Internet of Things intelligent water meter ad hoc network communication method based on parameter dynamic evaluation

Abstract

The invention relates to the technical field of Internet of Things communication, in particular to an Internet of Things intelligent water meter ad hoc network communication method based on parameter dynamic evaluation, which comprises the following steps: acquiring a state beacon packet; calculating a comprehensive score of each neighbor node according to the NB-IoT signal quality parameter, the Bluetooth link quality parameter, the remaining power percentage and the historical relay task times; judging whether the comprehensive score exceeds a threshold value or not according to the comprehensive score, if the comprehensive score exceeds a first threshold value, selecting a target node, establishing Bluetooth GATT connection, transmitting the water meter data to the target node, uploading the water meter data to a cloud server in batches in a preset low-power-consumption time window, and if the comprehensive score is lower than the first threshold value but higher than a second threshold value, transmitting the water meter data to the cloud server; and if so, starting a limited flooding broadcast mechanism. Therefore, the problems that in the prior art, power consumption is high, replacement cost is high, operation and maintenance are inconvenient, cost is increased by depending on a fixed relay, no load balance exists, and manpower is increased by on-site reading are solved.

Description

Technical Field

[0001] This application relates to the field of wireless communication technologies, and particularly to an Internet of Things intelligent water meter self-organizing network communication method based on parameter dynamic evaluation. Background Art

[0002] In the current process of smart water service construction, Internet of Things remote water meters, as key devices for realizing automatic collection and transmission of water usage data, have been widely used. Currently, most common Internet of Things remote water meters on the market rely on a single remote communication technology, such as NB-IoT technology, to complete the upload of water usage data. This mode can basically meet the data transmission requirements in a conventional environment. However, in a complex environment, many problems are exposed. In areas such as underground wells, corridor wells, and other areas with dense pipelines and metals, or in places with weak network coverage, the signal is easily blocked or there is weak coverage, which may lead to communication interruption between the water meter and the remote server, resulting in the inability to upload water usage data in a timely manner, seriously affecting the efficiency and accuracy of water service management.

[0003] Regarding the above communication interruption problem, traditional solutions have many drawbacks. First, adopting the passive retransmission method, the communication module will continuously and repeatedly attempt to establish a connection. This process not only has a high power consumption but also a low connection success rate. The long-term high-power operation is extremely likely to cause excessive loss of the water meter battery, shorten the battery life, and increase the equipment maintenance cost. Second, although changing the communication method can improve the communication situation to a certain extent, on-site equipment replacement involves a series of complex operations such as hardware replacement and software debugging, with high costs. At the same time, it is difficult to uniformly coordinate the equipment replacement time in different regions, and the operation and maintenance are extremely inconvenient, bringing huge management pressure to water service enterprises. Third, relying on fixed relay nodes can enhance signal transmission, but it requires additional deployment of dedicated relay equipment, which undoubtedly increases the installation and operation and maintenance costs. Moreover, the existing solutions lack an effective load balancing mechanism, resulting in some relay nodes being overloaded and local batteries being depleted prematurely, affecting the stability of the entire system. Fourth, the method of on-site meter reading by maintenance personnel not only consumes a large amount of manpower and increases the labor cost, but also the process of meter reading, sorting, and management of data is cumbersome and prone to human errors, further increasing the management cost. Summary of the Invention

[0004] This application provides an Internet of Things intelligent water meter self-organizing network communication method based on parameter dynamic evaluation to solve the problems in the prior art, such as high power consumption, high replacement cost and inconvenient operation and maintenance, increased cost due to reliance on fixed relays and lack of load balancing, and increased manpower for on-site meter reading.

[0005] The first aspect of the present application provides an IoT smart water meter self-organizing network communication method based on dynamic parameter evaluation, comprising the following steps: obtaining a status beacon packet, wherein the status beacon packet includes the NB-IoT signal quality parameters, Bluetooth link quality parameters, remaining power percentage and historical relay task times of the neighboring node; calculating a comprehensive score for each neighboring node based on the NB-IoT signal quality parameters, the Bluetooth link quality parameters, the remaining power percentage and the historical relay task times; judging whether a threshold is exceeded based on the comprehensive score, wherein if the comprehensive score exceeds a first threshold, selecting a target node and establishing a Bluetooth GATT connection, transmitting water meter data to the target node, and uploading the water meter data in batches to a cloud server within a preset low power consumption time window; if the comprehensive score is lower than the first threshold but higher than the second threshold, starting a restricted flooding broadcast mechanism.

[0006] Optionally, a comprehensive score for each neighbor node is calculated based on the NB-IoT signal quality parameters, the Bluetooth link quality parameters, the remaining power percentage and the number of historical relay tasks, including: normalizing the NB-IoT signal quality parameters, the Bluetooth link quality parameters, the remaining power percentage and the number of historical relay tasks; weighted summing the normalized parameters according to preset weight coefficients to obtain a comprehensive score, wherein the preset weight coefficients include the NB-IoT signal quality parameter weight, the Bluetooth link quality parameter weight, the remaining power percentage weight and the historical relay task number weight, and the sum of the weight coefficients is 1.

[0007] Optionally, the dynamic optimization of the preset weight coefficients includes: locally recording the performance indicators of historical relay tasks, the performance indicators including data packet success rate, transmission delay, and power consumption rate; encrypting and uploading the performance indicators to the cloud, and training new weights based on a reinforcement learning model in the cloud; and sending the optimized weights to all nodes in the network, which will take effect after the nodes verify the signature.

[0008] Optionally, the triggering conditions for the dynamic optimization include: the average data packet success rate of the entire network for three consecutive natural days is lower than 95%; the proportion of node pairs with a remaining power difference of more than 30% accounts for more than 20% of the total number of node pairs in the entire network; the network topology change rate per unit time exceeds 10%, wherein the formula for the network topology change rate is: .

[0009] Optionally, the reinforcement learning model satisfies a state space, an action space and a reward function, wherein the state space includes node density, average NB signal strength and historical load imbalance, the action space is a weight coefficient adjustment step less than or equal to 0.05, and the formula of the reward function is: ; Among them, R is the reward value.

[0010] Optionally, the dynamic optimization of the preset weight coefficients further includes a federated learning mechanism, including the following contents: Each water meter node calculates a model gradient based on local historical relay data, and encrypts the model gradient using a homomorphic encryption algorithm; The encrypted gradient is uploaded to the cloud through the NB-IoT channel, and the cloud uses the aggregated private key to decrypt and aggregate all node gradients; The cloud uses aggregated gradients to update the global model and verifies the performance of the new model through a sandbox environment; If the new model causes the average data success rate of the entire network to drop by more than 5%, or the difference in power consumption between nodes increases by more than 10%, it will automatically roll back to the previous version of the model; The optimized model parameters are sent to each node through a digital signature, and the node updates the local weight coefficient after verifying the validity of the signature.

