupation continuity, generating an aggregation passing unit representing the overall passing behavior; constructing a risk state mechanism coupled with a gate area operation load state, executing constrained risk state transition judgment on the aggregation passing unit, and determining a risk state level; the passing requests in the processing cycle are bound to the corresponding verification path instances in the door lock gateway according to the risk state hierarchy, and a verification set associated with the selected verification path is formed. The method has the advantage that the stability is improved.

Description

Technical Field

[0001] This invention relates to the field of edge processing, specifically to a smart door lock gateway system and method for campus dormitories based on edge processing. Background Technology

[0002] With the advancement of intelligent management in campus dormitories, smart door lock systems with identity verification and anomaly recognition functions are commonly deployed at dormitory entrances to automate the management of personnel entry and exit. Existing dormitory smart door lock systems typically monitor personnel passage behavior in real time at the entrance area and judge whether there are any abnormalities in the passage process based on preset behavior judgment rules or thresholds. In practical applications, during specific periods such as after class or evening return to dormitories, there will be dense passage scenarios at the entrance area of ​​campus dormitories in a short period of time. In such scenarios, the spatial distance between people is significantly reduced, and the passage trajectories overlap frequently. Some people may exhibit short-term, non-malicious abnormal behavior characteristics due to crowding or interaction during the passage process. However, the analysis of the above passage behavior is usually based on the assumption of single-person passage or static anomaly judgment rules, without fully considering the phased changes of behavioral characteristics in dense passage scenarios. It is easy to misjudge short-term behavioral deviations that occur during normal passage as abnormal events, thereby triggering passage blocking or verification failure processing, affecting the normal passage efficiency of personnel during peak hours, and leading to an increase in the system's misjudgment rate and probability of failure. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a campus dormitory smart door lock gateway system and method based on edge processing, which has the advantage of improved stability and solves the problems mentioned in the background technology.

[0004] To achieve the aforementioned goal of improving stability, this invention provides the following technical solution: a smart door lock gateway method for campus dormitories based on edge processing, comprising the following steps:

[0005] The length of the door lock processing queue, the passage trigger count, and the duration of door occupancy are obtained and mapped to the door area operating load vector. The state evolution sequence of the door area operating load is determined by the combination relationship of the direction consistency and change amplitude between adjacent load vectors.

[0006] Based on the state evolution sequence, the transition interval of the gate region's operating load from steady state to congested state is identified. When the gate region's operating state is in the transition interval, the arriving passage requests are encapsulated in a time window to form a corresponding processing cycle.

[0007] Within the corresponding processing cycle, based on the temporal adjacency relationship of the access requests and the continuity of gate occupancy, multiple access requests are merged to generate an aggregated access unit that represents the overall access behavior.

[0008] Based on the aggregated access unit, a risk state mechanism coupled with the gate area's operating load status is constructed. Combining the gate area's operating load decline trend, processing queue change characteristics, and edge computing resource status, a constrained risk state transition judgment is performed on the aggregated access unit to determine the risk state level.

[0009] Within the processing cycle, access requests are bound to the corresponding verification path instance within the door lock gateway according to the risk status level, forming a verification set associated with the selected verification path.

[0010] Preferably, the process of mapping to the gate region operating load vector is as follows:

[0011] The door lock queue length, number of access triggers, and door occupancy time are collected in real time through edge computing nodes. The collected data is standardized and stored in time series.

[0012] The vectorization mapping mechanism is used to project the data onto the multi-dimensional gate load space, and a vector representation reflecting the real-time load characteristics of the gate is generated synchronously through timestamps.

[0013] By combining the historical access records and abnormal event database of the gate area, the similarity between the vector representation and the historical typical load pattern is calculated to obtain the weighted matching value of each vector dimension, and a weighted gate area operation load vector is generated.

[0014] Preferably, the process for determining the state evolution sequence of the gate zone operating load is as follows:

[0015] Based on the gate zone operating load vector, calculate the direction cosine and amplitude change rate of each pair of adjacent load vectors to form a direction consistency matrix and an amplitude change matrix.

[0016] The directional consistency matrix and the amplitude change matrix are mapped into a multi-dimensional state evolution vector through a weighted fusion algorithm, and the current metric vector is adjusted by combining the similarity scores of historical typical load patterns.

[0017] The adjusted metric vectors are arranged in chronological order to form a continuous sequence. An adaptive threshold determination algorithm is then used to jointly determine the magnitude of metric mutations, continuous change trends, and abnormal weights to generate a state evolution sequence of the gate region's operating load.

[0018] Preferably, the process of identifying the transition intervals from steady state to congestion state in the gate region operating load based on the state evolution sequence is as follows:

[0019] The state evolution sequence is divided into overlapping sliding windows in chronological order. The load trend direction, change magnitude and risk weight in each window are statistically analyzed, and the load mutation rate, continuous growth duration and risk-weighted cumulative value in the window are calculated.

[0020] Using a sliding window peak detection algorithm, the time period in which the weighted risk cumulative value exceeds a preset dynamic threshold is identified and preliminarily marked as a potential transition interval;

[0021] The initially identified transition intervals are matched with the historical congestion pattern database of the gate area to calculate the similarity score;

[0022] By combining the continuity of adjacent time windows, the start and end times of the transition interval are optimized and corrected to obtain the final determined transition interval.

