Intelligent operation and maintenance system of smart community

Abstract

This invention belongs to the field of intelligent operation and maintenance technology, specifically relating to an intelligent operation and maintenance system for a smart community. The system includes: a resident return flow sensing module for acquiring access control event flow density values; a peak interference prediction module for generating a channel interference vector based on the access control event flow density values ​​and inputting it into an interference prediction model to obtain a predicted ZigBee channel interference level; an interference adaptive scheduling module for generating an anti-interference scheduling parameter set when the predicted interference level reaches the interference initiation threshold; a differentiated polling module for polling event-type and periodic-type sensor nodes separately according to the anti-interference scheduling parameter set; and an operation and maintenance integrity guarantee module for restoring the polling interval when the predicted interference level drops below the interference recovery threshold and sequentially retransmitting data frames in the local buffer queue to the upper-layer operation and maintenance platform according to the received timestamp. By predicting interference and implementing differentiated scheduling, reliable transmission of operation and maintenance data is achieved under channel saturation scenarios.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation and maintenance technology, specifically relating to an intelligent operation and maintenance system for smart communities. Background Technology

[0002] The intelligent operation and maintenance system of a smart community deploys edge nodes in buildings to aggregate data from sensors such as access control, elevators, water supply and drainage, and fire protection devices, enabling real-time monitoring of community infrastructure. These sensors generally transmit data based on the ZigBee protocol, operating in the 2.4GHz ISM unlicensed band. This shares spectrum resources with home WiFi routers and Bluetooth devices within residential buildings, posing a risk of co-channel interference. Existing solutions mitigate this by switching to low-interference channels when channel quality deteriorates, but their effectiveness depends on the continuous availability of usable low-interference channels within the frequency band.

[0003] When residents return home during peak hours, a large number of home routers in the building simultaneously enter high-intensity usage. All available ZigBee channels in the 2.4GHz band become saturated at the same time, leaving no available channels for switching, rendering the aforementioned solutions ineffective. Simultaneously, the concentrated return of residents directly triggers a surge in access control record generation, continuous elevator call responses, and the activation of water supply and drainage equipment during peak water usage, resulting in a high degree of overlap between the peak of maintenance data generation and the peak of channel interference. However, existing transmission scheduling schemes select nodes solely based on current channel quality, failing to predict channel saturation trends in this scenario or differentiate the timeliness differences of different types of maintenance data. This leads to frequent data loss during the periods when reliable transmission is most crucial, severely impairing the critical period perception capabilities of the intelligent maintenance system. Summary of the Invention

[0004] (1) Technical problems to be solved The purpose of this invention is to provide an intelligent operation and maintenance system for smart communities to solve the problem that the existing ZigBee transmission scheduling scheme fails to deliver reliable data during peak hours and causes frequent loss of operation and maintenance data during critical periods.

[0005] (2) Technical solution To achieve the above objectives, in one aspect, the present invention provides an intelligent operation and maintenance system for smart communities, the system comprising: The resident return flow sensing module is used to count the cumulative number of resident entry events reported by the access control sensor nodes per unit time within the access control statistics window, and obtain the access control event flow density value.

[0006] The peak interference prediction module is used to continuously update the access control statistics window with a sliding step size and repeatedly trigger the resident return flow sensing module. The access control event flow density value obtained after each update is used to predict the ZigBee channel interference level through the interference prediction model.

[0007] An interference adaptive scheduling module is used to generate an anti-interference scheduling parameter set when the predicted value of the ZigBee channel interference level reaches the interference initiation threshold.

[0008] The differentiated polling module is used to poll the event-type sensor nodes and the periodic sensor nodes respectively according to the anti-interference scheduling parameter set. The data frames received from the event-type sensor nodes in the first polling interval and the data frames received from the periodic sensor nodes in the second polling interval are respectively appended with a receiving timestamp and written into the local cache queue to obtain the local cache queue with timestamp.

[0009] The operation and maintenance integrity assurance module is used to restore the polling interval of the event-type sensor node and the polling interval of the periodic sensor node to the polling interval before the anti-interference scheduling parameter set was generated when the predicted value of the ZigBee channel interference level drops below the interference recovery threshold, and to retransmit the data frames in the local buffer queue with timestamps to the upper-layer operation and maintenance platform in the order of their received timestamps.

[0010] Furthermore, the peak interference prediction module includes: The change in access control event flow density is obtained by calculating the change in access control event flow density value relative to the previous update of access control statistical window after each sliding step size update; the link quality indicator value of each ZigBee channel is continuously collected according to the preset sampling period, and the link quality indicator value of the next sampled link quality indicator value is subtracted from the link quality indicator value of the previous sampled link quality indicator value and divided by the preset sampling period to obtain the link quality indicator change rate.

[0011] A channel interference vector is generated based on the access control event flow density value, the change in access control event flow density, the link quality indicator value, and the link quality indicator change rate; the channel interference vector is then input into the interference prediction model to obtain the predicted value of the ZigBee channel interference level.

[0012] Furthermore, the peak interference prediction module also includes: From the access control event flow density value sequence, access control event flow density change sequence, link quality indicator value sequence, and link quality indicator change rate sequence continuously recorded and stored according to a preset sampling period during the historical operation of the edge node, the access control event flow density value, access control event flow density change, link quality indicator value, and link quality indicator change rate are extracted respectively, and normalized to generate a channel interference vector.

[0013] Furthermore, the construction of the interference prediction model includes: Based on the normalized change in access control event flow density at the current moment, the resident density channel attenuation baseline value is calculated through logarithmic density coupling; the resident density channel attenuation baseline value... The calculation formula is: .

[0014] in, This represents the normalized change in access control event flow density at the current moment. This is the resident density coupling coefficient.

[0015] Based on the deviation between the resident density channel attenuation baseline value and the normalized link quality indicator value at the current time, and combined with the changing direction of the access control event flow density, the link quality asymmetric response correction amount is calculated through asymmetric response analysis; the link quality asymmetric response correction amount... The calculation formula is: .

[0016] in, This is the normalized link quality indicator value at the current moment; This represents the normalized change in access control event flow density at the current moment. It is a symbolic function.

[0017] The predicted channel interference level is obtained by adding the baseline value of channel attenuation due to resident density to the link quality asymmetric response correction.

[0018] Furthermore, the step of obtaining the resident density coupling coefficient includes: By jointly analyzing the degree of synchronization between access control event flow density and link quality indication at sampling times in the historical acquisition sequence of edge nodes, the channel coupling stability weight is obtained; the channel coupling stability weight The calculation formula is: .

[0019] in, The normalized rate of change of the link quality indicator at the current moment; , This represents the maximum value of the normalized change in access control event flow density. This represents the minimum value of the normalized change in access control event flow density. , This represents the maximum value of the normalized link quality indicator change rate. This is the minimum value of the normalized link quality indicator change rate.