[0011] Optionally, the preset low power consumption time window is determined in the following manner: the cloud server uniformly sends the time reference and local activation time period for the whole network synchronization; each node activates the networking function according to the time reference and the local activation time period, and the time error is controlled within ±1 minute.

[0012] Optionally, the initiation of the restricted flood broadcast mechanism includes: setting the initial TTL value of the flood broadcast to 2; assigning a unique message identifier to the data packet; the relay node deduplicates the data packet according to the unique message identifier, and reduces the TTL value by 1 after forwarding, and stops forwarding when the TTL value is 0.

[0013] Optionally, before obtaining the status beacon packet, the following steps are included: after the IoT smart water meter node fails to upload NB-IoT or Cat.1 data, the Bluetooth module is activated and a fault notification is broadcast, wherein the upload failure includes three consecutive connection timeouts, a data packet loss rate >50%, or a permanent connection rejection error code is received; the Bluetooth scanning parameters are configured to filter the scanned Bluetooth devices and only retain the neighbor node beacon packets carrying specific service UUIDs.

[0014] The second aspect of the present application provides an Internet of Things intelligent water meter self-organizing communication system based on parameter dynamic evaluation, including: an acquisition module for acquiring status beacon packets, where the status beacon packets include NB-IoT signal quality parameters, Bluetooth link quality parameters, remaining battery percentage, and historical relay task times of neighbor nodes; a calculation module for calculating a comprehensive score for each neighbor node according to the NB-IoT signal quality parameters, the Bluetooth link quality parameters, the remaining battery percentage, and the historical relay task times; a processing module for determining whether the comprehensive score exceeds a threshold, where if the comprehensive score exceeds a first threshold, a target node is selected and a Bluetooth GATT connection is established, water meter data is transmitted to the target node, and is batch uploaded to a cloud server within a preset low-power time window, and if the comprehensive score is lower than the first threshold but higher than a second threshold, a restricted flooding broadcast mechanism is started.

[0015] Thus, the present application has at least the following beneficial effects: In the embodiments of the present application, by acquiring parameters such as NB-IoT signal quality, Bluetooth link quality, remaining battery, and historical relay task times of neighbor nodes and calculating a comprehensive score, the optimal relay node can be accurately selected, avoiding the problem of unbalanced load of traditional fixed relays; based on a grading relay mechanism that dynamically switches between Bluetooth GATT direct connection and restricted flooding broadcast based on a score threshold, combined with batch data upload within a preset low-power time window, the daily average power consumption is controlled at 95 μAh, reducing the power consumption by more than 40% compared with the traditional passive retransmission scheme, and at the same time avoiding the high cost of replacing communication modules on site; through dynamic evaluation and adaptive relay, there is no need to deploy dedicated relay nodes, reducing hardware and operation and maintenance costs, and the data upload success rate is increased to 99.6%, fundamentally solving the problems of high power consumption, high operation and maintenance costs, relay load imbalance, and low manual meter reading efficiency in the prior art.

[0016] Additional aspects and advantages of the present application will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of the embodiments in conjunction with the drawings, where: Figure 1 is a flowchart of an Internet of Things intelligent water meter self-organizing communication method based on parameter dynamic evaluation provided according to an embodiment of the present application; Figure 2 is a schematic diagram from main channel communication failure to successful Bluetooth relay according to an embodiment of the present application; Figure 3It is a block diagram example of an Internet of Things intelligent water meter self-organizing network communication system based on parameter dynamic evaluation provided by an embodiment of the present application; Figure 4 It is an architecture diagram of an Internet of Things intelligent water meter self-organizing network communication system based on multi-parameter dynamic evaluation provided by an embodiment of the present application; Figure 5 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0018] The embodiments of the present application will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present application, but should not be construed as limiting the present application.

[0019] An Internet of Things intelligent water meter self-organizing network communication method based on parameter dynamic evaluation according to an embodiment of the present application will be described below with reference to the accompanying drawings. In view of the problem of high power consumption mentioned in the above background art, the present application provides an Internet of Things intelligent water meter self-organizing network communication method based on parameter dynamic evaluation. In this method, by obtaining parameters such as the NB-IoT signal quality, Bluetooth link quality, remaining battery power, and historical relay task times of neighbor nodes and calculating a comprehensive score, the optimal relay node can be accurately selected to avoid the load imbalance problem of traditional fixed relays; based on the score threshold, a hierarchical relay mechanism that dynamically switches between Bluetooth GATT direct connection and restricted flooding broadcast is adopted, combined with batch data upload in a preset low-power time window, the daily average power consumption is controlled within 95 μAh, reducing the power consumption by more than 40% compared with the traditional passive retransmission scheme, and at the same time avoiding the high cost of on-site replacement of communication modules; through dynamic evaluation and adaptive relay, there is no need to deploy dedicated relay nodes, reducing hardware and operation and maintenance costs, and the data upload success rate is increased to 99.6%, fundamentally solving the problems of high power consumption, high operation and maintenance costs, relay load imbalance, and low manual meter reading efficiency in the prior art.

[0020] An Internet of Things intelligent water meter self-organizing network communication method based on parameter dynamic evaluation according to an embodiment of the present application will be described below with reference to the accompanying drawings.

[0021] Specifically, Figure 1 It is a flowchart of an Internet of Things intelligent water meter self-organizing network communication method based on parameter dynamic evaluation provided by an embodiment of the present application.

[0022] As Figure 1 shown, the Internet of Things intelligent water meter self-organizing network communication method based on parameter dynamic evaluation includes the following steps: In step S101, a status beacon packet is obtained.

[0023] Among them, the status beacon packet may include the NB-IoT signal quality parameters, Bluetooth link quality parameters, remaining battery percentage, and historical relay task times of neighbor nodes. Among them, the NB-IoT signal quality parameters may include the reference signal received power and the signal-to-interference-plus-noise ratio, and the Bluetooth link quality parameters may include the received signal strength indication and the BLE connection success rate.

[0024] It can be understood that by obtaining the status beacon packet in the embodiment of the present application, a comprehensive and accurate data basis is provided for subsequent analysis.

[0025] In the embodiment of the present application, before obtaining the status beacon packet, it includes: after the NB-IoT or Cat.1 data upload of the Internet of Things intelligent water meter node fails, activating the Bluetooth module and broadcasting a fault notification; configuring Bluetooth scanning parameters, filtering the scanned Bluetooth devices, and only retaining the neighbor node beacon packets carrying specific service UUIDs.