[0023] Based on the final determined transition interval, a quantitative feature set for determining gate congestion risk is generated.

[0024] Preferably, the process for forming the corresponding processing cycle is as follows:

[0025] Based on the quantitative feature set for determining gate congestion risk, each transition interval and additional risk indicators are used as a reference for dividing the time window.

[0026] Each passage request arriving within the jump interval is clustered according to its temporal proximity and the risk weight of the jump interval to form an initial request group;

[0027] A time window is allocated to each request group, and the window length is adaptively adjusted based on the sudden congestion magnitude and duration of the transition interval and the length of the gate processing queue.

[0028] Requests within the time window are packaged into processing cycle instances. Each instance contains a request identifier, arrival time, gate occupancy time, and priority label, and is associated with the risk indicators of the corresponding transition interval to form a corresponding processing cycle.

[0029] Preferably, the process of generating aggregated traffic units that represent the overall traffic behavior is as follows:

[0030] The requests within the corresponding processing period are sorted in order of timestamp, and the congestion risk index attached to the transition interval is used to perform a preliminary weighted sort of the requests.

[0031] Analyze the time interval of requests within the same processing cycle, the overlap of gate occupancy, and historical typical passage patterns, and determine the aggregation conditions through multi-condition fusion;

[0032] Requests that meet the aggregation conditions are merged to form aggregated passage units, and the number of requests, cumulative occupancy time, time distribution characteristics, aggregation density and risk weighting index within the unit are calculated.

[0033] The aggregated access unit is coupled with the gate area operating load status, the risk level of the transition interval, and the edge computing resource usage to generate an aggregated access unit that represents the overall access behavior.

[0034] Preferably, the process of constructing a risk state mechanism coupled with the gate area's operating load state is as follows:

[0035] By taking the aggregated passage unit as input and combining it with the corresponding processing cycle instance and the risk index of the transition interval, a multi-dimensional state vector is constructed.

[0036] Based on the multidimensional state vector and state machine node rules, a jump decision is performed on each aggregation unit to dynamically select the next risk node;

[0037] By introducing continuity constraints and historical accumulated risks, state transitions are smoothed out.

[0038] The smoothed jump result is updated to the risk state mechanism instance to form a reusable risk state mechanism coupled with the gate area's operating load state.

[0039] Preferably, the process for determining the risk status level is as follows:

[0040] Based on the risk status mechanism, obtain the real-time risk status sequence for each aggregated passage unit;

[0041] Based on the trend of load decline in the gate area, the load change in the next cycle is predicted. A constrained risk transfer matrix is ​​established by combining the processing queue length and the available computing power of the edge computing nodes.

[0042] The risk transition matrix is ​​used to determine and dynamically adjust the instantaneous risk state sequence, and the final risk state level of each aggregated passage unit is output.

[0043] Preferably, the process of forming a verification set associated with the selected verification path is as follows:

[0044] Based on the final risk status level and historical jump trajectory of the aggregation unit to which each access request belongs, and combined with the access request attributes, the risk priority weight is calculated, and a weighted verification priority ranking table is generated.

[0045] Based on the sorting table and the constrained risk transfer matrix, each passage request is dynamically assigned to the corresponding verification path instance;

[0046] By combining the temporal proximity of requests within the aggregation unit and the continuity of gate load, the verification path is merged or split as necessary to generate a multi-level verification set within the processing cycle.

[0047] A smart door lock gateway system for campus dormitories based on edge processing includes:

[0048] Load acquisition module: Real-time acquisition of door lock queue length, number of passage triggers and door occupancy time, and mapping it into door zone operation load vector to form a state evolution sequence;

[0049] Transition identification module: Identifies the transition interval from steady state to congestion state based on the load state evolution sequence, and encapsulates the arriving requests into a processing cycle by time windowing;

[0050] Behavior aggregation module: Merges multiple passage requests based on time adjacency and gate occupancy continuity within the processing cycle to generate aggregated passage units that represent the overall passage behavior;

[0051] Risk assessment module: Performs risk state transition assessment based on the load change trend of aggregated access units and gate areas, processing queue characteristics, and edge computing status to determine the risk level;

[0052] Path binding module: Binds access requests within the processing cycle to the corresponding verification path instance according to the risk status level, forming a verification set associated with the selected verification path.

[0053] Compared with the prior art, the present invention provides a campus dormitory smart door lock gateway system and method based on edge processing, which has the following beneficial effects:

[0054] This invention introduces edge processing technology into the smart door lock gateway for campus dormitories, enabling real-time perception and dynamic analysis of access behavior in the door area. By acquiring the door lock processing queue length, access trigger count, and door occupancy duration, it constructs a door area operating load vector and generates a state evolution sequence. This accurately identifies peak periods of high population density and transition intervals from steady state to congestion state, effectively avoiding misjudgments caused by traditional single-person static judgment. Employing a time windowing encapsulation and aggregated access unit generation method, it can comprehensively represent access behavior, fully considering the spatiotemporal adjacency and interaction relationships between people, thereby reducing anomaly judgment errors in peak access scenarios. Combining the door area load decline trend, processing queue change characteristics, and risk state mechanism constructed from edge computing resource status, it can dynamically assess access risks and manage verification paths in layers, achieving intelligent binding of access requests and constrained risk transfer judgment. This significantly improves access efficiency during peak periods, reduces system misjudgment rate and probability of failure, while ensuring the stability and security of the door lock system, providing efficient and reliable technical support for the intelligent management of campus dormitory entrances and exits. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the method of the present invention;

[0056] Figure 2 This is a schematic diagram of the structure of the present invention. Detailed Implementation

[0057] 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.