[0020] Based on the channel coupling stability weights, a weighted least squares method is used to fit the logarithmic coupling relationship between resident density and ZigBee channel attenuation. An objective function for fitting resident density attenuation is established. One-dimensional numerical optimization is performed within the range where the resident density coupling coefficient is greater than zero to obtain the resident density coupling coefficient that minimizes the objective function. for: .

[0021] in, This represents all sampling periods during the historical operation of the edge node.

[0022] Furthermore, the interference adaptive scheduling module includes The progressive anti-interference response submodule is used to divide the predicted range of ZigBee channel interference levels above the interference initiation threshold into a first interference interval and a second interference interval. The lower boundary value of the second interference interval is higher than the lower boundary value of the first interference interval, and the lower boundary value of the first interference interval is equal to the interference initiation threshold.

[0023] When the predicted interference level enters the first interference interval, an anti-interference scheduling parameter set is generated. The polling interval for event-type sensor nodes is the first polling interval, and the polling interval for periodic sensor nodes is the second polling interval.

[0024] When the predicted interference level value enters the second interference range, the polling interval of the event-type sensor node in the anti-interference scheduling parameter set is updated from the first polling interval to the third polling interval, where the third polling interval is less than the first polling interval, and the polling status of the periodic sensor node is updated to stop polling.

[0025] Furthermore, the operation and maintenance integrity assurance module includes: After the predicted ZigBee channel interference level drops below the interference recovery threshold, the predicted ZigBee channel interference level output by the peak interference prediction module is continuously read at the channel recovery detection cycle. When the predicted interference level is lower than the interference recovery threshold for M consecutive channel recovery detection cycles, the current time is recorded as the channel recovery time. The static buffer duration of each data frame is obtained by subtracting the reception timestamp of each data frame in the local buffer queue with timestamps from the channel recovery time.

[0026] Remove data frames whose static cache duration exceeds the maximum cache duration corresponding to each data frame, and retain the remaining data frames to obtain an effective cache queue.

[0027] Data frames in the valid buffer queue are sent sequentially to the upper-layer operations and maintenance platform according to their received timestamps, with an acknowledgment pending after each frame is sent. If an acknowledgment is received within the waiting time limit, the sent data frame is deleted and the next data frame is sent; if no acknowledgment is received, the data frame is recorded in the retransmission record table. After all data frames in the valid buffer queue have been sent, the data frames in the retransmission record table are retransmitted to the upper-layer operations and maintenance platform in order of their received timestamps from earliest to latest, until the retransmission record table is empty.

[0028] Furthermore, the operation and maintenance integrity assurance module includes: The operation and maintenance event overflow protection submodule is used to calculate the maximum cache time limit for a data frame by multiplying the polling interval of the source node corresponding to the data frame by a preset cache time limit multiplier. For data frames originating from event-type sensor nodes, the maximum cache time limit is calculated by multiplying the first polling interval by the preset cache time limit multiplier, and for data frames originating from periodic sensor nodes, the maximum cache time limit is calculated by multiplying the second polling interval by the preset cache time limit multiplier. The dynamic cache duration is obtained by subtracting the receiving timestamp of each data frame from the current system time, and data frames whose dynamic cache duration exceeds the corresponding maximum cache time limit are identified as timeout data frames.

[0029] Read the device type identifier from the timeout data frame; for those originating from event-type sensor nodes, extract the device number field value, event type code field value, and original event timestamp field value from the data frame body of the timeout data frame, generate an overflow record, write it to the edge node, and then delete the data frame; for timeout data frames originating from periodic sensor nodes, delete them directly.

[0030] After all timed-out data frames have been processed, the occupancy rate of the current number of data frames in the local cache queue with timestamps relative to the maximum capacity of the queue is calculated. When the occupancy rate is lower than the polling recovery occupancy threshold, the polling permission status of the differentiated polling module is updated to allowed, and normal polling scheduling is restored according to the historical polling interval of each node.

[0031] Furthermore, the operation and maintenance integrity assurance module includes: The channel-aware retransmission submodule is used by edge nodes to maintain a sliding window of the real-time link quality indicator value of the ZigBee channel, and dynamically determines the retransmission start threshold and retransmission pause threshold based on the statistics of the sliding window. When the real-time value of the link quality indicator reaches the retransmission start threshold, all data frames with device type identifier of event-type sensor node are extracted from the local cache queue with timestamp, arranged in ascending order according to the original event timestamp, and sent to the upper-layer operation and maintenance platform frame by frame; after each frame is sent, the real-time value of the link quality indicator is read. If the real-time value of the link quality indicator is lower than the retransmission pause threshold, the sending stops and the monitoring state is re-entered; otherwise, the next frame is sent. An adaptive acknowledgment timeout timer is started for each sent data frame. The duration of the adaptive acknowledgment timeout timer is determined based on the round-trip delay statistics of acknowledgment responses collected historically by the edge node. When an acknowledgment response is received before the timeout timer reaches zero, the sent data frame is deleted from the queue. When no acknowledgment response is received before the timeout timer reaches zero, the retransmission count is accumulated. When the retransmission count exceeds the preset maximum number of retransmissions, the reception timestamp of this data frame is updated to the current system time and the retransmission count is reset. The updated data frame re-participates in the buffer duration determination process.

[0032] (3) Beneficial effects Compared with the prior art, the beneficial effects of the present invention are: 1. By quantifying ZigBee channel interference and differentiating the scheduling of sensor nodes at the edge nodes, and after interference recovery, the cached data is retransmitted to the upper-layer operation and maintenance platform in the order of the received timestamp, thus realizing reliable transmission of critical operation and maintenance data in channel saturation scenarios.

[0033] 2. By using the channel-aware retransmission submodule to opportunistically retransmit the data frames cached by the sensor nodes during channel quality improvement, the integrity of the operation and maintenance data transmission is further improved. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the module composition of an intelligent operation and maintenance system for a smart community according to Embodiment 1 of the present invention. Detailed Implementation

[0035] 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. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0036] Before providing examples, it is necessary to describe the application scenarios of this invention. This invention is applicable to the intelligent operation and maintenance of residential buildings.

[0037] Example 1: As Figure 1 As shown, this embodiment provides an intelligent operation and maintenance system for a smart community, the system comprising: The resident return flow sensing module is used to count the cumulative number of resident entry events reported by the access control sensor nodes per unit time within the access control statistics window, and obtain the access control event flow density value.

[0038] The peak interference prediction module is used to continuously update the access control statistics window with a sliding step size and repeatedly trigger the resident return flow sensing module. The access control event flow density value obtained after each update is used to predict the ZigBee channel interference level through the interference prediction model.

[0039] An interference adaptive scheduling module is used to generate an anti-interference scheduling parameter set when the predicted value of the ZigBee channel interference level reaches the interference initiation threshold.

[0040] The differentiated polling module is used to poll the event-type sensor nodes and the periodic sensor nodes respectively according to the anti-interference scheduling parameter set. The data frames received from the event-type sensor nodes in the first polling interval and the data frames received from the periodic sensor nodes in the second polling interval are respectively appended with a receiving timestamp and written into the local cache queue to obtain the local cache queue with timestamp.