[0026] Among them, the upload failure may include three consecutive connection timeouts, a packet loss rate > 50%, or receiving a permanent connection rejection error code, and the fault notification may include the node ID, data digest, and emergency level.

[0027] It can be understood that in the embodiment of the present application, by automatically activating the Bluetooth module and broadcasting a fault notification when the NB-IoT / Cat.1 upload fails, an adaptive communication link redundancy mechanism is constructed: when network anomalies such as three consecutive connection timeouts and a packet loss rate exceeding 50% are detected, trigger a Bluetooth fault broadcast, carry key information such as the node ID, data digest, and emergency level, broadcast continuously at 100ms intervals for 30 seconds, and at the same time configure Bluetooth scanning parameters with a 300ms scanning window / 1000ms interval and filter specific UUID beacon packets, and available relay nodes in the vicinity can be quickly discovered within 30 seconds; this mechanism can achieve automatic switching of the communication link without manual intervention. Compared with the traditional on-site replacement module solution, the fault response time is shortened from 30 minutes to 30 seconds, the single-device operation and maintenance cost is reduced by more than 80%, and due to the low-power consumption characteristics of Bluetooth relay (broadcast mode power consumption < 10mA), the high-power consumption problem of repeated NB-IoT retransmission is avoided, while ensuring that the data upload success rate is increased from 60% to more than 95%.

[0028] It should be noted that in the Internet of Things intelligent water meter system, the fault detection and mode switching mechanism operates efficiently: when the water meter node fails to report data three times in a row, the Bluetooth module will be quickly activated to broadcast a fault notification packet containing key information such as the node ID, data summary, and emergency level to inform surrounding nodes of its own fault status in a timely manner. At the same time, the system ensures the stability and reliability of communication through the neighbor node dynamic discovery strategy. Normal nodes will periodically broadcast status beacons, which cover important information such as network signal quality, remaining battery power, and historical relay times; while fault nodes use the BLE scanning function to obtain data such as the network signal quality, remaining battery power, historical relay times, and Bluetooth signal strength of normal nodes, and filter low-quality nodes based on this information to select better-quality neighbor nodes.

[0029] In step S102, the comprehensive score of each neighbor node is calculated according to the NB-IoT signal quality parameter, Bluetooth link quality parameter, remaining battery percentage, and historical relay task times.

[0030] It can be understood that in the embodiment of the present application, by quantifying multi-dimensional parameters such as NB-IoT signal quality, Bluetooth link quality, node remaining battery power, and historical relay task times, a standardized scoring model is formed, providing a scientific basis for the selection of relay nodes, breaking through the limitations of traditional single-link quality assessment, realizing the comprehensive quantification of the reliability of relay nodes, and avoiding the blindness of fixed relay deployment.

[0031] In the embodiment of the present application, calculating the comprehensive score of each neighbor node according to the NB-IoT signal quality parameter, Bluetooth link quality parameter, remaining battery percentage, and historical relay task times includes: the weight of the NB-IoT signal quality parameter, the weight of the Bluetooth link quality parameter, the weight of the remaining battery percentage, and the weight of the historical relay task times; performing weighted summation on the normalized parameters according to the preset weight coefficients to obtain the comprehensive score.

[0032] Among them, the preset weight coefficients include the weight of the NB-IoT signal quality parameter, the weight of the Bluetooth link quality parameter, the weight of the remaining battery percentage, and the weight of the historical relay task times, and the sum of the weight coefficients is 1.

[0033] It can be understood that in the embodiments of the present application, the weights of NB-IoT signal quality parameters, Bluetooth link quality parameters, remaining battery percentage weights, and historical relay task times weights are normalized, and the values of each parameter are mapped to a unified numerical range to eliminate the dimension difference, so that different parameters can be measured under the same standard; by flexibly adjusting the influence proportion of each parameter through weights, an adaptive relay node evaluation model is formed, which solves the problems of insufficient evaluation dimension of traditional single parameter and rigid weight configuration, realizes the scientific and quantitative selection of relay nodes, balances communication reliability and node energy consumption, and avoids relay load imbalance and resource waste.

[0034] It should be noted that the formula for calculating the comprehensive score is: Among them, is the normalization function, RSRP is the NB-IoT signal quality, α is the weight of the NB-IoT signal quality parameter, is the Bluetooth link quality parameter, β is the weight of the Bluetooth link quality parameter, is the remaining battery percentage, γ is the weight of the remaining battery percentage, is the historical relay task times, δ is the weight of the historical relay task times; For the NB-IoT signal quality parameter, the defined effective range is [-150dBm, -60dBm], and the normalization formula is: For the Bluetooth link quality parameter, the defined effective range is [-95dBm, -60dBm], and the normalization formula is: .

[0035] In the embodiments of the present application, the dynamic optimization of the preset weight coefficient includes: locally recording the performance indicators of historical relay tasks; encrypting and uploading the performance indicators to the cloud, and the cloud trains new weights based on the reinforcement learning model; sending the optimized weights to all network nodes, and the nodes take effect and update after verifying the signature.

[0036] Among them, the performance indicators may include packet success rate, transmission delay, and power consumption rate.

[0037] It is understandable that the embodiment of the present application records historical relay task performance indicators in the local device, covering key information such as task completion time, data transmission success rate, signal strength changes, etc., to provide basic data for subsequent analysis and optimization, and local records can achieve fast data acquisition, reduce transmission delays, and improve the efficiency of local decision-making and adjustment of equipment. Then, these performance indicators are uploaded to the cloud in an encrypted manner, which not only ensures the security of data during transmission, prevents sensitive information of equipment and user water data from being stolen or tampered with, but also facilitates the cloud to use powerful computing and storage capabilities to centrally manage and analyze a large amount of equipment data and explore potential rules and problems. Afterwards, the cloud is based on the reinforcement learning model, based on the uploaded historical performance indicator data, continuously learns and tries different weight combinations, and intelligently optimizes the weight coefficient based on the feedback results such as relay task success rate and power consumption, and can also perceive changes in the Internet of Things environment in real time, such as signal strength, network congestion, etc., and dynamically adjust the weight to keep the equipment relay performance at a better level. Finally, the optimized weights are distributed to all nodes in the network. After the nodes verify the signatures, the updates take effect and the whole network is synchronized to prevent the overall network performance from being affected by unreasonable parameter configuration of some devices. At the same time, the signature verification mechanism provides security for the weight distribution process, ensuring that the data source is reliable and complete, thereby comprehensively improving the relay performance, stability and security of the IoT system in different scenarios.