[0058] Example 1: Please refer to Figure 1 As shown in the figure, a campus dormitory smart door lock gateway method based on edge processing according to an embodiment of the present invention includes the following steps:

[0059] S1: Obtain the length of the door lock processing queue, the passage trigger count, and the duration of door occupancy, map them to the door zone operating load vector, and determine the state evolution sequence of the door zone operating load by combining the directional consistency and change amplitude of adjacent load vectors.

[0060] The process of mapping S1 to the gate region operating load vector is as follows:

[0061] The door lock queue length, number of access triggers, and door occupancy time are collected in real time through edge computing nodes. The collected data is standardized and stored in time series.

[0062] Edge computing nodes deployed at the door lock gateway collect multi-source data in real time, including door lock processing queue length, number of access triggers, and door occupancy time, through sensor interfaces. These data are sampled at uniform time intervals. The collected raw data first undergoes noise filtering, anomaly removal, and unit standardization. For example, the queue length is normalized, the number of access triggers is standardized using mean and variance, and the door occupancy time is proportionalized based on the average occupancy time of the door area. This results in a real-time data sequence with uniform dimensions that can be directly compared, ensuring the reliability and computability of the data. This provides sufficient data support for load vector generation and avoids the limitations of simply piling up conventional data collection methods.

[0063] The vectorization mapping mechanism is used to project the data onto the multi-dimensional gate load space, and a vector representation reflecting the real-time load characteristics of the gate is generated synchronously through timestamps.

[0064] Based on the historical access records and abnormal event database of the gate area, the currently generated vector representation is compared with the historical typical load patterns through a dimension-by-dimensional similarity calculation. The similarity calculation adopts the weighted cosine similarity or Euclidean distance weighting method, and combines the contribution of each dimension to abnormal events in the historical data to generate weight coefficients. Finally, the vector value of each dimension is multiplied by the corresponding weight to obtain the weighted gate area operating load vector. The vector not only reflects the current gate area load, but also incorporates the historical load patterns and the impact of abnormal behavior, realizing the early quantitative assessment of potential risks. It provides effective support for the phased changes of behavioral characteristics in peak dense passage scenarios, avoiding the defects of simply stacking conventional indicators.

[0065] By combining the historical access records and abnormal event database of the gate area, the similarity between the vector representation and the historical typical load pattern is calculated to obtain the weighted matching value of each vector dimension, and a weighted gate area operation load vector is generated.

[0066] Standardized multi-source traffic data is projected onto the multi-dimensional load space of the gate area through a vector mapping mechanism. Each dimension corresponds to different load characteristics of the gate area, such as queuing pressure, traffic density, and occupancy intensity. During the mapping process, a timestamp synchronization method is used to align data from different sensors, so that the gate area state at the same point in time can form a complete vector representation in the multi-dimensional space. The vector not only contains the current instantaneous state, but also superimposes the average load trend within a short-time sliding window, thereby reflecting the dynamic change characteristics of the gate area load. Vector modeling overcomes the problem that traditional single-dimensional statistical indicators cannot describe the complex behavior of peak traffic, and provides a quantitative basis for accurately characterizing the gate area load.

[0067] The process of determining the state evolution sequence of the gate zone operating load in S1 is as follows:

[0068] Based on the gate zone operating load vector, calculate the direction cosine and amplitude change rate of each pair of adjacent load vectors to form a direction consistency matrix and an amplitude change matrix.

[0069] The directional consistency matrix and the amplitude change matrix are mapped into a multi-dimensional state evolution vector through a weighted fusion algorithm, and the current metric vector is adjusted by combining the similarity scores of historical typical load patterns.

[0070] The adjusted metric vectors are arranged in chronological order to form a continuous sequence. An adaptive threshold determination algorithm is then used to jointly determine the magnitude of metric mutations, continuous change trends, and abnormal weights to generate a state evolution sequence of the gate region's operating load.

[0071] S2: Based on the state evolution sequence, identify the transition interval of the gate region's operating load from steady state to congested state. When the gate region's operating state is in the transition interval, encapsulate the arriving passage requests in a time window to form the corresponding processing cycle.

[0072] The process of identifying the transition interval from steady state to congested state of the gate operating load in S2 based on the state evolution sequence is as follows:

[0073] The state evolution sequence is divided into overlapping sliding windows in chronological order. The load trend direction, change magnitude and risk weight in each window are statistically analyzed, and the load mutation rate, continuous growth duration and risk-weighted cumulative value in the window are calculated.