[0041] The operation and maintenance integrity assurance module is used to restore the polling interval of the event-type sensor node and the polling interval of the periodic sensor node to the polling interval before the anti-interference scheduling parameter set was generated when the predicted value of the ZigBee channel interference level drops below the interference recovery threshold, and to retransmit the data frames in the local buffer queue with timestamps to the upper-layer operation and maintenance platform in the order of their received timestamps.

[0042] For example, edge nodes are deployed in the building's low-voltage electrical room, aggregating data reported by access control sensor nodes on each floor via a ZigBee coordinator. The resident return flow sensing module continuously counts the total number of resident entry events reported by the access control sensor nodes within a 10-minute access control statistics window and divides this count by 10 minutes to obtain the access control event flow density value. Taking a 32-story residential building as an example, the access control event flow density value during normal off-peak hours is approximately 0.5 to 1 times / minute, typically rising to 3 to 7 times / minute during weekday evening peak hours. The access control statistics window is continuously updated with a 1-minute sliding step. After each slide, the resident return flow sensing module recounts the total number of entry events within the window and calculates a new access control event flow density value, which is then input into the interference prediction model. The interference prediction model outputs a ZigBee channel interference level prediction value. Through the continuous sliding of the access control statistics window, the edge nodes can detect the increasing density trend of the interference level prediction value in advance. Taking the evening rush hour on a weekday as an example, at 18:02, the access control event flow density was 5.2 times / minute and the interference level prediction value was 5 (below the interference initiation threshold of 6). At 18:03, the density value rose to 6.1 times / minute and the interference level prediction value jumped to 8 (above the interference initiation threshold of 6). There is about a 1 to 2 minute window from the jump in the ZigBee channel interference level prediction value to the actual effectiveness of the scheduling parameter set, which reserves sufficient time for the interference adaptive scheduling module to issue parameters and adjust polling.

[0043] It should be noted that a hysteresis band is formed between the interference initiation threshold and the interference recovery threshold to prevent the predicted interference level from repeatedly triggering the defense mode entry and exit when it fluctuates near the critical point. The predicted interference level of the ZigBee channel is represented by integers from 0 to 10, where 0 corresponds to the best historical channel condition and 10 corresponds to the most severe historical interference condition. The interference initiation threshold is set to 6, indicating that the channel has significantly degraded but is not yet fully saturated. The interference recovery threshold is set to 4, forming a two-level hysteresis band between the two. After the evening peak in residential buildings, home WiFi routers do not immediately exit high-intensity usage; they typically fluctuate slightly between levels 4 and 6 at the tail end of the peak, and the hysteresis band width covers this aftershock range.

[0044] The differentiated polling module appends the current system time as a reception timestamp to each data frame successfully received from event-driven and periodic sensor nodes before writing it into the local buffer queue. This reception timestamp determines the data frame transmission order in subsequent retransmission stages and coexists with the original event timestamp of the data frame (recording the time the sensor node detected the event, for the upper-layer operations and maintenance platform to reconstruct the event sequence). After the predicted ZigBee channel interference level drops below the interference recovery threshold, the operations and maintenance integrity assurance module restores the polling interval of each node after confirming stable channel recovery and retransmits the data frames accumulated in the timestamp-bearing local buffer queue to the upper-layer operations and maintenance platform in order of reception timestamp from earliest to latest. After the channel recovery time is determined, the operations and maintenance event overflow preservation submodule first cleans up timed-out data frames in the timestamp-bearing local buffer queue to obtain a valid buffer queue, and then starts the main transmission process according to the reception timestamp order. The channel-aware retransmission submodule operates independently during interference and does not overlap with the main transmission process. The peak interference prediction module includes: The change in access control event flow density is obtained by calculating the change in access control event flow density value relative to the previous update of access control statistical window after each sliding step size update; the link quality indicator value of each ZigBee channel is continuously collected according to the preset sampling period, and the link quality indicator value of the next sampled link quality indicator value is subtracted from the link quality indicator value of the previous sampled link quality indicator value and divided by the preset sampling period to obtain the link quality indicator change rate.

[0045] A channel interference vector is generated based on the access control event flow density value, the change in access control event flow density, the link quality indicator value, and the link quality indicator change rate; the channel interference vector is then input into the interference prediction model to obtain the predicted value of the ZigBee channel interference level.

[0046] For example, after each update of the access control statistics window, the peak interference prediction module subtracts the access control event flow density value obtained in the previous update from the access control event flow density value obtained in the current update to obtain the change in access control event flow density. Taking the evening peak on a weekday as an example: from 18:02 to 18:12, 52 entry events were counted in the access control statistics window, with an access control event flow density value of 5.2 times / minute; from 18:03 to 18:13, 61 entry events were counted in the access control statistics window, with an access control event flow density value of 6.1 times / minute; the change in access control event flow density is 6.1 minus 5.2, which equals 0.9 times / minute. The reason for introducing the change in access control event flow density is that residents' concentrated return home is not instantaneous, but rather gradually accumulates from sporadic to dense over a period of about 30 to 50 minutes. In the early stages of the upward process, the absolute value of the access control event flow density may still be at an intermediate level. Based solely on its absolute value, it is impossible to distinguish between the two states of "currently rising rapidly and will enter a peak" and "long-term stability at this level without further increase." The change in access control event flow density can directly reflect this trend difference, providing additional time-series slope information for the interference prediction model.

[0047] Link quality indication (LMI) values ​​for each ZigBee channel are continuously collected with a sampling period of 1 second. The difference between two consecutive LMI values ​​is divided by 1 second to obtain the LMI change rate. If the LMI value of a channel at time t is 78 and the LMI value at time t+1 is 72, the LMI change rate is 72 minus 78 divided by 1 second, which equals -6. Because the link quality indicator value is dimensionless. The access control event flow density value is 6.1 times / minute, the access control event flow density change is 0.9 times / minute, the link quality indicator value is 78, and the link quality indicator change rate is negative 6. Both are used as input vectors to the interference prediction model to obtain the predicted value of the ZigBee channel interference level.

[0048] The peak interference prediction module also includes: From the access control event flow density value sequence, access control event flow density change sequence, link quality indicator value sequence, and link quality indicator change rate sequence continuously recorded and stored according to a preset sampling period during the historical operation of the edge node, the access control event flow density value, access control event flow density change, link quality indicator value, and link quality indicator change rate are extracted respectively, and normalized to generate a channel interference vector.