[0038] Specifically, the training process: a) Each node generates a training sample every day and uploads it to the cloud.

[0039] b) Cloud-based centralized training: Experience pool construction: Aggregate no less than 10,000 pieces of data to form a sample pool.

[0040] The core steps of the reinforcement learning model are: calculate The advantage function is: in, is the discount factor, is the immediate reward obtained at time step t, Status The value function of Status The value function of The Clip mechanism is used to limit the range of policy changes, where: The loss function is: in, To calculate the expectation for time step t, is the importance sampling ratio, is the clipping function, is the cropping parameter; The training is stopped when the reward fluctuation is < 5% for 10 consecutive iterations.

[0041] c) Model issuance and effectiveness: The difference between the new and old model parameters is quantized into INT8, with a compression rate of >80%. The model file is signed by ECC-SECP256R1 and sent to all nodes through the NB-IoT network or Cat.1 network. After the node receives it, the signature is verified in the MCU security isolation area, and the model weight is updated atomically.

[0042] In the embodiment of the present application, the triggering conditions for dynamic optimization include: the average data packet success rate of the entire network for three consecutive natural days is less than 95%; the proportion of node pairs with a remaining power difference of more than 30% accounts for more than 20% of the total number of node pairs in the entire network; the network topology change rate per unit time exceeds 10%.

[0043] The formula for the network topology change rate is: .

[0044] It is understandable that the dynamic optimization triggering conditions set in the embodiment of the present application are, first, that the average data packet success rate of the entire network for three consecutive natural days is less than 95%. The data packet success rate is the core indicator for measuring the communication performance of the system. It can keenly capture the frequent errors or losses in the data transmission process, avoid the failure of important information to be transmitted in a timely and accurate manner, and ensure the quality of data transmission. At the same time, it can perceive the impact of network environment changes on the system in advance, adjust parameters in time to optimize the communication path, and enhance the system's adaptive ability and stability. The second is that the proportion of node pairs with a remaining power difference of more than 30% accounts for more than 20% of the total number of node pairs in the entire network. The large difference in the remaining power of different nodes in the Internet of Things will cause some nodes to run out of power prematurely, affecting life and stability. It can accurately detect the problem of power imbalance, balance node energy consumption through dynamic optimization, reasonably allocate tasks, extend system operation time, reduce operation and maintenance costs, and optimize resource allocation. Third, the network topology change rate per unit time exceeds 10%. The IoT network topology may change rapidly due to factors such as device movement, increase or decrease, and signal interference. It can timely perceive the dynamic changes in the network structure, re-plan the communication path, and select more stable and efficient nodes for data relay to ensure the normal operation of the system when the network topology changes frequently, enhance robustness, and enable it to better respond to emergencies.

[0045] Specifically, the dynamic trigger mechanism: Trigger conditions and optimization actions: If the regional data success rate is <95% (for 3 days), the α weight is increased and the δ weight is reduced; if the node power difference is >30%, the γ weight is increased to trigger federated learning retraining; if the network topology change rate is >10%, the experience pool is reset and a new round of reinforcement learning training is started.

[0046] In the embodiments of the present application, the reinforcement learning model satisfies the state space, action space, and reward function. Among them, the state space includes node density, average NB signal strength, and historical load imbalance degree. The action space is that the weight coefficient adjustment step size is less than or equal to 0.05. The formula of the reward function is: ; where R is the reward value.

[0047] It can be understood that the reinforcement learning model in the embodiments of the present application constructs a state space covering node density, average NB signal strength, and historical load imbalance degree, and an action space where the weight coefficient adjustment step size is less than or equal to 0.05. In the state space, the node density reflects the network topology characteristics and affects the formulation of relay strategies. For example, short paths and reliable nodes are selected in high-density areas, and nodes are flexibly selected in low-density areas to expand coverage. The average NB signal strength measures the communication quality, and the weight adjustment can be guided according to its change. When the signal is weak, the weight of the signal enhancement parameter is increased, and when the signal is strong, the weights of other parameters are optimized. The historical load imbalance degree reveals the system load problem, prompting the model to predict in advance and adjust the strategy to achieve load balance. In terms of the action space, restricting the weight coefficient adjustment step size can ensure the smoothness of the adjustment, avoid large fluctuations in performance, achieve a smooth transition, and can also improve the training accuracy, enabling the model to explore finely in the parameter space, reducing the possibility of falling into the local optimal trap, and making it easier to find the global optimal solution.

[0048] Specifically, the reinforcement learning model architecture is adopted, and the PPO (Proximal Policy Optimization) algorithm is used to realize the dynamic optimization of the weight coefficient. The model architecture includes the following components: State space: includes a 6-dimensional vector [node density, average RSRP, average RSSI, power variance, load imbalance degree, data success rate] Action space: 4-dimensional continuous vector: [Δα, Δβ, Δγ, Δδ], the adjustment step size is ±0.05, and the sum constraint is 0 Reward function: ; where R is the reward value.

[0049] Policy network: 3-layer fully connected network (input 6 → 64 → 32 → 4), activation function: ReLU in the middle layer, Tanh in the output layer; Value network: 3-layer fully connected network (input 6 → 64 → 32 → 1), outputting the state value estimate.

[0050] In an embodiment of the present application, the dynamic optimization of the preset weight coefficient also includes a federated learning mechanism, including the following contents: each water meter node calculates the model gradient based on the local historical relay data, and encrypts the model gradient using a homomorphic encryption algorithm; the encrypted gradient is uploaded to the cloud through the NB-IoT channel, and the cloud uses an aggregated private key to decrypt and aggregate all node gradients; the cloud uses the aggregated gradient to update the global model, and verifies the performance of the new model through a sandbox environment; if the new model causes the average data success rate of the entire network to drop by more than 5%, or the power consumption difference between nodes increases by more than 10%, it will automatically roll back to the previous version of the model; the optimized model parameters are sent to each node through a digital signature, and the node verifies the validity of the signature and updates the local weight coefficient.

[0051] Among them, local historical relay data can be the data generated and recorded by the water meter node when performing relay tasks in the past, including relay time, transmission data volume, signal status and other information. The sandbox environment is an isolated, virtual test environment. It is used to perform performance testing and verification of new models without affecting the actual production system. It can simulate various real-life scenarios and data inputs to evaluate the performance of new models.