[0074] The evolution sequence of the load state of the gate area is divided into overlapping sliding windows according to time order. Each window covers several consecutive time points, and statistical analysis is performed on the load trend direction, change magnitude and risk weight within the window. For example, by calculating the rate of continuous load growth, the number of local peak occurrences and the cumulative value of risk weight, the load mutation rate and continuous growth duration of each window can be obtained. A dynamic window length adaptive adjustment mechanism can be adopted to ensure that rapidly changing load characteristics can be captured during peak dense traffic, while avoiding misjudging fluctuations as transitions under low traffic pressure. By quantifying the load change trend, a reliable analytical basis is provided for the identification of transition intervals.

[0075] Using a sliding window peak detection algorithm, the time period in which the weighted risk cumulative value exceeds a preset dynamic threshold is identified and preliminarily marked as a potential transition interval;

[0076] Peak detection is performed on the weighted cumulative risk value of each sliding window. A dynamic threshold algorithm is used, with the threshold dynamically adjusted based on historical congestion peaks and real-time gate load fluctuations. The time period when the cumulative value exceeds the threshold is identified and initially marked as a potential transition interval. The calculation method of the dynamic threshold includes considering the load average, standard deviation and maximum continuous growth value in the recent period, so as to adapt to the peak or valley traffic environment and ensure that the change of gate load from steady state to congestion state can be accurately captured under dense personnel or short-term fluctuation conditions, without misjudgment due to occasional abnormal actions.

[0077] The initially identified transition intervals are matched with the historical congestion pattern database of the gate area to calculate the similarity score;

[0078] The initially identified potential transition intervals are compared with the historical congestion pattern database of the gate area. Through similarity calculation, such as weighted cosine similarity, dynamic time warping (DTW), or Euclidean distance weighting, the similarity between the current load change and the typical historical congestion pattern is evaluated, and a similarity score is generated. The similarity score is used to help determine the authenticity of the transition interval, avoid misjudging short-term occasional high load as congestion transition, and provide historical reference for risk assessment. This enhances the system's ability to identify phased load changes and goes beyond the limitations of simple threshold judgment.

[0079] By combining the continuity of adjacent time windows, the start and end times of the transition interval are optimized and corrected to obtain the final determined transition interval.

[0080] By combining the temporal continuity of adjacent sliding windows, the start and end times of the initially calibrated transition intervals are optimized and corrected. For example, when consecutive windows show a load growth trend, the transition intervals can be expanded or merged. When a single window experiences a sudden high load but the loads of the windows before and after it are stable, the time period can be shortened or discarded. The optimization rules combine time intervals, load increments, and historical pattern matching results to ensure the accuracy of the final transition intervals with quantitative logic, effectively eliminating the interference of short-term fluctuations and making the transition intervals more consistent with the actual gate congestion development trend.

[0081] Based on the final determined transition interval, a quantitative feature set for determining gate congestion risk is generated;

[0082] Based on the final determined transition interval, a quantitative feature set for gate congestion risk assessment is extracted, including multi-dimensional indicators such as peak amplitude, duration of continuous growth, cumulative risk weight, time period location, and historical similarity score. These feature sets are used for subsequent aggregation of traffic units, risk status assessment, and dynamic verification of path allocation, realizing a closed loop from data perception to decision control. In this way, traffic behavior during peak hours can be accurately quantified, providing direct support for reducing misjudgment rate and ensuring traffic efficiency.

[0083] The process of forming the corresponding processing cycle in S2 is as follows:

[0084] Based on the quantitative feature set for determining gate congestion risk, each transition interval and additional risk indicators are used as a reference for dividing the time window.

[0085] Based on the quantitative feature set for gate congestion risk assessment, each identified transition interval and its additional risk indicators are used as a reference for time window division. For example, the start and end times of the initial window are calculated based on the cumulative risk value, continuous growth duration, and peak amplitude of the transition interval to ensure that the time window can cover the entire high-risk phase while taking into account the boundary transition of the low-risk phase. By applying the quantification of risk features to window division, the limitation of traditional fixed time slices in not being able to reflect peak traffic dynamics is avoided, ensuring that the processing cycle division is closely coupled with the actual gate load status.

[0086] Each passage request arriving within the jump interval is clustered according to its temporal proximity and the risk weight of the jump interval to form an initial request group;

[0087] Within each transition interval, arriving passage requests are clustered according to time proximity and transition interval risk weights. This includes using a weighted time proximity clustering algorithm to group requests with similar times and risk levels into the same request group, thus forming an initial request set. By considering the order of request arrival and risk weights, passage requests that accumulate during peak periods can be managed as a whole, avoiding congestion or misjudgment that may result from processing a single request, while also providing the basic data structure for processing cycle generation.

[0088] A time window is allocated to each request group, and the window length is adaptively adjusted based on the sudden congestion magnitude and duration of the transition interval and the length of the gate processing queue.

[0089] A time window is allocated to each request group. The window length is adaptively adjusted based on the sudden congestion magnitude and duration of the transition interval and the gate processing queue length. For example, when the sudden congestion magnitude is large and the queue length continues to grow, the window length can be automatically extended to cover more passing requests. When the queue drops rapidly or the load fluctuation is not significant, the window length can be shortened to improve processing efficiency. Adaptive allocation ensures that the processing cycle can be flexibly adjusted according to the actual gate status, avoiding congestion omissions or waste of resources during off-peak periods caused by fixed windows.