[0049] For example, if the access control event flow density value, access control event flow density change, link quality indicator value, and link quality indicator change rate are directly concatenated into an input vector, the absolute magnitude of the link quality indicator value is much higher than the other three parameters. During training, the interference prediction model will assign a disproportionate weight to the link quality indicator value, while the contribution of the access control event flow density change, which is only in the single digits, to the prediction result will be significantly compressed. Min-Max normalization is used to map the four quantities to the interval between 0 and 1. In the historical sequence of edge nodes, the minimum access control event flow density value is 0 times / minute, and the maximum is 9.4 times / minute; the minimum access control event flow density change sequence is -2.8 times / minute, and the maximum is 2.6 times / minute; the minimum link quality indicator value sequence is 31, and the maximum is 253; the minimum link quality indicator change rate sequence is -28. The maximum value is 14 The four raw values ​​at the current moment are 6.1 times / minute, 0.9 times / minute, 78, and -6. The normalized access control event flow density value is (6.1-0) / (9.4-0)=0.649, the normalized access control event flow density change is (0.9-(-2.8)) / (2.6-(-2.8))=0.685, the normalized link quality indicator value is (78-31) / (253-31)=0.212, and the normalized link quality indicator change rate is ((-6)-(-28)) / (14-(-28))=0.524. The four normalized values ​​are concatenated to form a four-dimensional input vector [0.649,0.685,0.212,0.524], which is then input into the interference prediction model.

[0050] The construction of the interference prediction model includes: Based on the normalized change in access control event flow density at the current moment, the resident density channel attenuation baseline value is calculated through logarithmic density coupling; the resident density channel attenuation baseline value... The calculation formula is: .

[0051] in, This represents the normalized change in access control event flow density at the current moment. This is the resident density coupling coefficient.

[0052] Based on the deviation between the resident density channel attenuation baseline value and the normalized link quality indicator value at the current time, and combined with the changing direction of the access control event flow density, the link quality asymmetric response correction amount is calculated through asymmetric response analysis; the link quality asymmetric response correction amount... The calculation formula is: .

[0053] in, This is the normalized link quality indicator value at the current moment; This represents the normalized change in access control event flow density at the current moment. It is a symbolic function.

[0054] The predicted channel interference level is obtained by adding the baseline value of channel attenuation due to resident density to the link quality asymmetric response correction.

[0055] For example, consider an edge node of a 32-story residential building. The historical minimum access control event flow density value sequence is 0 times / minute, and the maximum is 9.4 times / minute. The current access control event flow density value is 6.1 times / minute, and the normalized access control event flow density value is (6.1-0) / (9.4-0) = 0.649. Resident density coupling coefficient. =4.2, substituting into the calculation, the baseline value of channel attenuation due to resident density is ln(1+4.2×0.649) / ln(1+4.2)=0.797. This reflects the diminishing marginal effect of resident density on channel interference. When the building is sparsely populated, the increase of a small number of residents has a significant impact on channel interference; when a large number of people are present, the additional impact of the same number of residents on the channel tends to saturate.

[0056] The current link quality indicator (BMI) is 78. The historical BMI sequence shows a minimum of 31 and a maximum of 253. The normalized BMI is (78-31) / (253-31) = 0.212. (1-0.212) = 0.788 indicates the interference level reflected in the current BMI. The baseline attenuation value for the resident density channel is 0.797, which is higher than 0.788, indicating that the BMI has not yet fully reflected the expected interference corresponding to the density level. After residents enter the building, it takes time for them to disperse in the corridors on each floor, and the decrease in the BMI lags behind the increase in the access control event flow density. However, once residents reach their apartments, the corridor line of sight recovers rapidly, and the recovery rate of the BMI is faster than the decrease in the density value within the access control statistical window. This asymmetry is captured by the sign function of the access control event flow density change in the link quality asymmetric response correction. The current access control event flow density change is +0.9 times / minute, the normalized access control event flow density change is positive, the sign function value is +1, and the link quality asymmetric response correction is (0.797-0.788)×(+1)=+0.009.

[0057] Channel interference level prediction =0.797 + 0.009 = 0.806. Multiplying 0.806 by 10 and rounding down yields an interference level of 8, which is higher than the interference initiation threshold of 6. Based on this, the interference adaptive scheduling module triggers the second interference interval response. If only the resident density channel attenuation baseline value of 0.797 is used, multiplying it by 10 and rounding down also yields an interference level of 7. In this embodiment, the mapping results of the two differ by one level, which exactly crosses the boundary between the first and second interference intervals. The link quality asymmetric response correction amount causes the predicted value to trigger a stronger scheduling response one level earlier in the early stage of density increase, giving the differentiated polling module an advance of about 1 to 2 minutes.

[0058] The steps for obtaining the resident density coupling coefficient include: By jointly analyzing the degree of synchronization between access control event flow density and link quality indication at sampling times in the historical acquisition sequence of edge nodes, the channel coupling stability weight is obtained; the channel coupling stability weight The calculation formula is: .

[0059] in, The normalized rate of change of the link quality indicator at the current moment; , This represents the maximum value of the normalized change in access control event flow density. This represents the minimum value of the normalized change in access control event flow density. , This represents the maximum value of the normalized link quality indicator change rate. This is the minimum value of the normalized link quality indicator change rate.

[0060] Based on the channel coupling stability weights, a weighted least squares method is used to fit the logarithmic coupling relationship between resident density and ZigBee channel attenuation. An objective function for fitting resident density attenuation is established. One-dimensional numerical optimization is performed within the range where the resident density coupling coefficient is greater than zero to obtain the resident density coupling coefficient that minimizes the objective function. for: .

[0061] in, This represents all sampling periods during the historical operation of the edge node.

[0062] For example, the historical data of the edge nodes of the aforementioned 32-story residential building is used. During the dynamic transition period, i.e., the process of residents entering or leaving in large numbers, the access control event flow density value and the link quality indicator value have not yet reached a response balance. If the transition period samples are used with equal weights in the fitting, this imbalance will be systematically introduced into the estimation of the crowd density coupling coefficient, leading to… This deviates from the true steady-state logarithmic coupling relationship between resident density and channel attenuation. Channel coupling stability weights are calculated for each sampling time in the historical acquisition sequence to ensure that samples from stable periods dominate the fitting process.

[0063] Taking the evening rush hour at 18:07:00 as an example, The value is (0.9) / (2.6-(-2.8))=0.167. for =0.143, the channel coupling stability weight is 1 / (1+0.167+0.143)=0.714, both variables are changing rapidly, and the weights are too low. At a stable moment during off-peak hours, 0.012 The value is 0.009, and the channel coupling stability weight is 1 / (1+0.012+0.009)=0.979, which is close to 1.

[0064] Substituting all historical sampling times into the population density decay fitting objective function, and using the golden section method... One-dimensional numerical optimization is performed within the range (0,20), using two adjacent iterations. The absolute value of the difference between the values ​​being less than 0.001 was used as the convergence criterion, and convergence was achieved after 23 iterations. It is stable within the interval (4.1, 4.3), and eventually takes... =4.2.

[0065] The interference adaptive scheduling module includes: The progressive anti-interference response submodule is used to divide the predicted range of ZigBee channel interference levels above the interference initiation threshold into a first interference interval and a second interference interval. The lower boundary value of the second interference interval is higher than the lower boundary value of the first interference interval, and the lower boundary value of the first interference interval is equal to the interference initiation threshold.