[0052] It is understandable that the embodiment of the present application introduces a federated learning mechanism in the dynamic optimization of the preset weight coefficient. Each water meter node calculates the model gradient based on the local historical relay data, encrypts it using the homomorphic encryption algorithm, and uploads it to the cloud through the NB-IoT channel, which not only reduces the amount of data transmission, reduces the bandwidth pressure, but also ensures the security of data during transmission. The cloud uses the aggregated private key to decrypt and aggregate all node gradients to comprehensively reflect the system operation status, and then accurately update the global model to improve the generalization and accuracy of the model. The cloud verifies the performance of the new model in the sandbox environment. If the new model causes the average data success rate of the entire network to drop by more than 5% or the difference in power consumption between nodes increases by more than 10%, the system automatically rolls back to the previous version to correct the problem in the model update in time. The optimized model parameters are digitally signed and sent to each node. After the node verifies the validity of the signature, it updates the local weight coefficient to ensure that the parameter source is reliable and complete, and achieves synchronous optimization of the entire network, which effectively improves the performance and reliability of the IoT water meter system.

[0053] Specifically, the federated learning process is: a) Node uses local dataset Compute the gradient: in, represents the local gradient calculated by the i-th node, is the gradient operator, is the loss function, are model parameters, is the local data set of the i-th node; Differential Privacy: Adding Gaussian Noise , satisfying ; b) Gradient Encryption Upload: Homomorphic Encryption: Using the Paillier algorithm, the public key pk is distributed by the cloud, and the node calculates .

[0054] Compressed Transmission: The gradient value is scaled by 100 times and converted to INT16, and the data volume of a single node changes from 2KB to 0.5KB.

[0055] c) Cloud Aggregation and Update: Secure Aggregation: Calculate the mean of the encrypted gradients , where is the total number of nodes, i is the index of the node, is the operation of encrypting using the public key , is the local gradient calculated by the i-th node; Model Update: , where is the learning rate, is the operation of decrypting using the private key sk, is the model parameter updated at time step t + 1, is the model parameter at time step t; d) Anomaly Detection: Gradient Clipping: Limit the gradient norm of a single node

[0056] Byzantine Fault Tolerance: Using the Krum algorithm, eliminate malicious gradients that deviate from the mean by more than 3σ.

[0057] In step S103, it is judged whether the comprehensive score exceeds the threshold. Among them, if the comprehensive score exceeds the first threshold, the target node is selected and a Bluetooth GATT connection is established, the water meter data is transmitted to the target node, and it is batch uploaded to the cloud server within a preset low-power time window. If the comprehensive score is lower than the first threshold but higher than the second threshold, the restricted flooding broadcast mechanism is started.

[0058] Among them, the water meter data may include flow data, pressure data, water quality data, and device status data. The thresholds include the first threshold and the second threshold. Among them, the first threshold can be specifically calibrated, such as 0.8, etc., and the second threshold can be specifically calibrated, such as 0.4, etc.

[0059] It can be understood that the embodiments of the present application formulate a differentiated data transmission strategy based on the judgment mechanism of the comprehensive score and the threshold. When the comprehensive score exceeds the first threshold, it indicates that the network or node state is good. At this time, the target node is selected and a Bluetooth GATT connection is established. Utilizing its high-efficiency and low-power consumption characteristics to ensure the fast and accurate transmission of data between nodes. At the same time, data is batch uploaded to the cloud in a preset low-power consumption time window, which not only ensures the timeliness of data but also reduces the node energy consumption and extends the battery life. If the comprehensive score is lower than the first threshold but higher than the second threshold, it indicates that there are certain problems with the network or node. The restricted flooding broadcast mechanism is started. By reasonably restricting the broadcast range and frequency, while ensuring that as much data as possible is received, enhancing the network adaptability and reliability, it balances the data transmission requirements and network load, prevents network congestion, and maintains the stable operation of the network.

[0060] It should be noted that when transmitting the water meter data to the target node, the water meter data is temporarily stored in the local non-volatile memory and batch uploaded to the cloud server through the NB-IoT module within the preset low-power consumption time window. Specifically, after receiving the data, the data is stored in the Flash and marked with the "to be forwarded" status. All the water meter nodes in the network activate the communication module at 01:00-02:00 every day to batch upload the data.

[0061] In the embodiments of the present application, the preset low-power consumption time window is determined in the following way: The cloud server uniformly distributes the time reference for the whole network synchronization and the local activation time period; each node activates the networking function according to the time reference and the local activation period, and the time error is controlled within ±1 minute.

[0062] It can be understood that in the embodiments of the present application, the cloud server uniformly distributes the time reference for the whole network synchronization and the local activation time period, and each node activates the networking function at the specified time according to this, and the time error is strictly controlled within ±1 minute. Firstly, it ensures the whole network synchronization. The unified time reference makes each node follow the same clock precisely, and the accurate activation period ensures the coordinated operation of the system, improving the overall operation efficiency; secondly, it optimizes the energy consumption management. Concentrating the key operations in a specific low-power consumption time window avoids the ineffective wake-up of the nodes and extends the battery life; finally, it enhances the stability and reliability, reduces the node communication conflicts, prevents network congestion and data loss, and at the same time ensures that the data is collected within the same time range, improving the accuracy and integrity of the data.

[0063] In the embodiments of the present application, starting the restricted flooding broadcast mechanism includes: setting the initial value of the TTL of the flooding broadcast to 2; allocating a unique message identifier for the data packet; the relay node performs duplicate removal processing on the data packet according to the unique message identifier, and subtracts 1 from the TTL value after forwarding. When the TTL value is 0, the forwarding stops.

[0064] It can be understood that in the embodiments of the present application, by setting the initial TTL value to 2, an upper limit of "number of hops" is set for data packet propagation. When the TTL value is decremented by 1 after each relay node forwarding until it reaches 0, the broadcast range is effectively controlled, avoiding network congestion caused by infinite propagation of data packets. A unique message identifier is assigned to the data packet, enabling the relay node to determine whether the data packet has been processed based on this, and deduplicating before forwarding to prevent repeated transmission, improving transmission efficiency and reducing waste of network resources. At the same time, the TTL value decrement mechanism not only precisely controls the propagation range but also facilitates network management. The unique identifier also provides convenience for data traceability and management, comprehensively improving the reliability, manageability, and overall performance stability of system data transmission.

[0065] Specifically, the scores of the scanned nodes are sorted from high to low, and the node with the highest score and a score > 0.8 is preferentially selected to establish a GATT connection for direct data transmission; if there is no suitable node, a restricted flooding broadcast with TTL = 2 is initiated, and MsgID is used for deduplication to avoid loops.