[0090] The requests within the time window are packaged into processing cycle instances. Each instance contains a request identifier, arrival time, gate occupancy time, and priority label. It is also associated with the risk indicators of the corresponding transition interval to form a corresponding processing cycle.

[0091] The system packages passage requests within a time window into processing cycle instances. Each instance contains a request identifier, arrival time, gate occupancy time, and priority label, and is associated with risk indicators of the corresponding transition interval. This includes creating a data structure instance in the door lock gateway for each processing cycle to store request information and risk characteristics. This data structure can also be used for subsequent aggregated passage unit analysis and dynamic verification of path allocation, ensuring the integrity of the processing cycle in terms of data structure and its correlation with the gate load status. This enables the system to efficiently manage peak passage requests and reduce misjudgment rate and congestion risk.

[0092] S3: Within the corresponding processing cycle, based on the temporal adjacency relationship of the access requests and the continuity of gate occupancy, multiple access requests are merged to generate an aggregated access unit that represents the overall access behavior.

[0093] The process of generating aggregated traffic units representing the overall traffic behavior in S3 is as follows:

[0094] The requests within the corresponding processing period are sorted in order of timestamp, and the congestion risk index attached to the transition interval is used to perform a preliminary weighted sort of the requests.

[0095] All passage requests within the corresponding processing period are sorted according to timestamp order, and combined with the congestion risk indicators attached to the transition interval, the requests are initially weighted and sorted. Each request can be assigned a weight value. The weight calculation method takes into account the closeness of the request arrival time, the gate occupation time, and the risk level of the current transition interval. For example, requests in high-risk time periods have a larger weight, while requests in low-risk intervals have a relatively smaller weight. This provides a priority processing order for subsequent aggregation, ensuring that the system can prioritize potentially congested critical requests during peak passage periods, rather than simply processing them in chronological order.

[0096] Analyze the time interval of requests within the same processing cycle, the overlap of gate occupancy, and historical typical passage patterns, and determine the aggregation conditions through multi-condition fusion;

[0097] The system calculates whether the time interval between adjacent requests is less than a preset threshold, whether there is overlap in the gate occupancy time, and the similarity between the current request sequence and the historical peak passage pattern. Only when multiple conditions are met simultaneously will the system include these requests in the same aggregation unit, thereby avoiding misjudging occasional short-term overlaps as overall passage behavior. By fusing multi-dimensional features to determine aggregation conditions, the system ensures the accuracy and practicality of the aggregated passage unit.

[0098] Requests that meet the aggregation conditions are merged to form aggregated passage units, and the number of requests, cumulative occupancy time, time distribution characteristics, aggregation density and risk weighting index within the unit are calculated.

[0099] Requests that meet the aggregation conditions are merged to form aggregated passage units. During the aggregation process, core quantitative indicators are calculated for each unit, including the number of requests within the unit, the cumulative gate occupancy time, the request time distribution characteristics, the aggregation density, and the risk weighting indicator. The aggregation density can be calculated by dividing the number of requests per unit time by the window length, and the risk weighting indicator is the cumulative value of the weight of each request. Through these indicators, the aggregated passage unit can fully characterize the overall passage behavior within the processing cycle and can be directly used for risk status determination and verification path allocation.

[0100] The aggregated access unit is coupled with the gate area operating load status, the risk level of the transition interval, and the edge computing resource usage to generate an aggregated access unit that represents the overall access behavior.

[0101] A data structure is established in the edge computing node to bind the aggregated access unit with the real-time load vector, the risk level of the transition interval, and the computing resource usage information, thereby achieving multi-dimensional state coupling. Through this coupling process, the final aggregated access unit representing the overall access behavior is generated. This aggregated access unit not only provides an accurate quantitative representation of the overall access behavior, but also provides a direct basis for the system to perform intelligent scheduling, risk control, and verification path selection during peak periods, thereby effectively improving access efficiency and security.

[0102] S4: Based on the aggregated access unit, construct a risk state mechanism coupled with the gate area's operating load status, and combine the gate area's operating load decline trend, processing queue change characteristics, and edge computing resource status to perform constrained risk state transition judgment on the aggregated access unit and determine the risk state level.

[0103] The process of constructing a risk state mechanism coupled with the gate zone operating load state in S4 is as follows:

[0104] By taking the aggregated passage unit as input and combining it with the corresponding processing cycle instance and the risk index of the transition interval, a multi-dimensional state vector is constructed.

[0105] The system collects key behavioral parameters of each passage unit in the gate area during the processing cycle, including the number of passage requests, gate occupancy time, queue length, and trigger event type. Then, it maps these parameters to predefined risk indicators for transition intervals to form a risk feature set for each unit in the current cycle. Finally, it combines the risk features into a multi-dimensional state vector in a predetermined order, which serves as the basic input for state transition determination. This allows each aggregated passage unit to not only reflect its real-time operating load but also its potential risk level, providing quantifiable data support for the risk state mechanism.