[0066] When the predicted interference level enters the first interference interval, an anti-interference scheduling parameter set is generated. The polling interval for event-type sensor nodes is the first polling interval, and the polling interval for periodic sensor nodes is the second polling interval.

[0067] When the predicted interference level value enters the second interference range, the polling interval of the event-type sensor node in the anti-interference scheduling parameter set is updated from the first polling interval to the third polling interval, where the third polling interval is less than the first polling interval, and the polling status of the periodic sensor node is updated to stop polling.

[0068] For example, the predicted interference level of the ZigBee channel is represented by integers from 0 to 10, with an interference initiation threshold of 6. This corresponds to the interference prediction model determining that the link quality indicator value at the next sampling time will fall into the historical low-to-medium level, indicating that the channel has significantly degraded but is not yet fully saturated. The interference recovery threshold is 4, forming a two-level hysteresis band between the two. After the evening peak in residential buildings, home WiFi routers do not immediately cease high-intensity use; the link quality indicator value typically fluctuates slightly between level 4 and 6 at the tail end of the peak.

[0069] The first interference range is set at 6 to 7, and the second interference range is set at 8 to 10. The middle boundary is set at level 8 because the interference prediction model for level 8 predicts that the link quality indicator value will be lower than the degradation baseline value during training. That is, the channel will remain within the quality range corresponding to the interference saturation sample in the next 5 minutes. Although the predicted mean value has deviated from the normal range at levels 6 to 7, it has not yet fallen below the degradation baseline value. Although the channel baseline quality has deteriorated, it still has basic transmission capability. Differential frequency reduction can maintain the collection of operation and maintenance data.

[0070] When the predicted interference level of the ZigBee channel enters the first interference range, an anti-interference scheduling parameter set is generated. The polling interval for event-type sensor nodes is set to a first polling interval of 500 milliseconds, and the polling interval for periodic sensor nodes is set to a second polling interval of 30 seconds. Under normal conditions, the polling interval for event-type sensor nodes is 200 milliseconds. When interference intensifies, the probability of channel collisions increases, and high-frequency polling at 200 milliseconds will generate a large number of collision retransmissions, resulting in an actual effective throughput lower than the level after moderate frequency reduction. Therefore, it is adjusted to 500 milliseconds. The polling interval for periodic sensor nodes is increased from the normal 10 seconds to 30 seconds, rather than shortened. This is because periodic sensor nodes collect continuous quantities such as water pressure and temperature. During the stage of intensified interference, moderately reducing the collection frequency can free up channel capacity for event-type sensor nodes. 30 seconds is still within the acceptable range of data validity for periodic sensor nodes and does not affect the judgment of continuous monitoring trends.

[0071] When the predicted interference level of the ZigBee channel enters the second interference range, periodic sensor nodes stop polling, and the polling interval of event-driven sensor nodes is compressed from 500 milliseconds to a third polling interval of 200 milliseconds. 200 milliseconds is the minimum reliable interval for a single data frame transmission plus acknowledgment in the ZigBee protocol at 2.4 GHz. Below this value, channel occupancy of adjacent data frames overlaps, and the probability of collision increases sharply. 200 milliseconds is the highest polling frequency that event-driven sensor nodes can stably maintain when exclusively occupying the channel under conditions of complete channel saturation. The third polling interval of 200 milliseconds is the same as the normal polling interval of event-driven sensor nodes before the generation of the anti-interference scheduling parameter set. However, under normal conditions, periodic sensor nodes occupy the channel simultaneously, while at this time, the periodic sensor nodes have stopped polling, and the channel resources are exclusively occupied by the event-driven sensor nodes, resulting in a higher actual success rate under the same polling interval.

[0072] The operation and maintenance integrity assurance module includes: After the predicted ZigBee channel interference level drops below the interference recovery threshold, the predicted ZigBee channel interference level output by the peak interference prediction module is continuously read at the channel recovery detection cycle. When the predicted interference level is lower than the interference recovery threshold for M consecutive channel recovery detection cycles, the current time is recorded as the channel recovery time. The static buffer duration of each data frame is obtained by subtracting the reception timestamp of each data frame in the local buffer queue with timestamps from the channel recovery time.

[0073] Remove data frames whose static cache duration exceeds the maximum cache duration corresponding to each data frame, and retain the remaining data frames to obtain an effective cache queue.

[0074] Data frames in the valid buffer queue are sent sequentially to the upper-layer operations and maintenance platform according to their received timestamps, with an acknowledgment pending after each frame is sent. If an acknowledgment is received within the waiting time limit, the sent data frame is deleted and the next data frame is sent; if no acknowledgment is received, the data frame is recorded in the retransmission record table. After all data frames in the valid buffer queue have been sent, the data frames in the retransmission record table are retransmitted to the upper-layer operations and maintenance platform in order of their received timestamps from earliest to latest, until the retransmission record table is empty.

[0075] For example, the output judgment of the interference prediction model is that the predicted value of the ZigBee channel interference level drops below the interference recovery threshold. This reflects that the interference trend inferred from the access control event flow density value has subsided. However, home WiFi routers do not immediately switch to low-power mode after residents stop using them intensively. The interference in the 2.4GHz band will continue to oscillate for several minutes after the peak ends. With M=5 and the channel recovery detection period=1 minute, the subsequent retransmission process is triggered only if the predicted value of the ZigBee channel interference level is below the interference recovery threshold of 4 for 5 consecutive minutes. The aftershock of the 2.4GHz band in residential buildings usually lasts for 2 to 4 minutes. During this period, the predicted value of the interference level may repeatedly cross the interference recovery threshold. With M=2, it is very likely that the retransmission will be triggered prematurely before the aftershock has completely subsided. If the channel deteriorates again during the retransmission process, the retransmission data frame will be lost. With M=5, the duration of most aftershocks is covered. When the value is below the threshold for 5 consecutive minutes, the channel is basically stable. The detection period of 1 minute is consistent with the sliding step size of the resident return flow sensing module. The edge node does not need to maintain a separate sampling clock for this judgment.

[0076] The buffer duration for each data frame is calculated based on the channel recovery time (channel recovery time minus the receiving timestamp). This reflects the dwell time of the data frame before it reaches the upper-layer operation and maintenance platform after the interference ends, rather than the total dwell time of the data frame at the edge node. Buffering data frames during interference is a normal design behavior. If the total duration is used for timeout judgment, almost all buffered data frames will be judged to have timed out when the interference duration is long, rendering the retransmission process meaningless. The maximum buffer duration for each data frame is calculated and determined by the operation and maintenance event overflow protection submodule based on the source node polling interval and the preset buffer duration multiplier. The maximum buffer duration for event-type sensor nodes is 60 minutes. The channel recovery time is 19:51:00. The received timestamp of a data frame from an event-type sensor node is 18:52:37, and the buffer duration is 58 minutes and 23 seconds, which is less than 60 minutes. This data frame is retained in the effective buffer queue. The received timestamp of another data frame is 18:43:11, and the buffer duration is 67 minutes and 49 seconds, which is more than 60 minutes. It is removed from the local buffer queue with timestamps. Unacknowledged data frames are collected and retransmitted uniformly after being recorded in the retransmission record table, rather than being retransmitted immediately. The reason is that immediate retransmission would disrupt the orderly sending rhythm of the effective buffer queue, causing the data frames received by the upper-layer operation and maintenance platform to be interleaved between the main transmission and retransmission, increasing the complexity of reconstructing the timing. There are 43 data frames in the effective buffer queue. The data frames are sent one by one in the order of their received timestamps. If the 7th and 19th data frames do not receive an acknowledgment within the 3-second waiting time limit, they are recorded in the retransmission record table. After all 43 data frames have been sent, the 7th and 19th data frames in the retransmission record table are retransmitted to the upper-level operation and maintenance platform in order of their received timestamps from earliest to latest, until the retransmission record table is empty.