[0066] According to the IoT intelligent water meter self-organizing network communication method based on parameter dynamic evaluation proposed in the embodiments of the present application, by obtaining parameters such as the NB-IoT signal quality, Bluetooth link quality, remaining battery power, and historical relay task times of neighbor nodes and calculating a comprehensive score, the optimal relay node can be accurately selected, avoiding the problem of unbalanced load of traditional fixed relays; based on a grading relay mechanism that dynamically switches between Bluetooth GATT direct connection and restricted flooding broadcast based on a score threshold, combined with batch data upload in a preset low-power time window, the daily average power consumption is controlled within 95 μAh, reducing the power consumption by more than 40% compared with traditional passive retransmission schemes, and at the same time avoiding the high cost of on-site replacement of communication modules; through dynamic evaluation and adaptive relay, there is no need to deploy dedicated relay nodes, reducing hardware and operation and maintenance costs, and the data upload success rate is increased to 99.6%, fundamentally solving the problems of high power consumption, high operation and maintenance costs, unbalanced relay load, and low manual meter reading efficiency in the prior art.

[0067] The following will elaborate on the IoT intelligent water meter self-organizing network communication method based on parameter dynamic evaluation through a specific embodiment, and the specific content is as follows: I. Scenario and Requirement Analysis Intelligent water meters are often deployed in complex environments such as underground pipelines and residential building water wells. In such scenarios, there are many challenges: firstly, metal manhole covers, concrete structures, etc. will seriously block signals, resulting in a significant attenuation of the NB-IoT signal, and the signal reception power is often less than or equal to -110 dBm; secondly, there are strict requirements for low power consumption, and the battery life of the water meter needs to reach 8 years or more, and the daily average power consumption needs to be controlled within 100 μAh; thirdly, there are extremely high requirements for data reliability, and the daily water usage data reporting success rate must not be less than 99.5%.

[0068] II. Hardware Design and Deployment Hardware Selection and Configuration Select a low-power microcontroller unit based on the Cortex-m0 core. This control unit can support both low-power modes and full-speed operation, meeting the requirements of different working scenarios.

[0069] Equip with an NB-IoT communication module, supporting B3, B5, and B8 frequency bands to ensure communication stability and compatibility.

[0070] Adopt a low-power magnetoresistive metering sensor to accurately collect water usage data while reducing energy consumption.

[0071] In terms of power management, use ER26500 batteries with an expected lifespan of up to 10 years, and be equipped with a power management unit to detect and control the power consumption of key modules, further optimizing power consumption.

[0072] Have a waterproof design. Through PCB potting and sealing design combined with structural waterproofing, it reaches the IP68 protection level, adapting to harsh and humid environments.

[0073] Installation and Networking When deploying nodes, install 8 - 16 Internet of Things intelligent water meters in each pipe well.

[0074] After completing the installation of the water meters, establish a water meter ID file in the management platform and perform corresponding configurations.

[0075] The water meters automatically join the NB-IoT network, and the access point name is configured by the operator.

[0076] III. Fault Recovery and Relay Communication As Figure 2 shown, Fault Detection and Switching Trigger Conditions: When NB-IoT fails to camp on the network within 120 seconds, or fails to send data continuously for 3 times, and the RSRP remains below -120 dBm for 24 hours (this threshold can be dynamically and adaptively adjusted according to the environment), it is determined that a fault has occurred.

[0077] Switching Actions: Once the fault condition is triggered, immediately turn off the NB-IoT radio frequency, activate the Bluetooth Low Energy module to broadcast an encrypted data payload fault notification packet, and start the scanning function during the networking window period to obtain the beacon packets of normal water meters.

[0078] As Figure 2 shown, Neighbor Discovery and Dynamic Scoring Status Beacon Broadcast: Normally operating water meters will periodically broadcast beacon packets, which contain information such as the NB-IoT signal quality parameters, Bluetooth link quality parameters, remaining battery percentage, and historical relay task times of the water meter.

[0079] Neighbor Discovery and Scoring: The faulty water meter obtains the beacon packets broadcast by the normal water meters through scanning and calculates the score according to relevant algorithms.

[0080] As Figure 2 shown, the hierarchical relay strategy Direct Connection Mode: The faulty water meter preferentially establishes a BLE Generic Attribute Profile connection with the neighbor water meter with the highest score. The connection parameters are set to an interval of 100 ms, a delay of 0 ms, and then the data to be transmitted is sent.

[0081] Flooding Mode: When no node score is greater than or equal to 0.8, or the direct connection mode fails, the flooding mode is started. In this mode, a time-to-live value of 2 is added to the data packet, and the message ID is generated through the hash function Hash(water meter ID + timestamp). Each time a relay node forwards a data packet, the TTL value is decremented by 1. At the same time, to avoid network congestion, each MsgID is only allowed to be forwarded once.

[0082] IV. Data Aggregation and Upload As Figure 2 shown, the smart water meter waits until the reporting time window determined by the network time protocol server to synchronize the time, and batches and uploads the data that conforms to the protocol format.

[0083] V. Dynamic Optimization and Maintenance Reinforcement Learning Weight Optimization Training Data: Information such as the daily relay success rate, average delay, and power consumption rate (mAh / packet) is locally recorded; the data of 10,000 nodes is aggregated in the cloud for training the proximal policy optimization model, and the objective function is set to maximize the success rate and power balance.

[0084] Model Download: The weight file is updated every 7 days (the file size increment Δ ≤ 2 KB), and the node realizes seamless activation through the hot update method.

[0085] Federated Learning and Privacy Protection Gradient Encryption: After calculating the gradient locally, the node uploads it using the Paillier homomorphic encryption algorithm to ensure data privacy.

[0086] Data Desensitization: Sensitive information such as the water meter ID is removed, and only the anonymized performance indicators are uploaded.

[0087] Operation and Maintenance Management Remote Diagnosis: By issuing commands through NB-IoT, the status of the water meter can be queried, including information such as power, signal strength, and historical load.

[0088] Firmware Upgrade: The differential over-the-air download upgrade method is adopted, with each packet size ≤ 512 B. At the same time, a CRC32 checksum and rollback mechanism are equipped to ensure the stability and reliability of the upgrade process.

[0089] In summary, in the complex deployment environment, in response to challenges such as signal occlusion, low power consumption, and high data reliability requirements, the embodiments of the present application perform hardware selection and configuration (low-power MCUs, NB-IoT modules, etc.) and installation and networking; in terms of fault recovery and relay communication, the fault detection trigger conditions and switching actions are clarified, and hierarchical relaying (direct connection, flooding mode) is achieved through neighbor discovery and dynamic scoring; data is uploaded in batches within a specific time window; in terms of dynamic optimization and maintenance, the weights are optimized by means of reinforcement learning, privacy is protected by means of federated learning, and remote diagnosis and differential OTA firmware upgrade are also available.