[0106] Based on the multidimensional state vector and state machine node rules, a jump decision is performed on each aggregation unit to dynamically select the next risk node;

[0107] The multidimensional state vector is mapped to a pre-defined set of nodes in the state machine. Each node corresponds to a different risk level and a possible risk transition path. Then, by weighting and calculating the features of each dimension in the vector and judging the threshold, the next risk node that best matches the current cycle is determined. Combined with the real-time load change trend, the optimal node is dynamically selected to ensure that the risk state can respond to the fluctuation of the gate load in a timely manner. This ensures that the risk state mechanism can follow the changes in the gate's operating state in real time and intelligently, rather than relying on simple superposition of historical experience or fixed rules.

[0108] By introducing continuity constraints and historical accumulated risks, state transitions are smoothed out.

[0109] The current cycle's transition result is integrated with the historical risk status of previous cycles to calculate the cumulative risk weight. The risk level of the current node is then adjusted according to the weight. A smoothing function or moving weighted algorithm is used to process the transition result to ensure that the risk status changes smoothly over a continuous time series. Finally, the smoothed transition result is marked as the final risk node of the current cycle. This not only improves the stability of risk assessment but also ensures the continuity and reusability of the mechanism's output, avoiding misjudgments caused by transient fluctuations.

[0110] The smoothed jump results are updated to the risk state mechanism instance to form a reusable risk state mechanism coupled with the gate area's operating load state;

[0111] The final risk node of each aggregated passage unit is stored in a risk status instance, and the corresponding multi-dimensional feature vector, jump judgment rules and smoothing parameters are recorded. Then, the updated status instance is synchronized to the gate area management system through the interface module to achieve real-time linkage with the actual load status. Finally, the historical risk status and jump trajectory are archived for subsequent analysis and optimization of risk judgment rules. This mechanism can not only continuously reflect the changes in gate area operating load, but also support reuse in different times and scenarios, thereby improving the intelligent management capabilities of the system.

[0112] The process for determining the risk status level in S4 is as follows:

[0113] Based on the risk status mechanism, obtain the real-time risk status sequence for each aggregated passage unit;

[0114] The system collects real-time operational load data for each aggregation unit, such as the number of access requests, gate occupancy time, and processing queue length. Then, it uses a risk status mechanism to map this data, quantifying the risk status of each unit at different time points into sequence values. It also generates a continuous real-time risk status sequence for each aggregation unit to describe its risk change trend in the current period. This ensures that the risk status of each unit not only reflects its real-time load status but also provides a data basis for hierarchical determination, realizing the quantification and traceability of risk status.

[0115] Based on the trend of load decline in the gate area, the load change in the next cycle is predicted. A constrained risk transfer matrix is ​​established by combining the processing queue length and the available computing power of the edge computing nodes.

[0116] Trend analysis is performed on the gate load sequence of the current period. Weighted moving average or regression prediction methods are used to estimate the possible load level of the next period. The predicted load is combined with resource constraints such as the available computing power of edge computing nodes and the length of processing queues to define the allowable risk jump range and probability distribution. A constrained transition matrix from each risk state node to other possible nodes is formed for jump determination. This ensures that the risk state jump reflects both load changes and system resource constraints, and avoids unreasonable risk jumps caused by fluctuations in a single indicator.

[0117] The risk transition matrix is ​​used to determine and dynamically adjust the instantaneous risk state sequence, and the final risk state level of each aggregated passage unit is output.

[0118] Each node in the real-time risk status sequence is matched with the risk transition matrix, and the weight and probability of each possible transition are calculated. Dynamic judgment is performed based on the transition weight value to select the risk node that best meets the predicted load and resource constraints for the next cycle. Historical accumulated risk and smoothing constraints are introduced to optimize and adjust the transition results, ensuring the continuity and stability of the risk level over time. The adjusted risk node is stored in the risk status instance and associated with the load characteristics, transition history, and prediction data of the aggregation unit. The final risk level is synchronized to the gate area management system through the interface module to realize the visualization and real-time application of the risk status. Finally, the historical risk status and level adjustment results are archived for strategy optimization and security analysis, ensuring that each aggregation unit has a quantifiable and traceable risk status level in any cycle, which not only meets the real-time control requirements of the system but also provides data support for multi-cycle risk management.

[0119] S5: Within the processing cycle, access requests are bound to the corresponding verification path instance according to the risk status level within the door lock gateway, forming a verification set associated with the selected verification path.

[0120] The process of forming a verification set associated with the selected verification path in S5 is as follows:

[0121] Based on the final risk status level and historical jump trajectory of the aggregation unit to which each access request belongs, and combined with the access request attributes, the risk priority weight is calculated, and a weighted verification priority ranking table is generated.

[0122] The final risk status level and historical jump trajectory are quantified to form a dynamic risk index for each request. The attributes of the access request, such as personnel identity level, urgency or special permissions, are combined with the dynamic risk index, and a weighted algorithm is used to calculate the comprehensive priority weight. Finally, all requests are sorted according to the priority weight to generate a weighted verification priority ranking table, which provides a decision basis for verification path allocation. This ensures that the ranking of each access request not only reflects the risk level but also takes into account the access attributes, achieving a dual balance between risk and business needs.