[0077] The operation and maintenance integrity assurance module includes: The operation and maintenance event overflow protection submodule is used to calculate the maximum cache time limit for a data frame by multiplying the polling interval of the source node corresponding to the data frame by a preset cache time limit multiplier. For data frames originating from event-type sensor nodes, the maximum cache time limit is calculated by multiplying the first polling interval by the preset cache time limit multiplier, and for data frames originating from periodic sensor nodes, the maximum cache time limit is calculated by multiplying the second polling interval by the preset cache time limit multiplier. The dynamic cache duration is obtained by subtracting the receiving timestamp of each data frame from the current system time, and data frames whose dynamic cache duration exceeds the corresponding maximum cache time limit are identified as timeout data frames.

[0078] Read the device type identifier from the timeout data frame; for those originating from event-type sensor nodes, extract the device number field value, event type code field value, and original event timestamp field value from the data frame body of the timeout data frame, generate an overflow record, write it to the edge node, and then delete the data frame; for timeout data frames originating from periodic sensor nodes, delete them directly.

[0079] After all timed-out data frames have been processed, the occupancy rate of the current number of data frames in the local cache queue with timestamps relative to the maximum capacity of the queue is calculated. When the occupancy rate is lower than the polling recovery occupancy threshold, the polling permission status of the differentiated polling module is updated to allowed, and normal polling scheduling is restored according to the historical polling interval of each node.

[0080] For example, the duration of peak-hour interference in residential buildings typically ranges from 30 to 50 minutes. The maximum buffer time for event-type sensor nodes needs to cover this range with a margin, and is set at 60 minutes. The first polling interval is 500 milliseconds, and 60 minutes is converted to 3,600,000 milliseconds. Dividing 3,600,000 by 500 gives the preset buffer time multiplier of 7200. The preset buffer time multiplier is based on the polling interval rather than a directly fixed buffer time, allowing the maximum buffer time to adaptively float with the polling interval of each node. The polling interval is configured by maintenance personnel according to the actual needs of the building. Once the preset buffer time multiplier is determined, the buffer time does not need to be recalibrated due to adjustments in the polling interval. The maximum buffer time for event-type sensor nodes is 500 milliseconds × 7200 = 60 minutes; the maximum buffer time for periodic sensor nodes is 30 seconds × 7200 = 60 hours. The 60-minute maximum buffer time limit for event-driven sensor nodes covers the vast majority of interference duration periods; the 60-hour maximum buffer time limit for periodic sensor nodes is a safety recovery boundary set to deal with abnormal scenarios with significant delays in channel recovery, rather than an actual judgment condition that is expected to be triggered during normal operation.

[0081] It should be noted that the dynamic buffer duration obtained by subtracting the receiving timestamp from the current system time in the operation and maintenance event overflow protection submodule is different from the buffer duration obtained by subtracting the receiving timestamp from the channel recovery time in the main transmission process of the operation and maintenance integrity assurance module. The former is used to determine whether the data frame has been backed up in the queue for too long and needs to be actively cleared, while the latter is used to determine whether the data frame still has retransmission value after the channel is restored. They should not be used interchangeably.

[0082] The current system time is 20:15:30. The received timestamp of a data frame from an event-type sensor node is 19:12:18, and the dynamic buffer duration is 63 minutes and 12 seconds, exceeding the maximum buffer duration of 60 minutes for event-type sensor nodes. Therefore, it is determined to be a timeout data frame. The device number field value "E-07", the event type code field value "0x03", and the original event timestamp field value "19:12:11.508" are extracted from the data frame body. An overflow record "E-07|0x03|19:12:11.508" is generated, appended to the timeout overflow log in the edge node's built-in non-volatile memory, and then the data frame is deleted. The reason why timeout data frames from event-type sensor nodes are written to non-volatile memory instead of being directly discarded is that event-type sensor nodes record discrete events such as access control anomalies, elevator malfunctions, and smoke detector triggers. Each record corresponds to a real operational event. Directly discarding the data would permanently eliminate this event from the records of the upper-level operational platform. Writing the data to non-volatile memory ensures that the logs can still be read after the edge node is powered off and restarted. Operations personnel can manually re-enter the missing events by reading the timeout overflow logs. Periodic sensor nodes record periodic readings of continuous quantities such as water pressure and temperature. The loss of a single reading does not affect the reconstruction of the continuous monitoring trend. After all timeout data frames have been processed, the maximum queue capacity is 512 data frames, the polling recovery occupancy threshold is 60%, and the local cache queue with timestamps has 284 data frames remaining. The occupancy rate is 284 / 512, which is approximately 55.5%, lower than 60%. The polling permission status of the differentiated polling module is changed from paused to allowed, and normal polling scheduling is restored according to the historical polling interval of each node.

[0083] The polling resumption threshold is set based on the requirement that sufficient capacity must be freed up in the queue after timeout data frame cleanup before polling can resume. If the occupancy rate is too high (e.g., 90%), the queue is still nearly full, and there is almost no space to write new data frames after polling resumes. If the occupancy rate is too low (e.g., 20%), it means that polling must wait for the retransmission process to consume a large number of valid cached data frames before resuming, during which time new events from event-driven sensor nodes cannot be collected. Resuming polling at 60% corresponds to 40% queue capacity remaining. This 40% is sufficient to accommodate continuous writing of new data frames under off-peak polling frequencies, while the remaining valid cached data frames in the queue are still within the processing range of the retransmission process. With a maximum queue capacity of 512 data frames, 60% corresponds to 307 data frames. After timeout data frame cleanup, 284 data frames remain, representing an occupancy rate of approximately 55.5%, which is below 60%. The polling permission status of the differentiated polling module changes from paused to allowed, and normal polling scheduling resumes based on the historical polling intervals of each node.

[0084] It should be noted that after the channel recovery time is determined, the operation and maintenance event overflow protection submodule first cleans up the timed-out data frames in the local buffer queue with timestamps to obtain the valid buffer queue, and then starts the main transmission process according to the received timestamp order; the channel-aware retransmission submodule runs independently during the interference period and does not overlap with the main transmission process.