[0090] Next, a self-organizing network communication system for Internet of Things intelligent water meters based on parameter dynamic evaluation according to an embodiment of the present application will be described with reference to the accompanying drawings.

[0091] Figure 3 It is a block diagram of a self-organizing network communication system for Internet of Things intelligent water meters based on parameter dynamic evaluation according to an embodiment of the present application.

[0092] As Figure 3 shown, the self-organizing network communication system 10 for Internet of Things intelligent water meters based on parameter dynamic evaluation includes: an acquisition module 100, a calculation module 200, and a processing module 300.

[0093] Among them, the acquisition module 100 is used to acquire status beacon packets, where the status beacon packets include NB-IoT signal quality parameters, Bluetooth link quality parameters, remaining battery percentage, and historical relay task times of neighbor nodes; the calculation module 200 is used to calculate the comprehensive score of each neighbor node according to the NB-IoT signal quality parameters, Bluetooth link quality parameters, remaining battery percentage, and historical relay task times; the processing module 300 is used to determine whether the comprehensive score exceeds a threshold. If the comprehensive score exceeds the first threshold, a target node is selected and a Bluetooth GATT connection is established, the water meter data is transmitted to the target node, and is uploaded to the cloud server in batches within a preset low-power time window. If the comprehensive score is lower than the first threshold but higher than the second threshold, a restricted flooding broadcast mechanism is started.

[0094] It should be noted that the foregoing explanation of the embodiment of the self-organizing network communication method for Internet of Things intelligent water meters based on parameter dynamic evaluation is also applicable to the self-organizing network communication system for Internet of Things intelligent water meters based on parameter dynamic evaluation of this embodiment, and will not be repeated here.

[0095] The IoT intelligent water meter self-organizing network communication system based on parameter dynamic evaluation proposed in the embodiments of the present application can accurately select the optimal relay node by obtaining parameters such as the NB-IoT signal quality, Bluetooth link quality, remaining battery power, and historical relay task times of neighbor nodes and calculating the comprehensive score, avoiding the problem of unbalanced load of traditional fixed relays; based on the score threshold, it dynamically switches the hierarchical relay mechanism of Bluetooth GATT direct connection and restricted flood broadcast, and combines the batch data upload with the preset low-power time window to control the daily power consumption within 95 μAh, reducing the power consumption by more than 40% compared with the traditional passive retransmission scheme, and at the same time avoiding the high cost of on-site replacement of communication modules; through dynamic evaluation and adaptive relay, there is no need to deploy dedicated relay nodes, reducing hardware and operation and maintenance costs, and the data upload success rate is increased to 99.6%, fundamentally solving the problems of high power consumption, high operation and maintenance costs, unbalanced relay load, and low manual meter reading efficiency in the prior art.

[0096] The following elaborates on the IoT intelligent water meter self-organizing network communication system based on parameter dynamic evaluation through a specific embodiment, as Figure 4 shown, the content is as follows: The core functions of the intelligent water meter self-organizing network communication system, and each layer of modules work together to form a closed loop: Terminal interaction layer: The PC and mobile App serve as user interfaces to realize the remote monitoring and management of water meter data, supporting users to view real-time data, configure parameters, and receive alarm information.

[0097] Intelligent decision-making layer: Dynamic scoring engine: Calculate the comprehensive score of nodes based on multi-dimensional parameters (NB-IoT signal strength, Bluetooth quality, battery power, relay history), providing a quantitative basis for the selection of relay nodes.

[0098] Federated learning module: Optimize the relay strategy under the premise of ensuring data privacy through distributed gradient calculation and encrypted aggregation, improving the overall performance of the network.

[0099] Security guarantee system: Implement data encryption transmission, digital signature verification, and model rollback mechanism to ensure the security and reliability of the communication process.

[0100] Communication protocol layer: BLE protocol stack: Responsible for fault notification broadcast, neighbor node discovery, GATT connection establishment, and data relay, supporting the hierarchical relay strategy (direct connection mode and restricted flood mode).

[0101] CoAP protocol: A lightweight application layer protocol that optimizes the data transmission efficiency of low-power devices and realizes the time window synchronization mechanism.

[0102] Hardware support layer: Dual-mode communication module: NB-IoT / Cat.1 works in coordination with Bluetooth and automatically switches to Bluetooth relay mode when the main link fails.

[0103] Low-power MCU: Runs scoring algorithms, federated learning models, and protocol stacks to balance computing performance and power consumption.

[0104] Power management system: Monitors the battery status in real time, dynamically adjusts communication strategies according to the battery level, and extends the device life to over 8 years.

[0105] In summary, in the embodiments of the present application, the terminal interaction layer provides a remote monitoring and management interface for users; the intelligent decision-making layer realizes quantitative evaluation, policy optimization, and security guarantee of relay nodes through dynamic scoring, federated learning, and security guarantee; the communication protocol layer uses BLE and CoAP protocols to achieve fault broadcasting, relaying, and low-power data transmission; the dual-mode communication module, low-power MCU, and power management system in the hardware support layer cooperate to ensure communication and power consumption control, solving the problems of signal, power consumption, and reliability in the deployment of intelligent water meters.

[0106] Figure 5 This is a schematic structural diagram of the electronic device provided by the embodiments of the present application. The electronic device may include: A memory 501, a processor 502, and a computer program stored on the memory 501 and executable on the processor 502.

[0107] When the processor 502 executes the program, it implements the method for self-organizing communication of the Internet of Things intelligent water meter based on parameter dynamic evaluation provided in the above embodiments.

[0108] Further, the electronic device further includes: A communication interface 503 for communication between the memory 501 and the processor 502.

[0109] The memory 501 is used to store a computer program executable on the processor 502.

[0110] The memory 501 may include a high-speed RAM memory and may also include non-volatile memory, such as at least one disk memory.

[0111] If the memory 501, the processor 502, and the communication interface 503 are implemented independently, the communication interface 503, the memory 501, and the processor 502 can be interconnected via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 only a thick line is used to represent it in Figure 5 , but it does not mean that there is only one bus or one type of bus.

[0112] Optionally, in a specific implementation, if the memory 501, the processor 502, and the communication interface 503 are integrated on a single chip, the memory 501, the processor 502, and the communication interface 503 can communicate with each other through an internal interface.

[0113] The processor 502 may be a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0114] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0115] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include at least one such feature. In the description of the present application, the meaning of "N" is at least two, such as two, three, etc., unless otherwise specifically defined.