[0123] Based on the sorting table and the constrained risk transfer matrix, each passage request is dynamically assigned to the corresponding verification path instance;

[0124] High-priority requests in the sorting table are mapped according to the node jump paths allowed by the risk transfer matrix to determine their allocable set of verification paths. Based on the current load status of the gate area and the available computing power of the edge computing nodes, the request allocation order and path instances are dynamically adjusted to ensure that high-risk requests are verified in a timely manner, while low-risk requests can be processed with a delay. Finally, the allocation results are updated to the verification path instances. Each path instance contains the corresponding request set and processing order, realizing reasonable scheduling of verification resources and ensuring that the verification path of each request not only meets the risk status constraints but also dynamically responds to changes in the gate area load, achieving real-time optimized allocation.

[0125] By combining the temporal proximity of requests within the aggregation unit and the continuity of gate load, the verification path is merged or split as necessary to generate a multi-level verification set within the processing cycle.

[0126] The system analyzes the timestamp distribution of requests in each verification path and the request density within the aggregation unit. Requests that are close in time and have similar risks are merged into the same level verification set to reduce duplicate verification operations and resource waste. For paths with large time spans or significant load fluctuations, they are split into independent verification sets to ensure a balance between verification efficiency and risk coverage. Finally, the generated multi-level verification sets are recorded in the system instance, including the request members, priority order, and execution conditions of each set, to achieve traceable and reusable verification management. This enables the system to efficiently manage multi-level risk verification within a single processing cycle, while ensuring gate load continuity and system stability.

[0127] Example 2: Please refer to Figure 2 As shown, a campus dormitory smart door lock gateway system based on edge processing includes:

[0128] Load acquisition module: Real-time acquisition of door lock queue length, number of passage triggers and door occupancy time, and mapping it into door zone operation load vector to form a state evolution sequence;

[0129] Transition identification module: Identifies the transition interval from steady state to congestion state based on the load state evolution sequence, and encapsulates the arriving requests into a processing cycle by time windowing;

[0130] Behavior aggregation module: Merges multiple passage requests based on time adjacency and gate occupancy continuity within the processing cycle to generate aggregated passage units that represent the overall passage behavior;

[0131] Risk assessment module: Performs risk state transition assessment based on the load change trend of aggregated access units and gate areas, processing queue characteristics, and edge computing status to determine the risk level;

[0132] Path binding module: Binds access requests within the processing cycle to the corresponding verification path instance according to the risk status level, forming a verification set associated with the selected verification path.

[0133] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0134] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for a smart door lock gateway for campus dormitories based on edge processing, characterized in that, Includes the following steps: The length of the door lock processing queue, the number of passage triggers, and the duration of door occupancy are obtained and mapped to the door area operating load vector. The state evolution sequence of the door area operating load is determined by the combination relationship of the direction consistency and change amplitude between adjacent load vectors. Based on the state evolution sequence, the transition interval of the gate area operation load from steady state to congestion state is identified. When the gate area operation state is in the transition interval, the arriving passage requests are encapsulated in time window to form the corresponding processing cycle. Within the corresponding processing cycle, based on the temporal adjacency relationship of the access requests and the continuity of gate occupancy, multiple access requests are merged to generate an aggregated access unit that represents the overall access behavior. Based on the aggregated access unit, a risk state mechanism coupled with the gate area's operating load status is constructed. Combining the gate area's operating load decline trend, processing queue change characteristics, and edge computing resource status, a constrained risk state transition judgment is performed on the aggregated access unit to determine the risk state level. Within the processing cycle, access requests are bound to the corresponding verification path instance within the door lock gateway according to the risk status level, forming a verification set associated with the selected verification path.

2. The method for a smart door lock gateway for campus dormitories based on edge processing according to claim 1, characterized in that, The process of mapping to the gate region running load vector is as follows: The door lock queue length, number of access triggers, and door occupancy time are collected in real time through edge computing nodes. The collected data is standardized and stored in time series. The vectorization mapping mechanism is used to project the data onto the multi-dimensional gate load space, and a vector representation reflecting the real-time load characteristics of the gate is generated synchronously through timestamps. By combining the historical access records and abnormal event database of the gate area, the similarity between the vector representation and the historical typical load pattern is calculated to obtain the weighted matching value of each vector dimension, and a weighted gate area operation load vector is generated.

3. The method for a smart door lock gateway for campus dormitories based on edge processing according to claim 2, characterized in that, The process of determining the state evolution sequence of the gate zone operating load is as follows: Based on the gate zone operating load vector, calculate the direction cosine and amplitude change rate of each pair of adjacent load vectors to form a direction consistency matrix and an amplitude change matrix. The directional consistency matrix and the amplitude change matrix are mapped into a multi-dimensional state evolution vector through a weighted fusion algorithm, and the current metric vector is adjusted by combining the similarity scores of historical typical load patterns. The adjusted metric vectors are arranged in chronological order to form a continuous sequence. An adaptive threshold determination algorithm is then used to jointly determine the magnitude of metric mutations, continuous change trends, and abnormal weights to generate a state evolution sequence of the gate region's operating load.