[0085] The operation and maintenance integrity assurance module includes: The channel-aware retransmission submodule is used by edge nodes to maintain a sliding window of the real-time link quality indicator value of the ZigBee channel, and dynamically determines the retransmission start threshold and retransmission pause threshold based on the statistics of the sliding window. When the real-time value of the link quality indicator reaches the retransmission start threshold, all data frames with device type identifier of event-type sensor node are extracted from the local cache queue with timestamp, arranged in ascending order according to the original event timestamp, and sent to the upper-layer operation and maintenance platform frame by frame; after each frame is sent, the real-time value of the link quality indicator is read. If the real-time value of the link quality indicator is lower than the retransmission pause threshold, the sending stops and the monitoring state is re-entered; otherwise, the next frame is sent. An adaptive acknowledgment timeout timer is started for each sent data frame. The duration of the adaptive acknowledgment timeout timer is determined based on the round-trip delay statistics of acknowledgment responses collected historically by the edge node. When an acknowledgment response is received before the timeout timer reaches zero, the sent data frame is deleted from the queue. When no acknowledgment response is received before the timeout timer reaches zero, the retransmission count is accumulated. When the retransmission count exceeds the preset maximum number of retransmissions, the reception timestamp of this data frame is updated to the current system time and the retransmission count is reset. The updated data frame re-participates in the buffer duration determination process.

[0086] For example, the retransmission initiation threshold and retransmission pause threshold are dynamically determined based on the statistics of a sliding window of the real-time link quality indicator. The channel baseline quality varies across different buildings: the average link quality indicator value of the ZigBee channel in older buildings may remain between 150 and 170 during non-interference periods, while in newly built buildings it may remain between 210 and 230. Setting a fixed retransmission initiation threshold of 180 would be too lenient for older buildings and too delayed for newly built buildings; dynamically determining the threshold using sliding window statistics allows the retransmission triggering conditions of each building to adapt to its own channel baseline level.

[0087] The sliding window length is set to 20 detection cycles, with each detection cycle lasting 1 minute. This means the average link quality indicator value over the most recent 20 minutes is used as the current channel baseline quality. The preset retransmission start margin is 20, and the preset retransmission pause margin is 40. The retransmission start threshold is the current channel baseline quality minus 20, and the retransmission pause threshold is the current channel baseline quality minus 40. This creates a 20-unit hysteresis interval to prevent repeated start and pause switching when the link quality indicator value fluctuates slightly around the retransmission start threshold.

[0088] The average link quality indicator value within the current sliding window of a certain building is 188, the retransmission start threshold is 168, and the retransmission pause threshold is 148. The real-time link quality indicator reading is 172, which is higher than the retransmission start threshold of 168, triggering a retransmission. All 17 data frames with device type identifiers of event-type sensor nodes are extracted from the local buffer queue with timestamps and arranged in ascending order according to the original event timestamps to form the sequence of data frames to be retransmitted.

[0089] It should be noted that the retransmission order is based on the original event timestamp, while the order in the main transmission process of the operation and maintenance integrity assurance module is based on the receiving timestamp. The former deals with opportunistic retransmissions within the channel improvement window, aiming to complete the events as much as possible in chronological order of occurrence; the latter deals with systematic batch retransmissions after the channel has been stably restored, using the receiving timestamp to ensure that the retransmission order is consistent with the buffer order of the edge nodes.

[0090] After sending the 4th frame, the real-time value of the link quality indicator is read as 145, which is lower than the retransmission pause threshold of 148. Therefore, the transmission is stopped, and the remaining 13 frames are retained in the local buffer queue with timestamps. The system then re-enters the monitoring state.

[0091] The duration of the adaptive acknowledgment timeout timer is determined based on the historical acknowledgment response round-trip latency statistics collected from edge nodes. The timer duration is calculated as the mean round-trip latency plus twice the standard deviation. For example, the historical round-trip latency mean for a certain edge node is 320 milliseconds, and the standard deviation is 95 milliseconds; therefore, the acknowledgment timeout timer duration is 510 milliseconds. These statistics are periodically recalculated and the timer duration is updated using newly collected round-trip latency samples from the current month, ensuring continuous tracking of the current network status.

[0092] After the retransmission count exceeds the preset maximum number of retransmissions (3 times in this embodiment), the receiving timestamp of the data frame is updated to the current system time and the retransmission count is reset, so that the data frame re-participates in the buffer duration judgment with the current time as the starting point, obtains a complete maximum buffer time limit window and continues to wait for subsequent retransmission opportunities; if the waiting period expires again, the timeout overflow log writing process of the operation and maintenance event overflow protection submodule is entered.

[0093] Finally, it should be noted that although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent operation and maintenance system for a smart community, characterized in that, The system includes: The resident return flow sensing module is used to count the cumulative number of resident entry events reported by the access control sensor nodes per unit time within the access control statistics window, and obtain the access control event flow density value. The peak interference prediction module is used to continuously update the access control statistics window with a sliding step size and repeatedly trigger the resident return flow sensing module. The access control event flow density value obtained after each update is used to predict the ZigBee channel interference level prediction value through the interference prediction model. An interference adaptive scheduling module is used to generate an anti-interference scheduling parameter set when the predicted value of the ZigBee channel interference level reaches the interference initiation threshold. The differentiated polling module is used to poll the event-type sensor nodes and the periodic sensor nodes respectively according to the anti-interference scheduling parameter set. The data frames received from the event-type sensor nodes in the first polling interval and the data frames received from the periodic sensor nodes in the second polling interval are respectively appended with a receiving timestamp and written into the local cache queue to obtain the local cache queue with timestamp. The operation and maintenance integrity assurance module is used to restore the polling interval of the event-type sensor node and the polling interval of the periodic sensor node to the polling interval before the anti-interference scheduling parameter set was generated when the predicted value of the ZigBee channel interference level drops below the interference recovery threshold, and to retransmit the data frames in the local buffer queue with timestamps to the upper-layer operation and maintenance platform in the order of their received timestamps.

2. The intelligent operation and maintenance system for a smart community according to claim 1, characterized in that, The peak interference prediction module includes: The change in access control event flow density is obtained by calculating the change in access control event flow density value relative to the previous update of access control statistics window after each sliding step size update; the link quality indicator value of each ZigBee channel is continuously collected according to the preset sampling period, and the link quality indicator value of the next sampled link quality indicator value is subtracted from the link quality indicator value of the previous sampled link quality indicator value and divided by the preset sampling period to obtain the link quality indicator change rate. A channel interference vector is generated based on the access control event flow density value, the change in access control event flow density, the link quality indicator value, and the link quality indicator change rate; the channel interference vector is then input into the interference prediction model to obtain the predicted value of the ZigBee channel interference level.

3. The intelligent operation and maintenance system for a smart community according to claim 2, characterized in that, The peak interference prediction module also includes: From the access control event flow density value sequence, access control event flow density change sequence, link quality indicator value sequence, and link quality indicator change rate sequence continuously recorded and stored according to a preset sampling period during the historical operation of the edge node, the access control event flow density value, access control event flow density change, link quality indicator value, and link quality indicator change rate are extracted respectively, and normalized to generate a channel interference vector.