[0116] Any process or method description depicted in a flowchart or otherwise described herein can be understood to represent a module, segment, or portion of code including one or N executable instructions for implementing a customized logical function or process. The scope of the preferred embodiments of the present application includes additional implementations where functions may be executed in a substantially simultaneous manner or in a reverse order according to the involved functions, rather than in the order shown or discussed, which should be understood by those skilled in the art to which the embodiments of the present application pertain.

[0117] It should be understood that various parts of the present application can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0118] Those of ordinary skill in the art of the present technology can understand that all or part of the steps carried out in the methods of the above embodiments can be completed by instructing relevant hardware through a program. The said program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

Claims

1. An Internet of Things intelligent water meter self-organizing network communication method based on parameter dynamic evaluation, characterized in that, Includes the following: Obtain a status beacon packet, wherein the status beacon packet includes the NB-IoT signal quality parameters, Bluetooth link quality parameters, remaining power percentage, and historical relay task times of the neighboring node; Calculate a comprehensive score for each neighbor node based on the NB-IoT signal quality parameter, the Bluetooth link quality parameter, the remaining power percentage, and the number of historical relay tasks; Whether the comprehensive score exceeds the threshold is determined based on the comprehensive score. If the comprehensive score exceeds the first threshold, a target node is selected and a Bluetooth GATT connection is established. The water meter data is transmitted to the target node and uploaded to the cloud server in batches within a preset low-power time window. If the comprehensive score is lower than the first threshold but higher than the second threshold, a restricted flooding broadcast mechanism is started.

2. The method for self-organizing network communication of an IoT intelligent water meter based on parameter dynamic evaluation according to claim 1, wherein The comprehensive score of each neighbor node is calculated according to the NB-IoT signal quality parameter, the Bluetooth link quality parameter, the remaining power percentage, and the number of historical relay tasks, including: Normalizing the NB-IoT signal quality parameter, the Bluetooth link quality parameter, the remaining power percentage, and the number of historical relay tasks; The normalized parameters are weighted and summed according to the preset weight coefficients to obtain a comprehensive score. The preset weight coefficients include the NB-IoT signal quality parameter weight, the Bluetooth link quality parameter weight, the remaining power percentage weight and the historical relay task number weight, and the sum of the weight coefficients is 1.

3. The method for self-organizing network communication of an IoT intelligent water meter based on parameter dynamic evaluation according to claim 2, characterized in that, The dynamic optimization of the preset weight coefficients includes: Locally record the performance indicators of historical relay tasks, including data packet success rate, transmission delay, and power consumption rate; The performance indicators are encrypted and uploaded to the cloud, and the cloud trains new weights based on the reinforcement learning model; The optimized weights are sent to all nodes in the network, and the updates take effect after the nodes verify the signatures.

4. The method for self-organizing network communication of an IoT intelligent water meter based on parameter dynamic evaluation according to claim 3, characterized in that, The triggering conditions for the dynamic optimization include: the average data packet success rate of the entire network for three consecutive natural days is less than 95%; the proportion of node pairs with a remaining power difference of more than 30% accounts for more than 20% of the total number of node pairs in the entire network; the network topology change rate per unit time exceeds 10%, wherein the formula for the network topology change rate is: 。 5. The method for self-organizing network communication of an IoT intelligent water meter based on parameter dynamic evaluation according to claim 3, characterized in that, The reinforcement learning model satisfies the state space, action space and reward function, wherein the state space includes node density, average NB signal strength and historical load imbalance, the action space is a weight coefficient adjustment step less than or equal to 0.05, and the formula of the reward function is: ； Among them, R is the reward value.

6. The method for self-organizing network communication of an IoT intelligent water meter based on parameter dynamic evaluation according to claim 3, wherein The dynamic optimization of the preset weight coefficients also includes a federated learning mechanism, including the following contents: Each water meter node calculates a model gradient based on local historical relay data, and encrypts the model gradient using a homomorphic encryption algorithm; The encrypted gradient is uploaded to the cloud through the NB-IoT channel, and the cloud uses the aggregated private key to decrypt and aggregate all node gradients; The cloud uses aggregated gradients to update the global model and verifies the performance of the new model through a sandbox environment; If the new model causes the average data success rate of the entire network to decrease by more than 5% or the difference in power consumption between nodes to increase by more than 10%, it will automatically roll back to the previous version of the model; The optimized model parameters are sent to each node through digital signatures, and the local weight coefficients are updated after the nodes verify the validity of the signatures.

7. The method for self-organizing network communication of an IoT intelligent water meter based on parameter dynamic evaluation according to claim 1, wherein The preset low-power time window is determined in the following manner: The cloud server uniformly distributes the time reference for the entire network synchronization and the local activation time period; each node activates the networking function according to the time reference and the local activation period, and the time error is controlled within ±1 minute.

8. The method for self-organizing network communication of an IoT intelligent water meter based on parameter dynamic evaluation according to claim 1, characterized in that, The start of the restricted flooding broadcast mechanism includes: setting the initial value of the TTL for flooding broadcast to 2; assigning a unique message identifier to the data packet; the relay node performs duplicate removal on the data packet according to the unique message identifier, and decrements the TTL value by 1 after forwarding. When the TTL value is 0, forwarding stops.

9. The method for self-organizing network communication of an IoT intelligent water meter based on parameter dynamic evaluation according to claim 1, characterized in that Before obtaining the status beacon packet, it includes: After the IoT intelligent water meter node fails to upload data in NB-IoT or Cat.1, it activates the Bluetooth module and broadcasts a fault notification, where the upload failure includes three consecutive connection timeouts, a packet loss rate > 50%, or receiving a permanent connection rejection error code; Configure Bluetooth scan parameters to filter the scanned Bluetooth devices and only retain the neighbor node beacon packets carrying specific service UUIDs.

10. An Internet of Things intelligent water meter self-organizing network communication system based on parameter dynamic evaluation, characterized in that, It includes: An acquisition module for acquiring status beacon packets, where the status beacon packets include NB-IoT signal quality parameters, Bluetooth link quality parameters, remaining battery percentage, and historical relay task times of neighbor nodes; A calculation module for calculating a comprehensive score for each neighbor node according to the NB-IoT signal quality parameters, the Bluetooth link quality parameters, the remaining battery percentage, and the historical relay task times; A processing module for determining whether the comprehensive score exceeds a threshold. If the comprehensive score exceeds the first threshold, it selects a target node and establishes a Bluetooth GATT connection, transmits the water meter data to the target node, and batches it for upload to the cloud server within the preset low-power time window. If the comprehensive score is lower than the first threshold but higher than the second threshold, it starts the restricted flooding broadcast mechanism.

Images