4. The campus dormitory smart door lock gateway method based on edge processing according to claim 3, characterized in that, The process of identifying the transition interval from steady state to congested state of the gate region operating load based on the state evolution sequence is as follows: The state evolution sequence is divided into overlapping sliding windows in chronological order. Statistical analysis is performed on the load trend direction, change magnitude and risk weight in each window, and the load mutation rate, continuous growth duration and risk-weighted cumulative value in the window are calculated. Using a sliding window peak detection algorithm, the time period in which the weighted risk cumulative value exceeds a preset dynamic threshold is identified and preliminarily marked as a potential transition interval; The initially identified transition intervals are matched with the historical congestion pattern database of the gate area to calculate the similarity score; By combining the continuity of adjacent time windows, the start and end times of the transition interval are optimized and corrected to obtain the final determined transition interval. Based on the final determined transition interval, a quantitative feature set for determining gate congestion risk is generated.

5. The campus dormitory smart door lock gateway method based on edge processing according to claim 4, characterized in that, The process of forming the corresponding processing cycle is as follows: Based on the quantitative feature set for determining gate congestion risk, each transition interval and additional risk indicators are used as a reference for dividing the time window. Each passage request arriving within the jump interval is clustered according to its temporal proximity and the risk weight of the jump interval to form an initial request group; A time window is allocated to each request group, and the window length is adaptively adjusted based on the sudden congestion magnitude and duration of the transition interval and the length of the gate processing queue. Requests within the time window are packaged into processing cycle instances. Each instance contains a request identifier, arrival time, gate occupancy time, and priority label, and is associated with the risk indicators of the corresponding transition interval to form a corresponding processing cycle.

6. The method for a smart door lock gateway for campus dormitories based on edge processing according to claim 5, characterized in that, The process of generating aggregated traffic units that represent the overall traffic behavior is as follows: The requests within the corresponding processing period are sorted in order of timestamp, and the congestion risk index attached to the transition interval is used to perform a preliminary weighted sort of the requests. Analyze the time interval of requests within the same processing cycle, the overlap of gate occupancy, and historical typical passage patterns, and determine the aggregation conditions through multi-condition fusion; Requests that meet the aggregation conditions are merged to form aggregated passage units, and the number of requests, cumulative occupancy time, time distribution characteristics, aggregation density and risk weighting index within the unit are calculated. The aggregated access unit is coupled with the gate area operating load status, the risk level of the transition interval, and the edge computing resource usage to generate an aggregated access unit that represents the overall access behavior.

7. A campus dormitory smart door lock gateway method based on edge processing according to claim 6, characterized in that, The process of constructing a risk state mechanism coupled with the gate area's operating load state is as follows: By taking the aggregated passage unit as input and combining it with the corresponding processing cycle instance and the risk index of the transition interval, a multi-dimensional state vector is constructed. Based on the multidimensional state vector and state machine node rules, a jump decision is performed on each aggregation unit to dynamically select the next risk node; By introducing continuity constraints and historical accumulated risks, state transitions are smoothed out. The smoothed jump result is updated to the risk state mechanism instance to form a reusable risk state mechanism coupled with the gate area's operating load state.

8. A campus dormitory smart door lock gateway method based on edge processing according to claim 7, characterized in that, The process of determining the risk status level is as follows: Based on the risk status mechanism, obtain the real-time risk status sequence for each aggregated passage unit; Based on the trend of load decline in the gate area, the load change in the next cycle is predicted. A constrained risk transfer matrix is ​​established by combining the processing queue length and the available computing power of the edge computing nodes. The risk transition matrix is ​​used to determine and dynamically adjust the instantaneous risk state sequence, and the final risk state level of each aggregated passage unit is output.

9. A campus dormitory smart door lock gateway method based on edge processing according to claim 8, characterized in that, The process of forming a verification set associated with the selected verification path is as follows: Based on the final risk status level and historical jump trajectory of the aggregation unit to which each access request belongs, and combined with the access request attributes, the risk priority weight is calculated, and a weighted verification priority ranking table is generated. Based on the sorting table and the constrained risk transfer matrix, each passage request is dynamically assigned to the corresponding verification path instance; By combining the temporal proximity of requests within the aggregation unit and the continuity of gate load, the verification path is merged or split as necessary to generate a multi-level verification set within the processing cycle.

10. A campus dormitory smart door lock gateway system based on edge processing, applied to the method described in any one of claims 1-9, characterized in that, include: Load acquisition module: Real-time acquisition of door lock queue length, number of passage triggers and door occupancy time, and mapping it into door zone operation load vector to form a state evolution sequence; Transition identification module: Identifies the transition interval from steady state to congestion state based on the load state evolution sequence, and encapsulates the arriving requests into a processing cycle by time windowing; Behavior aggregation module: Merges multiple passage requests based on time adjacency and gate occupancy continuity within the processing cycle to generate aggregated passage units that represent the overall passage behavior; Risk assessment module: Performs risk state transition assessment based on the load change trend of aggregated access units and gate areas, processing queue characteristics, and edge computing status to determine the risk level; Path binding module: Binds access requests within the processing cycle to the corresponding verification path instance according to the risk status level, forming a verification set associated with the selected verification path.

Images