4. The intelligent operation and maintenance system for a smart community according to claim 3, characterized in that, The construction of the interference prediction model includes: Based on the normalized change in access control event flow density at the current moment, the resident density channel attenuation baseline value is calculated through logarithmic density coupling; the resident density channel attenuation baseline value... The calculation formula is: ； in, This represents the normalized change in access control event flow density at the current moment. The population density coupling coefficient; Based on the deviation between the resident density channel attenuation baseline value and the normalized link quality indicator value at the current time, and combined with the changing direction of the access control event flow density, the link quality asymmetric response correction amount is calculated through asymmetric response analysis; the link quality asymmetric response correction amount... The calculation formula is: ； in, This is the normalized link quality indicator value at the current moment; This represents the normalized change in access control event flow density at the current moment. It is a symbolic function; The predicted channel interference level is obtained by adding the baseline value of channel attenuation due to resident density to the link quality asymmetric response correction.

5. The intelligent operation and maintenance system for a smart community according to claim 4, characterized in that, The steps for obtaining the resident density coupling coefficient include: By jointly analyzing the degree of synchronization between access control event flow density and link quality indication at sampling times in the historical acquisition sequence of edge nodes, the channel coupling stability weight is obtained; the channel coupling stability weight The calculation formula is: ； in, The normalized rate of change of the link quality indicator at the current moment; , This represents the maximum value of the normalized change in access control event flow density. This represents the minimum value of the normalized change in access control event flow density. , This represents the maximum value of the normalized link quality indicator change rate. This represents the minimum normalized link quality indicator change rate. Based on the channel coupling stability weights, a weighted least squares method is used to fit the logarithmic coupling relationship between resident density and ZigBee channel attenuation. An objective function for fitting resident density attenuation is established. One-dimensional numerical optimization is performed within the range where the resident density coupling coefficient is greater than zero to obtain the resident density coupling coefficient that minimizes the objective function. for: ； in, This represents all sampling periods during the historical operation of the edge node.

6. The intelligent operation and maintenance system for a smart community according to claim 1, characterized in that, The interference adaptive scheduling module includes The progressive anti-interference response submodule is used to divide the predicted range of ZigBee channel interference levels above the interference initiation threshold into a first interference interval and a second interference interval. The lower boundary value of the second interference interval is higher than the lower boundary value of the first interference interval, and the lower boundary value of the first interference interval is equal to the interference initiation threshold. When the predicted interference level value enters the first interference interval, an anti-interference scheduling parameter set is generated. The polling interval of the event-type sensor node is the first polling interval, and the polling interval of the periodic sensor node is the second polling interval. When the predicted interference level value enters the second interference range, the polling interval of the event-type sensor node in the anti-interference scheduling parameter set is updated from the first polling interval to the third polling interval, where the third polling interval is less than the first polling interval, and the polling status of the periodic sensor node is updated to stop polling.

7. The intelligent operation and maintenance system for a smart community according to claim 6, characterized in that, The operation and maintenance integrity assurance module includes: After the predicted ZigBee channel interference level drops below the interference recovery threshold, the predicted ZigBee channel interference level output by the peak interference prediction module is continuously read at the channel recovery detection cycle. When the predicted interference level is lower than the interference recovery threshold for M consecutive channel recovery detection cycles, the current time is recorded as the channel recovery time. The static buffer duration of each data frame is obtained by subtracting the reception timestamp of each data frame in the local buffer queue with timestamp from the channel recovery time. Remove data frames whose static cache duration exceeds the maximum cache duration corresponding to each data frame, and retain the remaining data frames to obtain an effective cache queue; Data frames in the valid buffer queue are sent sequentially to the upper-layer operation and maintenance platform according to their received timestamps. After each data frame is sent, an acknowledgment is awaited. If an acknowledgment is received within the waiting time limit, the sent data frame is deleted and the next data frame is sent. If no acknowledgment is received, the data frame is recorded in the retransmission record table. After all data frames in the valid buffer queue have been sent, the data frames in the retransmission record table are retransmitted to the upper-layer operation and maintenance platform in order of their received timestamps from earliest to latest, until the retransmission record table is empty.

8. The intelligent operation and maintenance system for a smart community according to claim 7, characterized in that, The operation and maintenance integrity assurance module includes: The operation and maintenance event overflow protection submodule is used to calculate the maximum cache time limit for a data frame by multiplying the polling interval of the source node corresponding to the data frame by a preset cache time limit multiplier. For data frames originating from event-type sensor nodes, the maximum cache time limit is calculated by multiplying the first polling interval by the preset cache time limit multiplier, and for data frames originating from periodic sensor nodes, the maximum cache time limit is calculated by multiplying the second polling interval by the preset cache time limit multiplier. The dynamic cache duration is obtained by subtracting the receiving timestamp of each data frame from the current system time, and data frames whose dynamic cache duration exceeds the corresponding maximum cache time limit are identified as timeout data frames. Read the device type identifier from the timeout data frame; for those originating from event-type sensor nodes, extract the device number field value, event type code field value, and original event timestamp field value from the data frame body of the timeout data frame, generate an overflow record, write it to the edge node, and then delete the data frame; timeout data frames originating from periodic sensor nodes are deleted directly. After all timed-out data frames have been processed, the occupancy rate of the current number of data frames in the local cache queue with timestamps relative to the maximum capacity of the queue is calculated. When the occupancy rate is lower than the polling recovery occupancy threshold, the polling permission status of the differentiated polling module is updated to allowed, and normal polling scheduling is restored according to the historical polling interval of each node.

9. The intelligent operation and maintenance system for a smart community according to claim 8, characterized in that, The operation and maintenance integrity assurance module includes: The channel-aware retransmission submodule is used by edge nodes to maintain a sliding window of the real-time link quality indicator value of the ZigBee channel, and dynamically determines the retransmission start threshold and retransmission pause threshold based on the statistics of the sliding window. When the real-time value of the link quality indicator reaches the retransmission start threshold, all data frames with device type identifier of event-type sensor node are extracted from the local cache queue with timestamp, arranged in ascending order according to the original event timestamp, and sent to the upper-layer operation and maintenance platform frame by frame; after each frame is sent, the real-time value of the link quality indicator is read. If the real-time value of the link quality indicator is lower than the retransmission pause threshold, the sending stops and the monitoring state is re-entered; otherwise, the next frame is sent. An adaptive acknowledgment timeout timer is started for each sent data frame. The duration of the adaptive acknowledgment timeout timer is determined based on the round-trip delay statistics of acknowledgment responses collected historically by the edge node. When an acknowledgment response is received before the timeout timer reaches zero, the sent data frame is deleted from the queue. When no acknowledgment response is received before the timeout timer reaches zero, the retransmission count is accumulated. When the retransmission count exceeds the preset maximum number of retransmissions, the reception timestamp of this data frame is updated to the current system time and the retransmission count is reset. The updated data frame re-participates in the buffer duration determination process.

Images